Notes

LECTURE NOTES ON FUNDAMENTALS OF COMBUSTION
Joseph M. Powers Department of Aerospace and Mechanical Engineering University of Notre Dame Notre Dame, Indiana 46556-5637 USA updated 12 December 2011, 2:43pm
CC BY-NC-ND. 12 December 2011, J. M. Powers.
Contents
Preface 1 Introduction to kinetics 1.1 Isothermal, isochoric kinetics . . . . . . . . . . . . . . . . . . . . . 1.1.1 O O2 dissociation . . . . . . . . . . . . . . . . . . . . . . 1.1.1.1 Pair of irreversible reactions . . . . . . . . . . . . 1.1.1.1.1 Mathematical model . . . . . . . . . . . 1.1.1.1.2 Example calculation . . . . . . . . . . . 1.1.1.1.2.1 Species concentration versus time 1.1.1.1.2.2 Pressure versus time . . . . . . . 1.1.1.1.2.3 Dynamical system form . . . . . . 1.1.1.1.3 Eect of temperature . . . . . . . . . . . 1.1.1.2 Single reversible reaction . . . . . . . . . . . . . . 1.1.1.2.1 Mathematical model . . . . . . . . . . . 1.1.1.2.1.1 Kinetics . . . . . . . . . . . . . . 1.1.1.2.1.2 Thermodynamics . . . . . . . . . 1.1.1.2.2 Example calculation . . . . . . . . . . . 1.1.2 Zeldovich mechanism of NO production . . . . . . . . . . 1.1.2.1 Mathematical model . . . . . . . . . . . . . . . . 1.1.2.1.1 Standard model form . . . . . . . . . . . 1.1.2.1.2 Reduced form . . . . . . . . . . . . . . . 1.1.2.1.3 Example calculation . . . . . . . . . . . 1.1.2.2 Stiness, time scales, and numerics . . . . . . . . 1.1.2.2.1 Eect of temperature . . . . . . . . . . . 1.1.2.2.2 Eect of initial pressure . . . . . . . . . 1.1.2.2.3 Stiness and numerics . . . . . . . . . . 1.2 Adiabatic, isochoric kinetics . . . . . . . . . . . . . . . . . . . . . 1.2.1 Thermal explosion theory . . . . . . . . . . . . . . . . . . 1.2.1.1 One-step reversible kinetics . . . . . . . . . . . . 1.2.1.2 First law of thermodynamics . . . . . . . . . . . 1.2.1.3 Dimensionless form . . . . . . . . . . . . . . . . . 1.2.1.4 Example calculation . . . . . . . . . . . . . . . . 3 11 13 16 16 16 16 22 23 24 26 30 31 31 31 33 34 37 37 37 39 42 48 49 50 52 54 54 55 56 59 60
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS 1.2.1.5 High activation energy asymptotics . . . . . . . . . . . . . . Detailed H2 O2 N2 kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 66 71 71 74 75 75 79 82 84 85 91 93 93 98 101 105 109 115 116 116 119 119 119 120 122 123 127 129 129 131 131 136 139 141 141 143 143 150 152
1.2.2 2 Gas 2.1 2.2 2.3
mixtures Some general issues . . . . . . . . . . . . . . . . . . . Ideal and non-ideal mixtures . . . . . . . . . . . . . . Ideal mixtures of ideal gases . . . . . . . . . . . . . . 2.3.1 Dalton model . . . . . . . . . . . . . . . . . . 2.3.1.1 Binary mixtures . . . . . . . . . . . 2.3.1.2 Entropy of mixing . . . . . . . . . . 2.3.1.3 Mixtures of constant mass fraction . 2.3.2 Summary of properties for the Dalton mixture 2.3.3 Amagat model* . . . . . . . . . . . . . . . . .
3 Mathematical foundations of thermodynamics* 3.1 Exact dierentials and state functions . . . . . . . . . . . . . . 3.2 Two independent variables . . . . . . . . . . . . . . . . . . . . 3.3 Legendre transformations . . . . . . . . . . . . . . . . . . . . . 3.4 Heat capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Van der Waals gas . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Redlich-Kwong gas . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Compressibility and generalized charts . . . . . . . . . . . . . 3.8 Mixtures with variable composition . . . . . . . . . . . . . . . 3.9 Partial molar properties . . . . . . . . . . . . . . . . . . . . . 3.9.1 Homogeneous functions . . . . . . . . . . . . . . . . . . 3.9.2 Gibbs free energy . . . . . . . . . . . . . . . . . . . . . 3.9.3 Other properties . . . . . . . . . . . . . . . . . . . . . 3.9.4 Relation between mixture and partial molar properties 3.10 Frozen sound speed . . . . . . . . . . . . . . . . . . . . . . . . 3.11 Irreversibility in a closed multicomponent system . . . . . . . 3.12 Equilibrium in a two-component system . . . . . . . . . . . . 3.12.1 Phase equilibrium . . . . . . . . . . . . . . . . . . . . . 3.12.2 Chemical equilibrium: introduction . . . . . . . . . . . 3.12.2.1 Isothermal, isochoric system . . . . . . . . . . 3.12.2.2 Isothermal, isobaric system . . . . . . . . . . 3.12.3 Equilibrium condition . . . . . . . . . . . . . . . . . . 4 Thermochemistry of a single reaction 4.1 Molecular mass . . . . . . . . . . . . 4.2 Stoichiometry . . . . . . . . . . . . . 4.2.1 General development . . . . . 4.2.2 Fuel-air mixtures . . . . . . . 4.3 First law analysis of reacting systems
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
CONTENTS 4.3.1 4.3.2 4.3.3 Enthalpy of formation . . . . . . . . . . . . . . . . . . Enthalpy and internal energy of combustion . . . . . . Adiabatic ame temperature in isochoric stoichiometric 4.3.3.1 Undiluted, cold mixture . . . . . . . . . . . . 4.3.3.2 Dilute, cold mixture . . . . . . . . . . . . . . 4.3.3.3 Dilute, preheated mixture . . . . . . . . . . . 4.3.3.4 Dilute, preheated mixture with minor species Chemical equilibrium . . . . . . . . . . . . . . . . . . . . . . . Chemical kinetics of a single isothermal reaction . . . . . . . . 4.5.1 Isochoric systems . . . . . . . . . . . . . . . . . . . . . 4.5.2 Isobaric systems . . . . . . . . . . . . . . . . . . . . . . Some conservation and evolution equations . . . . . . . . . . . 4.6.1 Total mass conservation: isochoric reaction . . . . . . . 4.6.2 Element mass conservation: isochoric reaction . . . . . 4.6.3 Energy conservation: adiabatic, isochoric reaction . . . 4.6.4 Energy conservation: adiabatic, isobaric reaction . . . . 4.6.5 Non-adiabatic isochoric combustion . . . . . . . . . . . 4.6.6 Entropy evolution: Clausius-Duhem relation . . . . . . Simple one-step kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5 152 155 155 156 157 158 160 161 166 166 174 180 180 181 182 184 187 187 190 195 195 201 202 207 209 209 216 219 224 224 225 225 226 234 239 239 239 242 242 243
4.4 4.5
4.6
4.7
5 Thermochemistry of multiple reactions 5.1 Summary of multiple reaction extensions . . . . . . 5.2 Equilibrium conditions . . . . . . . . . . . . . . . . 5.2.1 Minimization of G via Lagrange multipliers 5.2.2 Equilibration of all reactions . . . . . . . . . 5.2.3 Zeldovichs uniqueness proof* . . . . . . . . 5.2.3.1 Isothermal, isochoric case . . . . . 5.2.3.2 Isothermal, isobaric case . . . . . . 5.2.3.3 Adiabatic, isochoric case . . . . . . 5.2.3.4 Adiabatic, isobaric case . . . . . . 5.3 Concise reaction rate law formulations . . . . . . . 5.3.1 Reaction dominant: J (N L) . . . . . . 5.3.2 Species dominant: J < (N L) . . . . . . . 5.4 Onsager reciprocity* . . . . . . . . . . . . . . . . . 5.5 Irreversibility production rate* . . . . . . . . . . . . 6 Reactive Navier-Stokes equations 6.1 Evolution axioms . . . . . . . . . . 6.1.1 Conservative form . . . . . . 6.1.2 Non-conservative form . . . 6.1.2.1 Mass . . . . . . . . 6.1.2.2 Linear momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6 6.1.2.3 Energy . . . 6.1.2.4 Second law . 6.1.2.5 Species . . . 6.1.2.6 Elements . . Mixture rules . . . . . . . . . Constitutive models . . . . . . Temperature evolution . . . . Shvab-Zeldovich formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 244 244 244 244 245 247 249 253 253 254 255 256 256 257 257 258 260 260 261 261 261 266 267 270 272 274 275 275 277 279 280 282 282 284 284 285
6.2 6.3 6.4 6.5
7 Simple solid combustion: Reaction-diusion 7.1 Simple planar model . . . . . . . . . . . . . . . . . . . . . . . 7.1.1 Model equations . . . . . . . . . . . . . . . . . . . . . 7.1.2 Simple planar derivation . . . . . . . . . . . . . . . . . 7.1.3 Ad hoc approximation . . . . . . . . . . . . . . . . . . 7.1.3.1 Planar formulation . . . . . . . . . . . . . . . 7.1.3.2 More general coordinate systems . . . . . . . 7.2 Non-dimensionalization . . . . . . . . . . . . . . . . . . . . . . 7.2.1 Diusion time discussion . . . . . . . . . . . . . . . . . 7.2.2 Final form . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.3 Integral form . . . . . . . . . . . . . . . . . . . . . . . 7.2.4 Innite Damkhler limit . . . . . . . . . . . . . . . . . o 7.3 Steady solutions . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3.1 High activation energy asymptotics . . . . . . . . . . . 7.3.2 Method of weighted residuals . . . . . . . . . . . . . . 7.3.2.1 One-term collocation solution . . . . . . . . . 7.3.2.2 Two-term collocation solution . . . . . . . . . 7.3.3 Steady solution with depletion . . . . . . . . . . . . . . 7.4 Unsteady solutions . . . . . . . . . . . . . . . . . . . . . . . . 7.4.1 Linear stability . . . . . . . . . . . . . . . . . . . . . . 7.4.1.1 Formulation . . . . . . . . . . . . . . . . . . . 7.4.1.2 Separation of variables . . . . . . . . . . . . . 7.4.1.3 Numerical eigenvalue solution . . . . . . . . . 7.4.1.3.1 Low temperature transients . . . . . 7.4.1.3.2 Intermediate temperature transients 7.4.1.3.3 High temperature transients . . . . . 7.4.2 Full transient solution . . . . . . . . . . . . . . . . . . 7.4.2.1 Low temperature solution . . . . . . . . . . . 7.4.2.2 High temperature solution . . . . . . . . . . .
8 Laminar ames: Reaction-advection-diusion 287 8.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 8.1.1 Evolution equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
CONTENTS 8.1.1.1 Conservative form . . . . . . . 8.1.1.2 Non-conservative form . . . . . 8.1.1.3 Formulation using enthalpy . . 8.1.1.4 Low Mach number limit . . . . 8.1.2 Constitutive models . . . . . . . . . . . . 8.1.3 Alternate forms . . . . . . . . . . . . . . 8.1.3.1 Species equation . . . . . . . . 8.1.3.2 Energy equation . . . . . . . . 8.1.3.3 Shvab-Zeldovich form . . . . . 8.1.4 Equilibrium conditions . . . . . . . . . . Steady burner-stabilized ames . . . . . . . . . 8.2.1 Formulation . . . . . . . . . . . . . . . . 8.2.2 Solution procedure . . . . . . . . . . . . 8.2.2.1 Model linear system . . . . . . 8.2.2.2 System of rst order equations 8.2.2.3 Equilibrium . . . . . . . . . . . 8.2.2.4 Linear stability of equilibrium . 8.2.2.5 Laminar ame structure . . . . 8.2.2.5.1 TIG = 0.2. . . . . . . . 8.2.2.5.2 TIG = 0.076. . . . . . . 8.2.3 Detailed H2 -O2-N2 kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7 288 288 288 289 290 292 292 292 293 295 297 298 301 301 302 303 303 305 305 309 310 315 315 315 317 317 318 319 319 319 320 321 321 321 321 322 323 323 329 331 333
8.2
9 Simple detonations: Reaction-advection 9.1 Reactive Euler equations . . . . . . . . . . . . 9.1.1 One-step irreversible kinetics . . . . . . 9.1.2 Thermicity . . . . . . . . . . . . . . . 9.1.3 Parameters for H2 -Air . . . . . . . . . 9.1.4 Conservative form . . . . . . . . . . . . 9.1.5 Non-conservative form . . . . . . . . . 9.1.5.1 Mass . . . . . . . . . . . . . . 9.1.5.2 Linear momenta . . . . . . . 9.1.5.3 Energy . . . . . . . . . . . . 9.1.5.4 Reaction . . . . . . . . . . . . 9.1.5.5 Summary . . . . . . . . . . . 9.1.6 One-dimensional form . . . . . . . . . 9.1.6.1 Conservative form . . . . . . 9.1.6.2 Non-conservative form . . . . 9.1.6.3 Reduction of energy equation 9.1.7 Characteristic form . . . . . . . . . . . 9.1.8 Rankine-Hugoniot jump conditions . . 9.1.9 Galilean transformation . . . . . . . . 9.2 One-dimensional, steady solutions . . . . . . .
8 9.2.1 9.2.2 Steady shock jumps . . . . . . . . . . . . . . . . . . . Ordinary dierential equations of motion . . . . . . . 9.2.2.1 Conservative form . . . . . . . . . . . . . . 9.2.2.2 Unreduced non-conservative form . . . . . . 9.2.2.3 Reduced non-conservative form . . . . . . . Rankine-Hugoniot analysis . . . . . . . . . . . . . . . 9.2.3.1 Rayleigh line . . . . . . . . . . . . . . . . . 9.2.3.2 Hugoniot curve . . . . . . . . . . . . . . . . Shock solutions . . . . . . . . . . . . . . . . . . . . . Equilibrium solutions . . . . . . . . . . . . . . . . . . 9.2.5.1 Chapman-Jouguet solutions . . . . . . . . . 9.2.5.2 Weak and strong solutions . . . . . . . . . . 9.2.5.3 Summary of solution properties . . . . . . . ZND solutions: One-step irreversible kinetics . . . . . 9.2.6.1 CJ ZND structures . . . . . . . . . . . . . . 9.2.6.2 Strong ZND structures . . . . . . . . . . . . 9.2.6.3 Weak ZND structures . . . . . . . . . . . . 9.2.6.4 Piston problem . . . . . . . . . . . . . . . . Detonation structure: Two-step irreversible kinetics . 9.2.7.1 Strong structures . . . . . . . . . . . . . . . 9.2.7.1.1 D > D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.7.1.2 D = D 9.2.7.2 Weak, eigenvalue structures . . . . . . . . . 9.2.7.3 Piston problem . . . . . . . . . . . . . . . . Detonation structure: Detailed H2 O2 N2 kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 333 333 334 337 337 338 339 345 346 347 348 350 351 353 358 361 361 362 369 369 373 373 374 376 383 384 385 385 385 386 387 387 387 388 389 389 390 390 390 392
9.2.3
9.2.4 9.2.5
9.2.6
9.2.7
9.2.8
10 Blast waves 10.1 Governing equations . . . . . . . . 10.2 Similarity transformation . . . . . . 10.2.1 Independent variables . . . 10.2.2 Dependent variables . . . . 10.2.3 Derivative transformations . 10.3 Transformed equations . . . . . . . 10.3.1 Mass . . . . . . . . . . . . . 10.3.2 Linear momentum . . . . . 10.3.3 Energy . . . . . . . . . . . . 10.4 Dimensionless equations . . . . . . 10.4.1 Mass . . . . . . . . . . . . . 10.4.2 Linear momentum . . . . . 10.4.3 Energy . . . . . . . . . . . . 10.5 Reduction to non-autonomous form 10.6 Numerical solution . . . . . . . . .
CONTENTS
10.6.1 Calculation of total energy . . . . . . . . . . . . . . . . . . . . . . . . 394 10.6.2 Comparison with experimental data . . . . . . . . . . . . . . . . . . . 397 10.7 Contrast with acoustic limit . . . . . . . . . . . . . . . . . . . . . . . . . . . 398 Bibliography 401
10
CONTENTS
Preface
These are lecture notes for AME 60636, Fundamentals of Combustion, a course taught since 1994 in the Department of Aerospace and Mechanical Engineering of the University of Notre Dame. Most of the students in this course are graduate students; the course is also suitable for interested undergraduates. The objective of the course is to provide background in theoretical combustion science. Most of the material in the notes is covered in one semester; some extra material is also included. Sections which need not be emphasized are marked with a . The goal of the notes is to provide a solid mathematical foundation in the physical chemistry, thermodynamics, and uid mechanics of combustion. The notes attempt to ll gaps in the existing literature in providing enhanced discussion of detailed kinetics models which are in common use for real physical systems. In addition, some model problems for paradigm systems are addressed for pedagogical purposes. Many of the unique aspects of these notes, which focus on some fundamental issues involving the thermodynamics of reactive gases with detailed nite rate kinetics, have arisen due to the authors involvement in a research project supported by the National Science Foundation, under Grant No. CBET0650843. Any opinions, ndings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reect the views of the National Science Foundation. The author is grateful for the support. Thanks are also extended to my former student Dr. Ashraf al-Khateeb, who generated the steady laminar ame plots for detailed hydrogen-air combustion. General thanks are also due to students in the course over the years whose interest has motivated me to nd ways to improve the material. The notes, along with information on the course itself, can be found on the world wide web at http://www.nd.edu/powers/ame.60636. At this stage, anyone is free to make copies for their own use. I would be happy to hear from you about errors or suggestions for improvement. Joseph M. Powers powers@nd.edu http://www.nd.edu/powers Notre Dame, Indiana; USA = $ CC BY: 12 December 2011
The content of this book is licensed under Creative Commons Attribution-Noncommercial-No Derivative Works 3.0.
11
12
CONTENTS
Chapter 1 Introduction to kinetics

Let us consider the reaction of N molecular chemical species composed of L elements via J chemical reactions. Let us assume the gas is an ideal mixture of ideal gases that satises Daltons1 law of partial pressures. The reaction will be considered to be driven by molecular collisions. We will not model individual collisions, but instead attempt to capture their collective eect. An example of a model of such a reaction is listed in Table 1.1. There we nd a N = 9 species, J = 37 step irreversible reaction mechanism for an L = 3 hydrogen-oxygen-argon mixture from Maas and Warnatz,2 with corrected fH2 from Maas and Pope.3 The model has also been utilized by Fedkiw, et al.4 We need not worry yet about fH2 , which is known as a collision eciency factor. The one-sided arrows indicate that each individual reaction is considered to be irreversible. Note that for nearly each reaction, a separate reverse reaction is listed; thus, pairs of irreversible reactions can in some sense be considered to model reversible reactions. In this model a set of elementary reactions are hypothesized. For the j th reaction we have the collision frequency factor aj , the temperature-dependency exponent j , and the activation energy E j . These will be explained in short order. Other common forms exist. Often reactions systems are described as being composed of reversible reactions. Such reactions are usually notated by two sided arrows. One such system is reported by Powers and Paolucci5 reported here in Table 1.2. Both overall models are complicated.
1 2
John Dalton, 1766-1844, English chemist. Maas, U., and Warnatz, J., 1988, Ignition Processes in Hydrogen-Oxygen Mixtures, Combustion and Flame, 74(1): 53-69. 3 Maas, U., and Pope, S. B., 1992, Simplifying Chemical Kinetics: Intrinsic Low-Dimensional Manifolds in Composition Space, Combustion and Flame, 88(3-4): 239-264. 4 Fedkiw, R. P., Merriman, B., and Osher, S., 1997, High Accuracy Numerical Methods for Thermally Perfect Gas Flows with Chemistry, Journal of Computational Physics, 132(2): 175-190. 5 Powers, J. M., and Paolucci, S., 2005, Accurate Spatial Resolution Estimates for Reactive Supersonic Flow with Detailed Chemistry, AIAA Journal, 43(5): 1088-1099.
13
14
CHAPTER 1. INTRODUCTION TO KINETICS
j Reaction 1 O2 + H OH + O 2 OH + O O2 + H 3 H2 + O OH + H 4 OH + H H2 + O 5 H2 + OH H2 O + H 6 H2 O + H H2 + OH 7 OH + OH H2 O + O 8 H2 O + O OH + OH 9 H + H + M H2 + M 10 H2 + M H + H + M 11 H + OH + M H2 O + M 12 H2 O + M H + OH + M 13 O + O + M O2 + M 14 O2 + M O + O + M 15 H + O2 + M HO2 + M 16 HO2 + M H + O2 + M 17 HO2 + H OH + OH 18 OH + OH HO2 + H 19 HO2 + H H2 + O2 20 H2 + O2 HO2 + H 21 HO2 + H H2 O + O 22 H2 O + O HO2 + H 23 HO2 + O OH + O2 24 OH + O2 HO2 + O 25 HO2 + OH H2 O + O2 26 H2 O + O2 HO2 + OH 27 HO2 + HO2 H2 O2 + O2 28 OH + OH + M H2 O2 + M 29 H2 O2 + M OH + OH + M 30 H2 O2 + H H2 + HO2 31 H2 + HO2 H2 O2 + H 32 H2 O2 + H H2 O + OH 33 H2 O + OH H2 O2 + H 34 H2 O2 + O OH + HO2 35 OH + HO2 H2 O2 + O 36 H2 O2 + OH H2 O + HO2 37 H2 O + HO2 H2 O2 + OH
aj 2.00 1014 1.46 1013 5.06 104 2.24 104 1.00 108 4.45 108 1.50 109 1.51 1010 1.80 1018 6.99 1018 2.20 1022 3.80 1023 2.90 1017 6.81 1018 2.30 1018 3.26 1018 1.50 1014 1.33 1013 2.50 1013 6.84 1013 3.00 1013 2.67 1013 1.80 1013 2.18 1013 6.00 1013 7.31 1014 2.50 1011 3.25 1022 2.10 1024 1.70 1012 1.15 1012 1.00 1013 2.67 1012 2.80 1013 8.40 1012 5.40 1012 1.63 1013
j 0.00 0.00 2.67 2.67 1.60 1.60 1.14 1.14 1.00 1.00 2.00 2.00 1.00 1.00 0.80 0.80 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.00 2.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Ej 70.30 2.08 26.30 18.40 13.80 77.13 0.42 71.64 0.00 436.08 0.00 499.41 0.00 496.41 0.00 195.88 4.20 168.30 2.90 243.10 7.20 242.52 1.70 230.61 0.00 303.53 5.20 0.00 206.80 15.70 80.88 15.00 307.51 26.80 84.09 4.20 132.71
Table 1.1: Units of aj are in appropriate combinations of cm, mol, s, and K so that i has units of mole cm3 s1 ; units of E j are kJ mol1 . Third body collision eciencies with M are fH2 = 1.00, fO2 = 0.35, and fH2 O = 6.5.
15
j 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Reaction H2 + O2 OH + OH OH + H2 H2 O + H H + O2 OH + O O + H2 OH + H H + O2 + M HO2 + M H + O2 + O2 HO2 + O2 H + O2 + N2 HO2 + N2 OH + HO2 H2 O + O2 H + HO2 OH + OH O + HO2 O2 + OH OH + OH O + H2 O H2 + M H + H + M O2 + M O + O + M H + OH + M H2 O + M H + HO2 H2 + O2 HO2 + HO2 H2 O2 + O2 H2 O2 + M OH + OH + M H2 O2 + H HO2 + H2 H2 O2 + OH H2 O + HO2
aj 1.70 1013 1.17 109 5.13 1016 1.80 1010 2.10 1018 6.70 1019 6.70 1019 5.00 1013 2.50 1014 4.80 1013 6.00 108 2.23 1012 1.85 1011 7.50 1023 2.50 1013 2.00 1012 1.30 1017 1.60 1012 1.00 1013
j 0.00 1.30 0.82 1.00 1.00 1.42 1.42 0.00 0.00 0.00 1.30 0.50 0.50 2.60 0.00 0.00 0.00 0.00 0.00
Ej 47780 3626 16507 8826 0 0 0 1000 1900 1000 0 92600 95560 0 700 0 45500 3800 1800
Table 1.2: Nine species, nineteen step reversible reaction mechanism for a hydrogen/oxygen/nitrogen mixture. Units of aj are in appropriate combinations of cm, mole, s, and K so that the reaction rate has units of mole/cm3 /s; units of E j are cal/mole. Third body collision eciencies with M are f5 (H2 O) = 21, f5 (H2 ) = 3.3, f12 (H2 O) = 6, f12 (H) = 2, f12 (H2 ) = 3, f14 (H2 O) = 20.
16
1.1
Isothermal, isochoric kinetics
For simplicity, we will rst focus attention on cases in which the temperature T and volume V are both constant. Such assumptions are known as isothermal and isochoric, respectively. A nice fundamental treatment of elementary reactions of this type is given by Vincenti and Kruger in their detailed monograph.6
1.1.1
One of the simplest physical examples is provided by the dissociation of O2 into its atomic component O. 1.1.1.1 Pair of irreversible reactions
O O2 dissociation
To get started, let us focus for now only on reactions 13 and 14 from Table 1.1 in the limiting case in which temperature T and volume V are constant. 1.1.1.1.1 Mathematical model The reactions describe oxygen dissociation and recombination in a pair of irreversible reactions: 13 : O + O + M O2 + M, 14 : O2 + M O + O + M, with a13 = 2.90 10 a14 = 6.81 10
18 17 2
(1.1) (1.2)
mole cm3
1
K , s
13 = 1.00,
E 13 = 0
kJ , mole kJ . mole
(1.3) (1.4)
mole cm3
K , s
14 = 1.00,
E 14 = 496.41
The irreversibility is indicated by the one-sided arrow. Though they participate in the overall hydrogen oxidation problem, these two reactions are in fact self-contained as well. So let us just consider that we have only oxygen in our box with N = 2 species, O2 and O, J = 2 reactions (those being 13 and 14), and L = 1 element, that being O. Recall that in the cgs system, common in thermochemistry, that 1 erg = 1 dyne cm = 7 10 J = 1010 kJ. Recall also that the cgs unit of force is the dyne and that 1 dyne = 1 g cm/s2 = 105 N. So for cgs we have E 13 = 0
6
erg , mole
E 14 = 496.41
kJ mole
1010 erg kJ
= 4.96 1012
erg . mole
(1.5)
Vincenti, W. G., and Kruger, C. H., 1965, Introduction to Physical Gas Dynamics, Wiley, New York.
1.1. ISOTHERMAL, ISOCHORIC KINETICS
17
The standard model for chemical reaction, which will be generalized and discussed in more detail later, induces the following two ordinary dierential equations for the evolution of O and O2 molar concentrations: dO = 2 a13 T 13 exp dt
=k13 (T ) =r13
E 13 RT
O O M +2 a14 T 14 exp
=k14 (T )
E 14 RT
=r14
O2 M ,
(1.6)
dO2 dt
= a13 T 13 exp
=k13 (T )
E 13 RT
=r13
O O M a14 T 14 exp
=k14 (T )
E 14 RT
=r14
O2 M .
(1.7)
Here we use the notation i as the molar concentration of species i. Also a common usage for molar concentration is given by square brackets, e.g. O2 = [O2 ]. The symbol M represents an arbitrary third body and is an inert participant in the reaction. The symbol R is the universal gas constant, for which R = 8.31441 J mole K 107 erg J = 8.31441 107 erg . mole K (1.8)
We also use the common notation of a temperature-dependent portion of the reaction rate for reaction j, kj (T ), where kj (T ) = aj T j exp Ej RT . (1.9)
Note that j = 1, . . . , J. The reaction rates for reactions 13 and 14 are dened as r13 = k13 O O M , r14 = k14 O2 M . (1.10) (1.11)
We will give details of how to generalize this form later. The system Eq. (1.6-1.7) can be written simply as dO = 2r13 + 2r14 , dt dO2 = r13 r14 . dt Even more simply, in vector form, Eqs. (1.12-1.13) can be written as d = r. dt (1.14) (1.12) (1.13)
18 Here we have taken = = r =
O O2
, ,
(1.15) (1.16) (1.17)
2 2 1 1 r13 r14 .
In general, we will have be a column vector of dimension N 1, will be a rectangular matrix of dimension N J of rank R, and r will be a column vector of length J 1. So Eqs. (1.12-1.13) take the form d O r13 2 2 = . (1.18) 1 1 r14 dt O2 Note here that the rank R of is R = L = 1. Let us also dene a stoichiometric matrix of dimension L N. The component of , li represents the number of element l in species i. Generally will be full rank, which will vary since we can have L < N, L = N, or L > N. Here we have L < N and is of dimension 1 2: = (1 2). = 0. Element conservation is guaranteed by insisting that be constructed such that (1.20) So we can say that each of the column vectors of lies in the right null space of . For our example, we see that Eq. (1.20) holds: = (1 2) Let us take as initial conditions O (t = 0) = O , 2 2 1 1 = (0 0). (1.21)
(1.19)
O2 (t = 0) = O2 .
(1.22) (1.23)
Now M represents an arbitrary third body, so here M = O2 + O . Thus, the ordinary dierential equations of the reaction dynamics reduce to dO = 2a13 T 13 exp dt E 13 O O O2 + O RT E 14 O2 O2 + O , +2a14 T 14 exp RT E 13 = a13 T 13 exp O O O2 + O RT E 14 O2 O2 + O . a14 T 14 exp RT
(1.24)
dO2 dt
(1.25)
19
Equations (1.24-1.25) with Eqs. (1.22) represent two non-linear ordinary dierential equations with initial conditions in two unknowns O and O2 . We seek the behavior of these two species concentrations as a function of time. Systems of non-linear equations are generally dicult to integrate analytically and generally require numerical solution. Before embarking on a numerical solution, we simplify as much as we can. Note that d dO + 2 O2 = 0, (1.26) dt dt d (1.27) + 2O2 = 0. dt O We can integrate and apply the initial conditions (1.22) to get O + 2O2 = O + 2O2 = constant. (1.28)
The fact that this algebraic constraint exists for all time is a consequence of the conservation of mass of each O element. It can also be thought of as the conservation of number of O atoms. Such notions always hold for chemical reactions. They do not hold for nuclear reactions. Standard linear algebra provides a robust way to nd the constraint of Eq. (1.28). We can use elementary row operations to cast Eq. (1.17) into a row-echelon form. Here our goal is to get a linear combination which on the right side has an upper triangular form. To achieve this add twice the second equation with the rst to form a new equation to replace the second equation. This gives d dt O O + 2O2 = 2 2 0 0 r13 r14 . (1.29)
Obviously the second equation is one we obtained earlier, d/dt(O + 2O2 ) = 0, and this induces our algebraic constraint. We also note the system can be recast as 1 0 1 2 This is of the matrix form d = U r. (1.31) dt Here L and L1 are N N lower triangular matrices of full rank N, and thus invertible. The matrix U is upper triangular of dimension N J and with the same rank as , R L. The matrix P is a permutation matrix of dimension N N. It is never singular and thus always invertable. It is used to eect possible row exchanges to achieve the desired form; often row exchanges are not necessary, in which case P = I, the N N identity matrix. Equation (1.31) can be manipulated to form the original equation via L1 P d = P1 L U r. dt = (1.32) d dt O O2 = 2 2 0 0 r13 r14 . (1.30)
20
What we have done is the standard linear algebra decomposition of = P1 L U. We can also decompose the algebraic constraint, Eq. (1.28), in a non-obvious way that is more readily useful for larger systems. We can write O2 = O2 Dening now O = O O , we can say O O2
=
1 O . 2 O
(1.33)
O O2
b =
1 1 2
=D
( O ) .
=
(1.34)
This gives the dependent variables in terms of a smaller number of transformed dependent variables in a way which satises the linear constraints. In vector form, the equation becomes = + D . (1.35)
Here D is a full rank matrix which spans the same column space as does . Note that may or may not be full rank. Since D spans the same column space as does , we must also have in general D = 0. We see here this is true: (1 2) 1 1 2 = (0). (1.37) (1.36)
We also note that the term exp(E j /RT ) is a modulating factor to the dynamics. Let us see how this behaves for high and low temperatures. First for low temperature, we have lim exp E j RT = 0. (1.38)
T 0
At high temperature, we have lim exp E j RT = 1. (1.39)
And lastly, at intermediate temperature, we have exp E j RT O(1) when T =O Ej R . (1.40)
A sketch of this modulating factor is given in Figure 1.1. Note

21
exp(- E j /(R T))

1
Ej/ R
Figure 1.1: Plot of exp(E j /R/T ) versus T ; transition occurs at T E j /R. for small T , the modulation is extreme, and the reaction rate is very small, for T E j /R, the reaction rate is extremely sensitive to temperature, and for T , the modulation is unity, and the reaction rate is limited only by molecular collision frequency. Now O and O2 represent molar concentrations which have standard units of mole/cm3 . So the reaction rates dO2 dO and dt dt 3 have units of mole/cm /s. Note that the argument of the exponential to be dimensionless. That is Ej RT = erg mole K 1 mole erg K dimensionless. (1.41)
Here the brackets denote the units of a quantity, and not molar concentration. Let us get units for the collision frequency factor of reaction 13, a13 . We know the units of the rate (mole/cm3 /s). Reaction 13 involves three molar species. Since 13 = 1, it also has an extra temperature dependency. The exponential of a unitless number is unitless, so we need not worry about that. For units to match, we must have mole cm3 s So the units of a13 are [a13 ] = mole cm3
2
= [a13 ]
mole cm3
mole cm3 K . s
mole cm3
K 1 .
(1.42)
(1.43)
22
For a14 we nd a dierent set of units! Following the same procedure, we get mole cm3 s So the units of a14 are [a14 ] = mole cm3
1
= [a14 ]
mole cm3
mole cm3
K 1 .
(1.44)
K . s
(1.45)
This discrepancy in the units of aj the molecular collision frequency factor is a burden of traditional chemical kinetics, and causes many diculties when classical non-dimensionalization is performed. With much eort, a cleaner theory could be formulated; however, this would require signicant work to re-cast the now-standard aj values for literally thousands of reactions which are well established in the literature. 1.1.1.1.2 Example calculation Let us consider an example problem. Let us take T = 5000 K, and initial conditions O = 0.001 mole/cm3 and O2 = 0.001 mole/cm3 . The initial temperature is very hot, and is near the temperature of the surface of the sun. This is also realizable in laboratory conditions, but uncommon in most combustion engineering environments. We can solve these in a variety of ways. I chose here to solve both Eqs. (1.24-1.25) without the reduction provided by Eq. (1.28). However, we can check after numerical solution to see if Eq. (1.28) is actually satised. Substituting numerical values for all the constants to get 2a13 T
13
exp
E 13 RT
= 2 2.9 10 = 1.16 1014
mole cm3 mole cm3 mole cm3
K s
(5000 K)1 exp(0), (1.46) (5000 K)1 , (1.47) (1.48)
1 , s K s
2a14 T 14 exp
E 14 RT
= 2 6.81 1018 exp
erg 4.96 1012 mole erg 8.31441 107 mole K (5000 K)
= 1.77548 1010 a13 T 13 exp a14 T

14
mole cm3
1 , s
exp
E 13 RT E 14 RT
= 5.80 1013 = 8.8774 10
mole cm3
9
1 , s
1
mole cm3
1 . s
(1.49)

O, O2 (mole/cm3) 0.00150 O2 0.00100
23
0.00070
0.00050 O
10
11
10
10
10
10
10
10
t s
Figure 1.2: Molar concentrations versus time for oxygen dissociation problem. Then the dierential equation system becomes dO = (1.16 1014 )2 (O + O2 ) + (1.77548 1010 )O2 (O + O2 ), (1.50) O dt dO2 (1.51) = (5.80 1013 )2 (O + O2 ) (8.8774 109 )O2 (O + O2 ), O dt mole , (1.52) O (0) = 0.001 cm3 mole O2 (0) = 0.001 . (1.53) cm3 These non-linear ordinary dierential equations are in a standard form for a wide variety of numerical software tools. Solution of such equations are not the topic of these notes. 1.1.1.1.2.1 Species concentration versus time A solution was obtained numerically, and a plot of O (t) and O2 (t) is given in Figure 1.2. Note that signicant reaction does not commence until t 1010 s. This can be shown to be very close to the time between molecular collisions. For 109 s < t < 108 s, there is a vigorous reaction. For t > 107 s, the reaction appears to be equilibrated. The calculation gives the equilibrium values e and O e 2 , as O mole , (1.54) t cm3 mole . (1.55) lim O2 = e 2 = 0.00127 O t cm3 Note that at this high temperature, O2 is preferred over O, but there are denitely O molecules present at equilibrium. lim O = e = 0.0004424 O
24
r 1.5 10 1.0 10 7.0 10 5.0 10
16
16
17
17
3.0 10
17
10
10
10
10
10
10
t s
Figure 1.3: Dimensionless residual numerical error r in satisfying the element conservation constraint in the oxygen dissociation example. We can check how well the numerical solution satised the algebraic constraint of element conservation by plotting the dimensionless residual error r r= O + 2O2 O 2O2 O + 2O2 , (1.56)
as a function of time. If the constraint is exactly satised, we will have r = 0. Any non-zero r will be related to the numerical method we have chosen. It may contain roundo error and have a sporadic nature. A plot of r(t) is given in Figure 1.3. Clearly the error is small, and has the character of a roundo error. In fact it is possible to drive r to be smaller by controlling the error tolerance in the numerical method. 1.1.1.1.2.2 Pressure versus time We can use the ideal gas law to calculate the pressure. Recall that the ideal gas law for molecular species i is Pi V = ni RT. (1.57)
Here Pi is the partial pressure of molecular species i, and ni is the number of moles of molecular species i. Note that we also have Pi = ni RT. V ni . V (1.58)
Note that by our denition of molecular species concentration that i =

(1.59)
1.1. ISOTHERMAL, ISOCHORIC KINETICS So we also have the ideal gas law as Pi = i RT.
25
(1.60)
Now in the Dalton mixture model, all species share the same T and V . So the mixture temperature and volume are the same for each species Vi = V , Ti = T . But the mixture pressure is taken to be the sum of the partial pressures:
N
P =
i=1
Pi .
(1.61)
Substituting from Eq. (1.60) into Eq. (1.61), we get

N N
P =
i=1
i RT = RT
i=1
i .
(1.62)
For our example, we only have two species, so P = RT (O + O2 ). The pressure at the initial state t = 0 is P (t = 0) = RT (O + O2 ), erg mole mole (5000 K) 0.001 + 0.001 3 mole K cm cm3 dyne = 8.31441 108 , cm2 = 8.31441 102 bar. = 8.31441 107 (1.64) , (1.65) (1.66) (1.67) (1.63)
This pressure is over 800 atmospheres. It is actually a little too high for good experimental correlation with the underlying data, but we will neglect that for this exercise. At the equilibrium state we have more O2 and less O. And we have a dierent number of molecules, so we expect the pressure to be dierent. At equilibrium, the pressure is P (t ) = lim RT (O + O2 ),
t
(1.68)
mole erg mole (5000 K) 0.0004424 , + 0.00127 3 mole K cm cm3 (1.69) dyne , (1.70) = 7.15 108 cm2 = 7.15 102 bar. (1.71) = 8.31441 107 The pressure has dropped because much of the O has recombined to form O2 . Thus there are fewer molecules at equilibrium. The temperature and volume have remained the same. A plot of P (t) is given in Figure 1.4.
26
P (dyne/cm2) 8.4 108 8.2 108 8. 108 7.8 108 7.6 108 7.4 108 7.2 108
11 10
10
10
10
10
10
10
t s
Figure 1.4: Pressure versus time for oxygen dissociation example. 1.1.1.1.2.3 Dynamical system form Now Eqs. (1.50-1.51) are of the standard form for an autonomous dynamical system: dy = f(y). dt (1.72)
Here y is the vector of state variables (O , O2 )T . And f is an algebraic function of the state variables. For the isothermal system, the algebraic function is in fact a polynomial. Equilibrium The dynamical system is in equilibrium when f(y) = 0. (1.73)
This non-linear set of algebraic equations can be dicult to solve for large systems. We will later see that for common chemical kinetics systems, such as the one we are dealing with, there is a guarantee of a unique equilibrium for which all state variables are physical. There are certainly other equilibria for which at least one of the state variables is non-physical. Such equilibria can be quite mathematically complicated. Solving Eq. (1.73) for our oxygen dissociation problem gives us symbolically from Eq. (1.61.7) 2a13 exp E 13 RT e e e T 13 + 2a14 exp O O M E 14 RT e 2 e T 14 = 0, O M (1.74)
1.1. ISOTHERMAL, ISOCHORIC KINETICS a13 T 13 exp E 13 RT e e e a14 T 14 exp O O M E 14 RT e 2 e = 0. O M
27
(1.75)
We notice that e cancels. This so-called third body will in fact never aect the equilibM rium state. It will however inuence the dynamics. Removing e and slightly rearranging M Eqs. (1.74-1.75) gives a13 T 13 exp a13 T 13 exp E 13 RT E 13 RT e e = a14 T 14 exp O O e e = a14 T 14 exp O O E 14 RT E 14 RT e 2 , O e 2 . O (1.76) (1.77)
These are the same equations! So we really have two unknowns for the equilibrium state e O and e 2 but seemingly only one equation. Note that rearranging either Eq. (1.76) or (1.77) O gives the result a14 T 14 exp e e O O = e 2 O a13 T 13 exp
E 14 RT E 13 RT
= K(T ).
(1.78)
That is, for the net reaction (excluding the inert third body), O2 O+O, at equilibrium the product of the concentrations of the products divided by the product of the concentrations of the reactants is a function of temperature T . And for constant T , this is the so-called equilibrium constant. This is a famous result from basic chemistry. It is actually not complete yet, as we have not taken advantage of a connection with thermodynamics. But for now, it will suce. We still have a problem: Eq. (1.78) is still one equation for two unknowns. We solve this be recalling we have not yet taken advantage of our algebraic constraint of element conservation, Eq. (1.28). Let us use this to eliminate e 2 in favor of e : O O e 2 = O So Eq (1.76) reduces to a13 T 13 exp E 13 RT e e = a14 T 14 exp O O E 14 RT 1 e + O2 . (1.80) O 2 O
=e O
2
1 O e + O2 . O 2
(1.79)
Equation (1.80) is one algebraic equation in one unknown. Its solution gives the equilibrium value e . It is a quadratic equation for e . Of its two roots, one will be physical. We note O O that the equilibrium state will be a function of the initial conditions. Mathematically this is because our system is really best posed as a system of dierential-algebraic equations. Systems which are purely dierential equations will have equilibria which are independent
28
f(O) (mole/cm3/s)
300 000 200 000 100 000
3 0.001 O (mole/cm )
0.004
0.003
0.002
0.001 100 000 200 000
Figure 1.5: Equilibria for oxygen dissociation example. of their initial conditions. Most of the literature of mathematical physics focuses on such systems of those. One of the foundational complications of chemical dynamics is the equilibria is a function of the initial conditions, and this renders many common mathematical notions from traditional dynamic system theory to be invalid Fortunately, after one accounts for the linear constraints of element conservation, one can return to classical notions from traditional dynamic systems theory. Consider the dynamics of Eq. (1.24) for the evolution of O . Equilibrating the right hand side of this equation, gives Eq. (1.74). Eliminating M and then O2 in Eq. (1.74) then substituting in numerical parameters gives the cubic algebraic equation This equation is cubic because we did not remove the eect of M . This will not aect the equilibrium, but will aect the dynamics. We can get an idea of where the roots are by plotting f (O ) as seen in Figure 1.5. Zero crossings of f (O ) in Figure 1.5 represent equilibria of the system, e , f (e ) = 0. The cubic equation has three roots O O e = 0.003 O mole , cm3 non-physical, (1.82) 33948.3 (1.78439 1011 )(O )2 (5.8 1013 )(O )3 = f (O ) = 0. (1.81)
mole , non-physical, (1.83) cm3 mole , physical. (1.84) e = 0.000442414 O cm3 Note the physical root found by our algebraic analysis is identical to that which was identied by our numerical integration of the ordinary dierential equations of reaction kinetics. e = 0.000518944 O Stability of equilibria We can get a simple estimate of the stability of the equilibria by considering the slope of f near f = 0. Our dynamic system is of the form dO (1.85) = f (O ). dt
29
Near the rst non-physical root at e = 0.003, a positive perturbation from equiO librium induces f < 0, which induces dO /dt < 0, so O returns to its equilibrium. Similarly, a negative perturbation from equilibrium induces dO /dt > 0, so the system returns to equilibrium. This non-physical equilibrium point is stable. Note stability does not imply physicality! Perform the same exercise for the non-physical root at e = 0.000518944. We nd O this root is unstable. Perform the same exercise for the physical root at e = 0.000442414. We nd this O root is stable. In general if f crosses zero with a positive slope, the equilibrium is unstable. Otherwise, it is stable. Consider a formal Taylor7 series expansion of Eq. (1.85) in the neighborhood of an equilibrium point 3 : O d df (O e ) = f (e ) + O O dt dO
=0 O =e O
(O e ) + . . . O
(1.86)
We nd df /dO by dierentiating Eq. (1.81) to get df = (3.56877 1011 )O (1.74 1014 )2 . O dO We evaluate df /dO near the physical equilibrium point at O = 0.004442414 to get df = (3.56877 1011 )(0.004442414) (1.74 1014 )(0.004442414)2, dO 1 = 1.91945 108 . (1.88) s Thus the Taylor series expansion of Eq. (1.24) in the neighborhood of the physical equilibrium gives the local kinetics to be driven by d (O 0.00442414) = (1.91945 108 ) (O 0.004442414) + . . . . dt So in the neighborhood of the physical equilibrium we have O = 0.0004442414 + A exp 1.91945 108 t . (1.89) (1.87)
(1.90)
Here A is an arbitrary constant of integration. The local time constant which governs the times scales of local evolution is where 1 = = 5.20983 109 s. (1.91) 8 1.91945 10 This nano-second time scale is very fast. It can be shown to be correlated with the mean time between collisions of molecules.
7
Brook Taylor, 1685-1731, English mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
30
1.1.1.1.3 Eect of temperature Let us perform four case studies to see the eect of T on the systems equilibria and it dynamics near equilibrium. T = 3000 K. Here we have signicantly reduced the temperature, but it is still higher than typically found in ordinary combustion engineering environments. Here we nd e = 8.9371 106 O mole , cm3 = 1.92059 107 s. (1.92) (1.93)
The equilibrium concentration of O dropped by two orders of magnitude relative to T = 5000 K, and the time scale of the dynamics near equilibrium slowed by two orders of magnitude. T = 1000 K. Here we reduce the temperature more. This temperature is common in combustion engineering environments. We nd e = 2.0356 1014 O mole , cm3 = 2.82331 101 s. (1.94) (1.95)
The O concentration at equilibrium is greatly diminished to the point of being dicult to detect by standard measurement techniques. And the time scale of combustion has signicantly slowed. T = 300 K. This is obviously near room temperature. We nd e = 1.14199 1044 O = 1.50977 1031 mole , cm3 s. (1.96) (1.97)
The O concentration is eectively zero at room temperature, and the relaxation time is eectively innite. As the oldest star in our galaxy has an age of 4.4 1017 s, we see that at this temperature, our mathematical model cannot be experimentally validated, so it loses its meaning. At such a low temperature, the theory becomes qualitatively correct, but not quantitatively predictive. T = 10000 K. Such high temperature could be achieved in an atmospheric re-entry environment. e = 2.74807 103 O mole , cm3 = 1.69119 1010 s. (1.98) (1.99)
At this high temperature, O become preferred over O2 , and the time scales of reaction become extremely small, under a nanosecond.
1.1. ISOTHERMAL, ISOCHORIC KINETICS 1.1.1.2 Single reversible reaction
31
The two irreversible reactions studied in the previous section are of a class that is common in combustion modeling. However, the model suers a defect in that its link to classical equilibrium thermodynamics is missing. A better way to model essentially the same physics and guarantee consistency with classical equilibrium thermodynamics is to model the process as a single reversible reaction, with a suitably modied reaction rate term. 1.1.1.2.1 Mathematical model
1.1.1.2.1.1 Kinetics For the reversible O O2 reaction, let us only consider reaction 13 from Table 1.2 for which 13 : O2 + M O + O + M. (1.100)
For this system, we have N = 2 molecular species in L = 1 elements reacting in J = 1 reaction. Here a13 = 1.85 10
11
mole cm3
(K)0.5 ,
13 = 0.5,
E 13 = 95560
cal . mole
(1.101)
Units of cal are common in chemistry, but we need to convert to erg, which is achieved via E 13 = 95560 cal mole 4.186 J cal 107 erg J = 4.00014 1012 erg . mole (1.102)
For this reversible reaction, we slightly modify the kinetics equations to dO = 2 a13 T 13 exp dt
=k13 (T ) =r13
E 13 RT
O2 M
1 , Kc,13 O O M
(1.103)
dO2 dt
= a13 T 13 exp
=k13 (T )
E 13 RT
O2 M
=r13
1 Kc,13 O O M
(1.104)
Here we have used equivalent denitions for k13 (T ) and r13 , so that Eqs. (1.103-1.104) can be written compactly as dO = 2r13 , dt dO2 = r13 . dt (1.105) (1.106)
32 In matrix form, we can simplify to d dt O O2
2 (r13 ). 1
=
(1.107)
Here the N J or 2 1 matrix is = 2 1 . (1.108)
Performing row operations, the matrix form reduces to d dt or 1 0 1 2 d dt O O2 = 2 0 (r13 ). (1.110) O O + 2O2 = 2 0 (r13 ), (1.109)
So here the N N or 2 2 matrix L1 is L1 = 1 0 1 2 . (1.111)
The N N or 2 2 permutation matrix P is the identity matrix. And the N J or 2 1 upper triangular matrix U is 2 . (1.112) U= 0 Note that = L U or equivalently L1 = U: 1 0 1 2
=L1
2 1
=
2 0
=U
(1.113)
Once again the stoichiometric matrix is = (1 2). And we see that = 0 is satised: (1 2) 2 1 = (0). (1.115) (1.114)
As for the irreversible reactions, the reversible reaction rates are constructed to conserve O atoms. We have d + 2O2 = 0. dt O
(1.116)
1.1. ISOTHERMAL, ISOCHORIC KINETICS Thus, we once again nd O + 2O2 = O + 2O2 = constant. As before, we can say O O2
=
33
(1.117)
O O2
b =
1 1 2
=D
( O ) .
=
(1.118)
This gives the dependent variables in terms of a smaller number of transformed dependent variables in a way which satises the linear constraints. In vector form, the equation becomes = + D . Once again D = 0. 1.1.1.2.1.2 Thermodynamics Equations (1.103-1.104) are supplemented by an expression for the thermodynamics-based equilibrium constant Kc,13 which is: Kc,13 = Po exp RT Go 13 RT . (1.120) (1.119)
Here Po = 1.01326 106 dyne/cm2 = 1 atm is the reference pressure. The net change of Gibbs8 free energy at the reference pressure for reaction 13, Go is dened as 13 Go = 2g o g o 2 . O O 13 (1.121)
We further recall that the Gibbs free energy for species i at the reference pressure is dened in terms of the enthalpy and entropy as g o = hi T so . i i
o o
(1.122)
It is common to nd hi and so in thermodynamic tables tabulated as functions of T . i We further note that both Eqs. (1.103) and (1.104) are in equilibrium when e 2 e = O M 1 e e e . Kc,13 O O M (1.123)
We rearrange Eq. (1.123) to nd the familiar Kc,13 =

8
e e O O = e 2 O
[products] . [reactants]
(1.124)
Josiah Williard Gibbs, 1839-1903, American mechanical engineer and the pre-eminent American scientist of the 19th century. CC BY-NC-ND. 12 December 2011, J. M. Powers.
34
If Kc,13 > 1, the products are preferred. If Kc,13 < 1, the reactants are preferred. Now, Kc,13 is a function of T only, so it is known. But Eq. (1.124) once again is one equation in two unknowns. We can use the element conservation constraint, Eq. (1.117) to reduce to one equation and one unknown, valid at equilibrium: Kc,13 e e O O = . 1 O2 + 2 (O e ) O (1.125)
Using the element constraint, Eq. (1.117), we can recast the dynamics of our system by modifying Eq. (1.103) into one equation in one unknown: dO = 2a13 T 13 exp dt E 13 RT
1 1 1 1 (O2 + (O O )) (O2 + (O O ) + O ) O O (O2 + (O O ) + O ) . 2 2 Kc,13 2

=O2 =M =M
(1.126)
1.1.1.2.2 Example calculation Let us consider the same example as the previous section with T = 5000 K. We need numbers for all of the parameters of Eq. (1.126). For O, we nd at T = 5000 K that erg o , (1.127) hO = 3.48382 1012 mole erg so = 2.20458 109 . (1.128) O mole K So go = O erg erg (5000 K) 2.20458 109 , mole mole K erg = 7.53908 1012 . mole 3.48382 1012 erg , mole erg . = 3.05406 109 mole K
(1.129)
For O2 , we nd at T = 5000 K that hO2 = 1.80749 1012 so 2 O So go 2 = O erg erg (5000 K) 3.05406 109 , mole mole K erg = 1.34628 1013 . mole 1.80749 1012
o
(1.130) (1.131)
(1.132)

O, O2 (mole/cm3) 0.00150 O2
35
0.00100 0.00070 0.00050
O 0.00030
10
11
10
10
10
t s 0.001
Figure 1.6: Plot of O (t) and O2 (t) for oxygen dissociation with reversible reaction. Thus, by Eq. (1.121), we have Go = 2(7.53908 1012 ) (1.34628 1013 ) = 1.61536 1012 13 Thus, by Eq. (1.120) we get for our system Kc,13 = 1.01326 106 dyne cm2 erg 7 8.31441 10 mole K (5000 K) exp = 1.187 104 erg . mole (1.133)
erg 1.61536 1012 mole erg 8.31441 107 mole K (5000 K)
(1.134) (1.135)
mole . cm3
Substitution of all numerical parameters into Eq. (1.126) and expansion yields the following dO = 3899.47 (2.23342 1010 )2 (7.3003 1012 )3 = f (O ), O (0) = 0.001.(1.136) O O dt A plot of the time-dependent behavior of O and O2 from solution of Eq. (1.136) is given in Figure 1.6. The behavior is similar to the predictions given by the pair of irreversible
36

f(O) (mole/cm3/s)
40 000 30 000 20 000 10 000 0.004 0.003 0.002 0.001 10 000 20 000
3 0.001 O (mole/cm )
Figure 1.7: Plot of f (O ) versus O for oxygen dissociation with reversible reaction. reactions in Fig. 1.1. Here direct calculation of the equilibrium from time integration reveals e = 0.000393328 O Using Eq. (1.117) we nd this corresponds to e 2 = 0.00130334 O mole . cm3 (1.138) mole . cm3 (1.137)
We note the system begins signicant reaction for t 109 s and is equilibrated for t 107 s. The equilibrium is veried by solving the algebraic equation f (O ) = 3899.47 (2.23342 1010 )2 (7.3003 1012 )3 = 0. O O This yields three roots: e = 0.003 O mole , cm3 non-physical, non-physical, physical, (1.140) (1.141) (1.142) (1.143) consistent with the plot given in Figure 1.7.
(1.139)
e = 0.000452678 O e O
mole , cm3 mole = 0.000393328 , cm3
37
Linearizing Eq. (1.136) in the neighborhood of the physical equilibrium yields the equation d (1.144) (O 0.000393328) = (2.09575 107 ) (O 0.000393328) + . . . dt This has solution O = 0.000393328 + A exp 2.09575 107 t . (1.145) Again, A is an arbitrary constant. Obviously the equilibrium is stable. Moreover the time constant of relaxation to equilibrium is 1 = 4.77156 108 s. (1.146) = 2.09575 107 This is consistent with the time scale to equilibrium which comes from integrating the full equation.
1.1.2
Zeldovich mechanism of N O production
Let us consider next a more complicated reaction system: that of NO production known as the Zeldovich9 mechanism. This is an important model for the production of a major pollutant from combustion processes. It is most important for high temperature applications. Related calculations and analysis of this system are given by Al-Khateeb, et al.10 1.1.2.1 Mathematical model
The model has several versions. One is 1: 2: N + NO N2 + O, N + O2 NO + O. (1.147) (1.148)
similar to our results for O2 dissociation, N2 and O2 are preferred at low temperature. As the temperature rises N and O begin to appear, and it is possible when they are mixed for NO to appear as a product. 1.1.2.1.1 with Standard model form Here we have the reaction of N = 5 molecular species N O N = N2 . O O2
(1.149)
Yakov Borisovich Zeldovich, 1915-1987, prolic Soviet physicist and father of thermonuclear weapons. Al-Khateeb, A. N., Powers, J. M., Paolucci, S., Sommese, A. J., Diller, J. A., Hauenstein, J. D., and Mengers, J. D., 2009, One-Dimensional Slow Invariant Manifolds for Spatially Homogeneous Reactive Systems, Journal of Chemical Physics, 131(2): 024118.
10
38
We have L = 2 with N and O as the 2 elements. The stoichiometric matrix of dimension L N = 2 5 is = 1 1 2 0 0 1 0 0 1 2 . (1.150)
Here, we have
The rst row of is for the N atom; the second row is for the O atom. And we have J = 2 reactions. The reaction vector of length J = 2 is Ta,1 a1 T 1 exp T N N O K1 N2 O r1 c,1 , r = = Ta,2 1 2 r2 N O2 Kc,2 N O O a2 T exp T k1 N N O K1 N2 O c,1 . = 1 k2 N O2 Kc,2 N O O k1 = a1 T 1 exp k2 Ta,1 T Ta,2 = a2 T 2 exp T , .
(1.151)
(1.152)
(1.153) (1.154)
In matrix form, the model can be written as N O 1 1 1 1 d N N = 1 0 dt 2 O 1 1 0 1 O2

=
r1 r2
(1.155)
Here the matrix has dimension N J which is 5 2. The model is of our general form d = r. dt (1.156)
We get 4 zeros because there are 2 reactions each with 2 element constraints.
Note that our stoichiometric constraint on element conservation for each reaction = 0 holds here: 1 1 1 1 0 0 1 1 2 0 0 . (1.157) 0 = = 1 0 0 1 0 0 1 2 1 1 0 1
39
1.1.2.1.2 Reduced form Here we describe non-traditional, but useful reductions, using standard techniques from linear algebra to bring the model equations into a reduced form in which all of the linear constraints have been explicitly removed. Let us perform a series of row operations to nd all of the linear dependencies. Our aim is to convert the matrix into an upper triangular form. The lower left corner of already has a zero, so there is no need to worry about it. Let us add the rst and fourth equations to eliminate the 1 in the 4, 1 slot. This gives 1 1 N O 1 1 N r1 d N = 1 (1.158) 0 2 r2 . dt 0 2 N O + O 0 1 O2 Next, add the rst and third equations to get 1 1 N O 1 1 N d N O + N = 0 1 2 dt 2 N O + O 0 0 1 O2
r1 r2
(1.159)
Now multiply the rst equation by 1 and add it to the second to get N O 1 1 + N 0 2 r1 d NO N O + N = 0 1 2 r2 . dt 0 2 N O + O 0 1 O2
(1.160)
Next add the second and fourth equations to get 1 1 N O 0 2 N O + N d = 0 1 N O + N2 dt 0 0 N + O 0 0 N O + N 2O2
Next multiply the fth equation by 2 and add it to the second to get 1 1 N O 0 2 N O + N r1 d = 0 1 N O + N2 r2 . dt 0 2 N O + O 0 0 N O + N 2O2
(1.161)
r1 r2
(1.162)
40
Rewritten, this becomes 1 0 1 1 1 1 0 1 1 1
Next multiply the third equation by 2 and add it to the second to get N O 1 1 0 2 N O + N r1 d N O + N + 2N = 0 0 2 r2 . dt 0 N + O 0 N O + N 2O2 0 0 0 0 2 0 0 N O 0 0 1 1 0 0 N 0 2 d 0 0 N2 = 0 0 dt 1 0 O 0 0 0 2 0 0 O2
=U
(1.163)
r1 r2
(1.164)
=L1
A way to think of this type of row echelon form is that it denes two free variables, those associated with the non-zero pivots of U: N O and N . The remain three variables N2 , O and O2 are bound variables which can be expressed in terms of the free variables. The last three of the ordinary dierential equations are homogeneous and can be integrated to form N O + N + 2N2 = C1 , N + O = C2 , N O + N 2O2 = C3 . (1.165) (1.166) (1.167)
The constants C1 , C2 and C3 are determined from the initial conditions on all ve state variables. In matrix form, we can say N O N C1 1 1 2 0 0 0 1 0 1 0 N = C2 . (1.168) 2 O C3 1 1 0 0 2 O2
Considering the free variables, N O and N , to be known, we move them to the right side to get 2 0 0 C1 N O N N2 0 1 0 O = . C2 N (1.169) 0 0 2 C3 + N O N O2 Solving, for the bound variables, we nd 1 1 N2 C 1 N O 2 N 2 1 2 . O = C2 N 1 1 1 2 C3 2 N O + 2 N O2 (1.170)
1.1. ISOTHERMAL, ISOCHORIC KINETICS We can rewrite this as
41
We can get a more elegant form by dening N O = N O and N = N . Thus we can say our state variables have the form 0 1 0 N O N 0 0 1 1 N O 1 1 N = C1 + (1.172) 2 2 2 2 N . O C2 0 1 1 1 O2 2 C3 1 2 2 By translating via N O = N O + N O and N = N + N and choosing the constants C1 , C2 and C3 appropriately, we can arrive at N O N O 1 N N 0 N = N2 + 1 2 2 O O 0 2 1 2 O2 O2
= b =
1 1 N2 2 C 2 1 O = C2 + 0 O2 1 1 C3 2 2
1 2 1
1 2
N O N
(1.171)
=D
0 1 1 2 1
1 2
N O N
=
(1.173)
This takes the form of = + D . (1.174)
Here the matrix D is of dimension N R, which here is 5 2. It spans the same column space as does the N J matrix which is of rank R. Here in fact R = J = 2, so D has the same dimension as . In general it will not. If c1 and c2 are the column vectors of D, we see that c1 c2 forms the rst column vector of and c1 c2 forms the second column vector of . Note that D = 0: 1 0 0 1 1 0 0 1 1 2 0 0 . (1.175) D= 2 1 = 2 0 0 1 0 0 1 2 0 1 1 1 2 2 Equations (1.165-1.167) can also be linearly combined in a way which has strong physical relevance. We rewrite the system as three equations in which the rst is identical to
42
Eq. (1.165); the second is the dierence of Eqs. (1.166) and (1.167); and the third is half of Eq. (1.165) minus half of Eq. (1.167) plus Eq. (1.166): N O + N + 2N2 = C1 , O + N O + 2O2 = C2 C3 , 1 (C1 C3 ) + C2 . N O + N + N2 + O + O2 = 2 (1.176) (1.177) (1.178)
Equation (1.176) insists that the number of nitrogen elements be constant; Eq. (1.177) demands the number of oxygen elements be constant; and Eq. (1.178) requires the number of moles of molecular species be constant. For general reactions, including the earlier studied oxygen dissociation problem, the number of moles of molecular species will not be constant. Here because each reaction considered has two molecules reacting to form two molecules, we are guaranteed the number of moles will be constant. Hence, we get an additional linear constraint beyond the two for element conservation. Note that since our reaction is isothermal, isochoric and mole-preserving, it will also be isobaric. 1.1.2.1.3 Example calculation Let us consider an isothermal reaction at T = 6000 K. (1.179)
The high temperature is useful in generating results which are easily visualized. It insures that there will be signicant concentrations of all molecular species. Let us also take as an initial condition N O = N = N2 = O = O2 = 1 106 mole . cm3 (1.180)
For this temperature and concentrations, the pressure, which will remain constant through the reaction, is P = 2.4942 106 dyne/cm2 . This is a little greater than atmospheric. Kinetic data for this reaction is adopted from Baulch, et al.11 The data for reaction 1 is a1 = 2.107 1013 For reaction 2, we have a2 = 5.8394 10
9
mole cm3
1
1 , s
1 = 0,
Ta1 = 0 K.
(1.181)
mole cm3
1 K 1.01 s
2 = 1.01,
Ta2 = 3120 K.
(1.182)
Here the so-called activation temperature Ta,j for reaction j is really the activation energy scaled by the universal gas constant: Ta,j =
11
Ej . R
(1.183)
Baulch, et al., 2005, Evaluated Kinetic Data for Combustion Modeling: Supplement II, Journal of Physical and Chemical Reference Data, 34(3): 757-1397. CC BY-NC-ND. 12 December 2011, J. M. Powers.
1.1. ISOTHERMAL, ISOCHORIC KINETICS Substituting numbers we obtain for the reaction rates k1 = (2.107 1013 )(6000)0 exp k2 = (5.8394 10 )(6000)
9 1.01
43
0 6000
= 2.107 1013
mole cm3
13
1 , s
1
(1.184) 1 . s (1.185)
exp
3120 6000
= 2.27231 10
mole cm3
We will also need thermodynamic data. The data here will be taken from the Chemkin database.12 Thermodynamic data for common materials is also found in most thermodynamic texts. For our system at 6000 K, we nd g o O = 1.58757 1013 N g o = 7.04286 1012 N g o 2 = 1.55206 1013 N g o = 9.77148 1012 O g o 2 = 1.65653 1013 O Thus for each reaction, we nd Go : j Go = g o 2 + g o g o g o O , (1.191) 1 N O N N 13 12 12 13 = 1.55206 10 9.77148 10 + 7.04286 10 + 1.58757 10 , 1.192) ( erg , (1.193) = 2.37351 1012 mole Go = g o O + g o g o g o 2 , (1.194) 2 N O N O 13 12 12 13 = 1.58757 10 9.77148 10 + 7.04286 10 + 1.65653 10 , 1.195) ( erg . (1.196) = 2.03897 1012 mole At 6000 K, we nd the equilibrium constants for the J = 2 reactions are Kc,1 = exp Go 1 , RT 2.37351 1012 = exp (8.314 107 )(6000) = 116.52, Go 2 , = exp RT (1.197) (1.198) (1.199) (1.200) erg , mole erg , mole erg , mole erg , mole erg . mole (1.186) (1.187) (1.188) (1.189) (1.190)
Kc,2
12
R. J. Kee, et al., 2000, The Chemkin Thermodynamic Data Base, part of the Chemkin Collection Release 3.6, Reaction Design, San Diego, CA. CC BY-NC-ND. 12 December 2011, J. M. Powers.
44
N, NO (mole/cm3) 1 10 5 5 10 6
1 10 6 5 10 7 NO
1 10 7 5 10 8 N 10 10 10 9 10 8 10 7 10 6 10 5 t s
Figure 1.8: NO and N concentrations versus time for T = 6000 K, P = 2.4942 106 dyne/cm2 Zeldovich mechanism. 2.03897 1012 , (8.314 107 )(6000) = 59.5861. = exp (1.201) (1.202)
Again, omitting details, we nd the two dierential equations governing the evolution of the free variables are dN O = 0.723 + 2.22 107 N + 1.15 1013 2 9.44 105 N O 3.20 1013 N N O , N dt (1.203) dN = 0.723 2.33 107 N 1.13 1013 2 + 5.82 105 N O 1.00 1013 N N O . N dt (1.204) Solving numerically, we obtain a solution shown in Fig. 1.8. The numerics show a relaxation to nal concentrations of
t
lim N O = 7.336 107

t
lim N = 3.708 108
mole , cm3 mole . cm3
(1.205) (1.206)
Equations (1.203-1.204) are of the form dN O = fN O (N O , N ), dt

(1.207)
1.1. ISOTHERMAL, ISOCHORIC KINETICS dN dt At equilibrium, we must have fN O (N O , N ) = 0, fN (N O , N ) = 0. We nd three nite roots to this problem: 1 : (N O , N ) = (1.605 106 , 3.060 108 ) = fN (N O , N ).
45 (1.208)
(1.209) (1.210)
mole , non-physical, (1.211) cm3 mole , non-physical, (1.212) 2 : (N O , N ) = (5.173 108 , 2.048 106 ) cm3 mole , physical. (1.213) 3 : (N O , N ) = (7.336 107 , 3.708 108 ) cm3
Obviously, because of negative concentrations, roots 1 and 2 are non-physical. Root 3 however is physical; moreover, it agrees with the equilibrium we found by direct numerical integration of the full non-linear equations. We can use local linear analysis in the neighborhood of each equilibria to rigorously ascertain the stability of each root. Taylor series expansion of Eqs. (1.207-1.208) in the neighborhood of an equilibrium point yields fN O d (N O e O ) = fN O |e + N dt N O
=0
(N O e O ) + N
fN O N
(N e ) + . . . , N (1.214)
fN d (N e ) = fN |e + N dt N O
=0
(N O e O ) + N
fN N
(N e ) + . . . . N
(1.215)
Evaluation of Eqs. (1.214-1.215) near the physical root, root 3, yields the system d dt N O 7.336 107 N 3.708 108 = 2.129 106 2.111 105
=J=
4.155 105 3.144 107

f e
N O 7.336 107 N 3.708 108
(1.216)
This is of the form d f ( e ) = dt ( e ) = J ( e ) . (1.217)
It is the eigenvalues of the Jacobian13 matrix J that give the time scales of evolution of the concentrations as well as determine the stability of the local equilibrium point. Recall that
13
after Carl Gustav Jacob Jacobi, 1804-1851, German mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
46
we can usually decompose square matrices via the diagonalization J = P1 P. (1.218)
Here P is the matrix whose columns are composed of the right eigenvectors of J, and is the diagonal matrix whose diagonal is populated by the eigenvalues of J. For some matrices (typically not those encountered after our removal of linear dependencies), diagonalization is not possible, and one must resort to the so-called near-diagonal Jordan form. This will not be relevant to our discussion, but could be easily handled if necessary. We also recall the eigenvector matrix and eigenvalue matrix are dened by the standard eigenvalue problem P J = P. (1.219)
We also recall that the components of are found by solving the characteristic polynomial which arises from the equation det (J I) = 0, (1.220)
where I is the identity matrix. With the decomposition Eq. (1.218), Eq. (1.217) can be rearranged to form d (P ( e )) = P ( e ). dt Taking z P ( e ), Eq. (1.221) reduces the diagonal form dz = z. dt This has solution for each component of z of z1 = C1 exp(1 t), z2 = C2 exp(2 t), . . . (1.224) (1.225) (1.226) (1.223) (1.222) (1.221)
Here, our matrix J, see Eq. (1.216), has two real, negative eigenvalues in the neighborhood of the physical root 3: 1 = 3.143 107 2 1 , s 1 = 2.132 106 . s (1.227) (1.228)
47
Thus we can conclude that the physical equilibrium is linearly stable. The local time constants near equilibrium are given by the reciprocal of the magnitude of the eigenvalues. These are 1 = 1/|1| = 3.181 108 s, 2 = 1/|2| = 4.691 107 s. (1.229) (1.230)
Evolution on these two time scales is predicted in Fig. 1.8. This in fact a multiscale problem. One of the major diculties in the numerical simulation of combustion problems comes in the eort to capture the eects at all relevant scales. The problem is made more dicult as the breadth of the scales expands. In this problem, the breadth of scales is not particularly challenging. Near equilibrium the ratio of the slowest to the fastest time scale, the stiness ratio , is = 2 4.691 107 s = = 14.75. 1 3.181 108 s (1.231)
Many combustion problems can have stiness ratios over 106 . This is more prevalent at lower temperatures. We can do a similar linearization near the initial state, nd the local eigenvalues, and the local time scales. At the initial state here, we nd those local time scales are 1 = 2.403 108 s, 2 = 2.123 108 s. (1.232) (1.233)
So initially the stiness, = (2.403 108 s)/(2.123 108 s) = 1.13 is much less, but the time scale itself is small. It is seen from Fig. 1.8 that this initial time scale of 108 s well predicts where signicant evolution of species concentrations commences. For t < 108 s, the model predicts essentially no activity. This can be correlated with the mean time between molecular collisionsthe theory on which estimates of the collision frequency factors aj are obtained. We briey consider the non-physical roots, 1 and 2. A similar eigenvalue analysis of root 1 reveals that the eigenvalues of its local Jacobian matrix are 1 = 1.193 107 2 = 5.434 106 Thus root 1 is a saddle and is unstable. For root 2, we nd 1 = 4.397 107 + i7.997 106 2 1 , s 1 = 4.397 107 i7.997 106 . s (1.236) (1.237) 1 , s (1.234) (1.235)
1 . s
48
-7
x 10 5 0 N (mole/cm 3)
saddle
3 sink
SIM
SIM
-5 -10 -15 -20
spira l source 2 -4 -3 -2 -1 0 1 2 x 10
-6
NO (mole/cm 3)
Figure 1.9: NO and N phase portraits for T = 6000 K, P = 2.4942 106 dyne/cm2 Zeldovich mechanism. The eigenvalues are complex with a positive real part. This indicates the root is an unstable spiral source. A detailed phase portrait is shown in Fig. 1.9. Here we see all three roots. Their local character of sink, saddle, or spiral source is clearly displayed. We see that trajectories are attracted to a curve labeled SIM for Slow Invariant Manifold. A part of the SIM is constructed by the trajectory which originates at root 1 and travels to root 3. The other part is constructed by connecting an equilibrium point at innity into root 3. Details are omitted here. 1.1.2.2 Stiness, time scales, and numerics
One of the key challenges in computational chemistry is accurately predicting species concentration evolution with time. The problem is made dicult because of the common presence

N, NO (mole/cm3) 10 6 NO
49
10 8
10 10
10 12
10 11
10 9
10 7
10 5
t s 0.001 0.1 10
Figure 1.10: N O and N versus time for Zeldovich mechanism at T = 1500 K, P = 6.23550 105 dyne/cm2 . of physical phenomena which evolve on a widely disparate set of time scales. Systems which evolve on a wide range of scales are known as sti, recognizing a motivating example in mass-spring-damper systems with sti springs. Here we will examine the eect of temperature and pressure on time scales and stiness. We shall also look simplistically how dierent numerical approximation methods respond to stiness.
1.1.2.2.1 Eect of temperature Let us see how the same Zeldovich mechanism behaves at lower temperature, T = 1500 K; all other parameters, including the initial species concentrations are the same as the previous high temperature example. The pressure however, lowers, and here is P = 6.23550105 dyne/cm2 , which is close to atmospheric pressure. For this case, a plot of species concentrations versus time is given in Figure 1.10. At T = 1500 K, we notice some dramatic dierences relative to the earlier studied T = 6000 K. First, we see the reaction commences in around the same time, t 108 s. For t 106 s, there is a temporary cessation of signicant reaction. We notice a long plateau in which species concentrations do not change over several decades of time. This is actually a pseudo-equilibrium. Signicant reaction recommences for t 0.1 s. Only around t 1 s does the system approach nal equilibrium. We can perform an eigenvalue analysis both at the initial state and at the equilibrium state to estimate the time scales of reaction. For this dynamical system which is two ordinary dierential equations in two unknowns, we will always nd two eigenvalues, and thus two time scales. Let us call them 1 and 2 . Both these scales will evolve with t.
50 At the initial state, we nd
1 = 2.37 108 s, 2 = 4.25 107 s.
(1.238) (1.239)
The onset of signicant reaction is consistent with the prediction given by 1 at the initial state. Moreover, initially, the reaction is not very sti; the stiness ratio is = 17.9. At equilibrium, we nd
t
lim N O = 4.6 109

t
lim N
mole , cm3 mole , = 4.2 1014 cm3
(1.240) (1.241)
and 1 = 7.86 107 s, 2 = 3.02 101 s. (1.242) (1.243)
The slowest time scale near equilibrium is an excellent indicator of how long the system takes to relax to its nal state. Note also that near equilibrium, the stiness ratio is large, = 2 /1 3.8 105 . This is known as the stiness ratio. When it is large, the scales in the problem are widely disparate and accurate numerical solution becomes challenging. In summary, we nd the eect of lowering temperature while leaving initial concentrations constant: lowers the pressure somewhat, slightly slowing down the collision time, and slightly slowing the fastest time scales, and slows the slowest time scales many orders of magnitude, stiening the system signicantly, since collisions may not induce reaction with their lower collision speed. 1.1.2.2.2 Eect of initial pressure Let us maintain the initial temperature at T = 1500 K, but drop the initial concentration of each species to N O = N = N2 = O2 = O = 108 With this decrease in number of moles, the pressure now is P = 6.23550 103 dyne . cm2 (1.245) mole . cm3 (1.244)
This pressure is two orders of magnitude lower than atmospheric. We solve for the species concentration proles and show the results of numerical prediction in Figure 1.11. Relative to the high pressure P = 6.2355 105 dyne/cm2 , T = 1500 K case, we notice some similarities

N, NO (mole/cm3) 10 8 NO
51
10 10
10 12
10 14 N 10 16
10 8
10 6
10 4
t s 0.01 1 100
Figure 1.11: N O and N versus time for Zeldovich mechanism at T = 1500 K, P = 6.2355 103 dyne/cm2 . and dramatic dierences. The overall shape of the time-proles of concentration variation is similar. But, we see the reaction commences at a much later time, t 106 s. For t 104 s, there is a temporary cessation of signicant reaction. We notice a long plateau in which species concentrations do not change over several decades of time. This is again actually a pseudo-equilibrium. Signicant reaction recommences for t 10 s. Only around t 100 s does the system approach nal equilibrium. We can perform an eigenvalue analysis both at the initial state and at the equilibrium state to estimate the time scales of reaction. At the initial state, we nd 1 = 2.37 106 s, 2 = 4.25 105 s. (1.246) (1.247)
The onset of signicant reaction is consistent with the prediction given by 1 at the initial state. Moreover, initially, the reaction is not very sti; the stiness ratio is = 17.9. Interestingly, by decreasing the initial pressure by a factor of 102 , we increased the initial time scales by a complementary factor of 102 ; moreover, we did not alter the stiness. At equilibrium, we nd
t
lim N O = 4.6 1011

t
lim N = 4.2 1016
mole , cm3 mole , cm3
(1.248) (1.249) (1.250)
52 and
1 = 7.86 105 s, 2 = 3.02 101 s.
(1.251) (1.252)
By decreasing the initial pressure by a factor of 102 , we decreased the equilibrium concentrations by a factor of 102 and increased the time scales by a factor of 102 , leaving the stiness ratio unchanged. In summary, we nd the eect of lowering the initial concentrations signicantly while leaving temperature constant lowers the pressure signicantly, proportionally slowing down the collision time, as well as the fastest and slowest time scales, and does not aect the stiness of the system. 1.1.2.2.3 Stiness and numerics The issue of how to simulate sti systems of ordinary dierential equations, such as presented by our Zeldovich mechanism, is challenging. Here a brief summary of some of the issues will be presented. The interested reader should consult the numerical literature for a full discussion. See for example the excellent text of Iserles.14 We have seen throughout this section that there are two time scales at work, and they are often disparate. The species evolution is generally characterized by an initial fast transient, followed by a long plateau, then a nal relaxation to equilibrium. We noted from the phase plane of Fig. 1.9 that the nal relaxation to equilibrium (shown along the green line labeled SIM) is an attracting manifold for a wide variety of initial conditions. The relaxation onto the SIM is fast, and the motion on the SIM to equilibrium is relatively slow. Use of common numerical techniques can often mask or obscure the actual dynamics. Numerical methods to solve systems of ordinary dierential equations can be broadly categorized as explicit or implicit. We give a brief synopsis of each class of method. We cast each as a method to solve a system of the form d = f(). dt (1.253)
Explicit: The simplest of these methods, the forward Euler method, discretizes Eq. (1.253) as follows: n+1 n = f(n ), (1.254) t so that n+1 = n + t f(n ). Explicit methods are summarized as
A. Iserles, 2008, A First Course in the Numerical Analysis of Dierential Equations, Cambridge University Press, Cambridge, UK. CC BY-NC-ND. 12 December 2011, J. M. Powers.
14
(1.255)
53
easy to program, since Eq. (1.255) can be solved explicitly to predict the new value n+1 in terms of the old values at step n. need to have t < f astest in order to remain numerically stable, able to capture all physics and all time scales at great computational expense for sti problems, requiring much computational eort for little payo in the SIM region of the phase plane, and thus inecient for some portions of sti calculations. Implicit: The simplest of these methods, the backward Euler method, discretizes Eq. (1.253) as follows: n+1 n = f(n+1 ), t so that n+1 = n + t f(n+1 ). Implicit methods are summarized as more dicult to program since a non-linear set of algebraic equations, Eq. (1.257), must be solved at every time step with no guarantee of solution, requiring potentially signicant computational time to advance each time step, capable of using very large time steps and remaining numerically stable, suspect to missing physics that occur on small time scales < t, and in general better performers than explicit methods. A wide variety of software tools exist to solve systems of ordinary dierential equations. Most of them use more sophisticated techniques than simple forward and backward Euler methods. One of the most powerful techniques is the use of error control. Here the user species how far in time to advance and the error that is able to be tolerated. The algorithm, which is complicated, selects then internal time steps, for either explicit or implicit methods, to achieve a solution within the error tolerance at the specied output time. A well known public domain algorithm with error control is provided by lsode.f, which can be found in the netlib repository.15 Let us exercise the Zeldovich mechanism under the conditions simulated in Fig. 1.11, T = 1500 K, P = 6.2355 103 dyne/cm2 . Recall in this case the fastest time scale near equilibrium is 1 = 7.86 105 s 104 s at the initial state, and the slowest time scale is
Hindmarsh, A. C., 1983,ODEPACK, a Systematized tic Computing, edited by R. S. Stepleman, et al., http://www.netlib.org/alliant/ode/prog/lsode.f.
15
(1.256)
(1.257)
Collection of North-Holland,
ODE Solvers, ScienAmsterdam, pp. 55-64.
54 Explicit Ninternal 106 105 104 103 102 101 100 100 100
CHAPTER 1. INTRODUCTION TO KINETICS Explicit tef f (s) 104 104 104 104 104 104 104 105 106 Implicit Ninternal 100 100 100 100 100 100 100 100 100 Implicit tef f (s) 102 101 100 101 102 103 104 105 106
t (s) 102 101 100 101 102 103 104 105 106
Table 1.3: Results from computing Zeldovich NO production using implicit and explicit methods with error control in dlsode.f. = 3.02 101 s at the nal state. Let us solve for these conditions using dlsode.f, which uses internal time stepping for error control, in both an explicit and implicit mode. We specify a variety of values of t and report typical values of number of internal time steps selected by dlsode.f, and the corresponding eective time step tef f used for the problem, for both explicit and implicit methods, as reported in Table 1.3. Obviously if output is requested using t > 104 s, the early time dynamics near t 4 10 s will be missed. For physically stable systems, codes such as dlsode.f will still provide a correct solution at the later times. For physically unstable systems, such as might occur in turbulent ames, it is not clear that one can use large time steps and expect to have delity to the underlying equations. The reason is the physical instabilities may evolve on the same time scale as the ne scales which are overlooked by large t.
1.2
Adiabatic, isochoric kinetics
It is more practical to allow for temperature variation within a combustor. The best model for this is adiabatic kinetics. Here we will restrict our attention to isochoric problems.
1.2.1
Thermal explosion theory
There is a simple description known as thermal explosion theory which provides a good explanation for how initially slow exothermic reaction induces a sudden temperature rise accompanied by a nal relaxation to equilibrium. Let us consider a simple isomerization reaction in a closed volume A B. (1.258)
Let us take A and B to both be calorically perfect ideal gases with identical molecular masses MA = MB = M and identical specic heats, cvA = cvB = cv ; cP A = cP B = cP . We can
1.2. ADIABATIC, ISOCHORIC KINETICS
55
consider A and B to be isomers of an identical molecular species. So we have N = 2 species reacting in J = 1 reactions. The number of elements L here is irrelevant. 1.2.1.1 One-step reversible kinetics
Let us insist our reaction process be isochoric and adiabatic, and commence with only A present. The reaction kinetics with = 0 are dA = a exp dt E RT
=k =r
1 , Kc B
(1.259)
dB dt
= a exp
E RT
=k
A
=r
1 , Kc B
(1.260)
A (0) = A , B (0) = 0. For our alternate compact linear algebra based form, we note that r = a exp and that d dt Performing the decomposition yields d dt Expanded, this is 1 0 1 1 d dt A B = 1 0 (r). A A + B = 1 0 (r). A B = 1 1 (r). E RT A 1 , Kc B
(1.261) (1.262)
(1.263)
(1.264)
(1.265)
(1.266)
Combining Eqs. (1.259-1.260) and integrating yields d ( + B ) = 0, dt A A + B = A , B = A A . (1.267) (1.268) (1.269)
56 Thus, Eq. (1.259) reduces to dA = a exp dt Scaling, Eq. (1.270) can be rewritten as d d(at) 1.2.1.2 A A = exp
E RT E 1 RTo T /To
1 A Kc A
(1.270)
1 A A Kc
A A
(1.271)
First law of thermodynamics
Recall the rst law of thermodynamics and neglecting potential and kinetic energy changes: dE = Q W. (1.272) dt Here E is the total internal energy. Because we insist the problem is adiabatic Q = 0. Because we insist the problem is isochoric, there is no work done, so W = 0. Thus we have dE = 0. (1.273) dt Thus, we nd E = Eo . Recall the total internal energy for a mixture of two calorically perfect ideal gases is E = nA eA + nB eB , nA nB = V eA + eB , V V = V (A eA + B eB ) , PA PB + B (hB = V A hA A B = V A hA RT + B (hB RT ,
o o
(1.274)
(1.275) (1.276) (1.277) , (1.278) (1.279)

o
= V A cP (T To ) + hTo ,A RT + B (cP (T To ) + hTo ,B RT = V (A + B )(cP (T To ) RT ) + A hTo ,A + B hTo ,B , = V (A + B )((cP R)T cP To ) + A hTo ,A + B hTo ,B , = V (A + B )((cP R)T (cP R + R)To ) + A hTo ,A + B hTo ,B , = V (A + B )(cv T (cv + R)To ) + A hTo ,A + B hTo ,B , = V (A + B )cv (T To ) + A (hTo ,A RTo ) + B (hTo ,B RTo ) , = V (A + B )cv (T To ) + A eo o ,A + B eo o ,B . T T
o o o o o o o o o
,(1.280) (1.281) (1.282) (1.283) (1.284) (1.285) (1.286)
1.2. ADIABATIC, ISOCHORIC KINETICS Now at the initial state, we have T = To , so Eo = V A eo o ,A + B eo o ,B . T T So, we can say our caloric equation of state is
57
(1.287)
E Eo = V (A + B )cv (T To ) + (A A )eo o ,A + (B B )eo o ,B , (1.288) T T = V (A + B )cv (T To ) + (A A )eo o ,A + (B B )eo o ,B . (1.289) T T As an aside, on a molar basis, we scale Eq. (1.289) to get e eo = cv (T To ) + (yA yAo )eo o ,A + (yB yBo )eo o ,B . T T (1.290)
And because we have assumed the molecular masses are the same, MA = MB , the mole fractions are the mass fractions, and we can write on a mass basis e eo = cv (T To ) + (YA YAo )eo o ,A + (YB YBo )eo o ,B . T T Returning to Eq. (1.289), our energy conservation relation, Eq. (1.274), becomes 0 = V (A + B )cv (T To ) + (A A )eo o ,A + (B B )eo o ,B . T T Now we solve for T 0 = (A + B )cv (T To ) + (A A )eo o ,A + (B B )eo o ,B , T T (1.293) (1.294) (1.295) (1.292) (1.291)
A A o B o eTo ,A + B eT ,B , A + B A + B o A A eo o ,A B B eo o ,B T T T = To + + . A + B cv A + B cv 0 = cv (T To ) +
Now we impose our assumption that B = 0, giving also B = A A , T = To + A A eo o ,A B eo o ,B T T , cv A A cv A eo o ,A eo o ,B T T . = To + A cv A

o o
(1.296) (1.297)
A :
In summary, realizing that hTo ,A hTo ,B = eo o ,A eo o ,B we can write T as a function of T T T = To + ( A A ) o o (hTo ,A hTo ,B ). A cv (1.298)
58
o o
We see then that if hTo ,A > hTo ,B , that as A decreases from its initial value of A that T will increase. We can scale Eq. (1.298) to form T To =1+ 1 A A hTo ,A hTo ,B cv To
o o
(1.299)
We also note that our caloric state equation, Eq. (1.290) can, for yAo = 1, yBo = 0 as e eo = cv (T To ) + (yA 1)eo o ,A + yB eo o ,B , T T = cv (T To ) + ((1 yB ) 1)eo o ,A + yB eo o ,B , T T o o = cv (T To ) yB (eTo ,A eTo ,B ). Similarly, on a mass basis, we can say, e eo = cv (T To ) YB (eo o ,A eo o ,B ). T T For this problem, we also have Kc = exp with Go = = = = So Kc = exp = exp = exp hTo ,A hTo ,B T (so o ,A so o ,B ) T T RT
o o o
(1.300) (1.301) (1.302)
(1.303)
Go RT
(1.304)
g o go , B A o o hB T so (hA T so ), B A o o o (hB hA ) T (sB so ), A o o o (hTo ,B hTo ,A ) T (sTo ,B so o ,A ). T
(1.305) (1.306) (1.307) (1.308)
, , .
(1.309) (1.310) (1.311)
cv To RT
hTo ,A hTo ,B T (so o ,A so o ,B ) T T cv To

o o
1 1 T 1 To
hTo ,A hTo ,B T (so o ,A so o ,B ) T T cv To To cv
Here we have used the denition of the ratio of specic heats, = cP /cv along with R = cP cv . So we can solve Eq. (1.270) by rst using Eq. (1.311) to eliminate Kc and then Eq. (1.298) to eliminate T .
1.2. ADIABATIC, ISOCHORIC KINETICS 1.2.1.3 Dimensionless form
59
Let us try writing dimensionless variables so that our system can be written in a compact dimensionless form. First lets take dimensionless time to be = at. Let us take dimensionless species concentration to be z with z= A . A T . To
o
(1.312)
(1.313)
Let us take dimensionless temperature to be with = (1.314)
Let us take dimensionless heat release to be q with hT ,A hTo ,B q= o . cv To Let us take dimensionless activation energy to be with = E . RTo (1.316)
o
(1.315)
And let us take the dimensionless entropy change to be with (so o ,A so o ,B ) T T = . cv So our equations become 1 dz z = exp (1 z) , d Kc = 1 + (1 z)q, 1 1 Kc = exp (q ) . 1 It is more common to consider the products. Let us dene for general problems = B B . = A + B A + B (1.321) (1.318) (1.319) (1.320) (1.317)
Thus is the mass fraction of product. For our problem, B = 0 so = B A = A . A A (1.322)
60 Thus,
= 1 z.
(1.323)
We can think of as a reaction progress variable as well. When = 0, we have = 0, and the reaction has not begun. Thus, we get 1 d (1 ) = exp , d Kc = 1 + q, 1 1 Kc = exp (q ) . 1 1.2.1.4 Example calculation (1.324) (1.325) (1.326)
Let us choose some values for the dimensionless parameters: = 20, = 0, q = 10, 7 = . 5 (1.327)
With these choices, our kinetics equations reduce to d = exp d 20 1 + 10 (1 ) exp 25 1 + 10 , (0) = 0. (1.328)
The right side of Eq. (1.328) is at equilibrium for values of which drive it to zero. Numerical root nding methods show this to occur at 0.920539. Near this root, Taylor series expansion shows the dynamics are approximated by d ( 0.920539) = 0.17993( 0.920539) + . . . d Thus the local behavior near equilibrium is given by = 0.920539 + C exp (0.17993 ) . (1.330) (1.329)
Here C is some arbitrary constant. Clearly the equilibrium is stable, with a time constant of 1/0.17993 = 5.55773. Numerical solution shows the full behavior of the dimensionless species concentration ( ); see Figure 1.12. Clearly the product concentration is small for some long period of time. At a critical time near = 2.7 106 , there is a so-called thermal explosion with a rapid increase in . Note that the estimate of the time constant near equilibrium is orders of magnitude less than the explosion time, 5.55773 << 2.7 106 . Thus, linear analysis here is a poor tool to estimate an important physical quantity, the ignition time. Once the ignition period is over, there is a rapid equilibration to the nal state. The dimensionless temperature plot is shown in Figure 1.13. The temperature plot is similar in behavior to
61
1.0
0.8
0.6
0.4
0.2
1. 106 2. 106 3. 106 4. 106 5. 106
Figure 1.12: Dimensionless plot of reaction product concentration versus time for adiabatic isochoric combustion with simple reversible kinetics.
10
1. 106 2. 106 3. 106 4. 106 5. 106
Figure 1.13: Dimensionless plot of temperature versus time for adiabatic, isochoric combustion with simple reversible kinetics.
62
the species concentration plot. At early time, the temperature is cool. At a critical time, the thermal explosion time, the temperature rapidly rises. This rapid rise, coupled with the exponential sensitivity of reaction rate to temperature, accelerates the formation of product. This process continues until the reverse reaction is activated to the extent it prevents further creation of product. 1.2.1.5 High activation energy asymptotics
Let us see if we can get an analytic prediction of the thermal explosion time, 2.7 106 . Such a prediction would be valuable to see how long a slowing reacting material might take to ignite. Our analysis is similar to that given by Buckmaster and Ludford in their Chapter 1.16 For convenience let us restrict ourselves to = 0. In this limit, Eqs. (1.324-1.326) reduce to d = exp d 1 + q (1 ) exp q ( 1)(1 + q) , (1.331)
with (0) = 0. The key trouble in getting an analytic solution to Eq. (1.331) is the presence of in the denominator of an exponential term. We need to nd a way to move it to the numerator. Asymptotic methods provide one such way. Now we recall for early time << 1. Let us assume takes the form = 1 + 2 2 + 3 3 + . . . (1.332)
Here we will assume 0 < << 1 and that 1 ( ) O(1), 2 ( ) O(1), . . ., and will dene in terms of physical parameters shortly. Now with this assumption, we have 1 1 = . 2 q + 3 q + . . . 1 + q 1 + q1 + 2 3 Long division of the term on the right side yields the approximation 1 = = 1 q1 + 2 (q 2 2 q2 ) + . . . , 1 1 + q = 1 q1 + O(2 ). So exp 1 + q exp (1 q1 + O(2 )) , exp() exp q1 + O(2 ) . (1.336) (1.337) (1.334) (1.335) (1.333)
We have moved from the denominator to the numerator of the most important exponential term.
16
J. D. Buckmaster and G. S. S. Ludford, 1983, Lectures on Mathematical Combustion, SIAM, Philadelphia.
1.2. ADIABATIC, ISOCHORIC KINETICS Now, let us take the limit of high activation energy by dening to be 1 .
63
(1.338)
Let us let the assume the remaining parameters, q and are both O(1) constants. When is large, will be small. With this denition, Eq. (1.337) becomes exp 1 + q exp 1 exp q1 + O(2 ) . (1.339)
With these assumptions and approximations, Eq. (1.331) can be written as 1 d (1 + . . .) = exp d exp q1 + O(2 ) q ( 1)(1 + q1 + . . .) . (1.340)
(1 1 . . .) (1 + . . .) exp
Now let us rescale time via = 1 exp 1 . (1.341)
With this transformation, the chain rule shows how derivatives transform: 1 d d d = = d d d exp With this transformation, Eq. (1.340) becomes 1 exp
1 1
d . d
(1.342)
1 d (1 + . . .) = exp d
exp q1 + O(2 ) q . ( 1)(1 + q1 + . . .) (1.343)
(1 1 . . .) (1 + . . .) exp
This simplies to d (1 + . . .) = exp q1 + O(2 ) d (1 1 . . .) (1 + . . .) exp q ( 1)(1 + q1 + . . .) . (1.344)
64
Retaining only O(1) terms in Eq. (1.344), we get d1 = exp (q1 ) . d (1.345)
This is supplemented by the initial condition 1 (0) = 0. Separating variables and solving, we get exp(q1 )d1 = d , 1 exp(q1 ) = + C. q Applying the initial condition gives 1 exp(q(0)) = C, q 1 = C. q (1.348) (1.349) (1.346) (1.347)
So 1 1 exp(q1 ) = , q q exp(q1 ) = q + 1, exp(q1 ) = q q1 1 (1.350) (1.351) (1.352) (1.353) . (1.354)
1 , q 1 = ln q , q 1 1 = ln q q q
For q = 10, a plot of 1 ( ) is shown in Fig. 1.14. We note at a nite that 1 begins to exhibit unbounded growth. In fact, it is obvious from Eq. (1.345) that as 1 , q that 1 . That is there exists a nite time for which 1 violates the assumptions of our asymptotic theory which assumes 1 = O(1). We associate this time with the ignition time, i : 1 i = . q
(1.355)
65
1
1.4 1.2 1.0 0.8 0.6 0.4 0.2
0.00
0.02
0.04
0.06
0.08
0.10
0.12
Figure 1.14: 1 versus for ignition problem. Let us return this to more primitive variables: 1 exp 1 i = i i 1 , q exp 1 = , q exp = . q (1.356) (1.357) (1.358)
For our system with = 20 and q = 10, we estimate the dimensionless ignition time as i = exp 20 = 2.42583 106 . (20)(10) (1.359)
This is a surprisingly good estimate, given the complexity of the problem. Recall the numerical solution showed ignition for 2.7 106 . In terms of dimensional time, ignition time prediction becomes ti = = exp , aq 1 a RTo E
o hTo ,A
(1.360) cv To o hTo ,B exp E RTo , (1.361)
Note the ignition is suppressed if the ignition time is lengthened, which happens when
66
CHAPTER 1. INTRODUCTION TO KINETICS he activation energy E is increased, since the exponential sensitivity is stronger than the algebraic sensitivity, the energy of combustion (hTo ,A hTo ,B ) is decreased because it takes longer to react to drive the temperature to a critical value to induce ignition, the collision frequency factor a is decreased, which suppresses reaction.
o o
1.2.2
Here is an example which uses multiple reactions for an adiabatic isothermal system is given. Consider the full time-dependency of a problem similar to the thermal explosion problem just considered. We choose a non-intuitive set of parameters for the problem. Our choices will enable a direct comparison to a detonation of the same mixture via the same reaction mechanism in a later chapter. A closed, xed, adiabatic volume, V = 0.3061251 cm3 , contains at t = 0 s a stoichiometric hydrogen-air mixture of 2 105 mole of H2 , 1 105 mole of O2 , and 3.76 105 mole of N2 at Po = 2.83230 106 P a and To = 1542.7 K.17 Thus the initial molar concentrations are H2 = 6.533 105 mole/cm3 , O2 = 3.267 105 mole/cm3 , H2 = 1.228 104 mole/cm3 . The initial mass fractions are calculated via Yi = Mi i /. They are YH2 = 0.0285, YO2 = 0.226, YN2 = 0.745. To avoid issues associated with numerical roundo errors at very early time for species with very small compositions, the minor species were initialized at a small non-zero value near machine precision; each was assigned a value of 1015 mole. The minor species all have i = 1.803 1016 mole/cm3 . They have correspondingly small initial mass fractions. We seek the reaction dynamics as the system proceeds from its initial state to its nal state. We use the reversible detailed kinetics mechanism of Table 1.2. This problem requires a detailed numerical solution. Such a solution was performed by solving the appropriate equations for a mixture of nine interacting species: H2 , H, O, O2 , OH, H2 O, HO2 , H2 O2 , and N2 . The dynamics of the reaction process are reected in Figs. 1.15-1.17
This temperature and pressure correspond to that of the same ambient mixture of H2 , O2 and N2 which was shocked from 1.01325 105 P a, 298 K, to a value associated with a freely propagating detonation. Relevant comparisons of reaction dynamics will be made in a later chapter CC BY-NC-ND. 12 December 2011, J. M. Powers.
17
Detailed H2 O2 N2 kinetics
67
10
10 Y
i
H O H2 O2 OH H2 O H2 O 2 HO2 N2
10
10
10
15 10 5 0
10
10 t (s)
10
Figure 1.15: Plot of YH2 (t), YH (t), YO (t), YO2 (t), YOH (t), YH2 O (t), YHO2 (t), YH2 O2 (t), YN2 (t), for adiabatic, isochoric combustion of a mixture of 2H2 + O2 + 3.76N2 initially at To = 1542.7 K, Po = 2.8323 106 P a.
4000 3500 3000 2500 T (K) 2000 1500 1000 500 0

10 5
10
10 t (s)
Figure 1.16: Plot of T (t), for adiabatic, isochoric combustion of a mixture of 2H2 +O2 +3.76N2 initially at To = 1542.7 K, Po = 2.8323 106 P a.
68
10
P (Pa) 10
6
10
10
10 t (s)
Figure 1.17: Plot of P (t), for adiabatic, isochoric combustion of a mixture of 2H2 + O2 + 3.76N2 initially at To = 1542.7 K, Po = 2.8323 106 P a. At early time, t < 107 s, the pressure, temperature, and major reactant species concentrations (H2 , O2 , N2 ) are nearly constant. However, the minor species, e.g. OH, HO2 , and the major product, H2 O, are undergoing very rapid growth, albeit with math fractions whose value remains small. In this period, the material is in what is known as the induction period. After a certain critical mass of minor species has accumulated, exothermic recombination of these minor species to form the major product H2 O induces the temperature to rise, which accelerates further the reaction rates. This is manifested in a thermal explosion. A common denition of the end of the induction period is the induction time, t = tind , the time when dT /dt goes through a maximum. Here one nds tind = 6.6 107 s. (1.362)
A close-up view of the species concentration proles is given in Fig. 1.18 At the end of the induction zone, there is a nal relaxation to equilibrium. The equilibrium mass fractions of each species are YO2 YH YOH YO Y H2 Y H2 O = = = = = = 1.85 102 , 5.41 104 , 2.45 102 , 3.88 103 , 3.75 103 , 2.04 101 , (1.363) (1.364) (1.365) (1.366) (1.367) (1.368)
69
10
10
H O H2 O2 OH H2 O H2 O 2
10
10
HO2 N2 2 4 t (s) 6 8 x 10 10
7
Figure 1.18: Plot near thermal explosion time of YH2 (t), YH (t), YO (t), YO2 (t), YOH (t), YH2 O (t), YHO2 (t), YH2 O2 (t), YN2 (t), for adiabatic, isochoric combustion of a mixture of 2H2 + O2 + 3.76N2 initially at at To = 1542.7 K, Po = 2.8323 106 P a. YHO2 = 6.84 105 , YH2 O2 = 1.04 105 , YN2 = 7.45 101 . (1.369) (1.370) (1.371)
We note that because our model takes N2 to be inert that its value remains unchanged. Other than N2 , the nal products are dominated by H2 O. The equilibrium temperature is 3382.3 K and 5.53 106 P a.
70
Chapter 2 Gas mixtures

One is often faced with mixtures of simple compressible substances, and it the thermodynamics of such mixtures upon which attention is now xed. Here a discussion of some of the fundamentals of mixture theory will be given. In general, thermodynamics of mixtures can be a challenging topic about which much remains to be learned. In particular, these notes will focus on ideal mixtures of ideal gases, for which results are often consistent with intuition. The chemical engineering literature contains a full discussion of the many nuances associated with non-ideal mixtures of non-ideal materials. Relevant background for this chapter is found in standard undergraduate texts.12 34 Some of these notes on mixtures are adaptations of material found in these texts, especially Borgnakke and Sonntag.
2.1
Some general issues
Consider a mixture of N components, each a pure substance, so that the total mass and total number of moles are
N
m = m1 + m2 + m3 + . . . mN =
i=1 N
mi , ni ,
mass, units= kg, moles, units= kmole.
(2.1) (2.2)
n = n1 + n2 + n3 + . . . nN =
i=1 23
Recall 1 mole = 6.02 10 . The mass fraction of component i is dened as Yi : mi Yi , mass fraction, dimensionless. m
1
(2.3)
Borgnakke, C., and Sonntag, R. E., 2009, Fundamentals of Thermodynamics, Seventh Edition, John Wiley, New York. 2 Sandler, S. I., 1998, Chemical and Engineering Thermodynamics, Third Edition, John Wiley, New York. 3 Smith, J. M., Van Ness, H. C., and Abbott, M., 2004, Introduction to Chemical Engineering Thermodynamics, Seventh Edition, McGraw-Hill, New York. 4 Tester, J. W. and Modell, M., 1997, Thermodynamics and Its Applications, Third Edition, Prentice Hall, Upper Saddle River, New Jersey.
71
72 The mole fraction of component i is dened as yi : yi ni , n
CHAPTER 2. GAS MIXTURES
mole fraction, dimensionless.
(2.4)
Now the molecular mass of species i is the mass of a mole of species i. It units are typically g/mole. This is an identical unit to kg/kmole. Molecular mass is sometimes called molecular weight, but this is formally incorrect, as it is a mass measure, not a force measure. Mathematically the denition of Mi corresponds to Mi mi , ni g kg = kmole mole . (2.5)
Then one gets mass fraction in terms of mole fraction as Yi = mi , m ni Mi = , m ni Mi , = N j=1 mj ni Mi , = N j=1 nj Mj = = =
ni M i n , N 1 j=1 nj Mj n ni M i n , N nj M j j=1 n
(2.6) (2.7) (2.8) (2.9) (2.10) (2.11) (2.12)
yi Mi . N j=1 yj Mj
Similarly, one nds mole fraction in terms of mass fraction by the following: yi = = = = ni , n (2.13) (2.14) (2.15) (2.16)
mi Mi mj , N j=1 Mj mi Mi m mj , N j=1 Mj m Yi Mi . Yj N j=1 Mj
2.1. SOME GENERAL ISSUES The mixture itself has a mean molecular mass: m , M n , n N ni Mi = , n i=1
N
73
(2.17) mi (2.18) (2.19) (2.20)
N i=1
=
i=1
yi Mi .
Example 2.1
Air is often modelled as a mixture in the following molar ratios: O2 + 3.76N2 Find the mole fractions, the mass fractions, and the mean molecular mass of the mixture. Take O2 to be species 1 and N2 to be species 2. Consider the number of moles of O2 to be n1 = 1 kmole, and N2 to be n2 = 3.76 kmole.
kg The molecular mass of O2 is M1 = 32 kmole . The molecular mass of N2 is M2 = 28 number of moles is n = 1 kmole + 3.76 kmole = 4.76 kmole. kg kmole .
(2.21)
The total
So the mole fractions are
y1 =
1 kmole = 0.2101. 4.76 kmole 3.76 kmole = 0.7899. y2 = 4.76 kmole

N
Note that
yi = 1.
i=1
(2.22)
That is to say y1 + y2 = 0.2101 + 0.7899 = 1. Now for the masses, one has m1 = n1 M1 = (1 kmole) 32 m2 = n2 M2 = (3.76 kmole) 28 So one has m = m1 + m2 = 32 kg + 105.28 kg = 137.28 kg. The mass fractions then are Y1 = m1 32 kg = = 0.2331, m 137.28 kg CC BY-NC-ND. 12 December 2011, J. M. Powers. kg kmole kg kmole = 32 kg, = 105.28 kg,
74
Y2 = Note that

m2 105.28 kg = = 0.7669. m 137.28 kg
N
Yi = 1.
i=1
(2.23)
That is Y1 + Y2 = 0.2331 + 0.7669 = 1. Now for the mixture molecular mass, one has M= Check against another formula.
N
m 137.28 kg kg = = 28.84 . n 4.76 kmole kmole
M=
i=1
yi Mi = y1 M1 + y2 M2 = (0.2101)(32) + (0.7899)(28) = 28.84
kg . kmole
Now postulates for mixtures are not as well established as those for pure substances. The literature has much controversial discussion of the subject. A strong advocate of the axiomatic approach, C. A. Truesdell, 1919-2000, proposed the following metaphysical principles for mixtures, which are worth considering.5 1. All properties of the mixture must be mathematical consequences of properties of the constituents. 2. So as to describe the motion of a constituent, we may in imagination isolate it from the rest of the mixture, provided we allow properly for the actions of the other constituents upon it. 3. The motion of the mixture is governed by the same equations as is a single body. Most important for the present discussion is the rst principle. When coupled with uid mechanics, the second two take on additional importance. The approach of mixture theory is to divide and conquer. One typically treats each of the constituents as a single material and then devises appropriate average or mixture properties from those of the constituents. The best example of this is air, which is not a single material, but is often treated as such.
2.2
Ideal and non-ideal mixtures

E = E(T, P, n1 , n2 , . . . , nN ). (2.24)
A general extensive property, such as E, for an N-species mixture will be such that
C. A. Truesdell, 1984, Rational Thermodynamics, Springer-Verlag, New York.
2.3. IDEAL MIXTURES OF IDEAL GASES
75
A partial molar property is a generalization of an intensive property, and is dened such that it is the partial derivative of an extensive property with respect to number of moles, with T and P held constant. For internal energy, then the partial molar internal energy is ei E ni .
T,P,nj ,i=j
(2.25)
Pressure and temperature are held constant because those are convenient variables to control in an experiment. One also has the partial molar volume vi = V ni .
T,P,nj ,i=j
(2.26)
It shall be soon seen that there are other natural ways to think of the volume per mole. Now in general one would expect to nd ei = ei (T, P, n1 , n2 , . . . , nN ), vi = vi (T, P, n1 , n2 , . . . , nN ). (2.27) (2.28)
This is the case for what is known as a non-ideal mixture. An ideal mixture is dened as a mixture for which the partial molar properties ei and vi are not functions of the composition, that is ei = ei (T, P ), vi = vi (T, P ), if ideal mixture, if ideal mixture. (2.29) (2.30)
An ideal mixture also has the property that hi = hi (T, P ), while for a non-ideal mixture hi = hi (T, P, n1 , . . . , nN ). Though not obvious, it will turn out that some properties of an ideal mixture will depend on composition. For example, the entropy of a constituent of an ideal mixture will be such that si = si (T, P, n1, n2 , . . . , nN ). (2.31)
2.3
Ideal mixtures of ideal gases
The most straightforward mixture to consider is an ideal mixture of ideal gases. Even here, there are assumptions necessary that remain dicult to verify absolutely.
2.3.1
Dalton model
The most common model for a mixture of ideal gases is the Dalton model. Key assumptions dene this model Each constituent shares a common temperature.
76 Each constituent occupies the entire volume.
Each constituent possesses a partial pressure which sums to form the total pressure of the mixture. The above characterize a Dalton model for any gas, ideal or non-ideal. One also takes for convenience Each constituent behaves as an ideal gas. The mixture behaves as a single ideal gas. It is more convenient to deal on a molar basis for such a theory. For the Dalton model, additional useful quantities, the species mass concentration i , the mixture mass concentration , the species molar concentration i , and the mixture molar concentration , can be dened. As will be seen, these denitions for concentrations are useful; however, they are not in common usage. Following Borgnakke and Sonntag, the bar notation, , will be reserved for properties which are mole-based rather than mass-based. As mentioned earlier, the notion of a partial molal property is discussed extensively in the chemical engineering literature and has implications beyond those considered here. For the Dalton model, in which each component occupies the same volume, one has Vi = V. The mixture mass concentration, also called the density is simply = The mixture molar concentration is = For species i, the equivalents are i = i mi , V ni , = V kg , m3 kmole . m3 (2.35) (2.36) n , V kmole m3 . (2.34) m , V kg m3 . (2.33) (2.32)
One can nd a convenient relation between species molar concentration and species mole fraction by the following operations ni n , (2.37) i = V n ni n , (2.38) = nV = yi . (2.39)
77
A similar relation exists between species molar concentration and species mass fraction via i = ni m Mi , V m Mi
=mi
(2.40)
m ni Mi , V mMi
=Yi
(2.41)
mi 1 , m Mi Yi = . Mi = The specic volumes, mass and molar, are similar. One takes v = vi V , m V , = mi v= V , n V vi = . ni
(2.42) (2.43)
(2.44) (2.45)
Note that this denition of molar specic volume is not the partial molar volume dened in V the chemical engineering literature, which takes the form vi = ni .
T,P,nj ,i=j
For the partial pressure of species i, one can say for the Dalton model
N
P =
i=1
Pi .
(2.46)
For species i, one has Pi V Pi

N
Pi
i=1 =P
= ni RT, ni RT = , V N ni RT = , V i=1
N
(2.47) (2.48) (2.49)
RT P = V
ni .
i=1 =n
(2.50)
So, for the mixture, one has P V = nRT. (2.51)
78 One could also say
n RT = RT. (2.52) V Here n is the total number of moles in the system. Additionally R is the universal gas constant with value P = R = 8.314472 kJ J = 8.314472 . kmole K mole K (2.53)
Sometimes this is expressed in terms of kb the Boltzmann6 constant and NA , Avogadros7 number: R = kb NA , (2.54) (2.55) (2.56) 1 , mole J = 1.3806505 1023 . K
NA = 6.02214199 1023 kb
Example 2.2
Compare the molar specic volume dened here with the partial molar volume from the chemical engineering literature. The partial molar volume vi , is given by vi = For the ideal gas, one has
N
V ni
.
T,P,nj ,i=j
(2.57)
PV V V ni
= RT = =
nk , nk , ,
(2.58) (2.59) (2.60) (2.61)

=1 =0
k=1 RT N k=1
P RT RT P
N k=1 ki
N nk k=1 ni
T,P,nj ,i=j
= vi
6 7
RT 1i + 2i + . . . + ii + . . . + N i P RT , P
P =0
,
=0
, (2.62) (2.63)
Ludwig Boltzmann, 1844-1906, Austrian physicist. Lorenzo Romano Amedeo Carlo Bernadette Avogadro di Quaregna e Cerreto, 1776-1856, Italian scien-
tist. CC BY-NC-ND. 12 December 2011, J. M. Powers.

= = V
N k=1
79
(2.64) (2.65)
nk
V . n
Here the so-called Kronecker8 delta function has been employed, which is much the same as the identity matrix: ki ki = = 0, 1, k = i, k = i. (2.66) (2.67)
Contrast this with the earlier adopted denition of molar specic volume vi = V . ni (2.68)
So, why is there a dierence? The molar specic volume is a simple denition. One takes the instantaneous volume V , which is shared by all species in the Dalton model, and scales it by the instantaneous number of moles of species i, and acquires a natural denition of molar specic volume consistent with the notion of a mass specic volume. On the other hand, the partial molar volume species how the volume changes if the number of moles of species i changes, while holding T and P and all other species mole numbers constant. One can imagine adding a mole of species i, which would necessitate a change in V in order to guarantee the P remain xed.
2.3.1.1
Binary mixtures
Consider now a binary mixture of two components A and B. This is easily extended to a general mixture of N components. First the total number of moles is the sum of the parts: n = nA + nB . Now, write the ideal gas law for each component: PA VA = nA RTA , PB VB = nB RTB . But by the assumptions of the Dalton model, VA = VB = V , and TA = TB = T , so PA V PB V One also has P V = nRT.
8
(2.69)
(2.70) (2.71)
= nA RT, = nB RT.
(2.72) (2.73) (2.74)
Leopold Kronecker, 1823-1891, German mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
80 Solving for n, nA and nB , one nds n = nA nB Now n = nA + nB , so one has PA V PB V PV = + . RT RT RT P = PA + PB . PV , RT PA V = , RT PB V = . RT
(2.75) (2.76) (2.77)
(2.78) (2.79)
That is the total pressure is the sum of the partial pressures. This is a mixture rule for pressure One can also scale each constituent ideal gas law by the mixture ideal gas law to get PA V nA RT , = PV nRT nA PA = , P n = yA , PA = yA P. Likewise PB = yB P. (2.84) Now, one also desires rational mixture rules for energy, enthalpy, and entropy. Invoke Truesdells principles on a mass basis for internal energy. Then the total internal energy E (with units J) for the binary mixture must be E = me = mA eA + mB eB , mA mB = m eA + eB , m m = m (YA eA + YB eB ) , e = YA eA + YB eB . For the enthalpy, one has H = mh = mA hA + mB hB , mA mB = m hA + hB , m m = m (YA hA + YB hB ) , h = YA hA + YB hB .
(2.80) (2.81) (2.82) (2.83)
(2.85) (2.86) (2.87) (2.88)
(2.89) (2.90) (2.91) (2.92)
81
It is easy to extend this to a mole fraction basis rather than a mass fraction basis. One can also obtain a gas constant for the mixture on a mass basis. For the mixture, one has PV PV mR = nR, T = (nA + nB )R, mA mB = R, + MA MB R R = mA + mB , MA MB = (mA RA + mB RB ) , mB mA RA + RB , R = m m R = (YA RA + YB RB ) . For the entropy, one has S = ms = = mA sA + mB sB , mB mA sA + sB , = m m m = m (YA sA + YB sB ) , s = Y A sA + Y B sB . (2.101) (2.102) (2.103) (2.104) = nRT mRT, (2.93) (2.94) (2.95) (2.96) (2.97) (2.98) (2.99) (2.100)
Note that sA is evaluated at T and PA , while sB is evaluated at T and PB . For a CPIG, one has sA = so o ,A + cP A ln T
so T,A
T To T To
RA ln
PA Po yA P Po
(2.105)
= so o ,A + cP A ln T Likewise sB = so o ,B + cP B ln T
so T,B
RA ln
(2.106)
T To
RB ln
yB P Po
(2.107)
Here the o denotes some reference state. As a superscript, it typically means that the property is evaluated at a reference pressure. For example, so denotes the portion of the T,A entropy of component A that is evaluated at the reference pressure Po and is allowed to vary with temperature T . Note also that sA = sA (T, P, yA) and sB = sB (T, P, yB ), so the entropy
82
of a single constituent depends on the composition of the mixture and not just on T and P . This contrasts with energy and enthalpy for which eA = eA (T ), eB = eB (T ), hA = hA (T ), hB = hB (T ) if the mixture is composed of ideal gases. Occasionally, one nds ho and ho A B used as a notation. This denotes that the enthalpy is evaluated at the reference pressure. However, if the gas is ideal, the enthalpy is not a function of pressure and hA = ho , hB = ho . A B If one is employing a calorically imperfect ideal gas model, then one nds for species i that yi P , i = A, B. (2.108) si = so Ri ln T,i Po 2.3.1.2 Entropy of mixing
Example 2.3
Initially calorically perfect ideal gases A and B are segregated within the same large volume by a thin frictionless, thermally conducting diaphragm. Thus, both are at the same initial pressure and temperature, P1 and T1 . The total volume is thermally insulated and xed, so there are no global heat or work exchanges with the environment. The diaphragm is removed, and A and B are allowed to mix. Assume A has mass mA and B has mass mB . The gases are allowed to have distinct molecular masses, MA and MB . Find the nal temperature T2 , pressure P2 , and the change in entropy. The ideal gas law holds that at the initial state VA1 = At the nal state one has V2 = VA2 = VB2 = VA1 + VB1 = (mA RA + mB RB ) Mass conservation gives m2 = m1 = mA + mB . One also has the rst law E2 E1 = = = = = Q W, 0, E1 , (2.112) (2.113) (2.114) (2.115) (2.116) (2.117) (2.118) (2.119) (2.120) (2.111) T1 . P1 (2.110) mA RA T1 , P1 VB1 = mB RB T1 . P1 (2.109)
E2 E1 E2
m2 e 2 (mA + mB )e2
mA eA1 + mB eB1 , mA eA1 + mB eB1 , mA (eA1 e2 ) + mB (eB1 e2 ), mA cvA (T1 T2 ) + mB cvB (T1 T2 ), mA cvA T1 + mB cvB T1 , mA cvA + mB cvB T1 .
0 = 0 = T2 = =
The nal pressure by Daltons law then is P2 = PA2 + PB2 , (2.121)

= = = = = mA RA T2 mB RB T2 + , V2 V2 mA RA T1 mB RB T1 + , V2 V2 (mA RA + mB RB ) T1 , V2 (mA RA + mB RB ) T1 , T (mA RA + mB RB ) P1 1 P1 .
83
(2.122) (2.123) (2.124) (2.125) (2.126)
So the initial and nal temperatures and pressures are identical. Now the entropy change of gas A is sA2 sA1 = = = cP A ln cP A ln cP A ln TA2 PA2 RA ln , TA1 PA1 yA2 P2 T2 RA ln , T1 yA1 P1 T1 yA2 P1 , RA ln T1 yA1 P1
=0
(2.127) (2.128) (2.129)
= = Likewise
RA ln
yA2 P1 (1)P1 RA ln yA2 .
(2.130) (2.131)
sB2 sB1 So the change in entropy of the mixture is S = = =
= RB ln yB2 .
(2.132)
mA RA ln yA2 mB RB ln yB2 , (nA MA )

=mA =RA
mA (sA2 sA1 ) + mB (sB2 sB1 ) R MA ln yA2 (nB MB )

=mB =RB
(2.133) (2.134) R MB ln yB2 , (2.135)
R(nA ln yA2 + nB ln yB2 ), R nA ln 0. nA nA + nB

0
+nB ln
nB nA + nB
0
(2.136)
(2.137)
(2.138)
For an N-component mixture, mixed in the same fashion such that P and T are constant, this extends to
N
S = R
nk ln yk ,
k=1
(2.139)
84
N
CHAPTER 2. GAS MIXTURES nk N i=1

0 N
= R
nk ln
k=1
ni
0,
(2.140)
= R
k=1 N
nk n ln yk , n nk ln yk , n yk ln yk ,
(2.141) (2.142) (2.143) (2.144) (2.145) (2.146) (2.147) (2.148)
= Rn = R
m M
k=1 N
k=1 N y ln ykk , k=1
= Rm
y y y = Rm (ln y1 1 + ln y2 2 + . . . + ln yNN ) , y y y = Rm ln (y1 1 y2 2 . . . yNN ) , N
= Rm ln s s = ln = R R
N
y yk k k=1 y yk k
k=1
Note that there is a fundamental dependency of the mixing entropy on the mole fractions. Since 0 yk 1, the product is guaranteed to be between 0 and 1. The natural logarithm of such a number is negative, and thus the entropy change for the mixture is guaranteed positive semi-denite. Note also that for the entropy of mixing, Truesdells third principle is not enforced. Now if one mole of pure N2 is mixed with one mole of pure O2 , one certainly expects the resulting homogeneous mixture to have a higher entropy than the two pure components. But what if one mole of pure N2 is mixed with another mole of pure N2 . Then we would expect no increase in entropy. However, if we had the unusual ability to distinguish N2 molecules whose origin was from each respective original chamber, then indeed there would be an entropy of mixing. Increases in entropy thus do correspond to increases in disorder. 2.3.1.3 Mixtures of constant mass fraction
If the mass fractions, and thus the mole fractions, remain constant during a process, the equations simplify. This is often the case for common non-reacting mixtures. Air at moderate values of temperature and pressure behaves this way. In this case, all of Truesdells principles can be enforced. For a CPIG, one would have e2 e1 = YA cvA (T2 T1 ) + YB cvB (T2 T1 ),
(2.149)
2.3. IDEAL MIXTURES OF IDEAL GASES = cv (T2 T1 ). where cv YA cvA + YB cvB . Similarly for enthalpy h2 h1 = YA cP A (T2 T1 ) + YB cP B (T2 T1 ), = cP (T2 T1 ). where cP Y A cP A + Y B cP B . For the entropy
85 (2.150)
(2.151)
(2.152) (2.153)
(2.154)
s2 s1 = YA (sA2 sA1 ) + YB (sB2 sB1 ), (2.155) T2 y A P2 T2 y B P2 = YA cP A ln RA ln + YB cP B ln RB ln , T1 y A P1 T1 y B P1 P2 T2 P2 T2 RA ln + YB cP B ln RB ln (2.156) , = YA cP A ln T1 P1 T1 P1 T2 P2 = cP ln R ln . (2.157) T1 P1 The mixture behaves as a pure substance when the appropriate mixture properties are dened. One can also take cP (2.158) = . cv Note that some intuitive denitions do not hold: with A = cP A /cvA , B = cP B /cvB , = YA A + YB B .
2.3.2
Summary of properties for the Dalton mixture model
Listed here is a summary of mixture properties for an N-component mixture of ideal gases on a mass basis:
N
M =
i=1 N
yiMi , i ,
i=1
(2.159) (2.160)
= v =
1
N 1 i=1 vi
1 = ,
(2.161)
86
N
e =
i=1 N
Yiei , Yihi ,
i=1 =R N N N
(2.162) (2.163)
h =
R R = = M
YiRi =
i=1 i=1
yi Mi Ri
N
R = M
=M
yi ,
i=1 =1
(2.164)
yj Mj
j=1
cv =
i=1
Yicvi , if ideal gas
(2.165) (2.166) (2.167) (2.168) (2.169) (2.170) (2.171) (2.172) (2.173) (2.174) (2.175) (2.176) (2.177)
cv = cP R,
N
cP =
i=1
Y i cP i , cP = cv
N N i=1 Yi cP i , N Yi cvi i=1
= s =
Y i si ,
i=1
Yi = Pi = i = vi = V = T = hi = hi =
yi Mi , M yi P, Yi , v 1 = , Yi i Vi , Ti , ho , if ideal gas, i Pi = ei + Pi vi = ei + Ri T , ei + i
if ideal gas T
hi = ho o ,i + T
c P i (T ) d T ,
(2.178)
To
if ideal gas

T
87 Pi , Po
si = so o ,i + T
To
c P i (T ) dT Ri ln T
if ideal gas
(2.179)
=so T,i
si = so Ri ln T,i
yi P Po
= so Ri ln T,i
if ideal gas
Pi , Po
(2.180)
if ideal gas
Pi = i Ri T = Ri T Yi =
if ideal gas N
Ri T , vi
(2.181)
P = RT = RT
i=1
RT Yi = , Mi v
(2.182)
if ideal gas N T
h =
i=1
Yiho o ,i + T
c P (T ) d T ,
(2.183)
To
if ideal gas
h = e+
N
P = e + P v = e + RT ,
if ideal gas T
(2.184) P Po
N y yi i i=1
s =
i=1
Yiso o ,i T
+
To
c P (T ) dT R ln T
if ideal gas
R ln
(2.185) These relations are not obvious. A few are derived in examples here.
Example 2.4
Derive the expression h = e + P/. Start from the equation for the constituent hi , multiply by mass fractions, sum over all species, and use properties of mixtures: hi Yi hi
N
= = =
ei +
Pi , i Pi , i
N
(2.186) (2.187) Yi
i=1
Yi ei + Yi
N
Yi hi
i=1
Yi ei +
i=1
Pi , i
(2.188)
88
N N N

i Ri T , i
Yi hi
i=1 =h
=
i=1
Yi ei +
i=1 =e N
Yi
(2.189)
e+T
i=1
Yi Ri ,
=R
(2.190)
= =
e + RT, P e+ .
(2.191) (2.192)
Example 2.5
Find the expression for mixture entropy of the ideal gas. si Yi si
N
= = =
so o ,i + T
T To
cP i (T ) dT Ri ln T
T To N
Pi Po
, Pi Po
N
(2.193) , Pi Po Pi Po , (2.194) (2.195)
Yi so o ,i + Yi T
N
cP i (T ) dT Yi Ri ln T
T
s=
i=1
Yi si
Yi so o ,i + T
i=1 N i=1
Yi
To T N To i=1
cP i (T ) dT T
Yi Ri ln
i=1 N
=
i=1
Yi so o ,i + T so o + T
T To
Yi cP i (T ) dT T
N
Yi Ri ln
i=1
(2.196)
cP (T ) dT T
Yi Ri ln
i=1
Pi Po
(2.197)
All except the last term are natural extensions of the property for a single material. Consider now the last term involving pressure ratios.
N
Yi Ri ln
i=1
Pi Po
R R
Yi Ri ln
i=1 N
Pi Po Pi Po
+ R ln
P P R ln Po Po P P ln Po Po Pi Po + ln ,
(2.198)
Yi
i=1 N
Ri ln R
N j=1
+ ln
(2.199)
Yi
i=1
R/Mi Yj R/Mj
ln
P P ln Po Po
(2.200)
N R i=1
Yi /Mi
N j=1
Yj /Mj
ln
Pi Po
+ ln
=yi
P P ln , Po Po
(2.201)

N
89
Pi Po P P ln Po Po P P + ln Po Po
R R
yi ln
i=1 N
+ ln ln
yi
(2.202)
ln
i=1 N
Pi Po Pi Po
N i=1
yi
(2.203)
R ln R ln R ln R ln R ln R ln
+ ln Pi Po
yi
i=1
Po P + ln P Po P Po (Pi )
i=1
(2.204)
Po P
+ ln 1
N
,
yi
(2.205) P Po
Po P
PN
i=1
yi
Po
yi
PN
i=1
yi
+ ln
(2.206)
i=1 N
Pi P yi P P
y yi i
+ ln
yi
P Po P Po
(2.207)
+ ln P Po
(2.208)
i=1 N
= So the mixture entropy becomes s = so o + T so o + T
+ ln
(2.209)
i=1
T To T
cP (T ) dT R ln T
N y yi i i=1 N
+ ln
P Po .
(2.210)
To
cP (T ) P R ln dT R ln Po T
y yi i i=1
(2.211)
classical entropy of a single body
nonTruesdellian
The extra entropy is not found in the theory for a single material, and in fact is not in the form suggested by Truesdells postulates. While it is in fact possible to redene the constituent entropy denition in such a fashion that the mixture entropy in fact takes on the classical form of a single material via the T denition si = so o ,i + To cP i (T )/T dT Ri ln (Pi /Po ) + Ri ln yi , this has the disadvantage of predicting T no entropy change after mixing two pure substances. Such a theory would suggest that this obviously irreversible process is in fact reversible.
On a molar basis, one has the equivalents

N
=
i=1
i = 1
N 1 i=1 v i
n = , V M = 1 V = = vM, n
(2.212) (2.213)
v =
90
N
e =
i=1 N
yiei = eM, yihi = hM,

i=1 N
(2.214) (2.215) (2.216) (2.217) (2.218) (2.219) (2.220) (2.221) (2.222) (2.223) (2.224)
h = cv =
i=1
yicvi = cv M,
cv = cP R,
N
cP =
i=1
y i cP i = cP M
N i=1 yi cP i , N i=1 yi cvi
cP = = cv
N
s =
i=1
yisi = sM,
i = yi = vi vi
i , Mi v 1 V = = = vi Mi , = ni yi i V V = = v = vM , = ni P,T,nj n
if ideal gas
Pi = yi P, P = RT = RT , v RT , vi
(2.225)
if ideal gas
Pi = i RT =
(2.226)
if ideal gas
h = e+ hi =
o hi ,
P = e + P v = e + RT = hM,
if ideal gas
(2.227) (2.228) (2.229)
if ideal gas, Pi = ei + Pi v i = ei + P vi = ei + RT = hi Mi , i
if ideal gas
hi = ei +

o hTo ,i T
91
hi =
+
To
cP i (T ) dT = hi Mi , c P i (T ) dT R ln T
if ideal gas
(2.230)
if ideal gas T
si = so o ,i + T
To
yi P , Po
(2.231)
=so T,i
si = so R ln T,i
N
yi P Po
T
= si Mi ,
(2.232)
if ideal gas
s =
i=1
yiso o ,i + T
To
c P (T ) dT R ln T
if ideal gas
P Po
R ln
y yi i i=1
= sM.
(2.233)
2.3.3
Amagat model*
The Amagat9 model is an entirely dierent paradigm than the Dalton model. It is not used as often. In the Amagat model, all components share a common temperature T , all components share a common pressure P , each component has a dierent volume. Consider, for example, a binary mixture of calorically perfect ideal gases, A and B. For the mixture, one has (2.234) P V = nRT, with n = nA + nB . For the components one has P VA = nA RT, P VB = nB RT. Then n = nA + nB reduces to P VA P VB PV = + . RT RT RT
9
(2.235)
(2.236) (2.237)
(2.238)
Emile Hilaire Amagat, 1841-1925, French physicist. CC BY-NC-ND. 12 December 2011, J. M. Powers.
92 Thus V = VA + VB , VA VB 1 = + . V V
(2.239) (2.240)
Chapter 3 Mathematical foundations of thermodynamics*

This chapter focuses on mathematical formalism which can be applied to thermodynamics. Understanding of calculus of many variables at an undergraduate level is sucient mathematical background for this chapter. Some details can be found in standard sources.1 2
3
3.1
Exact dierentials and state functions
This is known to be an exact dierential with the consequence that internal energy e is a function of the state of the system and not the details of any process which led to the state. As a counter-example, the work, w = P dv, (3.2) can be shown to be an inexact dierential so that the work is indeed a function of the process involved. Example 3.1
In thermodynamics, one is faced with many systems of the form of the well-known Gibbs equation: de = T ds P dv. (3.1)
Show the work is not a state function. If work were a state function, one might expect it to have the form w = w(P, v), provisional assumption, to be tested. (3.3) Abbott, M. M., and Van Ness, H. C., 1972, Thermodynamics, Schaums Outline Series in Engineering, McGraw-Hill, New York. See Chapter 3. 2 Borgnakke, C., and Sonntag, R. E, 2009, Fundamentals of Thermodynamics, Seventh Edition, John Wiley, New York. See Chapters 13 and 15. 3 Vincenti, W. G., and Kruger, C. H., 1965, Introduction to Physical Gas Dynamics, John Wiley, New York. See Chapter 3.
1
93
94
CHAPTER 3. MATHEMATICAL FOUNDATIONS OF THERMODYNAMICS*

In such a case, one would have the corresponding dierential form w = Now since w = P dv, one deduces that w v w P Integrating Eq. (3.5), one nds w = P v + f (P ), (3.7) where f (P ) is some function of P to be determined. Dierentiating Eq. (3.7) with respect to P , one gets w df (P ) =v+ . (3.8) P v dP Now use Eq. (3.6) to eliminate
w P v
w v
dv +
P
w P
dP.
v
(3.4)
=
P
P, 0.
(3.5) (3.6)
=
v
in Eq. (3.8) so as to obtain 0 df (P ) dP = = v+ v. df (P ) , dP (3.9) (3.10)
Equation (3.10) cannot be: a function of P only cannot be a function of v. So, w cannot be a state property: w = w(P, v). (3.11)
Consider now the more general form

N
1 dx1 + 2 dx2 + . . . + N dxN =

i=1
i dxi .
(3.12)
Here i and xi , i = 1, . . . , N, may be thermodynamic variables. This form is known in mathematics as a Pfa4 dierential form. As formulated, one takes at this stage xi : independent thermodynamic variables, i : thermodynamic variables which are functions of xi
4
Johann Friedrich Pfa, 1765-1825, German mathematician.
3.1. EXACT DIFFERENTIALS AND STATE FUNCTIONS
95
Now, if the dierential in Eq. (3.12), when set to a dierential dy, can be integrated to form the function y = y(x1, x2 , . . . , xN ), (3.13) the dierential is said to be exact. In such a case, one has
N
dy = 1 dx1 + 2 dx2 + . . . + N dxN =

i=1
i dxi .
(3.14)
Now, if the algebraic denition of Eq. (3.13) holds, what amounts to the denition of the partial derivative gives the parallel result that dy = y x1 dx1 +
xj ,j=1
y x2
dx2 + . . . +
xj ,j=2
y xN
dxN .
xj ,j=N
(3.15)
Now, combining Eqs. (3.14) and (3.15) to eliminate dy, one gets 1 dx1 + 2 dx2 + . . . + N dxN = y x1 dx1 +
xj ,j=1
y x2
dx2 + . . . +
xj ,j=2
y xN
dxN .
xj ,j=N
(3.16) Rearranging, one gets 0= y x1 1 dx1 + y x2 2 dx2 + . . . + y xN N dxN .
xj ,j=1
xj ,j=2
xj ,j=N
(3.17) Since the variables xi , i = 1, . . . , N, are independent, dxi , i = 1, . . . , N, are all independent in Eq. (3.17), and in general non-zero. For equality, one must require that each of the coecients be zero, so 1 = y x1 ,
xj ,j=1
2 =
y x2
,...,
xj ,j=2
N =
y xN
.
xj ,j=N
(3.18)
So when dy is exact, one says that each of the i and xi are conjugate to each other. From here on out, for notational ease, the j = 1, j = 2, . . . , j = N will be ignored in the notation for the partial derivatives. It becomes especially confusing for higher order derivatives, and is fairly obvious for all derivatives. If y and all its derivatives are continuous and dierentiable, then one has for all i = 1, . . . , N and k = 1, . . . , N that 2y 2y = . (3.19) xk xi xi xk Now from Eq. (3.18), one has k = y xk ,
xj
l =
y xl
.
xj
(3.20)
96
Taking the partial of the rst of Eq. (3.20) with respect to xl and the second with respect to xk , one gets k 2y 2y l = = , . (3.21) xl xj xl xk xk xj xk xl Since by Eq. (3.19) the order of the mixed second partials does not matter, one deduces from Eq. (3.21) that l k = (3.22) xl xj xk xj This is a necessary and sucient condition for the exact-ness of Eq. (3.12). It is a generalization of what can be found in most introductory calculus texts for functions of two variables. For the Gibbs equation, (3.1), de = P dv + T ds, one has y u, x1 v, x2 s, 1 P 2 T. (3.23)
and one expects the natural, or canonical form of e = e(v, s). (3.24)
Here, P is conjugate to v, and T is conjugate to s. Application of the general form of Eq. (3.22) to the Gibbs equation (3.1) gives then T v = P s .
v
(3.25)
Equation (3.25) is known as a Maxwell5 relation. Moreover, specialization of Eq. (3.20) to the Gibbs equation (3.1) gives P = If the general dierential dy = The path integral yB yA = e v ,
s
T =
e s
.
v
(3.26)
N i=1 B A
i dxi is exact, one also can show

N i=1
i dxi is independent of the path of the integral.
The integral around a closed contour is zero:

N
dy =
i=1
5
i dxi = 0.
(3.27)
James Clerk Maxwell, 1831-1879, Scottish physicist and mathematican.
3.1. EXACT DIFFERENTIALS AND STATE FUNCTIONS
97
The function y can only be determined to within an additive constant. That is, there is no absolute value of y; physical signicance is only ascribed to dierences in y. In fact now, other means, extraneous to this analysis, can be used to provide absolute values of key thermodynamic variables. This will be important especially for ows with reaction.
Example 3.2
Show the heat transfer q is not a state function. Assume all processes are fully reversible. The rst law gives de q = q w, = de + w, = de + P dv. Take now the non-canonical, although acceptable, form e = e(T, v). Then one gets de So q = = e e dv + dT + P dv, v T T v e e + P dv + dT. v T T v
M N
(3.28) (3.29) (3.30)
e v
dv +
T
e T
dT.
v
(3.31)
(3.32) (3.33)
M dv + N dT.
(3.34)
Now by Eq. (3.22), for q to be exact, one must have M T =

v
N v
,
T
(3.35) (3.36)
This reduces to
2e P + T v T
P T v
=
v
2e . vT
P T v
(3.37) = R/v.
This can only be true if So q is not exact.
= 0. But this is not the case; consider an ideal gas for which
Example 3.3
Show conditions for ds to be exact in the Gibbs equation. de = T ds P dv, (3.38) CC BY-NC-ND. 12 December 2011, J. M. Powers.
98

ds = = = de P + dv, T T 1 e e P dv + dT + dv, T v T T v T P 1 e 1 e + dv + dT. T v T T T T v
M N
(3.39) (3.40) (3.41)
Again, invoking Eq. (3.22), one gets then T 1 1 2u 2 T T v T 1 2 T e v e v

T
1 e T v 1 T 1 + T +
+
T
P T
=
v
1 e T T
,
v T
(3.42) (3.43) (3.44)
P T P T
P T2 v P 2 T v
= =
1 2u , T vT 0.
This is the condition for an exact ds. Experiment can show if it is true. For example, for an ideal gas, one nds from experiment that e = e(T ) and P v = RT , so one gets 0+ 1R 1 RT 2 T v T v 0 = = 0, 0. (3.45) (3.46)
So ds is exact for an ideal gas. In fact, the relation is veried for so many gases, ideal and non-ideal, that one simply asserts that ds is exact, rendering s to be path-independent and a state variable.
3.2
Two independent variables

f (x, y, z) = 0. (3.47)
Consider a general implicit function linking three variables, x, y, z:
In x y z space, this will represent a surface. If the function can be inverted, it will be possible to write the explicit forms x = x(y, z), y = y(x, z), z = z(x, y). (3.48)
Dierentiating the rst two of the Eqs. (3.48) gives dx = dy = x y y x dy +

z
x z y z
dz,
y
(3.49) (3.50)
dx +
z
dz
x
3.2. TWO INDEPENDENT VARIABLES Now use Eq. (3.50) to eliminate dy in Eq. (3.49): dx = x y y x y z y x
=0
99
dx +
z =dy
y z
dz +
x
x z
dz,
y
(3.51)
x y
y x
dx =
z
x y x y
+
x
x z
dz,
y
(3.52) x y y z +
x
0dx + 0dz =
1 dx +
x z
dz. (3.53)
y
=0
Since x and y are independent, so are dx and dy, and the coecients on each in Eq. (3.53) must be zero. Therefore from the coecient on dx in Eq. (3.53) x y y 1 = 0, z x z x y = 1, y z x z x 1 = y , y z x
z
(3.54) (3.55) (3.56)
and also from the coecient on dz in Eq. (3.53) x y y z +

x
x z x z
= 0,
y
(3.57) x y y z ,
x
(3.58) (3.59)
x z
y x
z y
= 1. dy x + dw z dz . dw
If one now divides Eq. (3.49) by a fourth dierential, dw, one gets dx = dw x y (3.60)
Demanding that z be held constant in Eq. (3.60) gives x w =

z
x y x y x y
y w ,
,
z
(3.61) (3.62) (3.63)
x w z y w z
= =
x w
w y
.
z
100
If x = x(y, w), one then gets dx = x y dy +

w
x w
dw.
y
(3.64)
Divide now by dy while holding z constant so x y =

z
x y
+
w
x w
w y
.
z
(3.65)
These general operations can be applied to a wide variety of thermodynamic operations. Example 3.4
Apply Eq. (3.65) to a standard P v T system and let x y T v Now by denition cv = so T e
e v s
=
z
T v
.
s
(3.66)
So T = x, v = y, and s = z. Let now e = w. So Eq. (3.65) becomes =

s
T v
+
e
T e ,
e v
.
s
(3.67)
e T =
v
(3.68) (3.69)
1 . cv P . cv
T v e
Now by Eq. (3.26), one has
= P , so one gets T v =
s
T v
(3.70) = 0, thus (3.71)
For an ideal gas, e = e(T ). Inverting, one gets T = T (e), and so T v dT dv dT T = P . cv
For an isentropic process in an ideal gas, one gets = = = ln T To T To = = RT P = , cv cv v R dv , cv v dv ( 1) , v vo ( 1) ln , v vo 1 . v (3.72) (3.73) (3.74) (3.75) (3.76)
3.3. LEGENDRE TRANSFORMATIONS
101
3.3
Legendre transformations
The Gibbs equation (3.1), de = P dv + T ds, is the fundamental equation of classical thermodynamics. It is a canonical form which suggests the most natural set of variables in which to express internal energy e are s and v: e = e(v, s). (3.77)
However, v and s may not be convenient for a particular problem. There may be other combinations of variables whose canonical form gives a more convenient set of independent variables for a particular problem. An example is the enthalpy: h e + P v. Dierentiating the enthalpy gives dh = de + P dv + vdP. Use now Eq. (3.79) to eliminate de in the Gibbs equation to give dh P dv vdP = P dv + T ds,
=de
(3.78)
(3.79)
(3.80) (3.81)
dh = T ds + vdP. So the canonical variables for h are s and P . One then expects h = h(s, P ).
(3.82)
This exercise can be systematized with the Legendre6 transformation,7 which denes a set of second order polynomial combinations of variables. Consider again the exact dierential Eq. (3.14): dy = 1 dx1 + 2 dx2 + . . . + N dxN . (3.83) For N independent variables xi and N conjugate variables i , by denition there are 2N 1 Legendre transformed variables: 1 = 1 (1 , x2 , x3 , . . . , xN ) = y 1 x1 , 2 = 2 (x1 , 2 , x3 , . . . , xN ) = y 2 x2 , . . . N = N (x1 , x2 , x3 , . . . , N ) = y N xN ,
6 7
(3.84) (3.85) (3.86)
Adrien-Marie Legendre, 1752-1833, French mathematician. Two dierentiable functions f and g are said to be Legendre transformations of each other if their rst derivatives are inverse functions of each other: Df = (Dg)1 . With some eort, not shown here, one can prove that the Legendre transformations of this section satisfy this general condition. CC BY-NC-ND. 12 December 2011, J. M. Powers.
102
CHAPTER 3. MATHEMATICAL FOUNDATIONS OF THERMODYNAMICS* 1,2 = 1,2 (1 , 2 , x3 , . . . , xN ) = y 1 x1 2 x2 , 1,3 = 1,3 (1 , x2 , 3 , . . . , xN ) = y 1 x1 3 x3 , . . .

N
(3.87) (3.88) (3.89) (3.90)
1,...,N = 1,...,N (1 , 2 , 3 , . . . , N ) = y
i xi .
i=1
Each is a new dependent variable. Each has the property that when it is known as a function of its N canonical variables, the remaining N variables from the original expression (the xi and the conjugate i ) can be recovered by dierentiation of . In general this is not true for arbitrary transformations. Example 3.5
Let y = y(x1 , x2 , x3 ). This has the associated dierential form dy = 1 dx1 + 2 dx2 + 3 dx3 . Choose now a Legendre transformed variable 1 z(1 , x2 , x3 ): z = y 1 x1 . Then dz = z 1 d1 +
x1 ,x2
(3.91)
(3.92) z x3
z x2
dx2 +
1 ,x3
dx3 .
1 ,x2
(3.93)
Now dierentiating Eq. (3.92), one also gets dz = dy 1 dx1 x1 d1 . Elimination of dy in Eq. (3.94) by using Eq. (3.91) gives dz = = 1 dx1 + 2 dx2 + 3 dx3 1 dx1 x1 d1 ,
=dy
(3.94)
(3.95) (3.96)
x1 d1 + 2 dx2 + 3 dx3 .
Thus from Eq. (3.93), one gets x1 = z 1 ,

x2 ,x3
2 =
z x2
,
1 ,x3
3 =
z x3
.
1 ,x2
(3.97)
So the original expression had three independent variables x1 , x2 , x3 , and three conjugate variables 1 , 2 , 3 . Denition of the Legendre function z with canonical variables 1 , x2 , and x3 allowed determination of the remaining variables x1 , 2 , and 3 in terms of the canonical variables.
For the Gibbs equation, (3.1), de = P dv + T ds, one has y = u, two canonical variables, x1 = v and x2 = s, and two conjugates, 1 = P and 2 = T . Thus N = 2, and one can
3.3. LEGENDRE TRANSFORMATIONS expect 22 1 = 3 Legendre transformations. They are
103
1,2
1 = y 1 x1 = h = h(P, s) = e + P v, enthalpy, (3.98) 2 = y 2 x2 = a = a(v, T ) = e T s, Helmholtz free energy, (3.99) = y 1 x1 2 x2 = g = g(P, T ) = e + P v T s, Gibbs free energy. (3.100)
It has already been shown for the enthalpy that dh = T ds + vdP , so that the canonical variables are s and P . One then also has dh = from which one deduces that T = h s ,
P
h s
ds +
P
h P
dP,
s
(3.101)
v=
h P
.
s
(3.102)
From Eq. (3.102), a second Maxwell relation can be deduced by dierentiation of the rst with respect to P and the second with respect to s: T P =
s
v s
.
P
(3.103)
The relations for Helmholtz8 and Gibbs free energies each supply additional useful relations including two new Maxwell relations. First consider the Helmholtz free energy a = e T s, da = de T ds sdT, = (P dv + T ds) T ds sdT, = P dv sdT. a v a v P T
8
(3.104) (3.105) (3.106) (3.107)
So the canonical variables for a are v and T . The conjugate variables are P and s. Thus da = So one gets P = and the consequent Maxwell relation =
v
dv +
T
a T
dT.
v
(3.108)
,
T
s = s v
a T
.
v
(3.109)
.
T
(3.110)
Hermann von Helmholtz, 1821-1894, German physician and physicist. CC BY-NC-ND. 12 December 2011, J. M. Powers.
104
For the Gibbs free energy g = e + P v T s,

=h
(3.111) (3.112) (3.113) (3.114) (3.115)
= h T s, dg = dh T ds sdT, = (T ds + vdP ) T ds sdT,

=dh
= vdP sdT.
So for Gibbs free energy, the canonical variables are P and T while the conjugate variables are v and s. One then has g = g(P, T ), which gives dg = So one nds v= g P g P v T Example 3.6
Canonical Form If h(s, P ) = cP To P Po
R/cP
dP +
T
g T
dT.
P
(3.116)
,
T
s = s P
g T
.
P
(3.117)
The resulting Maxwell function is then = .

T
(3.118)
exp
s cP
+ (ho cP To ) ,
(3.119)
and cP , To , R, Po , and ho are all constants, derive both thermal and caloric state equations P (v, T ) and e(v, T ). Now for this material h s h P Now since h s h P =
P
= To
P
P Po
R/cP
exp
R/cP 1
s cP exp
, s cP .
(3.120) (3.121)
=
s
RTo Po
P Po
T, v,
(3.122) (3.123)
=
s
3.4. HEAT CAPACITY

one has T v Dividing one by the other gives T v Pv = = P , R RT , = = To RTo Po P Po
R/cP
105
exp
R/cP 1
s cP exp
, s cP .
(3.124) (3.125)
P Po
(3.126) (3.127)
which is the thermal equation of state. Substituting from Eq. (3.124) into the canonical equation for h, Eq. (3.119), one also nds for the caloric equation of state h h = = cP (T To ) + ho . P v Po vo R R P v Po vo R R cP T + (ho cP To ) , (3.128) (3.129)
which is useful in itself. Substituting in for T and To , h = cP Using h e + P v we get e + P v = cP so e e e e e e = = = = = = cP cP 1 Pv 1 Po vo + uo , R R cP 1 (P v Po vo ) + eo , R cP 1 (RT RTo ) + eo , R (cP R) (T To ) + eo , (3.132) (3.133) (3.134) (3.135) (3.136) (3.137) + eo + Po vo (3.131) + ho . (3.130)
(cP (cP cv )) (T To ) + eo , cv (T To ) + eo .
So one canonical equation gives us all the information one needs. Oftentimes, it is dicult to do a single experiment to get the canonical form.
3.4
Heat capacity
cv = cP = e T h T ,
v
Recall that (3.138) (3.139)
.
P
106
Then perform operations on the Gibbs equation de = T ds P dv, e s = T , T v T v s . cv = T T v Likewise, dh = T ds + vdP, h s = T , T P T P s . cP = T T P One nds further useful relations by operating on the Gibbs equation: de = T ds P dv, e s = T P, v T v T P = T P. T v So one can then say e = e(T, v), e e dT + de = T v v P = cv dT + T T For an ideal gas, one has e v = T
T
(3.140) (3.141) (3.142)
(3.143) (3.144) (3.145)
(3.146) (3.147) (3.148)
(3.149) dv,
T
(3.150) dv. (3.151)
P T
P =T
R v
RT , v
(3.152) (3.153)
= 0.
Consequently, e is not a function of v for an ideal gas, so e = e(T ) alone. Since h = e + P v for an ideal gas reduces to h = e + RT h = e(T ) + RT = h(T ).
(3.154)
3.4. HEAT CAPACITY Now return to general equations of state. With s = s(T, v) or s = s(T, P ), one gets ds = ds = s T s T dT +
v
107
s dv, v T s dT + dP. P T
(3.155) (3.156)
Now using Eqs. (3.103,3.118,3.142,3.145) one gets cv P dT + T T cP v ds = dT T T ds = Subtracting one from the other, one nds v P cv cP dv + dT + T T v T v P dv + T dP. (cP cv )dT = T T v T P 0 = dP,
P
dv,
v
(3.157) (3.158)
dP.
P
(3.159) (3.160)
Now divide both sides by dT and hold either P or v constant. In either case, one gets cP cv = T Example 3.7
For an ideal gas nd cP cv . For the ideal gas, P v = RT , one has P T So cP cv = = = T RR , vP R2 , T RT R. (3.163) (3.164) (3.165) =
v
P T
v T
.
P
(3.161)
R , v
v T
=
P
R . P
(3.162)
This holds even if the ideal gas is calorically imperfect. That is cP (T ) cv (T ) = R. (3.166)
108
For the ratio of specic heats for a general material, one can use Eqs. (3.142) and (3.145) to get = cP cv = = = = = T T s T
s T P s T v
then apply Eq. (3.56) to get
(3.167) (3.168) (3.169) (3.170) (3.171)
T , then apply Eq. (3.58) to get P s v s T v P , P T T s v s s T v P T s , s T P T T s v s P v P T v s
The rst term can be obtained from P v T data. The second term is related to the isentropic sound speed of the material, which is also a measurable quantity. Example 3.8
For a calorically perfect ideal gas with gas constant R and specic heat at constant volume cv , nd expressions for the thermodynamic variable s and thermodynamic potentials e, h, a, and g, as functions of T and P . First get the entropy: de = T ds = T ds = ds = = ds s s0 s s0 cv = = = = = = = de + P dv, cv dT + P dv, P dT + dv, cv T T dv dT +R , cv T v dv dT + R , cv T v T v cv ln + R ln , T0 v0 T R RT /P ln + ln , T0 cv RT0 /P0 ln ln ln ln T T0 T T0 T T0 T T0 + ln
1+R/cv
T ds P dv,
(3.172) (3.173) (3.174) (3.175) (3.176) (3.177) (3.178) (3.179)

R/cv
T P0 T0 P + ln
, P0 P
R/cv
(3.180) , (3.181) , (3.182) (3.183)
1+(cp cv )/cv
+ ln
P0 P ,
(cp cv )/cv
+ ln
P0 P
3.5. VAN DER WAALS GAS

T T0
109
P0 P
1
s0 + cv ln
+ cv ln
(3.184)
Now, for the calorically perfect ideal gas, one has e = eo + cv (T To ). For the enthalpy, one gets h = = = = = = = e + P v, e + RT, eo + cv (T To ) + RT, (3.186) (3.187) (3.188) (3.189) (3.190) (3.191) (3.192) (3.185)
eo + cv (T To ) + RT + RTo RTo , eo + RTo +cv (T To ) + R(T To ),

=ho
ho + (cv + R)(T To ),
=cp
ho + cp (T To ).
For the Helmholtz free energy, one gets a = e T s, = eo + cv (T To ) T For the Gibbs free energy, one gets g = = h T s, ho + cp (T To ) T s0 + cv ln T T0 (3.195)
(3.193) s0 + cv ln T T0
+ cv ln
P0 P
(3.194)
+ cv ln
P0 P
(3.196)
3.5
Van der Waals gas
A van der Waals9 gas is a common model for a non-ideal gas. It can capture some of the behavior of a gas as it approaches the vapor dome. Its form is P (T, v) = RT a 2, vb v (3.197)
where b accounts for the nite volume of the molecules, and a accounts for intermolecular forces.
9
Johannes Diderik van der Waals, 1837-1923, Dutch thermodynamicist. CC BY-NC-ND. 12 December 2011, J. M. Powers.
110
Example 3.9
Find a general expression for e(T, v) if P (T, v) = Proceed as before: First we have de = recalling that e T = cv ,
v
a RT 2. vb v e v
(3.198)
e T
dT +
v
dv,
T
(3.199)
e v
=T
T
P T
P.
(3.200)
Now for the van der Waals gas, we have P T T P T

v
=
v
= =
R , vb RT P, vb a RT RT 2 vb vb v
(3.201) (3.202) = a . v2 (3.203)
So we have a , v2 T a e(T, v) = + f (T ). v = e v (3.204) (3.205)
Here f (T ) is some as-of-yet arbitrary function of T . To evaluate f (T ), take the derivative with respect to T holding v constant: e T =
v
df = cv . dT
(3.206)
Since f is a function of T at most, here cv can be a function of T at most, so we allow cv = cv (T ). Integrating, we nd f (T ) as

T
f (T ) = C +
To
cv (T )dT ,
(3.207)
where C is an integration constant. Thus e is

T
e(T, v) = C +
To
a cv (T )dT . v
(3.208)
Taking C = eo + a/vo , we get

T
e(T, v) = eo +
To
cv (T )dT + a
1 1 vo v
(3.209)

We also nd
T
111
h = e + P v = eo +
To T
cv (T )dT + a 1 1 vo v
1 1 vo v +
+ P v,
(3.210) (3.211)
h(T, v) = eo +
To
cv (T )dT + a
RT v a . vb v
Example 3.10
Thermodynamic process with a van der Waals Gas A van der Waals gas with R = a = J , kg K P a m6 , 150 kg 2 m3 0.001 , kg 200 (350 + 0.2(T 300)) J kg K , (3.212) (3.213) (3.214) (3.215)
b = cv =
begins at T1 = 300 K, P1 = 1 105 P a. It is isothermally compressed to state 2 where P2 = 1 106 P a. It is then isochorically heated to state 3 where T3 = 1000 K. Find w13 , q13 , and s3 s1 . Assume the surroundings are at 1000 K. Recall P = so at state 1 105 = Expanding, one gets
2 3 0.15 + 150v1 60100v1 + 100000v1 = 0.
RT a , v b v2
(3.216)
150 200 300 2 . v1 0.001 v1
(3.217)
(3.218)
This is a cubic equation which has three solutions: v1 v1 v1 = = = 0.598 m3 , kg physical, not physical, not physical. (3.219) (3.220) (3.221)
m3 kg m3 0.00125 + 0.0097i kg 0.00125 0.0097i
112

Now at state 2, P2 and T2 are known, so v2 can be determined: 106 = 200 300 150 2 . v2 0.001 v2 (3.222)
The physical solution is v2 = 0.0585 m3 /kg. Now at state 3 it is known that v3 = v2 and T3 . Determine P3 : P3 = 200 1000 150 = 3478261 43831 = 3434430 P a. 0.0585 0.001 0.05852
2 1
(3.223)
Now w13 = w12 + w23 =

v2
P dv +
3 2
P dv =
2 1
P dv since 2 3 is at constant volume. So (3.224) (3.225) (3.226) 1 1 0.0585 0.598 , (3.227) (3.228) (3.229) (3.230)
w13
= = = = = = =
RT a 2 dv, vb v v1 v2 v2 dv dv RT1 , a 2 v1 v b v1 v 1 1 v2 b +a , RT1 ln v1 b v2 v1 0.0585 0.001 200 300 ln + 150 0.598 0.001 140408 + 2313, J , 138095 kg kJ 138.095 . kg
The gas is compressed, so the work is negative. Since e is a state property:

T3
e3 e1 = Now
cv (T )dT + a
T1
1 1 v1 v3
(3.231)
1 cv = 350 + 0.2(T 300) = 290 + T. 5 so

T3
(3.232)
e3 e1
=
T1
1 290 + T 5
dT + a
1 1 v1 v3
(3.233) (3.234) , (3.235) (3.236) (3.237) (3.238)
= = = = =
1 1 1 2 T 2 T1 + a , 10 3 v1 v3 1 1 1 290 (1000 300) + 10002 3002 + 150 10 0.598 0.0585 203000 + 91000 2313, J , 291687 kg kJ . 292 kg 290 (T3 T1 ) +

Now from the rst law e3 e1 q13 q13 q13 = q13 w13 ,
113
(3.239) (3.240) (3.241) (3.242)
= e3 e1 + w13 , = 292 138, kJ = 154 . kg
The heat transfer is positive as heat was added to the system. Now nd the entropy change. Manipulate the Gibbs equation: T ds = de + P dv, P 1 de + dv, ds = T T a P 1 cv (T )dT + 2 dv + dv, ds = T v T 1 RT 1 a a ds = 2 cv (T )dT + 2 dv + T v T vb v cv (T ) R ds = dT + dv, T vb T3 v3 b cv (T ) dT + R ln , s3 s1 = T v1 b T1
1000
(3.243) (3.244) (3.245) dv, (3.246) (3.247) (3.248) (3.249) (3.250) (3.251) (3.252) (3.253)
= = = = =
v3 b 290 1 dT + R ln + , T 5 v1 b 300 1000 1 0.0585 0.001 290 ln + (1000 300) + 200 ln , 300 5 0.598 0.001 349 + 140 468, J 21 , kg K kJ 0.021 . kg K
Is the second law satised for each portion of the process? First look at 1 2 e2 e1 q12 q12 = = =
T1
q12 w12 ,
T2
(3.254) (3.255) 1 1 v1 v2 + RT1 ln v2 b v1 b +a 1 1 v2 v1 . (3.256)
e2 e1 + w12 , cv (T )dT + a
Recalling that T1 = T2 and canceling the terms in a, one gets q12 = RT1 ln v2 b v1 b = 200 300 ln 0.0585 0.001 0.598 0.001 v2 b v1 b = 140408 J . kg (3.257)
Since the process is isothermal, s2 s1 = R ln (3.258)
114

= 200 ln 0.0585 0.001 0.598 0.001 J = 468.0 kg K (3.259) (3.260)
Entropy drops because heat was transferred out of the system. Check the second law. Note that in this portion of the process in which the heat is transferred out of the system, that the surroundings must have Tsurr 300 K. For this portion of the process let us take Tsurr = 300 K. q12 s2 s1 ? (3.261) T J 140408 kg J , (3.262) 468.0 kg K 300 K J J 468.0 468.0 ok. (3.263) kg K kg K Next look at 2 3 q23 q23 since isochoric q23 = e3 e2 + w23 ,
T3
(3.264) 1 1 v2 v3
v3
=
T2 T3
cv (T )dT + a cv (T )dT, T2 1000 290 +

300
+
v2
P dv ,
(3.265) (3.266)
= =
T 5
dT = 294000
J . kg
(3.267)
Now look at the entropy change for the isochoric process:

T3
s3 s2
=
T2 T3
cv (T ) dT , T
(3.268) (3.269) (3.270) (3.271)
290 1 dT , + T 5 T2 1000 1 + (1000 300), = 290 ln 300 5 J = 489 . kg K = Entropy rises because heat transferred into system.
In order to transfer heat into the system we must have a dierent thermal reservoir. This one must have Tsurr 1000 K. Assume here that the heat transfer was from a reservoir held at 1000 K to assess the inuence of the second law. q23 s3 s2 ? (3.272) T J 294000 kg J 489 , (3.273) kg K 1000 K J J 294 , ok. (3.274) 489 kg K kg K
3.6. REDLICH-KWONG GAS
115
3.6
Redlich-Kwong gas
RT a . v b v(v + b)T 1/2
The Redlich10 -Kwong equation of state11 is P = (3.275)
It is modestly more accurate than the van der Waals equation in predicting material behavior. Example 3.11
For the case in which b = 0, nd an expression for e(T, v) consistent with the Redlich-Kwong state equation. Here the equation of state is now P = Proceeding as before, we have e v = T
T
RT a 2 1/2 . v v T
(3.276)
P P, T v R a = T + 2 3/2 v 2v T 3a . = 2 T 1/2 2v
(3.277) a RT 2 1/2 v v T , (3.278) (3.279)
Integrating, we nd e(T, v) = 3a + f (T ). 2vT 1/2 (3.280)
Here f (T ) is a yet-to-be-specied function of temperature only. Now the specic heat is found by the temperature derivative of e: cv (T, v) = e T =
v
3a df . + 4vT 3/2 dT
(3.281)
Obviously, for this material, cv is a function of both T and v. Let us dene cvo (T ) via df cvo (T ). dT Integrating, then one gets
T
(3.282)
f (T ) = C +
To 10 11
cvo (T ) dT .
(3.283)
Otto Redlich, 1896-1978, Austrian chemical engineer. Redlich, O., and Kwong, J. N. S., 1949, On the Thermodynamics of Solutions. V. An Equation of State. Fugacities of Gaseous Solutions, Chemical Reviews, 44(1): 233-244. CC BY-NC-ND. 12 December 2011, J. M. Powers.
116

Let us take C = eo + 3a/2/vo/To . Thus we arrive at the following expressions for cv (T, v) and e(T, v): cv (T, v) = e(T, v) = cvo (T ) +
T 1/2
3a , 4vT 3/2 3a cvo (T ) dT + 2 1

1/2 vo To
(3.284) 1 vT 1/2 (3.285)
eo +
To
3.7
Compressibility and generalized charts
A simple way to quantify the deviation from ideal gas behavior is to determine the so-called compressibility Z, where Pv . (3.286) Z RT For an ideal gas, Z = 1. For substances with a simple molecular structure, Z can be tabulated as functions of the so-called reduced pressure Pr and reduced temperature Tr . Tr and Pr are dimensionless variables found by scaling their dimensional counterparts by the specic materials temperature and pressure at the critical point, Tc and Pc : T P , Pr . (3.287) Tc Pc Often charts are available which give predictions of all reduced thermodynamic properties. These are most useful to capture the non-ideal gas behavior of materials for which tables are not available. Tr
3.8
Mixtures with variable composition
Consider now mixtures of N species. The focus here will be on extensive properties and molar properties. Assume that each species has ni moles, and the total number of moles is n = N ni . Now one might expect the extensive energy to be a function of the entropy, i=1 the volume and the number of moles of each species: E = E(V, S, ni ). The extensive version of the Gibbs law in which all of the ni are held constant is dE = P dV + T dS. Thus E V = P, E S = T.
V,ni
(3.288)
(3.289)
(3.290)
S,ni
3.8. MIXTURES WITH VARIABLE COMPOSITION
117
In general, since E = E(S, V, ni ), one should expect, for systems in which the ni are allowed to change that N E E E dV + dS + dni . (3.291) dE = V S,ni S V,ni ni S,V,nj i=1 Dening the new thermodynamics property, the chemical potential i , as i E ni ,
S,V,nj
(3.292)
one has the important Gibbs equation for multicomponent systems:

N
dE = P dV + T dS +
i dni .
i=1
(3.293)
Obviously, by its denition, i is on a per mole basis, so it is given the appropriate overline notation. In Eq. (3.293), the independent variables and their conjugates are x1 x2 x3 x4 xN +2 V, S, n1 , n2 , . . . = nN , = = = = 1 = P, 2 = T, 3 = 1 , 4 = 2 , . . . N +2 = N . (3.294) (3.295) (3.296) (3.297) (3.298)
Equation (3.293) has 2N +1 1 Legendre functions. Three are in wide usage: the extensive analog to those earlier found. They are H = E + P V, A = E T S, G = E + P V T S. (3.299) (3.300) (3.301)
A set of non-traditional, but perfectly acceptable additional Legendre functions would be formed from E 1 n1 . Another set would be formed from E + P V 2 n2 . There are many more, but one in particular is sometimes noted in the literature: the so-called grand potential, . The grand potential is dened as
N
E TS
i ni .
i=1
(3.302)
Dierentiating each dened Legendre function, Eqs. (3.299-3.302), and combining with Eq. (3.293), one nds
N
dH = T dS + V dP +
i=1
i dni ,
(3.303)
118

N
dA = SdT P dV + dG = SdT + V dP + d = P dV SdT
i dni ,
i=1 N
(3.304) (3.305) (3.306)
i dni ,
i=1 N
ni di .
i=1
Thus, canonical variables for H are H = H(S, P, ni). One nds a similar set of relations as before from each of the dierential forms: T = E S =
V,ni
H S = G P = ,
,
P,ni
(3.307) = V ,
T,i
P = V =
E V
S,ni
A V
(3.308) (3.309)
T,ni
H P
=
S,ni
,
T,ni
S = ni = i = E ni
A T i =
V,ni
G T
P,ni
,
V,i
(3.310) (3.311)
V,T,j
S,V,nj
H ni
=
S,P,nj
A ni
=
T,V,nj
G ni
.
T,P,nj
(3.312)
Each of these induces a corresponding Maxwell relation, obtained by cross dierentiation. These are T V T P P T V T i T i P
S,ni
= =
P S
,
V,ni
(3.313) (3.314) (3.315)
S,ni
V S S V
,
P,ni
=
V,ni
,
T,ni
P,ni
= = =
S P S ni
,
T,ni
(3.316) (3.317) (3.318)
,
V,nj
P,nj
T,nj
V ni
,
V,nj
3.9. PARTIAL MOLAR PROPERTIES l nk k nl P T nk i
119 = ,
T,P,nj
(3.319) (3.320) . (3.321)
T,P,nj
S V ni k
=
T,i
,
V,i
=
V,T,j ,j=k
V,T,j ,j=i
3.9
3.9.1
Partial molar properties

Homogeneous functions
f (x1 , . . . , xN ) = m f (x1 , . . . , xN ). (3.322) (3.323)
In mathematics, a homogeneous function f (x1 , . . . , xN ) of order m is one such that
If m = 1, one has f (x1 , . . . , xN ) = f (x1 , . . . , xN ). Thermodynamic variables are examples of homogeneous functions.
3.9.2
Gibbs free energy
Consider an extensive property, such as the Gibbs free energy G. One has the canonical form (3.324) G = G(T, P, n1 , n2 , . . . , nN ). One would like to show that if each of the mole numbers ni is increased by a common factor, say , with T and P constant, that G increases by the same factor : G(T, P, n1 , n2 , . . . , nN ) = G(T, P, n1 , n2 , . . . , nN ). (3.325)
Dierentiate both sides of Eq. (3.325) with respect to , while holding P , T , and nj constant, to get G(T, P, n1 , n2 , . . . , nN ) = G G d(n1 ) + (n1 ) nj ,P,T d (n2 ) = G (n1 )
nj ,P,T
nj ,P,T
G d(n2 ) + ...+ d (nN ) n2 + . . . +

nj ,P,T
nj ,P,T
d(nN ) , d nN ,
nj ,P,T
(3.326) (3.327)
n1 +
G (n2 )
G (nN )
This must hold for all , including = 1, so one requires G(T, P, n1, n2 , . . . , nN ) = G n1 n1 +
nj ,P,T
G n2
n2 + . . . +
nj ,P,T
G nN
nN ,
nj ,P,T
120
CHAPTER 3. MATHEMATICAL FOUNDATIONS OF THERMODYNAMICS* (3.328)

N
=
i=1
G ni
ni .
nj ,P,T
(3.329)
Recall now the denition partial molar property, the derivative of an extensive variable with respect to species ni holding nj , i = j, T , and P constant. Because the result has units per mole, an overline superscript is utilized. The partial molar Gibbs free energy of species i, g i is then G , (3.330) gi ni nj ,P,T so that G=
i=1 N
g i ni .
(3.331)
Using the denition of chemical potential, Eq. (3.312), one also notes then that
N
G(T, P, n2 , n2 , . . . , nN ) =
i=1
i ni .
(3.332)
The temperature and pressure dependence of G must lie entirely within i (T, P, nj ), which one notes is also allowed to be a function of nj as well. Consequently, one also sees that the Gibbs free energy per unit mole of species i is the chemical potential of that species: g i = i . (3.333)
Using Eq. (3.331) to eliminate G in Eq. (3.301), one recovers an equation for the energy:
N
E = P V + T S +
i ni .
i=1
(3.334)
3.9.3
Other properties
A similar result also holds for any other extensive property such as V , E, H, A, or S. One can also show that
N
=
i=1 N
ni ni
i=1 N
V ni E ni H ni
,
nj ,A,T
(3.335) (3.336)
E = H =
i=1
nj ,V,S
ni
,
nj ,P,S
(3.337)
3.9. PARTIAL MOLAR PROPERTIES

N
121 A ni S ni
A =
i=1 N
ni ni
i=1
,
nj ,T,V
(3.338) (3.339)
S =
.
nj ,E,T
Note that these expressions do not formally involve partial molar properties since P and T are not constant. Take now the appropriate partial molar derivatives of G for an ideal mixture of ideal gases to get some useful relations: G = H T S, G H S = T ni T,P,nj ni T,P,nj ni (3.340) .
T,P,nj
(3.341)
Now from the denition of an ideal mixture hi = hi (T, P ), so one has

N
H =
k=1
nk hk (T, P ),
N
(3.342) (3.343) (3.344)
H ni
T,P,nj
= ni
N
nk hk (T, P ) ,
k=1
=
k=1 N
nk hk (T, P ), ni
=ik
=
k=1
ik hk (T, P ),
(3.345) (3.346)
= hi (T, P ).
Here, the Kronecker delta function ki has been again used. Now for an ideal gas one further has hi = hi (T ). The analysis is more complicated for the entropy, in which
N
S =
k=1 N
nk so R ln T,k nk so R ln T,k nk so R ln T,k
Pk Po yk P Po P Po
, , R ln nk N q=1 nq ,
(3.347) (3.348) (3.349)
=
k=1 N
=
k=1
122

N
=
k=1
nk so R ln T,k
N
P Po P Po
nk ln
k=1
nk N q=1
nq
(3.350)
S ni
=
T,P,nj
ni
k=1
nk so R ln T,k
N
R ni
N
nk ln
T,P,nj k=1
nk N q=1
nq
(3.351)
=
k=1
nk ni
=ik
so R ln T,k
N
P Po nk N q=1
R =
ni
nk ln
T,P,nj k=1
nq
,
N
(3.352) nk ln nk N q=1 . (3.353)
so T,i
R ln
P Po
R ni
T,P,nj
k=1
nq
Evaluation of the nal term on the right side requires closer examination, and in fact, after tedious but straightforward analysis, yields a simple result which can easily be veried by direct calculation: ni
N
nk ln
T,P,nj k=1
nk N q=1
nq
= ln
ni N q=1
nq
(3.354)
So the partial molar entropy is in fact S ni = so R ln T,i = so R ln T,i = so R ln T,i = si . Thus, one can in fact claim for the ideal mixture of ideal gases that g i = hi T si . (3.359) P Po P Po Pi Po R ln R ln yi , , ni N q=1 , (3.355) (3.356) (3.357) (3.358)
T,P,nj
nq
3.9.4
Relation between mixture and partial molar properties
A simple analysis shows how the partial molar property for an individual species is related to the partial molar property for the mixture. Consider, for example, the Gibbs free energy.
3.10. FROZEN SOUND SPEED The mixture-averaged Gibbs free energy per unit mole is g= G . n
123
(3.360)
Now take a partial molar derivative and analyze to get g ni =

T,P,nj
1 G n ni
T,P,nj
G n n2 ni
,
T,P,nj N
(3.361) nk , (3.362) (3.363) (3.364) (3.365) (3.366)
1 G = n ni = = = 1 G n ni 1 G n ni
T,P,nj
G 2 n ni G n2 G n2
N
T,P,nj
k=1
T,P,nj
k=1 N
nk ni ik ,
,
T,P,nj
T,P,nj
k=1
1 G G 2, n ni T,P,nj n 1 g gi . = n n
Multiplying by n and rearranging, one gets gi = g + n A similar result holds for other properties. g ni .
T,P,nj
(3.367)
3.10
Frozen sound speed
Let us develop a relation for what is known as the frozen sound speed, denoted as c. This quantity is a thermodynamic property dened by the relationship c2 P .
s,ni
(3.368)
Later we shall see that this quantity describes the speed of acoustic waves, but at this point let us simply treat it as a problem in thermodynamics. The quantity is called frozen because the derivative is performed with all of the species mole numbers frozen at a constant value. Note that since v = 1/, c2 = P =
s,ni
P v
s,ni
dv 1 P = 2 d v
s,ni
= v 2
P v
.
s,ni
(3.369)
124
So we can say (c)2 = P v .

s,ni
(3.370)
Now we know that the caloric equation of state for a mixture of ideal gases takes the form, from Eq. (2.162):
N
e=
i=1
Yiei (T ).
(3.371)
Employing the ideal gas law, we could say instead

N
e=
i=1
Yi ei
P R
(3.372)
In terms of number of moles, ni , we could say

N
e=
i=1
Mi ni ei V
=Yi
P R
(3.373)
We can in fact generalize, and simply consider caloric equations of state of the form e = e(P, , ni ), or in terms of the specic volume v = 1/, e = e(P, v, ni). (3.375) (3.374)
These last two forms can be useful when temperature is unavailable, as can be the case for some materials. Now reconsider the Gibbs equation for a multicomponent system, Eq. (3.293
N
dE = P dV + T dS +
i dni .
i=1
(3.376)
Let us scale Eq. (3.376) by a xed mass m, recalling that e = E/m, v = V /m, and s = S/m, so as to obtain N i de = P dv + T ds + dni . (3.377) m i=1 Considering Eq. (3.377) with dv = 0 and dni = 0, and scaling by dP , we can say e P =T
v,ni
s P
.
v,ni
(3.378)
3.10. FROZEN SOUND SPEED Considering Eq. (3.377) with dni = 0, and dP = 0 and scaling by dv, we can say e v Applying now Eq. (3.59), we nd P v So we get P v =
s v P,ni . s P v,ni
125
P,ni
= P + T
s v
.
P,ni
(3.379)
s,ni
s P
v,ni
v s
P,ni
= 1.
(3.380)
(3.381)
s,ni
Substituting Eq. (3.381) into Eq. (3.370) we nd (c) =

2 s v P,ni . s P v,ni
(3.382)
Substituting Eqs. (3.378,3.379) into Eq. (3.382) we get (c)

2
= =
1 T
P+
e v P,ni
1 e T P v,ni e + v P,ni . e P v,ni
(3.383) (3.384)
So the sound speed c is c=v P+

e v P,ni
e P v,ni
(3.385)
Eq. (3.385) is useful because we can identify the frozen sound speed from data for e, P , and v alone. If e v > P, and e P > 0,
v,ni
(3.386)
P,ni
we will have a real sound speed. Other combinations are possible as well, but not observed in nature. Example 3.12
For an ideal mixture of calorically perfect ideal gases, nd the frozen sound speed. CC BY-NC-ND. 12 December 2011, J. M. Powers.
126

For a calorically perfect ideal gas, we have ei = ei (T ) = eo + cvi (T To ), Pv = eo + cvi To . R
N
(3.387) (3.388)
and e =
i=1 N
Yi ei , mi ei , m ni M i ei , m ni M i m eo + cvi Pv To R .
(3.389)
=
i=1 N
(3.390)
=
i=1 N
(3.391)
=
i=1
(3.392)
Thus we get the two relevant partial derivatives e v e P

N
=
P,ni i=1 N
ni M i P P cvi = m R R ni M i v v cvi = m R R
Yi cvi =
i=1 N
P cv , R vcv . R
(3.393)
=
v,ni i=1
Yi cvi =
i=1
(3.394)
Now substitute Eqs. (3.393, 3.394) into Eq. (3.385) to get c = = = = = = = = v P+ Pv +

cv R vcv R P cv R
, , ,
(3.395) (3.396) (3.397) (3.398) (3.399) (3.400) (3.401) (3.402)
P vcv R
RT +
cv R
RT cv R
RT RT RT RT
1+
cv R
cv R
R + cv , cv (cP cv ) + cv , cv cP , cv
RT .
3.11. IRREVERSIBILITY IN A CLOSED MULTICOMPONENT SYSTEM
127
3.11
Irreversibility in a closed multicomponent system
Consider a thermodynamic system closed to mass exchanges with its surroundings coming into equilibrium. Allow the system to be exchanging work and heat with its surroundings. Assume the temperature dierence between the system and its surroundings is so small that both can be considered to be at temperature T . If dQ is introduced into the system, then the surroundings suer a loss of entropy: dSsurr = dQ . T (3.403)
The systems entropy S can change via this heat transfer, as well as via other internal irreversible processes, such as internal chemical reaction. The second law of thermodynamics requires that the entropy change of the universe be positive semi-denite: dS + dSsurr 0. Eliminating dSsurr , one requires for the system that dS dQ . T (3.405) (3.404)
Consider temporarily the assumption that the work and heat transfer are both reversible. Thus, the irreversibilities must be associated with internal chemical reaction. Now the rst law for the entire system gives dE = dQ dW, = dQ P dV, dQ = dE + P dV. (3.406) (3.407) (3.408)
Note because the system is closed, there can be no species entering or exiting, and so there is no change dE attributable to dni . While within the system the dni may not be 0, the net contribution to the change in total internal energy is zero. A non-zero dni within the system simply re-partitions a xed amount of total energy from one species to another. Substituting Eq. (3.408) into Eq. (3.405) to eliminate dQ, one gets dS T dS dE P dV dE T dS + P dV 1 (dE + P dV ), T
=dQ
(3.409) (3.410) (3.411)
0, 0.
Eq. (3.411) involves properties only and need not require assumptions of reversibility for processes in its derivation. In special cases, it reduces to simpler forms. For processes which are isentropic and isochoric, the second law expression, Eq. (3.411), reduces to (3.412) dE|S,V 0.
128
Now using Eq. (3.293) to eliminate dS in Eq. (3.413), one can express the second law as 1 P 1 dE + dV T T T
=dS N
For processes which are isoenergetic and isochoric, the second law expression, Eq. (3.411), reduces to dS|E,V 0. (3.413)
i dni
i=1 E,V
0,
(3.414)
1 T
i=1
i dni 0.
(3.415)
irreversibility
The irreversibility associated with the internal chemical reaction must be the left side of Eq. (3.415). Now, while most standard texts focusing on equilibrium thermodynamics go to great lengths to avoid the introduction of time, it really belongs in a discussion describing the approach to equilibrium. One can divide Eq. (3.415) by a positive time increment dt to get 1 T
N
i
i=1
dni 0. dt
(3.416)
Since T 0, one can multiply Eq. (3.416) by T to get

N
i
i=1
dni 0. dt
(3.417)
This will hold if a model for dni is employed which guarantees that the left side of Eq. (3.417) dt is negative semi-denite. One will expect then for dni to be related to the chemical potentials dt i . Elimination of dE in Eq. (3.411) in favor of dH from dH = dE + P dV + V dP gives dH P dV V dP T dS + P dV
=dE
0,
(3.418) (3.419)
dH V dP T dS 0. Thus, one nds for isobaric, isentropic equilibration that dH|P,S 0. For the Helmholtz and Gibbs free energies, one analogously nds dA|T,V 0, dG|T,P 0.
(3.420)
(3.421) (3.422)
3.12. EQUILIBRIUM IN A TWO-COMPONENT SYSTEM
129
The expression of the second law in terms of dG is especially useful as it may be easy in an experiment to control so that P and T are constant. This is especially true in an isobaric phase change, in which the temperature is guaranteed to be constant as well. Now one has
N
G =
i=1 N
ni g i , ni i .
i=1 N i=1
(3.423) (3.424) i dni , holding T and P constant (3.425)
One also has from Eq. (3.305): dG = SdT + V dP + that

N
dG|T,P =
i=1
i dni .
Here the dni are associated entirely with internal chemical reactions. Substituting Eq. (3.425) into Eq. (3.422), one gets the important version of the second law which holds that
N
dG|T,P =
i=1
i dni 0.
(3.426)
In terms of time rates of change, one can divide Eq. (3.426) by a positive time increment dt > 0 to get G t
N
=
T,P i=1
dni 0. dt
(3.427)
3.12
Equilibrium in a two-component system
A major task of non-equilibrium thermodynamics is to nd a functional form for dni which dt guarantees satisfaction of the second law, Eq. (3.427) and gives predictions which agree with experiment. This will be discussed in more detail in the following chapter on thermochemistry. At this point, some simple examples will be given in which a na but useful functional ve dni form for dt is posed which leads at least to predictions of the correct equilibrium values. A much better model which gives the correct dynamics in the time domain of the system as it approaches equilibrium will be presented in the chapter on thermochemistry.
3.12.1
Phase equilibrium
Here, consider two examples describing systems in phase equilibrium.

130
Example 3.13
Consider an equilibrium two-phase mixture of liquid and vapor H2 O at T = 100 C, x = 0.5. Use the steam tables to check if equilibrium properties are satised. In a two-phase gas liquid mixture one can expect the following reaction: H2 O(l) H2 O(g) . (3.428)
That is one mole of liquid, in the forward phase change, evaporates to form one mole of gas. In the reverse phase change, one mole of gas condenses to form one mole of liquid. Because T is xed at 100 C and the material is a two phase mixture, the pressure is also xed at a constant. Here there are two phases at saturation; g for gas and l for liquid. Equation (3.426) reduces to l dnl + g dng 0. (3.429) Now for the pure H2 O if a loss of moles from one phase must be compensated by the addition to another. So one must have dnl + dng = 0. (3.430) Hence dng = dnl . So Eq. (3.429), using Eq. (3.431) becomes l dnl g dnl 0, (3.432) (3.433) (3.431)
dnl (l g ) 0.
At this stage of the analysis, most texts, grounded in equilibrium thermodynamics, assert that l = g , ignoring the fact that they could be dierent but dnl could be zero. That approach will not be taken here. Instead divide Eq. (3.433) by a positive time increment, dt 0 to write the second law as dnl ( g ) dt l dnl = (l g ), dt 0. (3.434)
One convenient, albeit na way to guarantee second law satisfaction is to let ve, 0, convenient, but nave model (3.435)
Here is some positive semi-denite scalar rate constant which dictates the time scale of approach to equilibrium. Note that Eq. (3.435) is just a hypothesized model. It has no experimental verication; in fact, other more complex models exist which both agree with experiment and satisfy the second law. For the purposes of the present argument, however, Eq. (3.435) will suce. With this assumption, the second law reduces to (l g )2 0, 0, (3.436) which is always true. Eq. (3.435) has two important consequences: Dierences in chemical potential drive changes in the number of moles. The number of moles of liquid, nl , increases when the chemical potential of the liquid is less than that of the gas, l < g . That is to say, when the liquid has a lower chemical potential than the gas, the gas is driven towards the phase with the lower potential. Because such a phase change is isobaric and isothermal, the Gibbs free energy is the appropriate variable to consider, and one takes = g. When this is so, the Gibbs free energy of the mixture, G = nl l + ng g is being driven to a lower value. So when dG = 0, the system has a minimum G.

The system is in equilibrium when the chemical potentials of liquid and gas are equal: l = g .
131
The chemical potentials, and hence the molar specic Gibbs free energies must be the same for each constituent of the binary mixture at the phase equilibrium. That is gl = gg . (3.437)
Now since both the liquid and gas have the same molecular mass, one also has the mass specic Gibbs free energies equal at phase equilibrium: gl = gg . This can be veried from the steam tables, using the denition g = h T s. From the tables gl gg = = hl T sl = 419.02 kJ kJ kJ = 68.6 ((100 + 273.15) K) 1.3068 , kg kg K kg kJ kJ kJ hg T sg = 2676.05 = 68.4 ((100 + 273.15) K) 7.3548 . kg kg K kg (3.439) (3.440) (3.438)
The two values are essentially the same; the dierence is likely due to table inaccuracies.
3.12.2
Chemical equilibrium: introduction
Here consider two examples which identify the equilibrium state of a chemically reactive system. 3.12.2.1 Isothermal, isochoric system
The simplest system to consider is isothermal and isochoric. The isochoric assumption implies there is no work in coming to equilibrium. Example 3.14
At high temperatures, collisions between diatomic nitrogen molecules induce the production of monatomic nitrogen molecules. The chemical reaction can be described by the model N2 + N2 2N + N2 . (3.441)
Here one of the N2 molecules behaves as an inert third body. An N2 molecule has to collide with something, to induce the reaction. Some authors leave out the third body and write instead N2 2N , but this does not reect the true physics as well. The inert third body is especially important when the time scales of reaction are considered. It plays no role in equilibrium chemistry. Consider 1 kmole of N2 and 0 kmole of N at a pressure of 100 kP a and a temperature of 6000 K. Using the ideal gas tables, nd the equilibrium concentrations of N and N2 if the equilibration process is isothermal and isochoric. CC BY-NC-ND. 12 December 2011, J. M. Powers.
132

The ideal gas law can give the volume. P1 V nN2 RT , V nN2 RT , = P1 kJ (1 kmole) 8.314 kmole = 100 kP a = 498.84 m3 . = (3.442) (3.443)
K
(6000 K)
(3.444) (3.445)
Initially, the mixture is all N2 , so its partial pressure is the total pressure, and the initial partial pressure of N is 0. Now every time an N2 molecule reacts and thus undergo a negative change, 2 N molecules are created and thus undergo a positive change, so dnN2 = 1 dnN . 2 (3.446)
This can be parameterized by a reaction progress variable , also called the degree of reaction, dened such that d d = dnN2 , 1 dnN . = 2 (3.447) (3.448)
As an aside, one can integrate this, taking = 0 at the initial state to get Thus nN 2 nN = = nN2 |t=0 , 2. (3.451) (3.452) = = nN2 |t=0 nN2 , 1 nN . 2 (3.449) (3.450)
One can also eliminate to get nN in terms of nN2 : nN = 2 ( nN2 |t=0 nN2 ) . Now for the reaction, one must have, for second law satisfaction, that N2 dnN2 + N dnN N2 (d) + N (2d) N2 + 2N d d N2 + 2N dt 0, 0, 0 0. (3.454) (3.455) (3.456) (3.457) (3.453)
In order to satisfy the second law, one can usefully, but na vely, hypothesize that the non-equilibrium reaction kinetics are given by d = k(N2 + 2N ), dt k 0, convenient, but nave model (3.458)
133
Note there are other ways to guarantee second law satisfaction. In fact, a more complicated model is well known to t data well, and will be studied later. For the present purposes, this na model will ve suce. With this assumption, the second law reduces to k(N2 + 2N )2 0, which is always true. Obviously, the reaction ceases when 2N = N2 . Away from equilibrium, for the reaction to go forward, one must expect have N2 + 2N 0,
d dt
k 0,
d dt
(3.459)
= 0, which holds only when (3.460) > 0, and then one must (3.461) (3.462)
2N N2 .
The chemical potentials are the molar specic Gibbs free energies; thus, for the reaction to go forward, one must have 2gN g N2 . (3.463) Substituting using the denitions of Gibbs free energy, one gets 2 h N T sN 2 hN T so R ln T,N yN P Po h N 2 T sN 2 , hN 2 T (3.464) (3.465) (3.466) (3.467) (3.468) (3.469)
o 2 h N T so T,N hN2 T sT,N2 o 2 hN T so T,N + hN2 T sT,N2 o 2 hN T so T,N + hN2 T sT,N2 o 2 hN T so T,N + hN2 T sT,N2
yN2 P , Po yN P yN2 P + RT ln , 2RT ln Po Po yN P yN2 P 2RT ln RT ln , Po Po 2 yN P 2 Po RT ln , 2 Po P yN2 2 yN P RT ln . yN2 Po so 2 R ln T,N
At the initial state, one has yN = 0, so the right hand side approaches , and the inequality holds. At equilibrium, one has equality.
o 2 h N T so T,N + hN2 T sT,N2
RT ln
2 yN P yN2 Po
(3.470)
Taking numerical values from Table A.9: 2 5.9727 105 kJ kJ (6000 K) 216.926 + kmole kg K kJ kJ 2.05848 105 (6000 K) 292.984 kmole kg K = 2.87635 =
ln KP
8.314 ln
kJ kmole K 2 yN P , yN2 Po
(6000 K) ln
2 yN P yN2 Po
(3.471)
134

0.0563399 =
KP 2 yN P , yN2 Po nN nN +nN2 nN2 nN +nN2 2
(3.472)
(nN + nN2 )
RT , Po V
(3.473)
= = = This is a quadratic equation for nN2 . It has two roots nN2 = 0.888154 kmole physical;
n2 RT N , nN2 Po V (2 ( nN2 |t=0 nN2 )) RT , nN 2 Po V

2 2
(3.474) (3.475)
(2 (1 kmole nN2 )) (8.314)(6000) . (3.476) nN 2 (100)(498.84)
nN2 = 1.12593 kmole,
non-physical
(3.477)
The second root generates more N2 than at the start, and also yields non-physically negative nN = 0.25186 kmole. So at equilibrium, the physical root is nN = 2(1 nN2 ) = 2(1 0.888154) = 0.223693 kmole. (3.478)
The diatomic species is preferred. Note in the preceding analysis, the term KP was introduced. This is the so-called equilibrium constant which is really a function of temperature. It will be described in more detail later, but one notes that it is commonly tabulated for some reactions. Its tabular value can be derived from the more fundamental quantities shown in this example. BSs Table A.11 gives for this reaction at 6000 K the value of ln KP = 2.876. Note that KP is fundamentally dened in terms of thermodynamic properties for a system which may or may not be at chemical equilibrium. Only at chemical equilibrium, can it can further be related to mole fraction and pressure ratios. The pressure at equilibrium is P2 = = = (nN2 + nN )RT , V (0.888154 kmole + 0.223693 kmole) 8.314 498.84 111.185 kP a. (3.479)
kJ kmole K
(6000 K)
(3.480) (3.481)
The pressure has increased because there are more molecules with the volume and temperature being equal. The molar concentrations i at equilibrium, are N N 2 = = 0.223693 kmole kmole mole = 4.484 104 = 4.484 107 , 3 3 498.84 m m cm3 kmole mole 0.888154 kmole = 1.78044 103 = 1.78044 106 . 3 3 498.84 m m cm3 (3.482) (3.483)
Now consider the heat transfer. One knows for the isochoric process that Q = U2 U1 . The initial energy is given by U1 = = nN2 uN2 , nN2 (hN2 RT ), (3.484) (3.485)

= = (1 kmole) 2.05848 105 1.555964 105 kJ. kJ kJ 8.314 kmole kmole K (6000 K) ,
135
(3.486) (3.487)
The energy at the nal state is U2 = = = nN 2 u N 2 + nN u N , kJ kJ (6000 K) 8.314 kmole kmole K kJ kJ (6000 K) , 8.314 +(0.223693 kmole) 5.9727 105 kmole kmole K (0.888154 kmole) 2.05848 105 2.60966 105 kJ. nN2 (hN2 RT ) + nN (hN RT ), (3.488) (3.489) (3.490) (3.491) (3.492)
= So
= 2.60966 105 kJ 1.555964 105 kJ, = 1.05002 105 kJ.
= U2 U1 ,
(3.493) (3.494) (3.495)
Heat needed to be added to keep the system at the constant temperature. This is because the nitrogen dissociation process is endothermic. One can check for second law satisfaction in two ways. Fundamentally, one can demand that Eq. (3.405), dS dQ/T , be satised for the process, giving
2
S2 S1 For this isothermal process, this reduces to
dQ . T
(3.496)
S2 S1 (nN2 sN2 + nN sN )|2 (nN2 sN2 + nN sN )|1 yN2 P Po yN2 P nN2 so 2 R ln T,N Po PN2 nN2 so 2 R ln T,N Po PN2 nN2 so 2 R ln T,N Po nN2 so 2 R ln T,N nN2 so 2 R ln T,N nN2 RT Po V nN2 RT Po V + nN + nN + nN + nN + nN + nN
=0
Q , T Q , T
(3.497)
(3.498)
so R ln T,N
yN P Po yN P so R ln T,N Po PN so R ln T,N Po PN so R ln T,N Po nN RT Po V nN RT Po V
Q , T
(3.499)
Q , T
(3.500)
nN2 so 2 R ln T,N
so R ln T,N so R ln T,N
2
1
Q . T
(3.501)
136

Now at the initial state nN = 0 kmole, and RT /Po /V has a constant value of
kJ 8.314 kmole K (6000 K) 1 RT = =1 , Po V (100 kP a)(498.84 m3 ) kmole
(3.502)
so Eq. (3.501) reduces to nN2 so 2 R ln T,N nN 2 1 kmole + nN so R ln T,N nN 2 so 2 T,N nN 1 kmole nN 2 R ln 1 kmole
2
Q , T (3.503)
((0.888143) (292.984 8.314 ln (0.88143)) + (0.223714) (216.926 8.314 ln (0.223714)))|2 ((1) (292.984 8.314 ln (1)))|1 19.4181 kJ kJ 17.5004 . K K
105002 , 6000 (3.504)
Indeed, the second law is satised. Moreover the irreversibility of the chemical reaction is 19.4181 17.5004 = +1.91772 kJ/K. For the isochoric, isothermal process, it is also appropriate to use Eq. (3.421), dA|T,V 0, to check for second law satisfaction. This turns out to give an identical result. Since by Eq. (3.300), A = U T S, A2 A1 = (U2 T2 S2 ) (U1 T1 S1 ). Since the process is isothermal, A2 A1 = U2 U1 T (S2 S1 ). For A2 A1 0, one must demand U2 U1 T (S2 S1 ) 0, or U2 U1 T (S2 S1 ), or S2 S1 (U2 U1 )/T . Since Q = U2 U1 for this isochoric process, one then recovers S2 S1 Q/T.
3.12.2.2
Isothermal, isobaric system
Allowing for isobaric rather than isochoric equilibration introduces small variation in the analysis. Example 3.15
Consider the same reaction N2 + N2 2N + N2 . (3.505) for an isobaric and isothermal process. That is, consider 1 kmole of N2 and 0 kmole of N at a pressure of 100 kP a and a temperature of 6000 K. Using the ideal gas tables, nd the equilibrium concentrations of N and N2 if the equilibration process is isothermal and isobaric. The initial volume is the same as from the previous example: V1 = 498.84 m3 . (3.506)
Note the volume will change in this isobaric process. Initially, the mixture is all N2 , so its partial pressure is the total pressure, and the initial partial pressure of N is 0. A few other key results are identical to the previous example: nN = 2 ( nN2 |t=0 nN2 ) , CC BY-NC-ND. 12 December 2011, J. M. Powers. (3.507)

and 2gN g N2 . Substituting using the denitions of Gibbs free energy, one gets 2 h N T sN 2 hN T so R ln T,N yN P Po h N 2 T sN 2 , hN 2 T
137
(3.508)
(3.509) (3.510) (3.511) (3.512) (3.513)
o 2 h N T so T,N hN2 T sT,N2 o 2 hN T so T,N + hN2 T sT,N2 o 2 hN T so T,N + hN2 T sT,N2
yN2 P , Po yN P yN2 P + RT ln , 2RT ln Po Po yN P yN2 P 2RT ln RT ln , Po Po 2 yN P 2 Po RT ln . 2 Po P yN2 so 2 R ln T,N
In this case Po = P , so one gets

RT ln
2 yN yN2
(3.514)
At the initial state, one has yN = 0, so the right hand side approaches , and the inequality holds. At equilibrium, one has equality.
= RT ln
2 yN yN2
(3.515)
Taking numerical values from Table A.9: 2 5.9727 105 kJ kJ + (6000 K) 216.926 kmole kg K kJ kJ 2.05848 105 (6000 K) 292.984 kmole kg K = 8.314
2 yN yN2
kJ kmole K ,
(6000 K) ln
2 yN yN2
2.87635 = ln
ln KP
(3.516)
0.0563399 =
KP
2 yN , yN2 nN nN +nN2 nN2 nN +nN2 2
(3.517)
(3.518)
= = =
n2 N , nN2 (nN + nN2 )

2
(3.519)
(2 ( nN2 |t=0 nN2 )) , (3.520) nN2 (2 ( nN2 |t=0 nN2 ) + nN2 )
(2 (1 kmole nN2 ))2 . (3.521) nN2 (2 (1 kmole nN2 ) + nN2 )
138

This is a quadratic equation for nN2 . It has two roots nN2 = 0.882147 kmole physical; nN2 = 1.11785 kmole, non-physical (3.522)
The second root generates more N2 than at the start, and also yields non-physically negative nN = 0.235706 kmole. So at equilibrium, the physical root is nN = 2(1 nN2 ) = 2(1 0.882147) = 0.235706 kmole. (3.523)
Again, the diatomic species is preferred. As the temperature is raised, one could show that the monatomic species would come to dominate. The volume at equilibrium is V2 (nN2 + nN )RT , P (0.882147 kmole + 0.235706 kmole) 8.314 = 100 kP a = 557.630 m3 . = (3.524)
kJ kmole K
(6000 K)
(3.525) (3.526)
The volume has increased because there are more molecules with the pressure and temperature being equal. The molar concentrations i at equilibrium, are N N 2 = = 0.235706 kmole kmole mole = 4.227 104 = 4.227 107 , 3 3 557.636 m m cm3 0.882147 kmole kmole mole = 1.58196 103 = 1.58196 106 . 557.636 m3 m3 cm3 (3.527) (3.528)
The molar concentrations are a little smaller than for the isochoric case, mainly because the volume is larger at equilibrium. Now consider the heat transfer. One knows for the isobaric process that Q = H2 H1 . The initial enthalpy is given by H1 = nN2 hN2 = (1 kmole) 2.05848 105 The enthalpy at the nal state is H2 = = = So Q = H2 H1 = 3.22389 105 kJ 2.05848 105 kJ = 1.16520 105 kJ. (3.533) Heat needed to be added to keep the system at the constant temperature. This is because the nitrogen dissociation process is endothermic. Relative to the isochoric process, more heat had to be added to maintain the temperature. This to counter the cooling eect of the expansion. Lastly, it is a straightforward exercise to show that the second law is satised for this process. nN2 hN2 + nN hN , (0.882147 kmole) 2.05848 105 3.22368 105 kJ. kJ kmole + (0.235706 kmole) 5.9727 105 (3.530) kJ (3.531) , kmole (3.532) kJ kmole = 2.05848 105 kJ. (3.529)
139
3.12.3
Equilibrium condition
The results of both of the previous examples, in which a functional form of a progress variables time variation, d/dt, was postulated in order to satisfy the second law gave a condition for equilibrium. This can be generalized so as to require at equilibrium that
N
i i = 0.
i=1
(3.534)
Here i represents the net number of moles of species i generated in the forward reaction. This negation of the term on the left side of Eq (3.534) is sometimes dened as the chemical anity, :
N
i i .
i=1
(3.535)
So in the phase equilibrium example, Eq. (3.534) becomes l (1) + g (1) = 0. In the nitrogen chemistry example, Eq. (3.534) becomes N2 (1) + N (2) = 0. This will be discussed in detail in the following chapter. (3.537) (3.536)
140
Chapter 4 Thermochemistry of a single reaction

This chapter will further develop notions associated with the thermodynamics of chemical reactions. The focus will be on single chemical reactions. Several sources can be consulted as references.12345
4.1
Molecular mass
The molecular mass of a molecule is a straightforward notion from chemistry. One simply sums the product of the number of atoms and each atoms atomic mass to form the molecular mass. If one denes L as the number of elements present in species i, li as the number of moles of atomic element l in species i, and Ml as the atomic mass of element l, the molecular mass Mi of species i
L
Mi =
l=1
Ml li.
(4.1)
In vector form, one would say MT = MT , or M = T M. (4.2)
Here M is the vector of length N containing the molecular masses, M is the vector of length L containing the elemental masses, and is the matrix of dimension L N containing the number of moles of each element in each species. Generally, is full rank. If N > L, has
Borgnakke, C., and Sonntag, R. E., 2009, Fundamentals of Thermodynamics, Seventh Edition, John Wiley, New York. See Chapters 14 and 15. 2 Abbott, M. M., and Van Ness, H. C., 1972, Thermodynamics, Schaums Outline Series in Engineering, McGraw-Hill, New York. See Chapter 7. 3 Kondepudi, D., and Prigogine, I., 1998, Modern Thermodynamics: From Heat Engines to Dissipative Structures, John Wiley, New York. See Chapters 4, 5, 7, and 9. 4 Turns, S. R., 2011, An Introduction to Combustion, Third Edition, McGraw-Hill, Boston. See Chapter 2. 5 Kuo, K. K., 2005, Principles of Combustion, Second Edition, John Wiley, New York. See Chapters 1, 2.
1
141
142
CHAPTER 4. THERMOCHEMISTRY OF A SINGLE REACTION
rank L. If N < L, has rank N. In any problem involving an actual chemical reaction, one will nd N L, and in most cases N > L. In isolated problems not involving a reaction, one may have N < L. In any case, M lies in the column space of T , which is the row space of . Example 4.1
Find the molecular mass of H2 O. Here, one has two elements H and O, so L = 2, and one species, so N = 1; thus, in this isolated problem, N < L. Take i = 1 for species H2 O. Take l = 1 for element H. Take l = 2 for element O. From the periodic table, one gets M1 = 1 kg/kmole for H, M2 = 16 kg/kmole for O. For element 1, there are 2 atoms, so 11 = 2. For element 2, there is 1 atom so 21 = 1. So the molecular mass of species 1, H2 O is M1 = = = = 11 , 21 M1 11 + M2 21 , kg kg (2) + 16 1 kmole kmole kg 18 . kmole ( M1 M2 ) (4.3) (4.4) (1), (4.5) (4.6)
Example 4.2
Find the molecular masses of the two species C8 H18 and CO2 . Here, for practice, the vector matrix notation is exercised. In a certain sense this is overkill, but it is useful to be able to understand a general notation. One has N = 2 species, and takes i = 1 for C8 H18 and i = 2 for CO2 . One also has L = 3 elements and takes l = 1 for C, l = 2 for H, and l = 3 for O. Now for each element, one has M1 = 12 kg/kmole, M2 = 1 kg/kmole, M3 = 16 kg/kmole. The molecular masses are then given by 11 12 ( M1 M2 ) = ( M1 M2 M3 ) 21 22 , (4.7) 31 32 8 1 = ( 12 1 16 ) 18 0 , (4.8) 0 2 = ( 114 44 ) . (4.9) That is for C8 H18 , one has molecular mass M1 = 114 kg/kmole. For CO2 , one has molecular mass M2 = 44 kg/kmole.
4.2. STOICHIOMETRY
143
4.2
4.2.1
Stoichiometry
General development
Stoichiometry represents a mass balance on each element in a chemical reaction. For example, in the simple global reaction 2H2 + O2 2H2 O, (4.10) one has 4 H atoms in both the reactant and product sides and 2 O atoms in both the reactant and product sides. In this section stoichiometry will be systematized. Consider now a general reaction with N species. This reaction can be represented by
N N
i i
i=1
i=1
i i .
(4.11)
Here i is the ith chemical species, i is the forward stoichiometric coecient of the ith reaction, and i is the reverse stoichiometric coecient of the ith reaction. Both i and i are to be interpreted as pure dimensionless numbers. In Equation (4.10), one has N = 3 species. One might take 1 = H2 , 2 = O2 , and 3 = H2 O. The reaction is written in more general form as (2)1 + (1)2 + (0)3 (0)1 + (0)2 + (2)3 , (2)H2 + (1)O2 + (0)H2 O (0)H2 + (0)O2 + (2)H2 O. Here, one has
1 = 2, 2 = 1, 3 = 0, 1 = 0, 2 = 0, 3 = 2.
(4.12) (4.13)
(4.14) (4.15) (4.16)
It is common and useful to dene another pure dimensionless number, the net stoichiometric coecients for species i, i . Here i represents the net production of number if the reaction goes forward. It is given by i = i i . For the reaction 2H2 + O2 2H2 O, one has
1 = 1 1 = 0 2 = 2, 2 = 2 2 = 0 1 = 1, 3 = 3 3 = 2 0 = 2.
(4.17)
(4.18) (4.19) (4.20)
With these denitions, it is possible to summarize a chemical reaction as

N
i i = 0.
i=1
(4.21)
144
In vector notation, one would say T = 0. For the reaction of this section, one might write the non-traditional form 2H2 O2 + 2H2 O = 0. (4.23) (4.22)
It remains to enforce a stoichiometric balance. This is achieved if, for each element, l = 1, . . . , L, one has the following equality:
N
li i = 0,
i=1
l = 1, . . . , L.
(4.24)
That is to say, for each element, the sum of the product of the net species production and the number of elements in the species must be zero. In vector notation, this becomes = 0. (4.25)
One may recall from linear algebra that this demands that lie in the right null space of . Example 4.3
Show stoichiometric balance is achieved for 2H2 O2 + 2H2 O = 0. Here again, the number of elements L = 2, and one can take l = 1 for H and l = 2 for O. Also the number of species N = 3, and one takes i = 1 for H2 , i = 2 for O2 , and i = 3 for H2 O. Then for element 1, H, in species 1, H2 , one has 11 = 2, H in H2 . (4.26) Similarly, one gets 12 13 21 22 23 In matrix form then,
N i=1
= 0, = 2, = 0, = 2, = 1,
H in O2 , H in H2 O, O in H2 , O in O2 , O in H2 O.
(4.27) (4.28) (4.29) (4.30) (4.31)
li i = 0 gives 2 0 0 2 2 1 1 2 = 3 0 0 . (4.32)
This is two equations in three unknowns. Thus, it is formally underconstrained. Certainly the trivial solution 1 = 2 = 3 = 0 will satisfy, but one seeks non-trivial solutions. Assume 3 has a known value 3 = . Then, the system reduces to 2 0 0 2 1 2 = 2 . (4.33)
4.2. STOICHIOMETRY
1 The inversion here is easy, and one nds 1 = , 2 = 2 . Or in vector form,
145
1 2 = 3 =
This simply corresponds to
Again, this amounts to saying the solution vector (1 , 2 , 3 )T lies in the right null space of the coecient matrix li . Here is any real scalar. If one takes = 2, one gets 1 2 2 = 1 , (4.36) 3 2 2H2 O2 + 2H2 O = 0. (4.37)
1 , 2 1 1 , 2 1
(4.34)
R1
(4.35)
which corresponds to the equally valid
If one takes = 4, one still achieves stoichiometric balance, with 1 4 2 = 2 , 3 4 4H2 + 2O2 4H2 O = 0.
(4.38)
(4.39)
In summary, the stoichiometric coecients are non-unique but partially constrained by mass conservation. Which set is chosen is to some extent arbitrary, and often based on traditional conventions from chemistry. But others are equally valid.
There is a small issue with units here, which will be seen to be dicult to reconcile. In practice, it will have little to no importance. In the previous example, one might be tempted to ascribe units of kmoles to i . Later, it will be seen that in classical reaction kinetics, i is best interpreted as a pure dimensionless number, consistent with the denition of this section. So in the context of the previous example, one would then take to be dimensionless as well, which is perfectly acceptable for the example. In later problems, it will be more useful to give the units of kmoles. Note that multiplication of by any scalar, e.g. kmole/(6.02 1026 ), still yields an acceptable result. Example 4.4
Balance an equation for hypothesized ethane combustion
1 C2 H6 + 2 O2 3 CO2 + 4 H2 O.
(4.40)
146

One could also say in terms of the net stoichiometric coecients 1 C2 H6 + 2 O2 + 3 CO2 + 4 H2 O = 0. (4.41)
Here one takes 1 = C2 H6 , 2 = O2 , 3 CO2 , 4 = H2 O. So there are N = 4 species. There are also L = 3 elements: l = 1 : C, l = 2 : H, l = 3 : O. One then has 11 12 13 14 21 22 23 24 31 32 33 34
N i=1
= = = = = = = = = = = =
2, 0, 1, 0, 6, 0, 0, 2, 0, 2, 2, 1,
C in C2 H6 , C in O2 , C in CO2 , C in H2 O, H in C2 H6 , H in O2 , H in CO2 , H in H2 O, O in C2 H6 , O in O2 , O in CO2 , O in H2 O,
(4.42) (4.43) (4.44) (4.45) (4.46) (4.47) (4.48) (4.49) (4.50) (4.51) (4.52) (4.53) (4.54)
So the stoichiometry equation,
li i = 0, is given by 1 1 0 0 0 2 2 = 0. 3 2 1 0 4
2 0 6 0 0 2
(4.55)
Here, there are three equations in four unknowns, so the system is underconstrained. There are many ways to address this problem. Here, choose the robust way of casting the system into row echelon form. This is easily achieved by Gaussian elimination. Row echelon form seeks to have lots of zeros in the lower left part of the matrix. The lower left corner has a zero already, so that is useful. Now, multiply the top equation by 3 and subtract the result from the second to get 1 2 0 1 0 0 0 0 3 2 2 = 0 . (4.56) 3 0 2 2 1 0 4 Next switch the last two equations to get 2 0 0 2 0 0 1 1 0 0 2 1 2 = 0. 3 3 2 0 4
(4.57)
Now divide the rst by 2, the second by 2 and the third by 3 to get unity in the diagonal: 1 0 0 1 0 1 2 1 2 0 1 1 = 0. 2 3 0 0 0 1 2 3 4 CC BY-NC-ND. 12 December 2011, J. M. Powers.
(4.58)
4.2. STOICHIOMETRY
147
So-called basic variables have non-zero coecients on the diagonal, so one can take the basic variables to be 1 , 2 , and 3 . The remaining variables are free variables. Here one takes the free variable to be 4 . So set 4 = , and rewrite the system as 1 1 0 2 1 0 0 1 1 2 = 1 . (4.59) 2 2 0 0 1 3 3 Solving, one nds 1 1 1 3 3 7 2 6 7 = 2 = 26 , 3 3 3 4 1
R1 .
(4.60)
Again one nds a non-unique solution in the right null space of . If one chooses = 6, then one gets 1 2 2 7 (4.61) = , 3 4 4 6 which corresponds to the stoichiometrically balanced reaction 2C2 H6 + 7O2 4CO2 + 6H2 O. In this example, is dimensionless. (4.62)
Example 4.5
Consider stoichiometric balance for a propane oxidation reaction which may produce carbon monoxide and hydroxyl in addition to carbon dioxide and water. The hypothesized reaction takes the form
1 C3 H8 + 2 O2 3 CO2 + 4 CO + 5 H2 O + 6 OH.
(4.63)
In terms of net stoichiometric coecients, this becomes 1 C3 H8 + 2 O2 + 3 CO2 + 4 CO + 5 H2 O + 6 OH = 0. There are N = 6 species and L = 3 elements. One then has 11 12 13 14 15 16 21 22 = = = = = = = = 3, 0, 1, 1, 0, 0, 8, 0, C in C3 H8 , C in O2 , C in CO2 , C in CO, C in H2 O, C in OH, H in C3 H8 , H in O2 , (4.65) (4.66) (4.67) (4.68) (4.69) (4.70) (4.71) (4.72) (4.64)
148

23 24 25 26 31 32 33 34 35 36 = = = = = = = = = = 0, 0, 2, 1, 0, 2, 2, 1, 1, 1, H in CO2 , H in CO, H in H2 O, H in OH, O in C3 H8 , O in O2 , O in CO2 , O in CO, O in H2 O, O in OH. 1 2 0 0 1 3 = 0. 4 1 0 5 6 (4.73) (4.74) (4.75) (4.76) (4.77) (4.78) (4.79) (4.80) (4.81) (4.82)
The equation = 0, then becomes 3 8 0 0 1 0 0 2 2 1 0 0 2 1 1
(4.83)
8 Multiplying the rst equation by 3 and adding it to the second gives 1 2 3 0 1 1 0 0 0 0 0 8 8 2 1 3 = 0 . 3 3 4 0 2 2 1 1 1 0 5 6
(4.84)
Trading the second and third rows gives
3 0 0
0 1 2 2 0 8 3
1 1 8 3
0 1 2
Take basic variables to be 1 , and 6 = 3 , and get 1 0 0
8 Dividing the rst row by 3, the second by 2 and the third by 3 gives 1 2 0 0 1 0 1 1 0 3 3 1 1 3 0 1 1 1 = 0. 2 2 2 4 3 0 0 1 1 4 3 0 8 5 6
1 2 0 0 1 3 = 0. 4 1 0 5 6
(4.85)
(4.86)
2 , and 3 and free variables to be 4 , 5 , and 6 . So set 4 = 1 , 5 = 2 , 0 1 0 1 3 1 2 3 1 1 2 = 2 2 2 . 3 3 1 3 1 + 4 2 + 8 3

1 3
(4.87)
4.2. STOICHIOMETRY
Solving, one nds 1 2 3 =
1 8 (22 3 ) 1 . 8 (41 102 73 ) 1 8 (81 + 62 + 33 )
149
(4.88)
1 Here, one nds three independent vectors in the right null space. To simplify the notation, take 1 = 8 , 2 3 2 = 8 , and 3 = 8 . Then, 1 0 2 1 2 4 10 7 3 1 8 + 2 6 + 3 3 . (4.90) = 4 8 0 0 5 0 8 0 6 0 0 8
For all the coecients, one then has 1 1 0 2 1 8 (22 3 ) 1 2 4 10 7 8 (41 102 73 ) 1 3 8 (81 + 62 + 33 ) 1 8 2 6 3 3 = = + + . 1 4 8 8 8 0 8 0 2 5 0 8 0 3 6 0 0 8
(4.89)
The most general reaction that can achieve a stoichiometric balance is given by
(22 3 )C3 H8 + (41 102 73 )O2 + (81 + 62 + 33 )CO2 + 81 CO + 82 H2 O + 83 OH = 0. (4.91) Rearranging, one gets (22 + 3 )C3 H8 + (41 + 102 + 73 )O2 (81 + 62 + 33 )CO2 + 81 CO + 82 H2 O + 83 OH. (4.92) This will be balanced for all 1 , 2 , and 3 . The values that are actually achieved in practice depend on the thermodynamics of the problem. Stoichiometry only provides some limitations. A slightly more familiar form is found by taking 2 = 1/2 and rearranging, giving (1 + 3 ) C3 H8 + (5 41 + 73 ) O2 (3 81 + 33 ) CO2 + 4 H2 O + 81 CO + 83 OH. (4.93)
One notes that often the production of CO and OH will be small. If there is no production of CO or OH, 1 = 3 = 0 and one recovers the familiar balance of C3 H8 + 5 O2 3 CO2 + 4 H2 O. (4.94)
One also notes that stoichiometry alone admits unusual solutions. For instance, if 1 = 100 2 = 1/2, and 3 = 1, one has 2 C3 H8 + 794 CO2 388 O2 + 4 H2 O + 800 CO + 8 OH. (4.95)
This reaction is certainly admitted by stoichiometry but is not observed in nature. To determine precisely which of the innitely many possible nal states are realized requires a consideration of the equilibrium condition N i i = 0. i=1 Looked at in another way, we can think of three independent classes of reactions admitted by the stoichiometry, one for each of the linearly independent null space vectors. Taking rst 1 = 1/4, 2 = 0, 3 = 0, one gets, after rearrangement 2CO + O2 2CO2 , (4.96)
150

as one class of reaction admitted by stoichiometry. Taking next 1 = 0, 2 = 1/2, 3 = 0, one gets C3 H8 + 5O2 3CO2 + 4H2 O, (4.97)
as the second class admitted by stoichiometry. The third class is given by taking 1 = 0, 2 = 0, 3 = 1, and is C3 H8 + 7O2 3CO2 + 8OH. (4.98) In this example, both and are dimensionless.
In general, one can expect to nd the stoichiometric coecients for N species composed of L elements to be of the following form:
N L
i =
k=1
Dik k ,
i = 1, . . . , N.
(4.99)
Here Dik is a dimensionless component of a full rank matrix of dimension N (N L) and rank N L, and k is a dimensionless component of a vector of parameters of length N L. The matrix whose components are Dik are constructed by populating its columns with vectors which lie in the right null space of li . Note that multiplication of k by any constant gives another set of i , and mass conservation for each element is still satised.
4.2.2
Fuel-air mixtures
In combustion with air, one often models air as a simple mixture of diatomic oxygen and inert diatomic nitrogen in the air:
air (O2 + 3.76N2 ).
(4.100)
The air-fuel ratio, A and its reciprocal, the fuel-air ratio, F , can be dened on a mass and mole basis. nair mair , Amole = . (4.101) Amass = mf uel nf uel Via the molecular masses, one has Amass = mair nair Mair Mair = = Amole . mf uel nf uel Mf uel Mf uel (4.102)
If there is not enough air to burn all the fuel, the mixture is said to be rich. If there is excess air, the mixture is said to be lean. If there is just enough, the mixture is said to be stoichiometric. The equivalence ratio, , is dened as the actual fuel-air ratio scaled by the stoichiometric fuel-air ratio: Fstoichiometric Factual = Astoichiometric Aactual (4.103)
4.2. STOICHIOMETRY
151
The ratio is the same whether F s are taken on a mass or mole basis, because the ratio of molecular masses cancel. Example 4.6
Calculate the stoichiometry of the combustion of methane with air with an equivalence ratio of = 0.5. If the pressure is 0.1 M P a, nd the dew point of the products. First calculate the coecients for stoichiometric combustion:
1 CH4 + 2 (O2 + 3.76N2 ) 3 CO2 + 4 H2 O + 5 N2 .
(4.104) (4.105)
Or 1 CH4 + 2 O2 + 3 CO2 + 4 H2 O + (5 + 3.762 )N2 = 0. Here one has N = 5 species and L = 4 elements. Adopting a slightly more intuitive procedure for variety, one writes a conservation equation for each element to get 1 + 3 41 + 24 22 + 23 + 4 3.762 + 5 In matrix form this becomes 1 4 0 0 0 0 2 3.76 1 0 2 0 0 2 1 0 1 0 0 2 0 0 3 = . 0 0 4 1 0 5 = 0, = 0, = 0, = 0, C H O N. (4.106) (4.107) (4.108) (4.109)
(4.110)
Now, one might expect to have one free variable, since one has ve unknowns in four equations. While casting the equation in row echelon form is guaranteed to yield a proper solution, one can often use intuition to get a solution more rapidly. One certainly expects that CH4 will need to be present for the reaction to take place. One might also expect to nd an answer if there is one mole of CH4 . So take 1 = 1. Realize that one could also get a physically valid answer by assuming 1 to be equal to any scalar. With 1 = 1, one gets 0 1 0 0 1 2 0 0 2 0 3 4 (4.111) = . 2 2 1 0 4 0 3.76 0 0 1 0 5 One easily nds the unique inverse does exist, and that the solution is 2 2 3 1 = . 4 2 5 7.52
(4.112)
If there had been more than one free variable, the four by four matrix would have been singular, and no unique inverse would have existed. In any case, the reaction under stoichiometric conditions is CH4 2O2 + CO2 + 2H2 O + (7.52 + (3.76)(2))N2 = 0, CH4 + 2(O2 + 3.76N2) CO2 + 2H2 O + 7.52N2. (4.113) (4.114)
152

For the stoichiometric reaction, the fuel-air ratio on a mole basis is Fstoichiometric = Now = 0.5, so Factual = Fstoichiometric , = (0.5)(0.1050), = 0.0525. (4.116) (4.117) (4.118) 1 = 0.1050. 2 + 2(3.76) (4.115)
By inspection, one can write the actual reaction equation as CH4 + 4(O2 + 3.76N2 ) CO2 + 2H2 O + 2O2 + 15.04N2. Check: Factual = 1 = 0.0525. 4 + 4(3.76) (4.120) (4.119)
For the dew point of the products, one needs the partial pressure of H2 O. The mole fraction of H2 O is 2 yH2 O = = 0.0499 (4.121) 1 + 2 + 2 + 15.04 So the partial pressure of H2 O is PH2 O = yH2 O P = 0.0499(100 kP a) = 4.99 kP a. (4.122)
From the steam tables, the saturation temperature at this pressure is Tsat = Tdew point = 32.88 C. If the products cool to this temperature in an exhaust device, the water could condense in the apparatus.
4.3
First law analysis of reacting systems
One can easily use the rst law to learn much about chemically reacting systems.
4.3.1
Enthalpy of formation
The enthalpy of formation is the enthalpy that is required to form a molecule from combining its constituents at P = 0.1 MP a and T = 298 K. Consider the reaction (taken here to be irreversible) C + O2 CO2 . (4.123)
4.3. FIRST LAW ANALYSIS OF REACTING SYSTEMS
153
In order to maintain the process at constant temperature, it is found that heat transfer to the volume is necessary. For the steady constant pressure process, one has E2 E1 = Q12 W12 ,
2
(4.124) (4.125) (4.126) (4.127) (4.128) (4.129) (4.130)
= Q12 Q12 Q12 = = = = =
P dV,
1
Q12 P (V2 V1 ), E2 E1 + P (V2 V1 ), H2 H1 , Hproducts Hreactants ,

products
ni hi
ni hi .
reactants
In this reaction, one measures that Q12 = 393522 kJ for the reaction of 1 kmole of C and O2 . That is the reaction liberates such energy to the environment. So measuring the heat transfer can give a measure of the enthalpy dierence between reactants and products. Assign a value of enthalpy zero to elements in their standard state at the reference state. kJ Thus, C and O2 both have enthalpies of 0 kmole at T = 298 K, P = 0.1 MP a. This enthalpy is designated, for species i, o o (4.131) hf,i = hTo ,i , and is called the enthalpy of formation. So the energy balance for the products and reactants, here both at the standard state, becomes Q12 = nCO2 hf,CO2 nC hf,C nO2 hf,O2 , 393522 kJ = (1 kmole)hf,CO2 (1 kmole) 0
o o o o o
(4.132) kJ kmole (1 kmole) 0 kJ . kmole (4.133)
kJ Thus, the enthalpy of formation of CO2 is hf,CO2 = 393522 kmole , since the reaction involved creation of 1 kmole of CO2 . Often values of enthalpy are tabulated in the forms of enthalpy dierences hi . These are dened such that
hi = hf,i + (hi hf,i ),

=hi
(4.134) (4.135)
o hf,i
+ hi .
Lastly, one notes for an ideal gas that the enthalpy is a function of temperature only, and so does not depend on the reference pressure; hence hi = hi ,
o
hi = hi ,
if ideal gas.
(4.136)
154
Example 4.7
(Borgnakke and Sonntag, p. 575). Consider the following irreversible reaction in a steady state, steady ow process conned to the standard state of P = 0.1 M P a, T = 298 K: CH4 + 2O2 CO2 + 2H2 O(). The rst law holds that Qcv =
products
(4.137)
ni hi
ni hi .
reactants
(4.138)
All components are at their reference states. Table A.10 gives properties, and one nds Qcv = = nCO2 hCO2 + nH2 O hH2 O nCH4 hCH4 nO2 hO2 , kJ kJ (1 kmole) 393522 + (2 kmole) 285830 kmole kmole kJ kJ (2 kmole) 0 , (1 kmole) 74873 kmole kmole 890309 kJ. (4.139)
(4.140) (4.141)
A more detailed analysis is required in the likely case in which the system is not at the reference state. Example 4.8
(adopted from Moran and Shapiro6 ) A mixture of 1 kmole of gaseous methane and 2 kmole of oxygen initially at 298 K and 101.325 kP a burns completely in a closed, rigid, container. Heat transfer occurs until the nal temperature is 900 K. Find the heat transfer and the nal pressure. The combustion is stoichiometric. Assume that no small concentration species are generated. The global reaction is given by CH4 + 2O2 CO2 + 2H2 O. (4.142) The rst law analysis for the closed system is slightly dierent: E2 E1 = Q12 W12 . Since the process is isochoric, W12 = 0. So Q12 = = = = =
6
(4.143)
E2 E1 ,
(4.144)
hCO2 + 2hH2 O hCH4 2hO2 3R(T2 T1 ), (4.147) o o o o (hCO2 ,f + hCO2 ) + 2(hH2 O,f + hH2 O ) (hCH4 ,f + hCH4 ) 2(hO2 ,f + hO2 )
(4.145) nCO2 eCO2 + nH2 O eH2 O nCH4 eCH4 nO2 eO2 , nCO2 (hCO2 RT2 ) + nH2 O (hH2 O RT2 ) nCH4 (hCH4 RT1 ) nO2 (hO2 RT1 ), (4.146)
Moran, M. J., and Shapiro, H. N., 2003, Fundamentals of Engineering Thermodynamics, Fifth Edition, John Wiley, New York. p. 619. CC BY-NC-ND. 12 December 2011, J. M. Powers.

3R(T2 T1 ),
155
(4.148) (4.149) (4.150)
= =
(393522 + 28030) + 2(241826 + 21937) (74873 + 0) 2(0 + 0) 3(8.314)(900 298), 745412 kJ.
For the pressures, one has P1 V1 V1 = = = = Now V2 = V1 , so P2 = = = 306.0 kP a. (nCO2 + nH2 O )RT2 , V2 (1 kmole + 2 kmole) 8.314 73.36 m3
kJ kg K
(nCH4 + nO2 )RT1 , (nCH4 + nO2 )RT1 , P1 (1 kmole + 2 kmole) 8.314 101.325 kP a 73.36 m3 .
kJ kg K
(4.151) (4.152) (298 K) , (4.153) (4.154)
(4.155) (900 K) , (4.156) (4.157)
The pressure increased in the reaction. This is entirely attributable to the temperature rise, as the number of moles remained constant here.
4.3.2
Enthalpy and internal energy of combustion
The enthalpy of combustion is the dierence between the enthalpy of products and reactants when complete combustion occurs at a given pressure and temperature. It is also known as the heating value or the heat of reaction. The internal energy of combustion is related and is the dierence between the internal energy of products and reactants when complete combustion occurs at a given volume and temperature. The term higher heating value refers to the energy of combustion when liquid water is in the products. Lower heating value refers to the energy of combustion when water vapor is in the product.
4.3.3
Adiabatic ame temperature in isochoric stoichiometric systems
The adiabatic ame temperature refers to the temperature which is achieved when a fuel and oxidizer are combined with no loss of work or heat energy. Thus, it must occur in a closed, insulated, xed volume. It is generally the highest temperature that one can expect to
156
achieve in a combustion process. It generally requires an iterative solution. Of all mixtures, stoichiometric mixtures will yield the highest adiabatic ame temperatures because there is no need to heat the excess fuel or oxidizer. Here four examples will be presented to illustrate the following points. The adiabatic ame temperature can be well over 5000 K for seemingly ordinary mixtures. Dilution of the mixture with an inert diluent lowers the adiabatic ame temperature. The same eect would happen in rich and lean mixtures. Preheating the mixture, such as one might nd in the compression stroke of an engine, increases the adiabatic ame temperature. Consideration of the presence of minor species lowers the adiabatic ame temperature. 4.3.3.1 Undiluted, cold mixture
Example 4.9
A closed, xed, adiabatic volume contains a stoichiometric mixture of 2 kmole of H2 and 1 kmole of O2 at 100 kP a and 298 K. Find the adiabatic ame temperature assuming the irreversible reaction 2H2 + O2 2H2 O. The volume is given by V = = = The rst law gives E2 E1 E2 E1 = = Q W, 0, 0, 0, 0, 0, 0, 0, 0, 0. (4.162) (4.163) (4.164) (4.165) (4.166) (4.167) (4.168) (4.169) (4.170) (4.171) (nH2 + nO2 )RT1 , P1 (2 kmole + 1 kmole) 8.314 100 kP a 74.33 m3 . (4.159)
kJ kmole K
(4.158)
(298 K)
(4.160) (4.161)
nH2 O eH2 O nH2 eH2 nO2 eO2 = (hH2 RT1 ) nO2 (hO2 RT1 ) = nH2 O (hH2 O RT2 ) nH2 2hH2 O 2 hH2 hO2 +R(2T2 + 3T1 ) =
=0 =0
2hH2 O + (8.314) ((2) T2 + (3) (298)) =

o hf,H2 O
241826 + hH2 O 8.314T2 + 3716.4 = 238110 + hH2 O 8.314T2 = CC BY-NC-ND. 12 December 2011, J. M. Powers.
+ hH2 O 8.314T2 + 3716.4 =
hH2 O 8.314T2 + 3716.4 =
157
At this point, one begins an iteration process, guessing a value of T2 and an associated hH2 O . When T2 is guessed at 5600 K, the left side becomes 6507.04. When T2 is guessed at 6000 K, the left side becomes 14301.4. Interpolate then to arrive at T2 = 5725 K. (4.172)
This is an extremely high temperature. At such temperatures, in fact, one can expect other species to co-exist in the equilibrium state in large quantities. These may include H, OH, O, HO2 , and H2 O2 , among others. The nal pressure is given by P2 nH2 O RT2 , V kJ (2 kmole) 8.314 kmole = 74.33 m3 = 1280.71 kP a. = (4.173)
K
(5725 K)
(4.174) (4.175)
The nal concentration of H2 O is H2 O = kmole 2 kmole = 2.69 102 . 74.33 m3 m3 (4.176)
4.3.3.2
Dilute, cold mixture
Example 4.10
Consider a variant on the previous example in which the mixture is diluted with an inert, taken here to be N2 . A closed, xed, adiabatic volume contains a stoichiometric mixture of 2 kmole of H2 , 1 kmole of O2 , and 8 kmole of N2 at 100 kP a and 298 K. Find the adiabatic ame temperature and the nal pressure, assuming the irreversible reaction 2H2 + O2 + 8N2 2H2 O + 8N2 . The volume is given by V = = = The rst law gives E2 E1 = Q W, 0, 0, (nH2 + nO2 + nN2 )RT1 , P1 (2 kmole + 1 kmole + 8 kmole) 8.314 100 kP a 3 272.533 m . (4.178)
kJ kmole K
(4.177)
(298 K)
(4.179) (4.180)
nH2 O eH2 O nH2 eH2 nO2 eO2
E2 E1 = + nN2 (eN2 2 eN2 1 ) =
158

= 0, 0, 0, 0, 0, 0, 0,
nH2 O (hH2 O RT2 ) nH2 (hH2 RT1 ) nO2 (hO2 RT1 ) + nN2 ((hN2 2 RT2 ) (hN2 1 RT1 ))
=0 =0 =hN2 =0
2hH2 O 2 hH2 hO2 +R(10T2 + 11T1 ) + 8( hN22 hN21 ) = 2hH2 O + (8.314) (10T2 + (11)(298)) + 8hN2 2
o 2hf,H2 O
= = = = =
2(241826) + 2hH2 O 83.14T2 + 27253.3 + 8hN2 2 456399 + 2hH2 O 83.14T2 + 8hN2 2
+ 2hH2 O 83.14T2 + 27253.3 + 8hN22
2hH2 O 83.14T2 + 27253.3 + 8hN2 2
At this point, one begins an iteration process, guessing a value of T2 and an associated hH2 O . When T2 is guessed at 2000 K, the left side becomes 28006.7. When T2 is guessed at 2200 K, the left side becomes 33895.3. Interpolate then to arrive at T2 = 2090.5 K. (4.181)
The inert diluent signicantly lowers the adiabatic ame temperature. This is because the N2 serves as a heat sink for the energy of reaction. If the mixture were at non-stoichiometric conditions, the excess species would also serve as a heat sink, and the adiabatic ame temperature would be lower than that of the stoichiometric mixture. The nal pressure is given by P2 = = = (nH2 O + nN2 )RT2 , V kJ (2 kmole + 8 kmole) 8.314 kmole 272.533 m3 637.74 kP a. (4.182)
K
(2090.5 K)
(4.183) (4.184)
The nal concentrations of H2 O and N2 are H2 O N2 = = 2 kmole kmole = 7.34 103 , 272.533 m3 m3 8 kmole kmole = 2.94 102 . 272.533 m3 m3 (4.185) (4.186)
4.3.3.3
Dilute, preheated mixture
Example 4.11
Consider a variant on the previous example in which the diluted mixture is preheated to 1000 K. One can achieve this via an isentropic compression of the cold mixture, such as might occur in an engine. To simplify the analysis here, the temperature of the mixture will be increased, while the pressure will be maintained. A closed, xed, adiabatic volume contains a stoichiometric mixture of 2 kmole of H2 , CC BY-NC-ND. 12 December 2011, J. M. Powers.
159
1 kmole of O2 , and 8 kmole of N2 at 100 kP a and 1000 K. Find the adiabatic ame temperature and the nal pressure, assuming the irreversible reaction 2H2 + O2 + 8N2 2H2 O + 8N2 . The volume is given by V = = = The rst law gives E2 E1 E2 E1 = = Q W, 0, 0, 0, 0, 0, 0, (nH2 + nO2 + nN2 )RT1 , P1 (2 kmole + 1 kmole + 8 kmole) 8.314 100 kP a 914.54 m3 . (4.188)
kJ kmole K
(4.187)
(1000 K)
(4.189) (4.190)
nH2 O eH2 O nH2 eH2 nO2 eO2 + nN2 (eN2 2 eN2 1 ) = nH2 O (hH2 O RT2 ) nH2 (hH2 RT1 ) nO2 (hO2 RT1 ) + nN2 ((hN2 2 RT2 ) (hN2 1 RT1 )) = 2(241826 + hH2 O ) 2(20663) 22703 + (8.314) (10T2 + (11)(1000)) + 8hN22 8(21463) = 2hH2 O 83.14T2 627931 + 8hN2 2 = 2hH2 O 2hH2 hO2 + R(10T2 + 11T1 ) + 8(hN2 2 hN2 1 ) =
At this point, one begins an iteration process, guessing a value of T2 and an associated hH2 O . When T2 is guessed at 2600 K, the left side becomes 11351. When T2 is guessed at 2800 K, the left side becomes 52787. Interpolate then to arrive at T2 = 2635.4 K. (4.191)
The preheating raised the adiabatic ame temperature. Note that the preheating was by 1000 298 = 702 K. The new adiabatic ame temperature is only 2635.4 2090.5 = 544.9 K greater. The nal pressure is given by P2 = = = (nH2 O + nN2 )RT2 , V kJ (2 kmole + 8 kmole) 8.314 kmole 914.54 m3 239.58 kP a. (4.192)
K
(2635.4 K)
(4.193) (4.194)
The nal concentrations of H2 O and N2 are H2 O N 2 = = kmole 2 kmole = 2.19 103 , 914.54 m3 m3 8 kmole kmole = 8.75 103 . 914.54 m3 m3 (4.195) (4.196)
160 4.3.3.4
CHAPTER 4. THERMOCHEMISTRY OF A SINGLE REACTION Dilute, preheated mixture with minor species
Example 4.12
Consider a variant on the previous example. Here allow for minor species to be present at equilibrium. A closed, xed, adiabatic volume contains a stoichiometric mixture of 2 kmole of H2 , 1 kmole of O2 , and 8 kmole of N2 at 100 kP a and 1000 K. Find the adiabatic ame temperature and the nal pressure, assuming reversible reactions. Here, the details of the analysis are postponed, but the result is given which is the consequence of a calculation involving detailed reactions rates. One can also solve an optimization problem to minimize the Gibbs free energy of a wide variety of products to get the same answer. In this case, the equilibrium temperature and pressure are found to be T = 2484.8 K, P = 227.89 kP a. kmole , m3 kmole 1.9 105 , m3 kmole 5.7 106 , m3 kmole , 3.6 105 m3 kmole , 5.9 105 m3 kmole 2.0 103 , m3 kmole 1.1 108 , m3 kmole 1.2 109 , m3 kmole , 1.7 109 m3 kmole 3.7 1010 , m3 kmole 1.5 1010 , m3 kmole 3.1 1010 , m3 kmole , 1.0 1010 m3 kmole , 3.1 106 m3 kmole 5.3 109 , m3 kmole 2.6 109 , m3 kmole , 1.7 109 m3 kmole . 8.7 103 m3 (4.197) Equilibrium species concentrations are found to be minor product minor product minor product minor product minor product major product trace product trace product trace product trace product trace product trace product trace product minor product trace product trace product trace product major product H2 H O O2 OH H2 O HO2 H2 O2 N N H N H2 N H3 N N H N O N O2 N 2 O HN O N 2 = 1.3 104 = = = = = = = = = = = = = = = = = (4.198) (4.199) (4.200) (4.201) (4.202) (4.203) (4.204) (4.205) (4.206) (4.207) (4.208) (4.209) (4.210) (4.211) (4.212) (4.213) (4.214) (4.215)
4.4. CHEMICAL EQUILIBRIUM
161
Note that the concentrations of the major products went down when the minor species were considered. The adiabatic ame temperature also went down by a signicant amount: 2635 2484.8 = 150.2 K. Some thermal energy was necessary to break the bonds which induce the presence of minor species.
4.4
Chemical equilibrium
Often reactions are not simply unidirectional, as alluded to in the previous example. The reverse reaction, especially at high temperature, can be important. Consider the four species reaction
1 1 + 2 2 3 3 + 4 4
(4.216)
In terms of the net stoichiometric coecients, this becomes 1 1 + 2 2 + 3 3 + 4 4 = 0. (4.217)
One can dene a variable , the reaction progress. Take the dimension of to be kmoles. When t = 0, one takes = 0. Now as the reaction goes forward, one takes d > 0. And a forward reaction will decrease the number of moles of 1 and 2 while increasing the number of moles of 3 and 4 . This will occur in ratios dictated by the stoichiometric coecients of the problem: dn1 dn2 dn3 dn4 = = = =
1 d, 2 d, +3 d, +4 d.
(4.218) (4.219) (4.220) (4.221)
Note that if ni is taken to have units of kmoles, i , and i are taken as dimensionless, then must have units of kmoles. In terms of the net stoichiometric coecients, one has dn1 dn2 dn3 dn4 = = = = 1 d, 2 d, 3 d, 4 d. (4.222) (4.223) (4.224) (4.225)
Again, for arguments sake, assume that at t = 0, one has n1 |t=0 n2 |t=0 n3 |t=0 n4 |t=0 = = = = n1o , n2o , n3o , n4o . (4.226) (4.227) (4.228) (4.229)
162
Then after integrating, one nds n1 n2 n3 n4 = = = = 1 + n1o , 2 + n2o , 3 + n3o , 4 + n4o . (4.230) (4.231) (4.232) (4.233)
One can also eliminate the parameter in a variety of fashions and parameterize the reaction one of the species mole numbers. Choosing, for example, n1 as a parameter, one gets n1 n1o = . (4.234) 1 Eliminating then one nds all other mole numbers in terms of n1 : n2 = 2 n3 n4 Written another way, one has n2 n2o n3 n3o n4 n4o n1 n1o = = = = . 1 2 3 4 For an N-species reaction,
N i=1
n1 n1o + n2o , 1 n1 n1o = 3 + n3o , 1 n1 n1o + n4o . = 4 1
(4.235) (4.236) (4.237)
(4.238)
i i = 0, one can generalize to say dni = = i d, ni = i + nio , ni nio = . i (4.239) (4.240) (4.241)
Note that
dni = i . (4.242) d Now, from the previous chapter, one manifestation of the second law is Eq. (3.426):
N
dG|T,P =
i=1
i dni 0.
(4.243)
Now, one can eliminate dni in Eq. (4.243) by use of Eq. (4.239) to get
N
dG|T,P =
i=1
i i d 0,
(4.244)
4.4. CHEMICAL EQUILIBRIUM G

N
163
=
T,P i=1
i i 0,
(4.245) (4.246)
= 0. Then for the reaction to go forward, one must require that the anity be positive: 0.
(4.247)
One also knows from the previous chapter that the irreversibility takes the form of Eq. (3.415): 1 T
N
1 d T 1 d T dt
i=1 N
i dni 0, i i 0, i i 0. i i , Eq. (4.250) can be written as
(4.248) (4.249) (4.250)
i=1 N
i=1 N i=1
In terms of the chemical anity, =
1 d 0. T dt
(4.251)
Now one straightforward, albeit na way to guarantee positive semi-deniteness of the ve, irreversibility and thus satisfaction of the second law is to construct the chemical kinetic rate equation so that d = k dt
N
i i = k,
i=1
k 0,
provisional, na assumption ve
(4.252)
This provisional assumption of convenience will be supplanted later by a form which agrees well with experiment. Here k is a positive semi-denite scalar. In general, it is a function of temperature, k = k(T ), so that reactions proceed rapidly at high temperature and slowly at low temperature. Then certainly the reaction progress variable will cease to change when the equilibrium condition
N
i i = 0,
i=1
(4.253)
is met. This is equivalent to requiring = 0. (4.254) Now, while Eq. (4.253) is the most compact form of the equilibrium condition, it is not the most commonly used form. One can perform the following analysis to obtain the form
164
in most common usage. Start by equating the chemical potential with the Gibbs free energy per unit mole for each species i: i = g i . Then employ the denition of Gibbs free energy for an ideal gas, and carry out a set of operations:
N
g i i = 0,
i=1 N
at equilibrium at equilibrium
(4.255) (4.256)
i=1
(hi T si )i = 0.
For the ideal gas, one can substitute for hi (T ) and si (T, P ) and write the equilibrium condition as N T T o yi P c P i (T ) o hT ,i + cP i (T ) dT T sTo ,i + dT R ln i = 0, o Po T To To i=1 o o =sT,i =hT,i
=hT,i =hT,i
o
=sT,i
(4.257)
Now writing the equilibrium condition in terms of the enthalpies and entropies referred to the standard pressure, one gets
N
i=1
hT,i T
so R ln T,i
N o hT,i
yi P Po T so T,i
N
i = 0,
N
(4.258) RT i ln
i=1 N
i=1
i =
yi P Po
i
(4.259)
=g o =o T,i T,i
g o i T,i
i=1 Go
= RT
i=1 N
ln
yi P Po
i
(4.260)
Go = RT
ln
i=1 N
yi P Po yi P Po
i
,
i
(4.261) , (4.262) (4.263)
= ln
i=1
exp
G RT
=
i=1
yi P Po
KP
4.4. CHEMICAL EQUILIBRIUM

N
165 yi P Po
PN
i=1
KP =
i=1
,
i n yi i , i=1
(4.264) (4.265)
KP = KP =
P Po
N
i=1
Pi Po
, at equilibrium. (4.266)
Here KP is what is known as the pressure-based equilibrium constant. It is dimensionless. Despite its name, it is not a constant. It is dened in terms of thermodynamic properties, and for the ideal gas is a function of T only: KP exp Go RT , generally valid. (4.267)
Only at equilibrium does the property KP also equal the product of the partial pressures as in Eq. (4.266). The subscript P for pressure comes about because it is also related to the product of the ratio of the partial pressure to the reference pressure raised to the net stoichiometric coecients. Also, the net change in Gibbs free energy of the reaction at the reference pressure, Go , which is a function of T only, has been dened as
N
Go
g o i . T,i
i=1
(4.268)
The term Go has units of kJ/kmole; it traditionally does not get an overbar. If Go > 0, one has 0 < KP < 1, and reactants are favored over products. If Go < 0, one gets KP > 1, and products are favored over reactants. One can also deduce that higher pressures P push the equilibrium in such a fashion that fewer moles are present, all else being equal. One can also dene Go in terms of the chemical anity, referred to the reference pressure, as Go = o . (4.269)
One can also dene another convenient thermodynamic property, which for an ideal gas is a function of T alone, the equilibrium constant Kc : Kc Po RT
PN
i=1
exp
Go RT
generally valid
(4.270)
This property is dimensional, and the units depend on the stoichiometry of the reaction. PN The units of Kc will be (kmole/m3 ) i=1 i . The equilibrium condition, Eq. (4.266), is often written in terms of molar concentrations and Kc . This can be achieved by the operations, valid only at an equilibrium state:
N
KP =
i=1
i RT Po
(4.271)
166
CHAPTER 4. THERMOCHEMISTRY OF A SINGLE REACTION G RT

o
exp Po RT
PN
i=1
= =
RT Po
N
PN
i=1
i N
i i ,
i=1
(4.272) (4.273)
exp
Kc
Go RT
i i ,
i=1 N
Kc =
i=1
i i .
at equilibrium
(4.274)
One must be careful to distinguish between the general denition of Kc as given in Eq. (4.270), and the fact that at equilibrium it is driven to also have the value of product of molar species concentrations, raised to the appropriate stoichiometric power, as given in Eq. (4.274).
4.5
Chemical kinetics of a single isothermal reaction
In the same fashion in ordinary mechanics that an understanding of statics enables an understanding of dynamics, an understanding of chemical equilibrium is necessary to understand to more challenging topic of chemical kinetics. Chemical kinetics describes the time-evolution of systems which may have an initial state far from equilibrium; it typically describes the path of such systems to an equilibrium state. Here gas phase kinetics of ideal gas mixtures that obey Daltons law will be studied. Important topics such as catalysis and solid or liquid reactions will not be considered. Further, this section will be restricted to strictly isothermal systems. This simplies the analysis greatly. It is straightforward to extend the analysis of this system to non-isothermal systems. One must then make further appeal to the energy equation to get an equation for temperature evolution. The general form for evolution of species is taken to be d dt i = i . (4.275)
Multiplying both sides of Eq. (4.275) by molecular mass Mi and using Eq. (2.43) to exchange i for mass fraction Yi then gives the alternate form i Mi dYi = . dt (4.276)
4.5.1
Isochoric systems
Consider the evolution of species concentration in a system which is isothermal, isochoric and spatially homogeneous. The system is undergoing a single chemical reaction involving
4.5. CHEMICAL KINETICS OF A SINGLE ISOTHERMAL REACTION N species of the familiar form of Eq. (4.21):
N
167
i i = 0.
i=1
(4.277)
Because the density is constant for the isochoric system, Eq. (4.275) reduces to di = i . dt (4.278)
Then, experiment, as well as a more fundamental molecular collision theory, shows that the evolution of species concentration i is given by
i
di = i aT exp dt
k(T )
E RT
k k
k=1 forward reaction r
1 1 Kc
reverse reaction
kk , k=1
N
isochoric system
(4.279)
This relation actually holds for isochoric, non-isothermal systems as well, which will not be considered in any detail here. Here some new variables are dened as follows: a: a kinetic rate constant called the collision frequency factor. Its units will depend on the actual reaction and could involve various combinations of length, time, and temperature. It is constructed so that di /dt has units of kmole/m3 /s; this requires it PN to have units of (kmole/m3 )(1 k=1 k ) /s/K . : a dimensionless parameter whose value is set by experiments, sometimes combined with guiding theory, to account for weak temperature dependency of reaction rates. E: the activation energy. It has units of kJ/kmole, though others are often used, and is t by both experiment and fundamental theory to account for the strong temperature dependency of reaction.
Note that in Eq. (4.279) that molar concentrations are raised to the k and k powers. As it does not make sense to raise a physical quantity to a power with units, one traditionally interprets the values of k , k , as well as k to be dimensionless pure numbers. They are also interpreted in a standard fashion: the smallest integer values that actually correspond to the underlying molecular collision which has been modeled. While stoichiometric balance can be achieved by a variety of k values, the kinetic rates are linked to one particular set which is dened by the community. Equation (4.279) is written in such a way that the species concentration production rate increases when
168
The net number of moles generated in the reaction, measured by i increases, The temperature increases; here, the sensitivity may be very high, as one observes in nature, The species concentrations of species involved in the forward reaction increase; this embodies the principle that the collision-based reaction rates are enhanced when there are more molecules to collide, The species concentrations of species involved in the reverse reaction decrease. Here, three intermediate variables which are in common usage have been dened. First one takes the reaction rate to be r aT exp E RT
N
k=1 forward reaction
k(T )
= aT exp
E RT
1 1 Kc 1 Kc
reverse reaction N
kk , k=1
N
(4.280)
k k
k=1
k(T ), Arrhenius rate
forward reaction
reverse reaction
law of mass action
k k . k=1
(4.281)
The reaction rate r has units of kmole/m3 /s. The temperature-dependency of the reaction rate is embodied in k(T ) is dened by what is known as an Arrhenius7 rate law: k(T ) aT exp E RT . (4.282)
This equation was advocated by vant Ho8 in 1884; in 1889 Arrhenius gave a physical justication. The units of k(T ) actually depend on the reaction. This is a weakness of the theory, PN and precludes a clean non-dimensionalization. The units must be (kmole/m3 )(1 k=1 k ) /s. In terms of reaction progress, one can also take r= 1 d . V dt (4.283)
The factor of 1/V is necessary because r has units of molar concentration per time and has units of kmoles. The over-riding importance of the temperature sensitivity is illustrated as part of the next example. The remainder of the expression involving the products of the
7 8
Svante Arrhenius, 1859-1927, Swedish physicist. Jacobus Henricus vant Ho, 1852-1922, Dutch chemist.
4.5. CHEMICAL KINETICS OF A SINGLE ISOTHERMAL REACTION
169
species concentrations is the dening characteristic of systems which obey the law of mass action. Though the history is complex, most attribute the law of mass action to Guldberg9 and Waage10 in 1864.11 Last, the overall molar production rate of species i, is often written as i , dened as i i r. As i is considered to be dimensionless, the units of i must be kmole/m3 /s. Example 4.13
Study the nitrogen dissociation problem considered in an earlier example in which at t = 0 s, 1 kmole of N2 exists at P = 100 kP a and T = 6000 K. Take as before the reaction to be isothermal and isochoric. Consider again the elementary nitrogen dissociation reaction N2 + N2 2N + N2 , which has kinetic rate parameters of a E In SI units, this becomes a E = = 7.0 1021 m3 K 1.6 1m 1000 mole cm3 K 1.6 = 7.0 1018 , mole s 100 cm kmole kmole s J kJ kJ 1000 mole cal 4.186 = 941550 . 224928.4 mole cal 1000 J kmole kmole
3
(4.284)
(4.285)
= 7.0 1021 = 1.6,
cm3 K 1.6 , mole s
(4.286) (4.287) (4.288)
cal = 224928.4 . mole
(4.289) (4.290)
At the initial state, the material is all N2 , so PN2 = P = 100 kP a. The ideal gas law then gives at t=0 P |t=0 N 2
t=0
= = =
PN2 |t=0 = N2 P |t=0 , RT
t=0
RT,
(4.291) (4.292) (4.293) (4.294)
100 kP a , kJ 8.314 kmole K (6000 K) kmole = 2.00465 103 . m3 Thus, the volume, constant for all time in the isochoric process, is V =
9
nN2 |t=0 1 kmole = N2 t=0 2.00465 103
kmole m3
= 4.9884 102 m3 .
(4.295)
Cato Maximilian Guldberg, 1836-1902, Norwegian mathematician and chemist. Peter Waage, 1833-1900, Norwegian chemist. 11 P. Waage and C. M. Guldberg, 1864, Studies Concerning Anity, Forhandlinger: Videnskabs-Selskabet i Christiania, 35. English translation: Journal of Chemical Education, 63(12): 1044-1047.
10
170

Now the stoichiometry of the reaction is such that dnN2 = 1 dnN , 2 1 (nN nN |t=0 ), 2
=0
(4.296) (4.297) (4.298) (4.299) (4.300) (4.301)
(nN2 nN2 |t=0 ) =

=1 kmole
nN nN V N
= = = =
2(1 kmole nN2 ), 1 kmole nN2 , 2 V V 1 kmole N 2 , 2 4.9884 102 m3 kmole N 2 . 2 2.00465 103 m3
Now the general equation for kinetics of a single reaction, Eq. (4.279), reduces for N2 molar concentration to dN2 dt = N2 aT exp E RT (N2 )N2 (N )N

1 ( )N2 (N )N Kc N2
(4.302)
Realizing that N2 = 2, N = 0, N2 = 1, and N = 2, one gets
dN2 dt
aT exp
E RT
2 2 1 N
1 2 N K c N 2
(4.303)
=k(T )
Examine the primary temperature dependency of the reaction k(T ) = = = aT exp E RT , T 1.6 exp 941550
kJ kmole kJ kmole K T
(4.304) , (4.305) (4.306)
7.0 1018
m3 K 1.6 kmole s
8.314
7.0 1018 exp T 1.6
1.1325 105 T
Figure 4.1 gives a plot of k(T ) which shows its very strong dependency on temperature. For this problem, T = 6000 K, so k(6000) = = 7.0 1018 1.1325 105 exp 60001.6 6000 3 m 40071.6 . kmole s
PN
(4.307) (4.308)
Now, the equilibrium constant Kc is needed. Recall Kc = For this system, since Kc = Po RT exp
N i=1
Po RT
i=1
exp
Go RT
(4.309)
i = 1, this reduces to RT , (4.310)
(2go g o 2 ) N N

k(T) (m /kmole/s) 10000 0.0001 1. x 10-12 1. x 10-20 1. x 10-28 1. x 10-36 1000 1500 2000 3000 5000 T (K) 7000 10000
3
171
Figure 4.1: k(T ) for Nitrogen dissociation example.

= = Po RT exp (2(hN T so ) (hN2 T so 2 )) T,N T,N RT
o o
(4.311)
100 (2(597270 (6000)216.926) (205848 (6000)292.984)) exp ,(4.312) (8.314)(6000) (8.314)(6000) kmole = 0.000112112 . (4.313) m3 The dierential equation for N2 evolution is then given by dN2 dt m3 = 40071.6 kmole 2 2 N 1 1 0.000112112
kmole m3
2 2.00465 103 N 2
kmole m3
N2
f (N2 )
(4.314) = f (N2 ). (4.315) The system is at equilibrium when f (N2 ) = 0. This is an algebraic function of N2 only, and can be plotted. Figure 4.2 gives a plot of f (N2 ) and shows that it has three potential equilibrium points. It is seen there are three roots. Solving for the equilibria requires solving 0 m3 = 40071.6 kmole 2 2 N 1 1 0.000112112
kmole m3
2 2.00465 103 N 2
kmole m3
N 2
. (4.316)
The three roots are N 2 = 0 kmole , m3 0.00178121

stable
kmole , m3
0.00225611
kmole . m3
(4.317)
unstable
unstable
By inspection of the topology of Fig. 4.2, the only stable root is 0.00178121 kmole . This root agrees with m3 the equilibrium value found in an earlier example for the same problem conditions. Small perturbations CC BY-NC-ND. 12 December 2011, J. M. Powers.
172
f ( N
) (kmole/m 3 /s) 2
unstable equilibrium
1
0.0005
0.001
0.0015
0.002
0.0025
-1
N2
(kmole/m 3 )
-2
stable, physical equilibrium
Figure 4.2: Forcing function, f (N2 ), which drives changes of N2 as a function of N2 in isothermal, isochoric problem.
from this equilibrium induce the forcing function to supply dynamics which restore the system to its original equilibrium state. Small perturbations from the unstable equilibria induce non-restoring dynamics. For this root, one can then determine that the stable equilibrium value of N = 0.000446882 kmole . m3 One can examine this stability more formally. Dene an equilibrium concentration eq2 such that N f (eq2 ) = 0. N Now perform a Taylor series of f (N2 ) about N2 = eq2 : N f (N2 ) f (eq2 ) + N
=0
(4.318)
df dN2
N2 =eq N
(N2 eq2 ) + N
1 d2 f ( eq2 )2 + . . . N 2 d2 2 N2 N
(4.319)
Now the rst term of the Taylor series is zero by construction. Next neglect all higher order terms as small so that the approximation becomes f (N2 ) Thus, near equilibrium, one can write dN2 df dt dN2 (N2 eq2 ). N (4.321) df dN2 (N2 eq2 ). N (4.320)
N2 =eq N2
N2 =eq N2
Since the derivative of a constant is zero, one can also write the equation as df d (N2 eq2 ) N dt dN2 CC BY-NC-ND. 12 December 2011, J. M. Powers. (N2 eq2 ). N (4.322)
N2 =eq N2
173
concentration (kmole/m 3 )
0.002
N2
0.0015 0.001 0.0005
0.001 0.002 0.003 0.004 0.005
t (s)
Figure 4.3: N2 (t) and N (t) in isothermal, isochoric nitrogen dissociation problem.
This has a solution, valid near the equilibrium point, of df eq t , (N2 N2 ) = C exp dN2 =eq N2 N2 df t . N2 = eq2 + C exp N dN2 =eq
N2 N2
(4.323)
(4.324) (4.325)
Here C is some constant whose value is not important for this discussion. If the slope of f is positive, that is, df > 0, unstable, (4.326) dN2 =eq
N2 N2
the equilibrium will be unstable. That is a perturbation will grow without bound as t . If the slope is zero, df = 0, neutrally stable, (4.327) dN2 =eq
N2 N2
the solution is stable in that there is no unbounded growth, and moreover is known as neutrally stable. If the slope is negative, df dN2 < 0,
N2 =eq N
2
asymptotically stable,
(4.328)
the solution is stable in that there is no unbounded growth, and moreover is known as asymptotically stable. A solution via numerical integration is found for Eq. (4.314). The solution for N2 , along with N is plotted in Fig. 4.3. Linearization of Eq. (4.314) about the equilibrium state gives rise to the locally linearly valid d (4.329) 0.00178121 = 1209.39(N2 0.00178121) + . . . dt N2 CC BY-NC-ND. 12 December 2011, J. M. Powers.
174

This has local asymptotically stable solution N2 = 0.00178121 + C exp (1209.39t) . (4.330)
Here C is some integration constant whose value is irrelevant for this analysis. The time scale of relaxation is the time when the argument of the exponential is 1, which is = 1 = 8.27 104 s. 1209.39 s1 (4.331)
One usually nds this time scale to have high sensitivity to temperature, with high temperatures giving fast time constants and thus fast reactions. The equilibrium values agree exactly with those found in the earlier example. Here the kinetics provide the details of how much time it takes to achieve equilibrium. This is one of the key questions of non-equilibrium thermodynamics.
4.5.2
Isobaric systems
The form of the previous section is the most important as it is easily extended to a computational grid with xed volume elements in uid ow problems. However, there is another important spatially homogeneous problem in which the formulation needs slight modication: isobaric reaction, with P equal to a constant. Again, in this section only isothermal conditions will be considered. In an isobaric problem, there can be volume change. Consider rst the problem of isobaric expansion of an inert mixture. In such a mixture, the total number of moles of each species must be constant, so one gets dni = 0, dt inert, isobaric mixture. (4.332)
Now carry out the sequence of operations, realizing the total mass m is also constant: 1 d (ni ) m dt d ni dt m d ni V dt V m d i dt 1 di i d 2 dt dt di dt
= 0, = 0, = 0, = 0, = 0, = i d . dt
(4.333) (4.334) (4.335) (4.336) (4.337) (4.338)
175
So a global density decrease of the inert material due to volume increase of a xed mass system induces a concentration decrease of each species. Extended to a material with a single reaction rate r, one could say either d di = i r + i , or dt dt i 1 = i r, generally valid, i = . (4.339) (4.340) (4.341)
d dt
Equation (4.340) is consistent with Eq. (4.275) and is actually valid for general systems with variable density, temperature, and pressure. However, in this section, it is required that pressure and temperature be constant. Now dierentiate the isobaric, isothermal, ideal gas law to get the density derivative.
N
P =
i=1 N
i RT, RT
i=1 N
(4.342) (4.343) (4.344) i d dt

N
0 = 0 =
i=1 N
di , dt
di , dt i r + , i ,
i=1
0 =
i=1 N
(4.345) (4.346) (4.347) (4.348) (4.349)
0 = r
i=1
1 d i + dt
d = dt = =
= =
r N i i=1 . N i i=1 r N i i=1 , N i i=1 r N i i=1 , P RT RT r N i=1 P RT r P
, .
N k=1
(4.350) (4.351) k = 0, the isobaric,
N k=1 k
Note that if there is no net number change in the reaction,
176
isothermal reaction also guarantees there would be no density or volume change. It is convenient to dene the net number change in the elementary reaction as n:
N
k .
k=1
(4.352)
Here n is taken to be a dimensionless pure number. It is associated with the number change in the elementary reaction and not the actual mole change in a physical system; it is, however, proportional to the actual mole change. Now use Eq. (4.351) to eliminate the density derivative in Eq. (4.339) to get di RT r = i r i dt P
N k=1 k
, k , k=1
N
(4.353)
RT i i = r P reaction eect = r i
reaction eect
(4.354)
expansion eect
yin
expansion eect
There are two terms dictating the rate change of species molar concentration. The rst, a reaction eect, is precisely the same term that drove the isochoric reaction. The second is due to the fact that the volume can change if the number of moles change, and this induces an intrinsic change in concentration. Note that the term i RT /P = yi, the mole fraction. Example 4.14
Study a variant of the nitrogen dissociation problem considered in an earlier example in which at t = 0 s, 1 kmole of N2 exists at P = 100 kP a and T = 6000 K. In this case, take the reaction to be isothermal and isobaric. Consider again the elementary nitrogen dissociation reaction N2 + N2 2N + N2 , which has kinetic rate parameters of a E In SI units, this becomes a E = = 7.0 1021 m3 K 1.6 1m 1000 mole cm3 K 1.6 = 7.0 1018 , mole s 100 cm kmole kmole s J kJ cal kJ 1000 mole 4.186 = 941550 224928.4 . mole cal 1000 J kmole kmole
3
(4.355)
(4.356)
= 7.0 1021 = 1.6, = 224928.4
cm3 K 1.6 , mole s
(4.357) (4.358)
cal . mole
(4.359)
(4.360) (4.361)
177
At the initial state, the material is all N2 , so PN2 = P = 100 kP a. The ideal gas law then gives at t=0 P N 2
t=0
= PN2 = N2 RT, P = , RT 100 kP a , = kJ 8.314 kmole K (6000 K) kmole . = 2.00465 103 m3
(4.362) (4.363) (4.364) (4.365)
Thus, the initial volume is V |t=0 = nN2 |t=0 1 kmole = i |t=0 2.00465 103
kmole m3
= 4.9884 102 m3 .
(4.366)
In this isobaric process, one always has P = 100 kP a. Now, in general P = RT (N2 + N ), therefore one can write N in terms of N2 : N = = = P N 2 , RT 100 kP a N2 , kJ 8.314 kmole K (6000 K) 2.00465 103 kmole m3 N 2 . (4.368) (4.369) (4.370) (4.367)
Then the equations for kinetics of a single isobaric isothermal reaction, Eq. (4.354) in conjunction with Eq. (4.280), reduce for N2 molar concentration to dN2 dt = aT exp E RT (N2 )N2 (N )N
=r

1 ( )N2 (N )N Kc N2
N2
N2 RT (N2 + N ) . P (4.371)
Realizing that N2 = 2, N = 0, N2 = 1, and N = 2, one gets
dN2 dt
aT exp
E RT
2 2 1 N
1 2 N K c N 2
N2 RT P
(4.372)
=k(T )
The temperature dependency of the reaction is unchanged from the previous reaction: k(T ) = = = aT exp E RT , T 1.6 exp .
kJ 941550 kmole kJ 8.314 kmole K T
(4.373) , (4.374) (4.375)
7.0 1018
m3 K 1.6 kmole s
7.0 1018 exp T 1.6
1.1325 105 T
178

For this problem, T = 6000 K, so k(6000) = = 7.0 1018 1.1325 105 exp 60001.6 6000 3 m . 40130.2 kmole s , (4.376) (4.377)
The equilibrium constant Kc is also unchanged from the previous example. Recall Kc = For this system, since Kc = = = = = Po RT Po RT Po RT exp exp exp
N i=1
Po RT
PN
i=1
exp
Go RT
(4.378)
i = n = 1, this reduces to , ,
o
(2go g o 2 ) N N RT (2go g o 2 ) N N RT
o
(4.379) (4.380) , (4.381)
(2(hN T so ) (hN2 T so 2 )) T,N T,N RT
100 exp (8.314)(6000) kmole . 0.000112112 m3
(2(597270 (6000)216.926) (205848 (6000)292.984)) ,(4.382) (8.314)(6000) (4.383)
The dierential equation for N2 evolution is then given by dN2 dt = m3 40130.2 kmole 1 f (N2 ). 2 2 N 1 1 0.000112112 (6000 K) 2.00465 103 kmole N2 m3 N 2
2
kmole m3
N2 8.314
kJ kmole K
100 kP a
, (4.384) (4.385)
The system is at equilibrium when f (N2 ) = 0. This is an algebraic function of N2 only, and can be plotted. Figure 4.4 gives a plot of f (N2 ) and shows that it has four potential equilibrium points. It is seen there are four roots. Solving for the equilibria requires solving 0 = 40130.2 1 m3 kmole 2 2 N 1 1 0.000112112 (6000 K) 2.00465 103 kmole N2 m3 N 2
2
kmole m3
N2 8.314
kJ kmole K
100 kP a
, (4.386)
The four roots are N2 = 0.002005 kmole , m3 0 kmole , m3 0.001583 kmole , m3 0.00254 kmole . m3 (4.387)
stable,nonphysical
unstable
stable,physical
unstable

f ( N
179
) (kmole/m 3 /s)
4 3 2 1
-0.002
-0.001 -1
0.001
0.002
0.003
N2
(kmole/m 3 )
stable, non-physical equilibrium
stable, physical equilibrium
Figure 4.4: Forcing function, f (N2 ), which drives changes of N2 as a function of N2 in isothermal, isobaric problem.
By inspection of the topology of Fig. 4.2, the only stable, physical root is 0.001583 kmole/m3 . Small perturbations from this equilibrium induce the forcing function to supply dynamics which restore the system to its original equilibrium state. Small perturbations from the unstable equilibria induce nonrestoring dynamics. For this root, one can then determine that the stable equilibrium value of N = 0.000421 kmole/m3 . A numerical solution via an explicit technique such as a Runge-Kutta integration is found for Eq. (4.386). The solution for N2 , along with N is plotted in Fig. 4.5. Linearization of Eq. (4.386) about the equilibrium state gives rise to the locally linearly valid d 0.001583 = 967.073(N2 0.001583) + . . . dt N2 This has local solution N2 = 0.001583 + C exp (967.073t) . (4.389) Again, C is an irrelevant integration constant. The time scale of relaxation is the time when the argument of the exponential is 1, which is = 1 = 1.03 103 s. 967.073 s1 (4.390) (4.388)
Note that the time constant for the isobaric combustion is about a factor 1.25 greater than for isochoric combustion under the otherwise identical conditions. The equilibrium values agree exactly with those found in the earlier example. Again, the kinetics provide the details of how much time it takes to achieve equilibrium.
180
0.002 concentration (kmole/m 3 )
0.0015
N2
0.001
0.0005
0.001
0.002
0.003
0.004
t (s) 0.005
Figure 4.5: N2 (t) and N (t) in isobaric, isothermal nitrogen dissociation problem.
4.6
Some conservation and evolution equations
Here a few useful global conservation and evolution equations are presented for some key properties. Only some cases are considered, and one could develop more relations for other scenarios.
4.6.1
Total mass conservation: isochoric reaction
One can easily show that the isochoric reaction rate model, Eq. (4.279), satises the principle of mixture mass conservation. Begin with Eq. (4.279) in a compact form, using the denition of the reaction rate r, Eq. (4.281), and perform the following operations:
d dt
di = i r, dt Yi = i r, Mi d (Yi ) = i Mi r, dt L d Ml li r, (Yi ) = i dt l=1

=Mi
(4.391) (4.392) (4.393) (4.394)
d (Yi ) = dt
l=1
Ml li i r,
(4.395)
4.6. SOME CONSERVATION AND EVOLUTION EQUATIONS d (Yi ) = dt i=1

N i=1 =1 N N L
181
i=1 l=1 L N
Ml li i r,
(4.396)
d dt
Yi =
l=1 i=1 L
Ml li i r,
N
(4.397)
d = r dt d = 0. dt
l=1
Ml
li i ,
i=1 =0
(4.398)
(4.399)
Note the term
N i=1
li i = 0 because of stoichiometry, Eq. (4.24).
4.6.2
Element mass conservation: isochoric reaction
Through a similar series of operations, one can show that the mass of each element, l = 1, . . . , L, in conserved in this reaction, which is chemical, not nuclear. Once again, begin with Eq. (4.281) and perform a set of operations, di = i r, dt d li i = li i r, dt (4.400) l = 1, . . . , L, (4.401) (4.402) (4.403) (4.404)
i=1
d l = 1, . . . , L, (li i ) = rli i , dt N d (li i ) = rlii , l = 1, . . . , L, dt i=1

N N
d dt d dt
li i
i=1 N
= r
i=1
li i ,
=0
l = 1, . . . , L,
li i
i=1
= 0,
l = 1, . . . , L.
(4.405)
The term N li i represents the number of moles of element l per unit volume, by the i=1 following analysis
N N
li i =
i=1 i=1
moles element l moles species i moles element l = l e . moles species i volume volume
(4.406)
182
Here the elemental mole density, l e , for element l has been dened. So the element concentration for each element remains constant in a constant volume reaction process: dl e = 0, l = 1, . . . , L. (4.407) dt One can also multiply by the elemental mass, Ml to get the elemental mass density, e : l e Ml l e , l l = 1, . . . , L.
(4.408)
Since Ml is a constant, one can incorporate this denition into Eq. (4.407) to get de l = 0, l = 1, . . . , L. (4.409) dt The element mass density remains constant in the constant volume reaction. One could also simply say since the elements density is constant, and the mixture is simply a sum of the elements, that the mixture density is conserved as well.
4.6.3
Energy conservation: adiabatic, isochoric reaction
Consider a simple application of the rst law of thermodynamics to reaction kinetics: that of a closed, adiabatic, isochoric combustion process in a mixture of ideal gases. One may be interested in the rate of temperature change. First, because the system is closed, there can be no mass change, and because the system is isochoric, the total volume is a non-zero constant; hence, dm = 0, (4.410) dt d (V ) = 0, (4.411) dt d V = 0, (4.412) dt d = 0. (4.413) dt For such a process, the rst law of thermodynamics is dE = Q W. (4.414) dt But there is no heat transfer or work in the adiabatic isochoric process, so one gets dE = 0, (4.415) dt d (me) = 0, (4.416) dt dm de = 0, (4.417) m +e dt dt
=0
de = 0. dt
(4.418)
4.6. SOME CONSERVATION AND EVOLUTION EQUATIONS
183
Thus for the mixture of ideal gases, e(T, 1 , . . . , N ) = eo . One can see how reaction rates aect temperature changes by expanding the derivative in Eq. (4.418) d dt
N N
Yiei
i=1
= 0,
(4.419) (4.420) (4.421) (4.422) (4.423)

N
i=1 N
d (Yiei ) = 0, dt = 0, = 0, = 0, ei
i=1
Yi
i=1 N
dYi dei + ei dt dt
Yi
i=1 N
dYi dei dT + ei dT dt dt dT dYi + ei dt dt dT dt

N
Yi cvi
i=1
i=1
Yi cvi = dT dt dT dt dT dt dT dt
dYi , dt d dt Mi i di , dt
(4.424)
=cv N
cv cv cv
= = = =
ei
i=1 N
(4.425) (4.426) (4.427) (4.428)
ei Mi
i=1 N
ei Mi i r,
i=1
N i=1
i ei
cv
If one denes the net energy change of the reaction as

N
E one then gets
i ei ,
i=1
(4.429)
dT rE = . (4.430) dt cv The rate of temperature change is dependent on the absolute energies, not the energy differences. If the reaction is going forward, so r > 0, and that is a direction in which the net molar energy change is negative, then the temperature will rise.
184
4.6.4
Energy conservation: adiabatic, isobaric reaction
Solving for the reaction dynamics in an adiabatic isobaric system requires some non-obvious manipulations. First, the rst law of thermodynamics says dE = dQdW . Since the process is adiabatic, one has dQ = 0, so dE + P dV = 0. Since it is isobaric, one gets d(E + P V ) = 0, or dH = 0. So the total enthalpy is constant. Then d H = 0, (4.431) dt d (mh) = 0, (4.432) dt dh = 0, (4.433) dt N d Yi hi = 0, (4.434) dt i=1
N
i=1 N
d (Yi hi ) = 0, dt
(4.435) (4.436) (4.437) (4.438) (4.439)
Yi
i=1 N
dYi dhi + hi = 0, dt dt
Yi
i=1 N
dYi dhi dT + hi = 0, dT dt dt
N
i=1
dT Y i cP i + dt
N
hi
i=1 N
dYi = 0, dt dYi = 0, dt
dT dt
Y i cP i +
i=1 =cP i=1
hi
dT cP + dt cP dT + dt
hi
i=1 N
d dt
i Mi d dt i
= 0, = 0.
(4.440) (4.441)
hi Mi
i=1
Now use Eq (4.340) to eliminate the term in Eq. (4.441) involving molar concentration derivatives to get cP dT + dt
N
hi
i=1
i r = 0, dT dt = r
N i=1 hi i
(4.442) . (4.443)
cP
185
So the temperature derivative is known as an algebraic function. If one denes the net enthalpy change as
N
H one gets that Eq. (4.443) transforms to
hi i ,
i=1
(4.444)
dT rH = . dt cP or cP dT = rH. dt
(4.445)
(4.446)
Equation (4.446) is in a form which can easily be compared to a form to be derived later when we add variable pressure and diusion eects. Now dierentiate the isobaric ideal gas law to get the density derivative.
N
P =
i=1 N
i RT, i R
i=1
(4.447)
N
0 = 0 = dT dt
dT + dt i +
RT
i=1 N
di , dt i d dt
N
(4.448) , i ,
i=1
T
i=1 N
i r +
(4.449) (4.450) (4.451)
i=1 N
1 dT 0 = T dt
1 T dT d dt = dt
i + r
i=1 i=1 N i=1 i r N i i=1
1 d i + dt
N i=1
One takes dT /dt from Eq. (4.443) to get d = dt

1 r T PN
i=1
hi i
cP
N i=1 i N i i=1
N i=1
(4.452)
Now recall from Eqs. (2.212) and (2.218) that = /M and cP = cP M, so cP = cP . Then Equation (4.452) can be reduced slightly:
=1 PN N hi i cP T
i=1
d = r dt
i=1 N i=1
i i
N i=1
i , (4.453)
186

N hi i N i=1 cP T i=1 N i=1 i N hi i=1 i cP T 1 P RT N
= r
(4.454)
= r = r
(4.455) (4.456) (4.457)
RT P
N
i
i=1
hi 1 , cP T
= rM
i=1
hi 1 , cP T
N where M is the mean molecular mass. Note for exothermic reaction i=1 i hi < 0, so exothermic reaction induces a density decrease as the increased temperature at constant pressure causes the volume to increase. Then using Eq. (4.457) to eliminate the density derivative in Eq. (4.339), and changing the dummy index from i to k, one gets an explicit expression for concentration evolution: N
di = i r + i rM dt
k
k=1 N
hk 1 , cP T
(4.458)
= r i + i M
=yi N k=1
k
k=1
= r i + yi
hk 1 cP T
hk 1 , cP T .
N k=1
(4.459)
(4.460)
Dening the change of enthalpy of the reaction as H number of the reaction as n N k , one can also say k=1 di = r i + yi dt H n cP T .
k hk , and the change of
(4.461)
Exothermic reaction, H < 0, and net number increases, n > 0, both tend to decrease the molar concentrations of the species in the isobaric reaction. Lastly, the evolution of the adiabatic, isobaric system, can be described by the simultaneous, coupled ordinary dierential equations: Eqs. (4.443, 4.452, 4.460). These require numerical solution in general. Note also that one could also employ a more fundamental treatment as a dierential algebraic system involving H = H1 , P = P1 = RT N i and i=1 Eq. (4.339).
187
4.6.5
Non-adiabatic isochoric combustion
Consider briey combustion in a xed nite volume in which there is simple convective heat transfer with the surroundings. In general, the rst law of thermodynamics is dE = Q W. dt (4.462)
Because the system is isochoric W = 0. And using standard relations from simple convective heat transfer, one can say that dE = hA(T T ). dt (4.463)
Here h is the convective heat transfer coecient and A is the surface area associated with the volume V , and T is the temperature of the surrounding medium. One can, much as before, write E in detail and get an equation for the evolution of T .
4.6.6
Entropy evolution: Clausius-Duhem relation
Now consider whether the kinetics law that has been posed actually satises the second law of thermodynamics. Consider again Eq. (3.415). There is an algebraic relation on the right side. If it can be shown that this algebraic relation is positive semi-denite, then the second law is satised, and the algebraic relation is known as a Clausius12 -Duhem13 relation. Now take Eq. (3.415) and perform some straightforward operations on it: dS|E,V 1 = T V T V T
N
i=1
i dni 0, dni 1 0, dt V di 0, dt E RT
N k k
(4.464)
irreversibility
dS dt
E,V
= =
i
i=1 N
(4.465) (4.466) 1 1 Kc kk 0,(4.467) k=1

N
i
i=1 N
V = T
i i aT exp
i=1 k(T )
k=1 forward reaction r
reverse reaction
12 13
Rudolf Julius Emanuel Clausius, 1822-1888, German physicist. Pierre Maurice Marie Duhem, 1861-1916, French physicist. CC BY-NC-ND. 12 December 2011, J. M. Powers.
188
CHAPTER 4. THERMOCHEMISTRY OF A SINGLE REACTION V T

N N
k
i i k(T )
i=1 N k k k=1
1
N
1 Kc
kk
k=1 N
0, i i 0,
(4.468) (4.469)
V = k(T ) T
k=1
1 1 Kc
kk
k=1 i=1
Change the dummy index from k back to i, V = k(T ) T V r, T d = . T dt =

N
i
i=1
1 1 Kc
i i
i=1
0,
(4.470) (4.471) (4.472)
Consider now the anity term in Eq. (4.470) and expand it so that it has a more useful form:
N N
i=1
i i = = =
g i i ,
i=1 N
(4.473) Pi Po ln
i=1
g o + RT ln T,i
i=1 N N
i , Pi Po
i
(4.474) , (4.475)
g o i T,i
i=1 =G
o
RT
Go = RT RT
=ln KP
ln
i=1 N
Pi Po Pi Po
i
, ,
(4.476)
= RT = RT
ln KP ln
(4.477) , (4.478) (4.479)

i
i=1 N
1 ln + ln KP i=1 1 KP
Po RT N
Pi Po
i
= RT ln
= RT ln
i=1 PN
Pi Po
i
,
N
i=1
Kc
i=1
i RT Po
(4.480)
4.6. SOME CONSERVATION AND EVOLUTION EQUATIONS 1 Kc

N
189
= RT ln
i i
i=1
(4.481)
Equation (4.481) is the common denition of anity. Another form can be found by employing the denition of Kc from Eq. (4.270) to get = RT ln = RT Po RT
o PN
i=1
exp Po RT Po RT

Go RT
i=1
i i
i=1
, ,
(4.482) (4.483) (4.484)
G + ln RT
PN
i N
i i
i=1
PN
= Go RT ln
N i=1 i i=1
i i
To see clearly that the entropy production rate is positive semi-denite, substitute Eq. (4.481) into Eq. (4.470) to get dS dt V = k(T ) T
N
i
i=1 N
E,V
1 1 Kc
i
i=1 N
RT ln 1 Kc
1 Kc
N
i i
i=1
0, (4.485)
= RV k(T )
i
i=1
1 Kc
i i
i=1
ln
i i
i=1
0.
(4.486)
Dene forward and reverse reaction coecients, R , and R , respectively, as

N
R k(T ) R k(T ) Kc
i i ,
i=1 N
(4.487) (4.488)
i i .
i=1
Both R and R have units of kmole/m3 /s. It is easy to see that r = R R . (4.489)
Note that since k(T ) > 0, Kc > 0, and i 0, that both R 0 and R 0. Since i = i i , one nds that 1 Kc
N
i
i=1
1 k(T ) = Kc k(T )
i
i=1
i i
R . R
(4.490)
190
Then Eq. (4.486) reduces to dS dt = RV R 1 R R ln R R R R 0. 0, (4.491) (4.492)
E,V
= RV (R R ) ln
Obviously, if the forward rate is greater than the reverse rate R R > 0, ln(R /R ) > 0, and the entropy production is positive. If the forward rate is less than the reverse rate, R R < 0, ln(R /R ) < 0, and the entropy production is still positive. The production rate is zero when R = R . Note that the anity can be written as = RT ln R R . (4.493)
And so when the forward reaction rate exceeds the reverse, the anity is positive. It is zero at equilibrium, when the forward reaction rate equals the reverse.
4.7
Simple one-step kinetics
A common model in theoretical combustion is that of so-called simple one-step kinetics. Such a model, in which the molecular mass does not change, is quantitatively appropriate only for isomerization reactions. However, as a pedagogical tool as well as a qualitative model for real chemistry, it can be valuable. Consider the reversible reaction A B. (4.494) where chemical species A and B have identical molecular masses MA = MB = M. Consider further the case in which at the initial state, no moles of A only are present. Also take the reaction to be isochoric and isothermal. These assumptions can easily be relaxed for more general cases. Specializing then Eq. (4.240) for this case, one has nA = nB = Thus nA = + no , nB = .
A + nAo ,
=1 =no
(4.495) (4.496)
B + nBo .
=1 =0
(4.497) (4.498)
4.7. SIMPLE ONE-STEP KINETICS
191
Now no is constant throughout the reaction. Scale by this and dene the dimensionless reaction progress as /no to get nA no = + 1, = . (4.499)
=yA
nB no
=yB
(4.500)
In terms of the mole fractions then, one has yA = 1 , yB = . The reaction kinetics for each species reduce to dA = r, dt dB = r, dt Addition of Eqs. (4.503) and (4.504) gives d ( + B ) = 0, dt A A + B = o , A + B = 1. o o
=yA =yB
(4.501) (4.502)
A (0) = B (0) = 0.
no o , V
(4.503) (4.504)
(4.505) (4.506) (4.507)
In terms of the mole fractions yi , one then has yA + yB = 1. The reaction rate r is then r = kA 1 1 B , Kc A 1 B /o , = ko A 1 o Kc A /o 1 yB = ko yA 1 , Kc y A 1 . = ko (1 ) 1 Kc 1 (4.509) (4.510) (4.511) (4.512) (4.508)
192
Now r = (1/V )d/dt = (1/V )d(no )/dt = (no /V )d()/dt = o d/dt. So the reaction dynamics can be described by a single ordinary dierential equation in a single unknown: o 1 d , = ko (1 ) 1 dt Kc 1 d 1 . = k(1 ) 1 dt Kc 1 (4.513) (4.514)
Equation (4.514) is in equilibrium when 1 1 = + ... 1 1 Kc 1 + Kc
(4.515)
As Kc , the equilibrium value of 1. In this limit, the reaction is irreversible. That is, the species B is preferred over A. Equation (4.514) has exact solution = 1 exp k 1 + 1+
1 Kc 1 Kc
t . (4.516)
For k > 0, Kc > 0, the equilibrium is stable. The time constant of relaxation is 1 . = 1 k 1 + Kc
(4.517)
For the isothermal, isochoric system, one should consider the second law in terms of the Helmholtz free energy. Combine then Eq. (3.421), dA|T,V 0, with Eq. (3.304), dA = SdT P dV + N i dni and taking time derivatives, one nds i=1
N
dA|T,V =
SdT P dV + dA dt
i dni
i=1 N T,V
0,
(4.518)
=
T,V i=1 N
i i
i=1
dni 0, dt di 0. dt
(4.519) (4.520)
V 1 dA = T dt T
This is exactly the same form as Eq. (4.486), which can be directly substituted into Eq. (4.520) to give 1 dA T dt dA dt
N T,V
= RV k(T )
i
i=1 N
1 1 Kc 1 Kc
i
i=1 N
ln
1 Kc 1 Kc
i i
i=1 N
0, (4.521)
= RV T k(T )
T,V i=1
i i
i=1
ln
i i
i=1
0. (4.522)
4.7. SIMPLE ONE-STEP KINETICS For the assumptions of this section, Eq. (4.522) reduces to dA dt = RT ko V (1 ) 1 = kno RT (1 ) 1 1 Kc 1 ln ln 1 Kc 1 0, 0.
193
(4.523) (4.524)
T,V
1 Kc 1
1 Kc 1
Since the present analysis is nothing more than a special case of the previous section, Eq. (4.524) certainly holds. One questions however the behavior in the irreversible limit, 1/Kc 0. Evaluating this limit, one nds
1/Kc 0
lim
dA dt
T,V
(4.525) Now, performing the distinguished limit as 1; that is the reaction goes to completion, one notes that all terms are driven to zero for small 1/Kc . Recall that 1 goes to zero faster than ln(1 ) goes to . Note that the entropy inequality is ill-dened for a formally irreversible reaction with 1/Kc = 0.
= kno RT (1 ) ln
>0
1 Kc
+(1 ) ln (1 ) ln(1 ) + . . . 0.
194
Chapter 5 Thermochemistry of multiple reactions

This chapter will extend notions associated with the thermodynamics of a single chemical reactions to systems in which many reactions occur simultaneously. Some background is in some standard sources.1 2 3
5.1
Summary of multiple reaction extensions
Consider now the reaction of N species, composed of L elements, in J reactions. This section will focus on the most common case in which J (N L), which is usually the case in large chemical kinetic systems in use in engineering models. While much of the analysis will only require J > 0, certain results will depend on J (N L). It is not dicult to study the complementary case where 0 < J < (N L). The molecular mass of species i is still given by Eq. (4.1):
L
Mi =
l=1
Ml li,
i = 1, . . . , N.
(5.1)
However, each reaction has a stoichiometric coecient. The j th reaction can be summarized in the following ways:
N i ij i=1 N N
i=1
i ij ,
j = 1, . . . , J,
(5.2) (5.3)
i ij
i=1
1
= 0,
j = 1, . . . , J.
Turns, S. R., 2011, An Introduction to Combustion, Third Edition, McGraw-Hill, Boston. Chapters 4-6. Kuo, K. K., 2005, Principles of Combustion, Second Edition, John Wiley, New York. Chapters 1 and 2. 3 Kondepudi, D., and Prigogine, I., 1998, Modern Thermodynamics: From Heat Engines to Dissipative Structures, John Wiley, New York. Chapters 16 and 19.
2
195
196
CHAPTER 5. THERMOCHEMISTRY OF MULTIPLE REACTIONS
Stoichiometry for the j th reaction and lth element is given by

N
liij = 0,
i=1
l = 1, . . . , L, j = 1, . . . , J.
(5.4)
The net change in Gibbs free energy and equilibrium constants of the j th reaction are dened by
N
Go j
g o ij , T,i
i=1
j = 1, . . . , J, j = 1, . . . , J, j = 1, . . . , J.
(5.5) (5.6) (5.7)
KP,j exp Kc,j Po RT

PN
i=1
Go j RT Go j RT
, ,
ij
exp
The equilibrium of the j th reaction is given by

N
i ij = 0,
i=1 N
j = 1, . . . , J, j = 1, . . . , J.
(5.8) (5.9)
g i ij = 0,
i=1
The multi-reaction extension for anity is

N
j =
i ij ,
i=1
j = 1, . . . , J.
(5.10)
In terms of the chemical anity of each reaction, the equilibrium condition is simply j = 0, j = 1, . . . , J. (5.11)
At equilibrium, then the equilibrium constraints can be shown to reduce to

N
KP,j =
i=1 N
Pi Po i
ij
ij
j = 1, . . . , J, j = 1, . . . , J.
(5.12) (5.13)
Kc,j =
i=1
5.1. SUMMARY OF MULTIPLE REACTION EXTENSIONS
197
For isochoric reaction, the evolution of species concentration i due to the combined eect of J reactions is given by
i
di = dt
ij aj T
j=1
exp
kj (T )
E j RT
k kj
rj =(1/V )dj /dt
1 1 Kc,j
k=1
reverse reaction
k kj ,
i = 1, . . . , N. (5.14)
The extension to isobaric reactions, not given here, is straightforward, and follows the same analysis as for a single reaction. Again, three intermediate variables which are in common usage have been dened. First one takes the reaction rate of the j th reaction to be rj aj T
j
exp
E j RT
k kj
kj (T )
1 1 Kc,j 1 Kc,j
k=1
reverse reaction N
kj
= aj T
exp
E j RT
kj (T ), Arrhenius rate
k kj
k=1
reverse reaction
k kj ,
j = 1, . . . , J,
(5.15)
j = 1, . . . , J,
(5.16)
law of mass action
1 dj . V dt
(5.17)
Here j is the reaction progress variable for the j th reaction. Each reaction has a temperature-dependent rate function kj (T ), which is kj (T ) aj T j exp E j RT , j = 1, . . . , J. (5.18)
The evolution rate of each species is given by i , dened now as

J
ij rj ,
j=1
i = 1, . . . , N.
(5.19)
So we can summarize Eq. (5.14) as di = i . dt (5.20)
198
It will be useful to cast this in terms of mass fraction. Using the denition Eq. (2.43), Eq. (5.20) can be rewritten as d dt Yi Mi = i . (5.21)
Because Mi is a constant, Eq. (5.21) can be recast as d (Yi ) = Mi i. dt The multi-reaction extension for mole change in terms of progress variables is
J
(5.22)
dni =
j=1
ij dj ,
i = 1, . . . , N.
(5.23)
One also has

N
dG|T,P =
i=1 N
i dni ,
J
(5.24) ik dk , (5.25) (5.26) (5.27)
=
i=1
i
N k=1 J
G j
=
p i=1 N
i
k=1 J
ik
k , j
=
i=1 N
i
j=1
ik kj ,
=
i=1
i ij , j = 1, . . . , J.
(5.28) (5.29)
= j ,
For a set of adiabatic, isochoric reactions, one can show dT = dt

J j=1 rj Uj
cv
(5.30)
where the energy change for a reaction Uj is dened as

N
Uj =
i=1
ui ij ,
j = 1, . . . , J.
(5.31)
5.1. SUMMARY OF MULTIPLE REACTION EXTENSIONS Similarly for a set of adiabatic, isobaric reactions, one can show dT = dt
J j=1 rj Hj
199
cP
(5.32)
where the enthalpy change for a reaction Hj is dened as

N
Hj =
i=1
hi ij ,
j = 1, . . . , J.
(5.33)
Moreover, the density and species concentration derivatives for an adiabatic, isobaric set can be shown to be d = M dt di = dt where
N J J N
rj
j=1 i=1
ij
hi 1 , cP T Hj nj cP T ,
(5.34)
rj ij + yi
j=1
(5.35)
nj =
k=1
kj .
(5.36)
In a similar fashion to that shown for a single reaction, one can further sum over all reactions and prove that mixture mass is conserved, element mass and number are conserved. Example 5.1
Show that element mass and number are conserved for the multi-reaction formulation. Start with Eq. (5.14) and expand as follows: di dt li di dt
J
=
j=1
ij rj ,
J
(5.37)
= li
j=1 J
ij rj ,
(5.38)
d (li i ) = dt
N i=1
li ij rj ,
j=1 N J
(5.39)
d (li i ) = dt
N
li ij rj ,
i=1 j=1 J N
(5.40)
d dt
li i
i=1 =l e
=
j=1 i=1
li ij rj ,
(5.41)
200

dl e dt dl e dt
J N
=
j=1
rj
i=1
li ij ,
=0
(5.42)
= 0,
l = 1, . . . , L, l = 1, . . . , L, l = 1, . . . , L.
(5.43) (5.44) (5.45)
d (Ml l e ) = 0, dt dl e = 0, dt
It is also straightforward to show that the mixture density is conserved for the multi-reaction, multicomponent mixture: d = 0. (5.46) dt
The proof of the Clausius-Duhem relationship for the second law is an extension of the single reaction result. Start with Eq. (4.464) and operate much as for a single reaction model. dS|E,V = 1 T
N
i=1
i dni 0, dni 1 0, dt V di 0, dt
J
(5.47)
irreversibility
dS dt
E,V
V = T = = = = V T V T V T V T
i
i=1 N
(5.48) (5.49) (5.50)
i
i=1 N
i
i=1 N J j=1
ij rj 0,
i=1 j=1 J
i ij rj 0,
N
(5.51)
rj
j=1 J i=1 N
i ij 0,
i ij
(5.52)
N i ij i=1 N i=1 N
V = T = V T
kj
j=1 J i=1 N
1 1 Kc,j 1 1 Kc,j
i ij 0, 1 Kc,j
N
(5.53) i ij
i=1
kj
j=1 i=1
i ij
i ij
i=1
RT ln
0,
5.2. EQUILIBRIUM CONDITIONS
201 (5.54)
= RV
kj
j=1 i=1
i ij
1 Kc,j
i ij
i=1
ln
1 Kc,j
i ij
i=1
0, (5.55)
Note that Eq. (5.52) can also be written in terms of the anities (see Eq. (5.10)) and reaction progress variables (see Eq. (5.17) as dS dt 1 = T
J
j
j=1
E,V
dj 0. dt
(5.56)
Similar to the argument for a single reaction, if one denes

N
Rj
= kj
i=1
i
N
ij
,
ij
(5.57) , (5.58)
R = j then it is easy to show that
kj Kc,j
i
i=1
rj = Rj R , j and dS dt
J
(5.59)
= RV
E,V j=1
Rj R ln j
Rj R j
0.
(5.60)
Since kj (T ) > 0, R > 0, and V 0, and each term in the summation combines to be positive semi-denite, one sees that the Clausius-Duhem inequality is guaranteed to be satised for multi-component reactions.
5.2
Equilibrium conditions
For multicomponent mixtures undergoing multiple reactions, determining the equilibrium condition is more dicult. There are two primary approaches, both of which are essentially equivalent. The most straightforward method requires formal minimization of the Gibbs free energy of the mixture. It can be shown that this actually nds the equilibrium associated with all possible reactions.
202
5.2.1
G ni
Minimization of G via Lagrange multipliers

N i=1
Recall Eq. (3.422), dG|T,P 0. Recall also Eq. (3.424), G =

P,T,nj
g i ni . Since i = g i =
N i=1
, one also has G =
N i=1
i ni . From Eq. (3.425), dG|T,P =
i dni . Now
one must also demand for a system coming to equilibrium that the element numbers are conserved. This can be achieved by requiring
N
i=1
li (nio ni ) = 0,
l = 1, . . . , L.
(5.61)
Here recall nio is the initial number of moles of species i in the mixture, and li is the number of moles of element l in species i. One can now use the method of constrained optimization given by the method of Lagrange multipliers to extremize G subject to the constraints of element conservation. The extremum will be a minimum; this will not be proved, but it will be demonstrated. Dene a set of L Lagrange4 multipliers l . Next dene an augmented Gibbs free energy function G , which is simply G plus the product of the Lagrange multipliers and the constraints:
L N
G =G+
l=1
l
i=1
li (nio ni ).
(5.62)
Now when the constraints are satised, one has G = G, so assuming the constraints can be satised, extremizing G is equivalent to extremizing G . To extremize G , take its dierential with respect to ni , with P , T and nj constant and set it to zero for each species: G ni =
T,P,nj
G ni
L T,P,nj
l li = 0.
l=1
i = 1, . . . , N.
(5.63)
=i
With the denition of the partial molar property i , one then gets
L
l li = 0,
l=1
i = 1, . . . , N.
(5.64)
Next, for an ideal gas, one can expand the chemical potential so as to get o T,i + RT ln
=i
Pi Po
l li = 0,
l=1
i = 1, . . . , N,
(5.65)
o T,i
Joseph-Louis Lagrange, 1736-1813, Italian-French mathematician.
+ RT ln
ni P
N k=1 nk =Pi
1 Po
l li = 0,
l=1
i = 1, . . . , N.
(5.66)
5.2. EQUILIBRIUM CONDITIONS Recalling that

N k=1 nk
203
= n, in summary then, one has N + L equations ni P n Po

N L
o T,i
+ RT ln
l li = 0,
l=1
i = 1, . . . , N, l = 1, . . . , L.
(5.67) (5.68)
i=1
li (nio ni ) = 0,
in N + L unknowns: ni , i = 1, . . . , N, l , l = 1, . . . , L. Example 5.2

Consider a previous example problem in which N2 + N2 2N + N2 . (5.69)
Take the reaction to be isothermal and isobaric with T = 6000 K and P = 100 kP a. Initially one has 1 kmole of N2 and 0 kmole of N . Use the extremization of Gibbs free energy to nd the equilibrium composition. First nd the chemical potentials at the reference pressure of each of the possible constituents. o = go = hi T so = h298,i + hi T so . T,i i i i For each species, one then nds kJ , kmole kJ = 472680 + 124590 (6000)(216.926) = 704286 . kmole = 0 + 205848 (6000)(292.984) = 1552056 RT ln ni P nPo
o o o
(5.70)
o 2 N o N
(5.71) (5.72)
To each of these one must add
to get the full chemical potential. Now P = Po = 100 kP a for this problem, so one only must consider kJ RT = 8.314(6000) = 49884 kmole . So, the chemical potentials are N2 N nN 2 , nN + nN 2 nN = 704286 + 49884 ln . nN + nN 2 = 1552056 + 49884 ln (5.73) (5.74)
Then one adds on the Lagrange multiplier and then considers element conservation to get the following coupled set of nonlinear algebraic equations: 1552056 + 49884 ln nN 2 2N nN + nN 2 nN 704286 + 49884 ln N nN + nN 2 nN + 2nN2 = = = 0, 0, 2. (5.75) (5.76) (5.77)
204

These non-linear equations are solved numerically to get nN 2 nN N = = = 0.88214 kmole, 0.2357 kmole, kJ . 781934 kmole (5.78) (5.79) (5.80)
These agree with results found in an earlier example problem.
Example 5.3
Consider a mixture of 2 kmole of H2 and 1 kmole of O2 at T = 3000 K and P = 100 kP a. Assuming an isobaric and isothermal equilibration process with the products consisting of H2 , O2 , H2 O, OH, H, and O, nd the equilibrium concentrations. Consider the same mixture at T = 298 K and T = 1000 K. The rst task is to nd the chemical potentials of each species at the reference pressure and T = 3000 K. Here one can use the standard tables along with the general equation o = go = hi T so = h298,i + hi T so . T,i i i i For each species, one then nds o 2 H o 2 O o 2 O H o OH o H o O kJ , kmole kJ = 0 + 98013 3000(284.466) = 755385 , kmole = 0 + 88724 3000(202.989) = 520242 kJ , kmole kJ = 38987 + 89585 3000(256.825) = 641903 , kmole kJ , = 217999 + 56161 3000(162.707) = 213961 kmole kJ = 249170 + 56574 3000(209.705) = 323371 . kmole = 241826 + 126548 3000(286.504) = 974790 RT ln ni P nPo (5.82) (5.83) (5.84) (5.85) (5.86) (5.87)
o o o
(5.81)
To each of these one must add
to get the full chemical potential. Now P = Po = 100 kP a for this problem, so one must only consider RT = 8.314(3000) = 24942 kJ/kmole. So, the chemical potentials are H2 O2 H2 O OH = = = = 520243 + 24942 ln nH2 nH2 + nO2 + nH2 O + nOH nO2 755385 + 24942 ln nH2 + nO2 + nH2 O + nOH nH2 O 974790 + 24942 ln nH2 + nO2 + nH2 O + nOH nOH 641903 + 24942 ln nH2 + nO2 + nH2 O + nOH + nH + nO + nH + nO + nH + nO + nH + nO , , , , (5.88) (5.89) (5.90) (5.91)

H O = = nH nH2 + nO2 + nH2 O + nOH + nH + nO nO 323371 + 24942 ln nH2 + nO2 + nH2 O + nOH + nH + nO 213961 + 24942 ln , .
205
(5.92) (5.93)
Then one adds on the Lagrange multipliers and then considers element conservation to get the following coupled set of nonlinear equations: nH2 2H nH2 + nO2 + nH2 O + nOH + nH + nO nO2 2O 755385 + 24942 ln nH2 + nO2 + nH2 O + nOH + nH + nO nH2 O 974790 + 24942 ln 2H O nH2 + nO2 + nH2 O + nOH + nH + nO nOH H O 641903 + 24942 ln nH2 + nO2 + nH2 O + nOH + nH + nO nH 213961 + 24942 ln H nH2 + nO2 + nH2 O + nOH + nH + nO nO O 323371 + 24942 ln nH2 + nO2 + nH2 O + nOH + nH + nO 2nH2 + 2nH2 O + nOH + nH 2nO2 + nH2 O + nOH + nO 520243 + 24942 ln = 0, = 0, = 0, = 0, = 0, = 0, = 4, = 2. (5.94) (5.95) (5.96) (5.97) (5.98) (5.99) (5.100) (5.101)
These non-linear algebraic equations can be solved numerically via a Newton-Raphson technique. The equations are sensitive to the initial guess, and one can use ones intuition to help guide the selection. For example, one might expect to have nH2 O somewhere near 2 kmole. Application of the Newton-Raphson iteration yields nH2 nO2 nH2 O nOH nH nO H O = = = = = = = = 3.19 101 kmole, 1.10 101 kmole, (5.102) (5.103) (5.104) (5.105) (5.106) (5.107) (5.108) (5.109)
1.50 100 kmole, 2.20 101 kmole, 5.74 10

2
kmole, kJ , 2.85 105 kmole kJ 4.16 105 . kmole
1.36 101 kmole,
At this relatively high value of temperature, all species considered have a relatively major presence. That is, there are no truly minor species. Unless a good guess is provided, it may be dicult to nd a solution for this set of non-linear equations. Straightforward algebra allows the equations to be recast in a form which sometimes converges more rapidly: nH2 + nH2 O + nOH + nH + nO nO2 + nH2 O + nOH + nH + nO = exp = exp 520243 24942 755385 24942 exp exp H 24942 O 24942
2
nH2 + nO2 nH2 + nO2
,
2
(5.110) (5.111)
206

nH2 O + nO2 + nH2 O + nOH + nH + nO nOH + nO2 + nH2 O + nOH + nH + nO nH + nO2 + nH2 O + nOH + nH + nO nO + nO2 + nH2 O + nOH + nH + nO 2nH2 + 2nH2 O + nOH + nH 2nO2 + nH2 O + nOH + nO 974790 24942 641903 24942 213961 24942 323371 24942 O H exp 24942 24942 O H exp , 24942 24942 H , 24942 O , 24942
2
nH2 nH2 nH2 nH2
= exp = exp = exp = exp = 4, = 2.
exp exp exp exp
, (5.112) (5.113) (5.114) (5.115) (5.116) (5.117)
Then solve these considering ni , exp (O /24942), and exp (H /24942) as unknowns. The same result is recovered, but a broader range of initial guesses converge to the correct solution. One can verify that this choice extremizes G by direct computation; moreover, this will show that the extremum is actually a minimum. In so doing, one must exercise care to see that element conservation is retained. As an example, perturb the equilibrium solution above for nH2 and nH such that nH2 nH = = 3.19 101 + , 1.36 101 2. (5.118) (5.119)
Leave all other species mole numbers the same. In this way, when = 0, one has the original equilibrium solution. For = 0, the solution moves o the equilibrium value in such a way that elements are conserved. Then one has G = N i ni = G(). i=1 The dierence G() G(0) is plotted in Fig. 5.1. When = 0, there is no deviation from the value predicted by the Newton-Raphson iteration. Clearly when = 0, G() G(0), takes on a minimum value, and so then does G(). So the procedure works. At the lower temperature, T = 298 K, application of the same procedure yields very dierent results: nH2 nO2 nH2 O nOH nH nO H O = 4.88 1027 kmole,
27
(5.120) (5.121) (5.122) (5.123) (5.124) (5.125) (5.126) (5.127)
= 1.67 1054 kmole, kJ , = 9.54 104 kmole kJ = 1.07 105 . kmole
= 2.22 1029 kmole, = 2.29 1049 kmole,
= 2.44 10 kmole, 0 = 2.00 10 kmole,
At the intermediate temperature, T = 1000 K, application of the same procedure shows the minor species become slightly more prominent: nH2 nO2 nH2 O nOH nH = 4.99 107 kmole,
7
(5.128) (5.129) (5.130) (5.131) (5.132)
= 2.44 10 kmole, = 2.00 100 kmole,
= 2.09 108 kmole, = 2.26 1012 kmole,

G() - G(0) (kJ) 40
207
30
20
10
-0.01
-0.005
0.005
0.01 (kmole)
Figure 5.1: Gibbs free energy variation as mixture composition is varied maintaining element conservation for mixture of H2 , O2 , H2 O, OH, H, and O at T = 3000 K, P = 100 kP a.
nO H O = 1.10 1013 kmole, kJ , = 1.36 105 kmole kJ . = 1.77 105 kmole (5.133) (5.134) (5.135)
5.2.2
Equilibration of all reactions
In another equivalent method, if one commences with a multi-reaction model, one can require each reaction to be in equilibrium. This leads to a set of algebraic equations for rj = 0, which from Eq. (5.16) leads to Kc,j = Po RT
PN
i=1
ij
exp
Go j RT
=
k=1
kj
j = 1, . . . , J.
(5.136)
With some eort it can be shown that not all of the J equations are linearly independent. Moreover, they do not possess a unique solution. However, for closed systems, only one of the solutions is physical, as will be shown in the following section. The others typically involve non-physical, negative concentrations.
208
Nevertheless, Eqs. (5.136) are entirely consistent with the predictions of the N + L equations which arise from extremization of Gibbs free energy while enforcing element number constraints. This can be shown by beginning with Eq. (5.66), rewritten in terms of molar concentrations, and performing the following sequence of operations: o T,i + RT ln P N Po k=1 nk /V ni /V i N k=1 P k Po i P Po RT Po
L
l li = 0,
l=1 L
i = 1, . . . , N, (5.137) i = 1, . . . , N, (5.138) i = 1, . . . , N, (5.139) i = 1, . . . , N, (5.140) i = 1, . . . , N, (5.141)
o + RT ln T,i
l li = 0,
l=1 L
o + RT ln T,i
l li = 0,
l=1 L
o + RT ln i T,i ij o + ij RT ln i T,i
l li = 0,
l=1 L
RT Po
ij
N
l li = 0,
l=1
j = 1, . . . , J,
N N
ij o T,i
i=1 =Go j
+
i=1
RT ij RT ln i Po
N
ij
i=1 l=1
l li = 0,
j = 1, . . . , J, (5.142)
Go j
+ RT
i=1
RT ij ln i Po
l
l=1 i=1
li ij = 0,
=0
j = 1, . . . , J, (5.143)
Go + RT j
i=1
ij ln i
RT Po
= 0,
j = 1, . . . , J. (5.144)
Here, the stoichiometry for each reaction has been employed to remove the Lagrange multipliers. Continue to nd
N
ln i
i=1 N
RT Po
ij
=
ij
Go j RT
, Go j RT Go j RT
j = 1, . . . , J, , , j = 1, . . . , J, j = 1, . . . , J,
(5.145) (5.146) (5.147)
exp
i=1
RT ln i Po
N
= exp
ij
i=1
RT i Po
= exp
5.2. EQUILIBRIUM CONDITIONS RT Po

PN
i=1
209 Go j RT
i=1
ij N
i
i=1 N
ij
= exp = Po RT
j = 1, . . . , J, Go j RT ,
(5.148) j = 1, . . . , J, (5.149)
PN
ij
i
i=1 N
ij
exp
=Kc,j
i ij = Kc,j ,
i=1
j = 1, . . . , J.
(5.150)
Thus, extremization of Gibbs free energy is consistent with equilibrating each of the J reactions.
5.2.3
Zeldovichs uniqueness proof*
Here a proof is given for the global uniqueness of the equilibrium point in the physically accessible region of composition space following a procedure given in a little known paper by the great Russian physicist Zeldovich.5 The proof follows the basic outline of Zeldovich, but the notation will be consistent with the present development. Some simplications were available to Zeldovich but not employed by him. For further background see Powers and Paolucci.6 5.2.3.1 Isothermal, isochoric case
Consider a mixture of ideal gases in a closed xed volume V at xed temperature T . For such a system, the canonical equilibration relation is given by Eq. (3.421), dA|T,V 0. So A must be always decreasing until it reaches a minimum. Consider then A. First, combining Eqs. (3.300) and (3.301), one nds A = P V + G. Now from Eq. (3.424) one can eliminate G to get
N
(5.151)
A = P V +
ni i .
i=1 N i=1
(5.152) ni , one gets (5.153)
From the ideal gas law, P V = nRT , and again with n =

N
A = nRT +
5
ni i ,
i=1
Zeldovich, Ya. B., 1938, A Proof of the Uniqueness of the Solution of the Equationsfor the Law of Mass Action, Zhurnal Fizicheskoi Khimii, 11: 685-687. 6 Powers, J. M., and Paolucci, S., 2008, Uniqueness of Chemical Equilibria in Ideal Mixtures of Ideal Gases, American Journal of Physics, 76(9): 848-855. CC BY-NC-ND. 12 December 2011, J. M. Powers.
210

N N
ni RT +
i=1 N i=1
ni o + RT ln T,i 1 + ln ni P nPo RT Po V ,
Pi Po
(5.154) (5.155)
= RT
i=1 N
ni ni
i=1
o T,i RT o T,i RT
= RT
1 + ln ni
(5.156)
For convenience, dene, for this isothermal isochoric problem no , the total number of moles at the reference pressure, which for this isothermal isochoric problem, is a constant: no So A = RT
i=1 N
Po V . RT 1 + ln ni no .
(5.157)
ni
o T,i RT
(5.158)
Now recall that the atomic element conservation, Eq. (5.61), demands that
N
i=1
li (nio ni ) = 0,
l = 1, . . . , L.
(5.159)
As dened earlier, li is the number of atoms of element l in species i; note that li 0. It is described by a L N non-square matrix, typically of full rank, L. Dening the initial number of moles of element l, l = N li nio , one can rewrite Eq. (5.159) as i=1
N
li ni = l ,
i=1
l = 1, . . . , L.
(5.160)
Equation (5.160) is generally under-constrained, and one can nd solutions of the form D1 N L D12 D11 n1o n1 D N D n2 n2o D21 . = . + . 1 + .22 2 + . . . + 2 . L N L . (5.161) . . . . . . . . . . DN 2 DN 1 nN o nN DN N L Here, Dik represents a dimensionless component of a matrix of dimension N (N L). Here, in contrast to earlier analysis, i is interpreted as having the dimensions of kmole. Each of the N L column vectors of the matrix whose components are Dik has length N and lies in the right null space of the matrix whose components are li . That is,
N
i=1
li Dik = 0,
k = 1, . . . , N L, l = 1, . . . , L.
(5.162)
211
There is an exception to this rule when the left null space of ij is of higher dimension than the right null space of li . In such a case, there are quantities conserved in addition to the elements as a consequence of the form of the reaction law. The most common exception occurs when each of the reactions also conserves the number of molecules. In such a case, there will be N L 1 free variables, rather than N L. One can robustly form Dij from the set of independent column space vectors of ij . These vectors are included in the right null space of li since N liij = 0. i=1 One can have Dik (, ). Each of these is a function of li . In matrix form, one can say D11 n1o n1 n2 n2o D21 . = . + . . . . . . . nN o nN DN 1 In index form, this becomes
N L
D12 D22 . . . DN 2
... ... . . .
D1 D2
N L
. . .
N L
. . . DN
N L
1 2 . . . . N L
(5.163)
ni = nio +
k=1
Dik k ,
i = 1, . . . , N.
(5.164)
It is also easy to show that the N L column space vectors in Dik are linear combinations of N L column space vectors that span the column space of the rank-decient N J components of ij . The N values of ni are uniquely determined once N L values of k are specied. That is, a set of independent k , k = 1, . . . , N L, is sucient to describe the system. This insures the initial element concentrations are always maintained. One can also develop a J-reaction generalization of the single reaction Eq. (4.99). Let kj , k = 1, . . . , N L, j = 1, . . . , J, be the extension of k . Then the appropriate generalization of Eq. (4.99) is
N L
ij =
k=1
Dik kj .
(5.165)
In Gibbs notation, one would say = D . One can nd the matrix by the following operations D = , D D = DT ,
T
(5.166)
(5.167) (5.168)
1
DT D
DT .
(5.169)
This calculation has only marginal utility.

212
Returning to the primary exercise, note that one can form the partial derivative of Eq. (5.164): ni p
N L
=
j k=1 N L
Dik
k p
,
j
i = 1, . . . , N; p = 1, . . . , N L, i = 1, . . . , N; p = 1, . . . , N L,
(5.170) (5.171) (5.172)
=
k=1
Dik kp ,
= Dip ,
i = 1, . . . , N; p = 1, . . . , N L.
Here kp is the Kronecker delta function. Next, return to consideration of Eq. (5.156). It is sought to minimize A while holding T and V constant. The only available variables are ni , i = 1, . . . , N. These are not fully independent, but they are known in terms of the independent p , p = 1, . . . , N L. So one can nd an extremum of A by dierentiating it with respect to each of the p and setting each derivative to zero: A p
N
= RT
j ,T,V i=1 N
ni p ni p
o T,i
j
RT o T,i
1 +
ni p ni p
ln
j
ni ni + ni ln o no p n ni + ni no
N N
= 0, (5.173) = 0, (5.174)
p = 1, . . . , N L,
q=1
= RT
i=1
RT
1 +
ln
j
ni nq ln o p nq n
= RT
i=1
Dip
o T,i RT o T,i RT o T,i RT
p = 1, . . . , N L, Dqp 1 iq ni = 0,
1 + Dip ln
ni + ni no
q=1
= RT
i=1 N
Dip
1 + Dip ln ni no
ni + Dip no
p = 1, . . . , N L, = 0,
(5.175)
p = 1, . . . , N L, = 0, p = 1, . . . , N L.
(5.176) (5.177)
= RT
i=1
Dip
+ ln
Following Zeldovich, one then rearranges Eq. (5.177) to dene the equations for equilibrium:
N
i=1 N
ni Dip ln o n ni ln o n
Dip
= =
Dip o T,i RT RT Dip o T,i
, ,
i=1 N
p = 1, . . . , N L, (5.178) p = 1, . . . , N L, (5.179)
i=1
i=1

N
213
Dip N
ln
i=1 N N L
ni no 1 no
= =
Dip o T,i RT Dip o T,i RT
i=1 N
p = 1, . . . , N L, (5.180) p = 1, . . . , N L. (5.181)
Dip
ln
i=1
nio +
k=1
Dik k
i=1
Equation (5.181) forms N L equations in the N L unknown values of p and can be solved by Newton iteration. Note that as of yet, no proof exists that this is a unique solution. Nor is it certain whether or not A is maximized or minimized at such a solution point. One also notices that the method of Zeldovich is consistent with a more rudimentary form. Rearrange Eq. (5.177) to get A p
N
=
j ,T,V i=1 N
Dip o + RT ln T,i Dip o + RT ln T,i Dip o + RT ln T,i i Dip .
ni no ni P nPo Pi Po
= 0, , ,
p = 1, . . . , N L,
(5.182) (5.183) (5.184) (5.185)
=
i=1 N
=
i=1 N
=
i=1
Now, Eq. (5.185) is easily found via another method. Recall Eq. (3.304), and then operate on it in an isochoric, isothermal limit, taking derivative with respect to p , p = 1, . . . , N L:
N
dA = SdT P dV +
N
i dni ,
i=1
(5.186) (5.187) (5.188) (5.189)
dA|T,V A p
=
i=1 N
i dni , i
i=1 N
=
j ,T,V
ni , p
=
i=1
i Dip .
One can determine whether such a solution, if it exists, is a maxima or minima by examining the second derivative, given by dierentiating Eq. (5.177). We nd the Hessian7 ,
7
after Ludwig Otto Hesse, 1811-1874, German mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
214 H, to be
2A = RT H= j p = RT
i=1 N
Dip Dip Dip
ni ln o , j n ln ni ln no , j j 1 ni , ni j
(5.190) (5.191) (5.192) (5.193)
i=1 N
= RT
i=1 N
= RT
i=1
Dip Dij . ni
Here j = 1, . . . , N L. Scaling each of the rows of Dip by a constant does not aect the rank. So we can say that the N (N L) matrix whose entries are Dip / ni has rank N L, and consequently the Hessian H, of dimension (N L) (N L), has full rank N L, and is symmetric. It is easy to show by means of singular value decomposition, or other methods, that the eigenvalues of a full rank symmetric square matrix are all real and non-zero. Now consider whether 2 A/j p is positive denite. By denition, it is positive denite if for an arbitrary vector zi of length N L with non-zero norm that the term T V , dened below, be always positive:
N L N L
T V =
j=1 p=1
2A zj zp > 0. j p
(5.194)
Substitute Eq. (5.193) into Eq. (5.194) to nd

N L N L N
T V
=
j=1 p=1 N
RT
i=1
Dip Dij zj zp , ni Dip Dij zj zp ,

N L
(5.195)
= RT
i=1 N
1 ni 1 ni 1 ni
N L N L
(5.196)
j=1 p=1 N L
= RT
i=1 N
j=1
Dij zj
p=1
Dip zp ,
N L
(5.197)
N L
= RT
i=1
j=1
Dij zj
p=1
Dip zp .
(5.198)
Dene now
N L
yi
j=1
Dij zj ,
i = 1, . . . , N.
(5.199)
5.2. EQUILIBRIUM CONDITIONS This yields

N
215
T V
= RT
i=1
2 yi . ni
(5.200)
Now, to restrict the domain to physically accessible space, one is only concerned with ni 0, T > 0, so for arbitrary yi , one nds T V > 0, so one concludes that the second mixed partial of A is positive denite globally. We next consider the behavior of A near a generic point in the physical space where A = A. If we let d represent , we can represent the Helmholtz free energy by the Taylor series 1 A() = A + d T A + dT H d + . . . , (5.201) 2 where A and H are evaluated at . Now if = eq , where eq denotes an equilibrium point, eq A = 0, A = A , and 1 A() Aeq = d T H d + . . . . (5.202) 2 Because H is positive denite in the entire physical domain, any isolated critical point will be a minimum. Note that if more than one isolated minimum point of A were to exist in the domain interior, a maximum would also have to exist in the interior, but maxima are not allowed by the global positive denite nature of H. Subsequently, any extremum which exists away from the boundary of the physical region must be a minimum, and the minimum is global. Global positive deniteness of H alone does not rule out the possibility of non-isolated multiple equilibria, as seen by the following analysis. Because it is symmetric, H can be orthogonally decomposed into H = QT Q, where Q is an orthogonal matrix whose columns are the normalized eigenvectors of H. Note that QT = Q1 . Also is a diagonal matrix with real eigenvalues on its diagonal. We can eect a volume-preserving rotation of axes by taking the transformation dw = Q d; thus, d = QT dw. Hence, 1 1 1 AAeq = (QT dw)T HQT dw = dwT QQT QQT dw = dwT dw. (5.203) 2 2 2 The application of these transformations gives in the neighborhood of equilibrium the quadratic form N L 1 eq AA = p (dwp )2 . (5.204) 2 p=1 For A to be a unique minimum, p > 0. If one or more of the p = 0, then the minimum could be realized on a line or higher dimensional plane, depending on how many zeros are present. The full rank of H guarantees that p > 0. For our problem the unique global minimum which exists in the interior will exist at a unique point.
216
Lastly, one must check the boundary of the physical region to see if it can form an extremum. Near a physical boundary given by nq = 0, one nds that A behaves as
N nq 0
lim A RT
ni
i=1
T,i RT
1 + ln
ni no
nite.
(5.205)
The behavior of A itself is nite because limnq 0 nq ln nq = 0, and the remaining terms in the summation are non-zero and nite. The analysis of the behavior of the derivative of A on a boundary of nq = 0 is more complex. We require that nq 0 for all N species. The hyperplanes given by nq = 0 dene a closed physical boundary in reduced composition space. We require that changes in nq which originate from the surface nq = 0 be positive. For such curves we thus require that near nq = 0, perturbations dk be such that
N L
dnq =
k=1
Dqk dk > 0.
(5.206)
We next examine changes in A in the vicinity of boundaries given by nq = 0. We will restrict our attention to changes which give rise to dnq > 0. We employ Eq. (5.182) and nd that
N L
dA =
p=1
A dp p
=
T,V,j=p i=1
o i
ni + RT ln o n
N L
p=1
Dip dp .
(5.207)
On the boundary given by nq = 0, the dominant term in the sum is for i = q, and so on this boundary N L nq Dqp dp . (5.208) lim dA = RT ln o nq 0 n p=1
>0
The term identied by the brace is positive because of Eq. (5.206). Because R and T > 0, we see that as nq moves away from zero into the physical region, changes in A are large and negative. So the physical boundary can be a local maximum, but never a local minimum in A. Hence, the only admissible equilibrium is the unique minimum of A found from Eq. (5.181); this equilibrium is found at a unique point in reduced composition space. 5.2.3.2 Isothermal, isobaric case
Next, consider the related case in which T and P are constant. For such a system, the canonical equilibration relation is given by Eq. (3.422) holds that dG|T,P 0. So G must be decreasing until it reaches a minimum. So consider G, which from Eq. (3.424) is
N
G =
i=1
ni i ,
(5.209)

N
217 Pi Po ni P nPo P Po
=
i=1 N
ni o + RT ln T,i ni o + RT ln T,i
i=1 N
, , + ni ln ni N k=1 .
(5.210) (5.211) (5.212)
= = RT
ni
i=1
o T,i RT
+ ln
nk
Next, dierentiate with respect to each of the independent variables p for p = 1, . . . , N L: G = RT p = RT

i=1 N N
i=1 N
ni p ni p
o T,i RT o T,i RT
+ ln + ln
P Po P Po ni N k=1 nk P Po
ni ln
ni N k=1
nk
(5.213)
+
q=1 N
nq p nq Dip
ni ln o T,i RT + ln
(5.214)
= RT
i=1 N
+
q=1 N
Dqp Dip Dip ni n Dip Dqp Dip

N
ni N k=1 nk o T,i RT o T,i RT
+ iq P Po P Po
1 + ln
ni N k=1 nk ni n
N
, Dqp ,
(5.215)
= RT
i=1 N
+ ln
ni + 1 + ln n + 1 + ln ni n
(5.216)
q=1
= RT
i=1 N
+ ln
RT = RT
i=1
q=1
Dqp , + ln P Po + 1 + ln ni n
(5.217)
o T,i RT
N
i=1 N
RT = RT
q=1
i=1
ni , n + ln P Po + 1 + ln ni n
N
(5.218) RT Dqp ,
o T,i RT
(5.219)
i=1
q=1
218

N
= RT
i=1 N
Dip Dip
o T,i RT o T,i RT
+ ln + ln
P Po P Po ni P nPo ni P nPo
+ 1 + ln + ln , , ni n
ni n ,
RT
i=1
Dip ,
(5.220) (5.221) (5.222) (5.223) (5.224)
= RT
i=1 N
=
i=1 N
Dip o + RT ln T,i Dip o + RT ln T,i i Dip ,
=
i=1 N
=
i=1
p = 1, . . . , N L.
Note this simple result is entirely consistent with a result that could have been deduced by commencing with the alternative Eq. (3.426), dG|T,P = N i dni . Had this simplication i=1 been taken, one could readily deduce that G p
N
=
j i=1 N
ni , p
p = 1, . . . N L, p = 1, . . . N L.
(5.225) (5.226)
=
i=1
i Dip ,
Now, to equilibrate, one sets the derivatives to zero to get

N
i=1
Dip o + RT ln T,i
N
ni P nPo ni P nPo ni P nPo

Dip
= 0, p = 1, . . . , N L,
N
(5.227) (5.228) (5.229)
ln
i=1 N
=
Dip
i=1 N
Dip Dip Dip
o T,i RT o T,i RT o T,i RT
, ,
ln
N
ln
i=1
i=1
= Dip =
i=1 N
nio +
N q=1
N L k=1
Dik k P Dqk k Po
nqo +
N L k=1
i=1
p = 1, . . . , N L. (5.230)
These N L non-linear algebraic equations can be solved for the N L unknown values of p via an iterative technique. One can extend the earlier analysis to show that the equilibrium is unique in the physically accessible region of composition space.
219
We can repeat our previous analysis to show that this equilibrium is unique in the physically accessible region of composition space. By dierentiating Eq. (5.226) it is seen that 2G = RT j p = RT
i=1 N N
i=1 N
Dip
ni , ln j n
N
(5.231) Dip ln n , j ,
N
ln ni Dip j Dip Dip Dip 1 ni ni j 1 ni ni j 1 ni ni j

N
(5.232) (5.233)
i=1
= RT
i=1 N
i=1 N
Dip Dip
1 n n j 1 n j Dip
= RT
i=1 N
nq
q=1
(5.234)
i=1 N
= RT
i=1 N
i=1 N N
q=1
1 nq n j .
(5.235)
= RT
i=1
1 Dip Dij ni n
i=1 q=1
Dip Dqj
(5.236)
Next consider the sum

N L N L
T P =
j=1 p=1
2G zj zp = RT j p
N L N L
j=1 p=1
i=1
1 Dip Dij ni n
i=1 q=1
Dip Dqj
zj zp . (5.237)
We use Eq. (5.199) and following a long series of calculations, reduce Eq. (5.237), to the positive denite form T P RT = n
N N
i=1 j=i+1
nj yi ni
ni yj nj
> 0.
(5.238)
It is easily veried by direct expansion that Eqs. (5.237) and (5.238) are equivalent. On the boundary ni = 0, and as for the isothermal-isochoric case, it can be shown that dG for changes with dni > 0. Thus, the boundary has no local minimum, and we can conclude that G is minimized in the interior and the minimum is unique. 5.2.3.3 Adiabatic, isochoric case
One can extend Zeldovichs proof to other sets of conditions. For example, consider a case which is isochoric and isoenergetic. This corresponds to a chemical reaction in an xed volume which is thermally insulated. In this case, one operates on Eq. (3.422):
N
dE
=0
= P dV +T dS +
=0
i dni ,
i=1
(5.239)
220

N
0 = T dS +
i=1
i dni , i dni ,
(5.240) (5.241)
dS =
1 T
1 = T = S j = = = = = = = 1 T 1 T 1 T 1 T 1 T 1 T 1 T 1 T
N
i=1 N
N L
i
i=1 k=1 N N L
Dik dk ,
(5.242) (5.243) (5.244) (5.245) (5.246) Pi Po ni P nPo ni RT To Po V To T To Dij , Dij , Dij , ni RTo Po V Dij , (5.247) (5.248) (5.249) (5.250)
i=1 k=1 N N L
i Dik dk , i Dik k , j
i=1 k=1 N N L
i=1 k=1 N
i Dik kj ,
i=1 N
i Dij , o + RT ln T,i
i=1 N
o + RT ln T,i
i=1 N
o + RT ln T,i
i=1 N
o + RT ln T,i
i=1
+ RT ln
i=1 N
o T,i + R ln T
T To
+R ln
ni RTo Po V
i (T )
= 2S k j =
i (T ) + R ln
i=1 N
ni RTo Po V
Dij ,
(5.251)
Dij , Dij ,
(5.252) (5.253)
i=1
R ni i (T ) T + T k ni k

N
221
i=1
R i (T ) T + Dik Dij . T k ni
(5.254)
Now, consider dT for the adiabatic system.

N
E = Eo =
q=1 N
nq eq (T ), eq nq dT + eq dnq , T q=1
=cvq N
(5.255)
dE = 0 =
(5.256)
0 =
q=1
(nq cvq dT + eq dnq ) ,

N q=1 eq dnq , N nq cvq q=1
(5.257) (5.258)
dT = = =
1 eq ncv q=1 1 ncv q=1
N L
p=1
Dqp dp ,
(5.259)
N N L
p=1
eq Dqp dp , eq Dqp p , k
(5.260)
T 1 = k ncv q=1 = = 1 ncv q=1

N
N N L
(5.261)
p=1
N N L
p=1
eq Dqp pk ,
(5.262)
1 eq Dqk . ncv q=1
(5.263)
Now return to Eq. (5.254), using Eq. (5.263) to expand: 2S k j

N
=
i=1 N
R i (T ) 1 eq Dqk Dik T ncv q=1 ni i (T ) 1 eq Dqk Dij T ncv q=1

N N N N
Dij ,
(5.264)
=
i=1
i=1
R Dik Dij , ni
N
(5.265)
1 ncv
i=1 q=1
i (T ) eq Dqk Dij R T
i=1
Dik Dij . ni
(5.266)
222
Now consider the temperature derivative of i (T ), where i (T ) is dened in Eq. (5.251): o T T,i , (5.267) + R ln i T To o di R 1 do T,i T,i = 2 + + . (5.268) dT T T dT T Now o = hT,i T so , T,i T,i = hTo ,i +
o T To
o
(5.269)
T
cP i (T )dT T
T
so o ,i + T
To
c P i (T ) dT , T
(5.270)
=hT,i
=so T,i
do T,i = cP i so o ,i T dT
T
= so o ,i T = so . T,i
To
c P i (T ) d T cP i , T To c P i (T ) dT , T
(5.271) (5.272) (5.273)
Note that Eq. 5.273 is a special case of the Gibbs equation given by Eq. 3.317. With this, one nds that Eq. 5.268 reduces to o so di R T,i T,i = 2 + , (5.274) dT T T T 1 (5.275) = 2 o + T so RT , T,i T,i T 1 (5.276) = 2 g o + T so RT , T,i T T,i = = = = = So substituting Eq. (5.281) into 2S k j = 1 o hT,i RT , 2 T 1 2 ei + Pi v i RT , T 1 2 ei + RT RT , T 1 2 ei . T Eq. (5.266), one gets 1 ncv T 2
N N
1 o h T so +T so RT , T,i T,i 2 T,i T

=go T,i
(5.277)
(5.278) (5.279) (5.280) (5.281)
i=1 q=1
ei eq Dqk Dij R
i=1
Dik Dij . ni
(5.282)
5.2. EQUILIBRIUM CONDITIONS Next, as before, consider the sum
223
N L N L
k=1 j=1
1 2S zk zj = k j ncv T 2 R =
N L N L N
k=1 j=1 i=1 q=1
ei eq Dqk Dij zk zj (5.283)
N L N L N
k=1 j=1 i=1 N
Dik Dij zk zj , ni eq Dqk zk ei Dij zj
1 ncv T 2
N N L N L
i=1 q=1 k=1 j=1
N N L N L
R =
i=1 k=1 j=1 N N
Dik zk Dij zj , ni
N L N L
(5.284)
1 ncv T 2
N
i=1 q=1
k=1
eq Dqk zk
N L
j=1
ei Dij zj , eq Dqk zk , (5.286)

N L 2
i=1
1 ni
N L
k=1
Dik zk ei Dij zj
j=1
Dij zj
N N L
(5.285)
1 = ncv T 2
N
N N L
i=1 j=1
q=1 k=1 N L
i=1
1 ni
N L
k=1
Dik zk ei Dij zj
2
1 = ncv T 2 = 1 ncv T 2
N N L
j=1 2
Dij zj
N
i=1 j=1 N
R
N 2 yi . ni
i=1
1 ni
k=1
Dik zk
(5.287) ,
ei yi
i=1
(5.288)
i=1
Since cv > 0, T > 0, R > 0, ni > 0, and the other terms are perfect squares, it is obvious that the second partial derivative of S < 0; consequently, critical points of S represent a maximum. Again near the boundary of the physical region, S ni ln ni , so limni 0 S 0. From Eq. (5.263), there is no formal restriction on the slope at the boundary. However, if a critical point is to exist in the physical domain in which the second derivative is guaranteed negative, the slope at the boundary must be positive everywhere. This combines to guarantee that if a critical point exists in the physically accessible region of composition space, it is unique.
224 5.2.3.4
CHAPTER 5. THERMOCHEMISTRY OF MULTIPLE REACTIONS Adiabatic, isobaric case
A similar proof holds for the adiabatic-isobaric case. Here the appropriate Legendre transformation is H = E + P V , where H is the enthalpy. We omit the details, which are similar to those of previous sections, and nd a term which must be negative semi-denite, HP :
N L N L
HP =
k=1 j=1
2S zk zj k j
N 2
(5.289) R n
N N
1 ncP T 2
hi yi
i=1
i=1 j=i+1
nj yi ni
ni yj nj
(5.290)
Because cP > 0 and ni 0, the term involving hi yi is a perfect square, and the term multiplying R is positive denite for the same reasons as discussed before. Hence, HP < 0, and the Hessian matrix is negative denite.
5.3
Concise reaction rate law formulations
One can employ notions developed in the Zeldovich uniqueness proof to obtain a more ecient representation of the reaction rate law for multiple reactions. There are two important cases: 1) J (N L); this is most common for large chemical kinetic systems, and 2) J < (N L); this is common for simple chemistry models. The species production rate is given by Eq. (5.14), which reduces to di 1 = dt V
J
ij
j=1
dj , dt
i = 1, . . . , N.
(5.291)
Now dierentiating Eq. (5.164), one obtains

N L
dni =
k=1
Dik dk ,
i = 1, . . . , N.
(5.292)
Comparing then Eq. (5.292) to Eq. (5.23), one sees that

J N L
ij dj =
j=1 k=1
Dik dk , Dik dk ,
i = 1, . . . , N,
(5.293)
1 V
ij dj =
j=1
1 V
N L
i = 1, . . . , N,
(5.294) (5.295)
k=1
5.3. CONCISE REACTION RATE LAW FORMULATIONS
225
5.3.1
Consider rst the most common case in which J (N L). One can say the species production rate is given 1 di = dt V
N L
Reaction dominant: J (N L)
J
k=1
dk = Dik dt
ij rj ,
j=1
i = 1, . . . , N.
(5.296)
One would like to invert and solve directly for dk /dt. However, Dik is non-square and has no inverse. But since N li Dip = 0, and N li ij = 0, L of the equations N equations i=1 i=1 in Eq. (5.296) are redundant. At this point, it is more convenient to go to a Gibbs vector notation, where there is an obvious correspondence between the bold vectors and the indicial counterparts: d 1 d = D = r, (5.297) dt V dt d = V DT r, (5.298) DT D dt d = V (DT D)1 DT r. (5.299) dt Because of the L linear dependencies, there is no loss of information in this matrix projection. This system of N L equations is the smallest number of dierential equations that can be solved for a general system in which J > (N L). Lastly, one recovers the original system when forming D d = V D (DT D)1 DT r. dt
=P
(5.300)
Here the N N projection matrix P is symmetric, has norm of unity, has rank of N L, has N L eigenvalues of value unity, and L eigenvalues of value zero. And, while in general, application of a projection matrix to r loses some of the information in r, because of the nature of the linear dependencies, no information is lost in Eq. (5.300) relative to the original Eq. (5.297). Note nally that it can be shown that D, of dimension N (N L), and , of dimension N J, share the same column space, which is of dimension (N L); consequently, both matrices map vectors into the same space.
5.3.2
Next consider the case in which J < (N L). This often arises in models of simple chemistry, for example one- or two-step kinetics. The fundamental reaction dynamics are most concisely governed by the J equations which form 1 d = r. (5.301) V dt
Species dominant: J < (N L)
226
However, r is a function of the concentrations; one must therefore recover as a function of reaction progress . In vector form, Eq. (5.291) is written as d 1 d = . dt V dt (5.302)
Take as an initial condition that the reaction progress is zero at t = 0 and that there are an appropriate set of initial conditions on the species concentrations : = 0, = o , t = 0, t = 0. (5.303) (5.304)
Then, since is a constant, Eq. (5.302) is easily integrated. After applying the initial conditions, Eq. (5.304), one gets = o + 1 . V (5.305)
Last, if J = (N L), either approach yields the same number of equations, and is equally concise.
5.4
Onsager reciprocity*
There is a powerful result from physical chemistry which speaks to how systems behave near equilibrium. The principle was developed by Onsager8 in the early twentieth century, and is known as the reciprocity principle. Where it holds, one can guarantee that on their approach to equilibrium, systems will approach the equilibrium in a non-oscillatory manner. It will be illustrated here. Note that Eq. (5.17) can be written as 1 dj = rj = Rj R , j = 1, . . . , J, j V dt R j = Rj 1 , j = 1, . . . , J. Rj Note further that the denition of anity gives exp Therefore, one can say 1 dj = rj = Rj 1 exp V dt
8
(5.306) (5.307)
j RT
R j = , Rj j RT
j = 1, . . . , J.
(5.308)
j = 1, . . . , J.
(5.309)
Lars Onsager, 1903-1976, Norwegian-American physicist.
5.4. ONSAGER RECIPROCITY*
227
Now, as each reaction comes to equilibrium, one nds that j 0, so a Taylor series expansion of rj yields j 1 dj + ... = rj Rj 1 1 V dt RT j Rj , j = 1, . . . , J. RT , j = 1, . . . , J, (5.310) (5.311)
Note that Rj > 0, while j can be positive or negative. Note also there is no summation on j. Take now the matrix R to be the diagonal matrix with Rj populating its diagonal: R1 0 R . . . 0 Adopt the vector notation r= 1 R . RT (5.313) 0 R2 . . . 0 ... ... .. . 0 0 0 . . . .
(5.312)
RJ
Now the entropy production is given for multi-component systems by Eq. (5.56): dS dt V = T
J
E,V
j=1
j rj 0.
(5.314)
Cast this entropy inequality into Gibbs notation: dS dt =

E,V
V T r 0. T
(5.315)
Now consider the denition of anity, Eq. (5.10), in Gibbs notation: = T . (5.316)
Now T is of dimension J N with rank typically N L. Because T is typically not of full rank, one nds only N L of the components of to be linearly independent. When one recalls that T maps vectors into the column space of T , one recognizes that can be represented as = C . (5.317) Here C is a J (N L)-dimensional matrix of full rank, N L, whose N L columns are populated by the linearly independent vectors which form the column space of T , and is a column vector of dimension (N L) 1. If J N L, one can explicitly solve for ,
228
starting by operating on both sides of Eq. (5.317) by CT : CT = CT C , T T C C = C , 1 = CT C CT , = C = C CT C

B
(5.318) (5.319) (5.320)

T
= C C
C ,
1
(5.321) (5.322)
CT T .
Here, in the recomposition of , one can employ the J J symmetric projection matrix B, which has N L eigenvalues of unity and J (N L) eigenvalues of zero. The matrix B has rank N L, and is thus not full rank. Substitute Eqs. (5.313, 5.317) into Eq. (5.315) to get
=
T
dS dt
E,V
V = C T = V R T
T
= 1 R C 0, RT =r
(5.323)
CT R C
L
0.
(5.324)
Since each of the entries of the diagonal R are guaranteed positive semi-denite in the physical region of composition space, the entropy production rate near equilibrium is also positive semi-denite. The constant square matrix L, of dimension (N L) (N L), is given by L CT R C. (5.325) The matrix L has rank N L and is thus full rank. Because R is diagonal with positive semidenite elements, L is symmetric positive semi-denite. Thus its eigenvalues are guaranteed to be positive and real. Note that o-diagonal elements of L can be negative, but that the matrix itself remains positive semi-denite. Onsager reciprocity simply demands that near equilibrium, the linearized version of the combination of the thermodynamic forces (here the anity ) and uxes (here the reaction rate r) be positive semi-denite. Upon linearization, one should always be able to nd a positive semi-denite matrix associated with the dynamics of the approach to equilibrium. Here that matrix is L, and by choices made in its construction, it has the desired properties. One can also formulate an alternative version of Onsager reciprocity using the projection matrix B, which from Eq. (5.322), is B C CT C
1
CT .
(5.326)
229
With a series of straightforward substitutions, it can be shown that the entropy production rate given by Eq. (5.324) reduces to dS dt =
E,V
V R
T T
BT R B
L
0.
(5.327)
One can also express the entropy generation directly in terms of the chemical potential rather than the anity by dening the J N matrix S as S B T , = C (CT C)1 C T .
Here, an alternative symmetric positive semi-denite matrix L, of dimension J J and rank N L, has been dened as L BT R B. (5.328)
(5.329) (5.330)
The matrix S has rank N L and thus is not full rank. With a series of straightforward substitutions, it can be shown that the entropy production rate given by Eq. (5.327) reduces to V T dS ST R S 0. (5.331) = dt E,V T R T
L
Here, an alternative symmetric positive semi-denite matrix L, of dimension N N and rank N L, has been dened as L ST R S. (5.332) Example 5.4
Find the matrices associated with Onsager reciprocity for the reaction mechanism given by H2 + O2 H2 + OH H + O2 H2 + O H +H 2OH H2 O2 H + OH 2OH, H + H2 O, O + OH, H + OH, H2 , O + H2 O, H + H, O + O, H2 O. (5.333) (5.334) (5.335) (5.336) (5.337) (5.338) (5.339) (5.340) (5.341)
Here there are N = 6 species (H, H2 , O, O2 , OH, H2 O), composed of L = 2 elements (H, O), reacting in J = 9 reactions. Here J N L, so the analysis of this section can be performed. Take species i = 1 as H, i = 2 as H2 ,. . ., i = N = 6 as H2 O. Take element l = 1 as H and element l = L = 2 as O. The full rank stoichiometric matrix , of dimension 2 6 = L N and rank 2 = L, is = 1 0 2 0 0 1 0 1 2 1 2 1 . (5.342)
230

The rank-decient matrix of stoichiometric coecients , of dimension 69 = N J and rank 4 = N L is 0 1 1 1 2 0 2 0 1 0 1 0 0 1 1 0 1 1 0 1 1 0 1 0 2 0 0 = (5.343) . 0 0 0 1 0 1 0 1 0 2 1 1 1 0 2 0 0 1 0 1 0 0 0 1 0 0 1 Check for stoichiometric balance: 0 1 1 1 2 0 2 0 1 1 1 0 1 1 0 1 1 0 1 0 0 0 0 0 1 0 1 0 2 1 1 1 0 2 0 0 1 0 0 0 1 0 0 0 0 0 . 0 1 0 0 2 0 , 1 0 0 1 0 1
1 0
2 0 0 1
0 1 2 1
2 1
(5.344)
0 0
0 0 0 0
0 0 0 0
0 0 0 0
(5.345)
So the number and mass of every element is conserved in every reaction; every vector in the column space of is in the right null space of . The detailed version of the reaction kinetics law is given by d dt = r, (5.346)
r1 r2 0 1 1 1 2 0 2 0 1 r 0 1 0 0 3 1 1 0 1 1 r4 0 1 1 0 1 0 2 0 0 = r , 0 0 0 1 0 5 1 0 1 0 r6 2 1 1 1 0 2 0 0 1 r7 0 1 0 0 0 1 0 0 1 r8 r9 r2 r3 + r4 2r5 + 2r7 r9 r1 r2 r4 + r5 r7 r3 r4 + r6 + 2r8 = . r1 r3 r8 2r1 r2 + r3 + r4 2r6 r9 r2 + r6 + r9
(5.347)
(5.348)
The full rank matrix D, of dimension 6 4 = N (N L) and rank 4 = N L, is composed of vectors in the right null space of . It is non-unique, as linear combinations of right null space vectors suce. It is equivalently composed by casting the N L linearly independent vectors of the column space of in its columns. Recall that some of the columns of are linearly dependent. In the present example, the rst N L = 4 column vectors of happen to be linearly dependent, and thus will not suce. Other sets are not; the last N L = 4 column vectors of happen to be linearly independent CC BY-NC-ND. 12 December 2011, J. M. Powers.

and thus suce for the present purposes. Take then 0 2 0 1 0 0 1 0 0 2 0 1 D= . 0 1 0 0 2 0 0 1 1 0 0 1
231
(5.349)
Since = 0 and D = 0, one concludes that the column spaces of both and D are one and the same. The non-unique concise version of the reaction kinetics law is given by d dt = V (DT D)1 DT r, 0 1 0 2 1 0 = V 0 2 0 1 0 0 (5.351) 1 1 0 0 0 0 2 1 0 0 0 1 1 0 1 r1 r2 1 r 0 3 r4 0 r , 0 5 r6 1 r7 1 r8 r9 1 0 2 0 0 0 2 1 0 0 0 1 1 0 0 1
It is easily veried by direct substitution that D is in the right null space of : 0 2 0 1 0 0 1 0 1 2 0 0 1 2 0 2 0 0 0 0 0 1 D= = 0 0 1 2 1 1 0 1 0 0 0 0 0 0 2 0 0 1 1 0 0 1
(5.350)
0 2 0 0 2 1 0 1 0 0 0 0 1 0 2 1 0 0 0 0 1 0 1 1 2 0 0 1 0 0
0 1 1 1 2 0 2 0 0 1 0 1 1 0 1 1 0 1 1 0 1 0 2 0 0 0 0 1 1 0 1 0 2 1 1 1 0 2 0 0 0 1 0 0 0 1 0 0 2r1 r3 r4 + r6 r + r2 + r4 r5 + r7 = V 1 . r1 + r3 + r8 2r1 + r2 + r3 + r4 + r9
(5.352)
(5.353)
The rank-decient projection matrix P, of dimension 6 6 = N N and rank 4 = N L, is P = D DT D 54 14 14 61 3 = 61 6 61 4

61 11 61 61 61 33 61 6 61 12 61 8 61 22 61 1 3 61 6 61 51 61 20 61 7 61 4 61
DT ,
(5.354)
4 61 8 61 7 61 14 61 50 61 15 61 11 61 22 61 4 61 8 61 15 61 35 61
6 61 12 61 20 61 21 61 14 61 8 61
(5.355)
232

The projection matrix P has 4 = N L eigenvalues of unity and 2 = L eigenvalues of zero. The anity vector , of dimension 9 1 = J 1, is given by = T , 0 1 1 1 2 1 1 0 1 1 0 1 1 0 0 = 0 1 0 1 0 2 1 1 1 0 0 1 0 0 0 0 1 0 1 2 0 1 1 1 0 1 1 1 1 0 1 1 1 1 0 = 2 1 0 0 0 0 1 0 2 0 0 0 2 1 0 0 0 2 1 0 1 0 0 0 1 2 + 4 25 1 + 2 + 5 6 1 3 + 4 5 1 + 2 + 3 5 = 21 2 . 3 + 25 6 21 + 2 23 + 4 1 + 5 6
(5.356) 1 0 2 0 1 0 1 0 0 2 1 0 2 0 3 , 0 0 1 0 4 5 2 0 0 1 6 1 0 0 1 0 1 1 0 2 0 0 3 , 1 4 5 0 6 0 1 T
(5.357)
(5.358)
(5.359)
The full rank matrix C, of dimension 9 4 = J (N L) and rank 4 = N L, is composed of the set of N L = 4 linearly independent column space vectors of T ; thus they also comprise the N L = 4 linearly independent row space vectors of . It does not matter which four are chosen, so long as they are linearly independent. In this case, the rst four column vectors of T suce: 0 1 1 1 C = 2 0 2 0 1 1 0 1 1 0 0 0 1 1 1 1 0 1 0 0 . 0 1 0 1 0 0 0 2 1 0 0 0
(5.360)
When J > N L, not all of the components of are linearly independent. In this case, one can form the reduced anity vector, , of dimension 4 1 = (N L) 1 via = (CT C)1 CT , (5.361)

= 1 0 1 0 1 0 1 1 0 1 0 1 1 0 2 0 1 1 1 1 0 0 1 0 0 0 0 2 1 1 0 2 0 0 1 0 0 0 0 1 0 0 2 1 0 0 0 2 1 1 0 0 0 2 + 4 25 1 + 2 + 5 6 1 3 + 4 5 2 0 1 1 + 2 + 3 5 1 0 0 21 2 , 0 2 0 3 + 25 6 0 1 0 21 + 2 23 + 4 1 + 5 6
233
1 1 1 2 0 1 1 0 1 1 0 1 1 0 0 0 1 0 1 0 0 1 1 1 0 0 1 0 1 1 2 0 0 1 1 0 1 1 0 1 1 0 0 0
(5.362)
1 5 + 6 2 25 + 26 . 3 + 25 6 4 + 45 26
(5.363)
The rank-decient projection matrix B, of dimension 9 9 = J J and rank 4 = N L, is found to be B = C (CT C)1 CT , 57 9 31
94 94 41 94 5 94 14 94 15 94 27 94 15 94 4 94 26 94 94 5 94 35 94 4 94 11 94 1 94 11 94 28 94 6 94 26 94 14 94 4 94 30 94 12 94 16 94 12 94 22 94 2 94 1 94 15 94 11 94 12 94 33 94 3 94 33 94 10 94 18 94 17 94 27 94 1 94 16 94 3 94 43 94 3 94 18 94 24 94 1 94 15 94 11 94 12 94 33 94 3 94 33 94 10 94 18 94 6 94 4 94 28 94 22 94 10 94 18 94 10 94 60 94 14 94 8 94 26 94 6 94 2 94 18 94 24 94 18 94 14 94 44 94
9 94 31 94 26 94 = 1 94 17 94 1 94 6 94 8 94
(5.364)
(5.365)
The projection matrix B has a set of 9 = J eigenvalues, 4 = N L of which are unity, and 5 = J (N L) of which are zero. One can recover by the operation = C = B T . The square full rank Onsager matrix L, of dimension 44 = (N L)(N L) and rank 4 = N L, is given by L = CT R C, 0 0 0 0 0 0 0 0 R9 (5.366)
0 1 1 1 2 0 1 1 0 1 1 0 0 0 1 1 0 1 1 0 1 0 0 0
R 1 0 0 2 0 1 0 1 0 0 0 0 2 0 0 0 1 0 0 0 0
0 R 2 0 0 0 0 0 0 0
0 0 R 3 0 0 0 0 0 0
0 0 0 R 4 0 0 0 0 0
0 0 0 0 R 5 0 0 0 0
0 0 0 0 0 R 6 0 0 0
0 0 0 0 0 0 R 7 0 0
0 0 0 0 0 0 0 R 8 0
234
0 1 1 1 2 0 2 0 1 R 2

1 0 1 1 0 0 0 1 1 1 1 0 1 0 0 , 0 1 0 1 0 0 0 2 1 0 0 0 R 9 R R 2R 2R 2 4 5 7 R + R + R + R + R 1 2 4 5 7 R 4 R 1 R 4 R + R + R + 4R 3 4 6 8 R 2R 3 8 R 3 R 4 R 3 R 1
+ R + R + 4R + 4R + 3 4 5 7 R R 2R 2R 2 4 5 7 R R 3 4 R 3
. R 2R 3 8 R + R + R 1 3 8 (5.368)
(5.367)
Obviously L is symmetric, and thus has all real eigenvalues. It is also positive semi-denite. With a similar eort, one can obtain the alternate rank-decient square Onsager matrices L and L. Recall that L, L, and L each have rank N L, while L has the smallest dimension, (N L) (N L), and so forms the most ecient Onsager matrix.
5.5
Irreversibility production rate*
This section will be restricted to isothermal, isochoric reaction. It is known that Gibbs free energy and irreversibility production rates reach respective minima at equilibrium. Consider the gradient of the irreversibility production rate in the space of species progress variables k . As discussed extensively by Prigogine9 , there is a tendency for systems to relax to a state which, near equilibrium, minimizes the production rate of irreversibility. Moreover, in the neighborhood of equilibrium, the irreversibility production rate can be considered a Lyapunov10 function. First, recall Eq. (5.226) for the gradient of Gibbs free energy with respect to the independent species progress variables, G/k = N i Dik . Now, beginning from Eq. (4.464), i=1 dene the dierential irreversibility dI as 1 dI = T
N
i dni .
i=1
(5.369)
In terms of an irreversibility production rate, one has dI 1 =I= dt T

9 10
i
i=1
dni . dt
(5.370)
Ilya Prigogine, 1917-2003, Russian-Belgian chemist. Aleksandr Mikhailovich Lyapunov, 1857-1918, Russian mathematician.
5.5. IRREVERSIBILITY PRODUCTION RATE* Take now the time derivative of Eq. (5.164) to get dni = dt
N L
235
k=1
Dik
dk . dt
(5.371)
Substitute from Eq. (5.371) into Eq. (5.370) to get 1 I = T = 1 T

N N L
i
i=1 N L
k=1 N dk
Dik
dk , dt
(5.372) (5.373)
k=1
dt
i=1
i Dik ,
k
G =
1 T
N L
k=1
dk G . dt k
(5.374)
Now Eq. (5.299) gives an explicit algebraic formula for dk /dt. Dene then the constitutive function k : dk = V (DT D)1 DT r. (5.375) k (1 , . . . , N L) dt So the irreversibility production rate is 1 I= T The gradient of this eld is given by I 1 = p T The Hessian of this eld is given by 1 2I = l p T
N L N L N L
k=1
G . k k
(5.376)
k=1
k G 2 G + k p k p k
(5.377)
k=1
2 k G k 2 G k 2G 3G + + + k l p k p l k l p k l p k
(5.378)
e Now at equilibrium, k = k , we have k = 0 as well as G/k = 0. Thus
I I k 2I l p
e k =k
= 0, = 0,
(5.379) (5.380)
N L
e k =k
e k =k
1 = T
k=1
k 2 G k 2G + p l k l p k
.
e k =k
(5.381)
236
It can be shown for arbitrary matrices that the rst term in the right hand side of Eq. (5.381) is the transpose of the second; moreover, for arbitrary matrices, the two terms are in general asymmetric. But, their sum is a symmetric matrix, as must be since the Hessian of I must be symmetric. But for arbitrary matrices, we nd no guarantee that the Hessian of I is positive semi-denite, as the eigenvalues can take any sign. However, it can be veried by direct expansion for actual physical systems that the two terms in the Hessian evaluated at equilibrium are identical. This must be attributable to extra hidden symmetries in the system. Moreover for actual physical systems, the eigenvalues of the Hessian of I are positive semi-denite. So one can say 2I l p = 2 T
N L
e k =k
k=1
k p
e k =k
2G l k
.
e k =k
(5.382)
Where does the hidden symmetry arise? The answer lies in the fact that the k and G are not independent. Using notions from Onsager reciprocity, let us see why the Hessian of I is indeed positive semi-denite. Consider, for instance, Eq. (5.324), which is valid in the neighborhood of equilibrium: V I= R T T L T valid only near equilibrium. (5.383)
Recall that by construction L is positive semi-denite. Now, near equilibrium, has a Taylor series expansion ( |eq ) + .... (5.384) = eq +
=0 eq
Now recall that at equilibrium that = 0. Let us also dene the Jacobian of as J . Thus Eq. (5.383) can be rewritten as
T V J ( |eq ) L J ( |eq ) , RT 2 V = ( |eq )T JT L J ( |eq ). RT 2 By inspection, then we see that the Hessian of I is
I =
(5.385) (5.386)
HI =
2V T J L J . RT 2
(5.387)
Moreover L has the same eigenvalues as the similar matrix JT L J , so HI is positive semi-denite. Returning to the less rened formulation, in Gibbs notation, we can write Eq. (5.382) as 2 HI = HG J, T
(5.388)
5.5. IRREVERSIBILITY PRODUCTION RATE*
237
where J is the Jacobian matrix of k . Consider the eigenvectors and eigenvalues of the various matrices here. The eigenvalues of J give the time scales of reaction in the neighborhood of equilibrium. They are guaranteed real and negative. The eigenvectors of J give the directions of fast and slow modes. Near equilibrium, the dynamics will relax to the slow mode, and the motion towards equilibrium will be along the eigenvector associated with the slowest time scale. Now, in the unlikely circumstance that HG were the identity matrix, one would have the eigenvalues of HI equal to the product of 2/T and the eigenvalues of HG . So they would be positive, as expected. Moreover, the eigenvectors of HI would be identical to those of J, so in this unusual case, the slow dynamics could be inferred from examining the slowest descent down contours of I. Essentially the same conclusion would be reached if HG had a diagonalization with equal eigenvalues on its diagonal. This would correspond to reactions proceeding at the same rate near equilibrium. However, in the usual case, the eigenvalues of HG are non-uniform. Thus the action of HG on J is to stretch it non-uniformly in such a fashion that HI does not share the same eigenvalues or eigenvectors. Thus it cannot be used to directly infer the dynamics. Now consider the behavior of I in the neighborhood of an equilibrium point. In the neighborhood of a general point k = k , I has a Taylor series expansion
N L
I=I
k =k
+
k=1
I k
(5.389) has the behavior Near equilibrium, the rst two terms of this Taylor series are zero, and I 1 I= 2
N L N L
k =k
1 k k + 2
N L N L
l=1 p=1
l l
2I l p
k =k
p p + . . .
l=1 p=1
(l
le )
2I l p
e k =k
e p p + . . .
(5.390)
Substituting from Eq. (5.382), we nd near equilibrium that 1 I = T

N L N L N L
l=1 p=1 k=1
(l le )
k p
e k =k
2G l k
e k =k
e p p + . . .
(5.391)
e Lastly, let us study whether I is a Lyapunov function. We can show that I > 0, p = p , e and I = 0, p = p . Now to determine whether or not the Lyapunov function exists, we must study dI/dt:
dI dt
N L
=
p=1 N L
I dp , p dt I p , p
(5.392)
=
p=1
(5.393)
238

N L
=
p=1
1 T
N L
k=1
k G 2 G + k p k p k
2 k G G k p + p p k p k
p, .
(5.394)
1 T
N L N L
(5.395)
p=1 k=1
At the equilibrium state, k = 0, so obviously dI/dt = 0, at equilibrium. Away from equilibrium, we know from our uniqueness analysis that the term 2 G/p k is positive semi-denite; so this term contributes to rendering dI/dt < 0. But the other term does not transparently contribute. Let us examine the gradient of dI/dt. It is dI m dt 1 = T +
N L N L
p=1 k=1
2 k G k 2G k G p p + p + m p k p m k p k m . (5.396)
2 p 2G 3G G k k + p k + p m p k m p k p k m
At equilibrium, all terms in the gradient of dI/dt are zero, so it is a critical point. Let us next study the Hessian of dI/dt to ascertain the nature of this critical point. Because there are so many terms, let us only write those terms which will be non-zero at equilibrium. In this limit, the Hessian is 2 dI n m dt 1 = T
N L N L
eq
p=1 k=1
k 2 G p k 2 G p + p m k n p n k m . (5.397)
p 2G k p 2G k + + m p k n n p k m
With some eort, it may be possible to show the total sum is negative semi-denite. This would guarantee that dI/dt < 0 away from equilibrium. If so, then it is true that I is a Lyapunov function in the neighborhood of equilibrium. Far from equilibrium, it is not clear whether or not a Lyapunov function exists.
Chapter 6 Reactive Navier-Stokes equations

Here we will present the compressible reactive Navier1 -Stokes2 equations for an ideal mixture of N gases. We will not give detailed derivations. We are guided in part by the excellent derivations given by Aris (1929-2005) 3 and Merk.4
6.1
6.1.1
Evolution axioms
Conservative form
The conservation of mass, linear momenta, and energy, and the entropy inequality for the mixture are expressed in conservative form as + (u) = 0, t t 1 e+ uu 2 (u) + (uu + P I ) = 0, t 1 + u e + u u + jq + (P I ) u = 0, 2 jq 0. (s) + su + t T (6.1) (6.2) (6.3) (6.4)
New variables here are the velocity vector u, the viscous shear tensor , the diusive heat ux vector jq . Note that these equations are precisely the same one would use for a single uid. We have neglected body forces.
Claude-Louis Navier, 1785-1836, French engineer. George Gabriel Stokes, 1819-1903, Anglo-Irish mathematician. 3 R. Aris, 1962, Vectors, Tensors, and the Basic Equations of Fluid Mechanics, Dover, New York. 4 H. J. Merk, 1959, The macroscopic equations for simultaneous heat and mass transfer in isotropic continuous and closed systems, Applied Scientic Research, 8(1): 73-99.
2 1
239
240
CHAPTER 6. REACTIVE NAVIER-STOKES EQUATIONS
The evolution of molecular species is dictated by the evolution axiom, an extension of Eq. (5.22) to additionally account for advection and diusion: (Yi ) + (Yi u + jm ) = Mi i, i t i = 1, . . . , N 1. (6.5)
Here, the diusive mass ux vector of species i is jm . Note that together, Eqs. (6.1) and the i N 1 of Eq. (6.5) form N equations for the evolution of the N species. We insist that the species diusive mass ux be constrained by
N
jm = 0. i
i=1
(6.6)
Recalling our earlier denitions of Mi and i , Eqs. (5.1) and (5.19), respectively, Eq. (6.5) can be rewritten as (Yi ) + (Yi u + jm ) = i t Let us sum Eq. (6.7) over all species to get (Yi ) + t i=1
N N N L J L J
l=1
Ml li
ij rj .
j=1
(6.7)
i=1
(Yi u +
N N
jm ) i
=
i=1 N l=1 L
Ml li
J
ij rj ,
j=1
(6.8)
i=1
=1
Yi + u
Yi +
i=1 =1 i=1
=0
jm i
=
i=1 l=1 j=1 J L
Ml li ij rj ,
N
(6.9)
=
j=1
rj
l=1
Ml
liij ,
i=1 =0
(6.10) (6.11)
+ (u) = 0. t
So, the summation over all species gives a redundancy with Eq. (6.1). We can get a similar relation for the elements. Let us multiply Eq. (6.5) by our stoichiometric matrix li to get (Yi ) + li (Yi u + jm ) = li Mi i , i t J li Yi li Yi jm + = li ij rj , u + li i Mi Mi Mi j=1 li (6.12) (6.13)
6.1. EVOLUTION AXIOMS

N
241
N
i=1
li Yi Mi
+
i=1
li Yi jm u + li i Mi Mi
N
=
i=1 J
li
j=1 N
ij rj , (6.14) li ij , (6.15)
i=1 =0
t t Ml t t
i=1 N
li Yi Mi li Yi Mi
+ +
i=1 N
liYi u+ Mi liYi u+ Mi liYi u+ Mi
i=1 N
jm li i Mi li jm i Mi jm i Mi jm i Mi
=
j=1
rj
i=1 N
i=1 N
= 0,
(6.16)
i=1 N
li Yi Mi
i=1 N
+ Ml
i=1 N
li
i=1 N
= 0, l = 1, . . . , L, (6.17)
Ml
i=1
li Yi Mi
Ml
i=1
li Yi u + Ml Mi
li
i=1
= 0, l = 1, . . . , L, (6.18)
Let us now dene the element mass fraction Yle , l = 1, . . . , L as

N
Yle Note that this can be expressed as Yle mass element l = moles element l
=Ml N
Ml
i=1
liYi . Mi
(6.19)
i=1
moles element l mass species i mole species i , (6.20) moles species i total mass mass species i
=li =Yi =1/Mi
mass element l . = total mass Similarly, we take the diusive element mass ux to be
N
(6.21)
je l
Ml
li
i=1
jm i , Mi
l = 1, . . . , L.
(6.22)
Substitute Eqs. (6.19,6.22) into Eq. (6.18) to get (Yle ) + (Yle u + je ) = 0. l t We also insist that
L
(6.23)
je = 0, l
l=1
(6.24)
242
so as to keep the total mass of each element constant. This is easily seen to be guaranteed. Sum Eq. (6.22) over all elements to get
L L N
je = l
l=1 l=1 N
Ml
L
li
i=1
jm i , Mi
(6.25) (6.26) (6.27)
= =
i=1 l=1 N jm i i=1
jm Ml li i , Mi
L
Mi
l=1
Ml li ,
=Mi
=
i=1 N
jm i Mi , Mi jm , i
(6.28) (6.29) (6.30)
=
i=1
= 0.
In summary, we have L 1 conservation equations for the elements, one global mass conservation equation, and N L species evolution equations, in general. These add to form N equations for the overall species evolution.
6.1.2
Non-conservative form
It is often convenient to have an alternative non-conservative form of the governing equations. Let us dene the material derivative as d = + u . dt t 6.1.2.1 Mass (6.31)
Using the product rule to expand the mass equation, Eq. (6.1), we get + (u) = 0, t + u + u = 0, t
=d/dt
(6.32) (6.33)
d + u = 0. dt
(6.34)
6.1. EVOLUTION AXIOMS 6.1.2.2 Linear momentum
243
We again use the product rule to expand the linear momentum equation, Eq. (6.2): u + u + u (u) + u u + P = 0, t t u + u u +u + (u) +P = 0, t t
=du/dt =0
(6.35) (6.36)
du + P = 0. dt
(6.37)
The key simplication was eected by using the mass equation Eq. (6.1). 6.1.2.3 Energy
Now use the product rule to expand the energy equation, Eq. (6.3). t 1 1 1 e + u u + u e + u u + e + u u 2 2 2 + (u) t
=0
1 1 e + u u + u e + u u + jq + (P u) ( u) = 0. 2 2
+ jq + (P u) ( u) = 0,
(6.38) (6.39)
We have once again used the mass equation, Eq. (6.1), to simplify. Let us expand more using the product rule: e + u e +u t
=de/dt
u + u u + jq + (P u) ( u) = 0, t + jq + P u : u = 0,
(6.40)
de +u dt
u + u u + P t
=0
(6.41)
de + jq + P u : u = 0. dt
(6.42)
We have used the linear momentum equation, Eq. (6.37) to simplify. From the mass equation, Eq. (6.34), we have u = (1/)d/dt, so the energy equation, Eq. (6.42), can also be written as P d de + jq : u = 0. dt dt (6.43)
244
This energy equation can be formulated in terms of enthalpy. Use the denition h = e + P/ to get an expression for dh/dt: P 1 d + dP, 2 de P d 1 dP dh = + , dt dt 2 dt dt de P d dh 1 dP 2 = , dt dt dt dt dh dP de P d = . dt dt dt dt dh = de So the energy equation, Eq. (6.43), in terms of enthalpy is 6.1.2.4 Second law dh dP + jq : u = 0. dt dt (6.48) (6.44) (6.45) (6.46) (6.47)
The second law in non-conservative form is, by inspection 6.1.2.5 Species ds jq + 0. dt T (6.49)
The species evolution equation in non-conservative form, is by inspection 6.1.2.6 Elements dYi + jm = Mi i, i dt i = 1, . . . , N 1. (6.50)
The element evolution equation in non-conservative form, is by inspection dYie + je = 0, l dt l = 1, . . . , L. (6.51)
6.2
Mixture rules
N
We adopt the following rules for the ideal mixture: P =

i=1
Pi ,
(6.52)
6.3. CONSTITUTIVE MODELS

N
245
1 =
i=1 N
Yi ,
N
(6.53) i , Mi (6.54) (6.55) (6.56) (6.57) (6.58) (6.59) (6.60) (6.61)
=
i=1 N
i =
i=1
e =
i=1 N
Yi ei , Yi hi ,
i=1 N
h = s =
i=1 N
Y i si , Yi cvi ,
i=1 N
cv = cP =
i=1
Y i cP i ,
V = Vi , T = Ti .
6.3
Constitutive models
The evolution axioms do not form a complete set of equations. Let us supplement these by a set of constitutive model equations appropriate for a mixture of calorically perfect ideal gases that react according the to the law of mass action with an Arrhenius kinetic reaction rate. We have seen many of these models before, and repeat them here for completeness. For the thermal equation of state, we take the ideal gas law for the partial pressures: Pi = RT i = RT So the mixture pressure is
N N
Yi i = RT = Ri T i . Mi Mi
(6.62)
P = RT
i=1
i = RT
i=1
Yi = RT Mi
i=1
i . Mi
(6.63)
For the ideal gas, the enthalpy and internal energy of each component is a function of T at most. We have for the enthalpy of a component
T
hi =
ho o ,i T
+
To
cP i(T )dT .
(6.64)
246 So the mixture enthalpy is
h =
i=1
Yi ho o ,i + T
cP i (T )dT .
(6.65)
To
We then use the denition of enthalpy to recover the internal energy of component i: ei = hi So the mixture internal energy is
N T
Pi = hi Ri T. i
(6.66)
e =
i=1 N
Yi
ho o ,i T
Ri T +
cP i(T )dT ,
T
(6.67) cP i (T )dT , (6.68) (6.69) (6.70) (6.71)
To
=
i=1 N
Yi ho o ,i Ri (T To ) Ri (To ) + T
T
To
=
i=1 N
Yi ho o ,i Ri (To ) + T Yi ho o ,i Ri (To ) + T
T
To T To
(cP i(T ) Ri )dT , cvi (T )dT ,
=
i=1 N
=
i=1
Yi eo o ,i + T
cvi (T )dT .
To
The mixture entropy is

N T
s =
i=1
Yi so o ,i T
+
To
cP (T ) dT T
Yi Ri ln
i=1
Pi Po
(6.72)
The viscous shear stress for an isotropic Newtonian uid which satises Stokes assumption is = 2 u + (u) 1 ( u)I . 2 3 (6.73)
Here is the mixture viscosity coecient which is determined from a suitable mixture rule averaging over each component. The energy ux vector jq is written as
N N
j = kT +
jm hi i
i=1
RT
i=1
T Di Mi
yi Mi + 1 yi M
P P
(6.74)
6.4. TEMPERATURE EVOLUTION
247
T Here k is a suitably mixture averaged thermal conductivity. The parameter Di is the socalled thermal diusion coecient. Recall yi is the mole fraction. We consider a mass diusion vector with multicomponent diusion coecient Dik : N
jm i
=
k=1,
Mi Dik Yk M k=i
yk Mk 1 yk M
P P
T Di
T . T
(6.75)
We adopt, as before, the reaction rate of creation of species i

J
i =
j=1
ij rj .
(6.76)
Here rj is given by the law of mass action:

N
rj = k j
k=1
kkj
1 Kc,j
kkj . ,
k=1
(6.77)
kj is given by the Arrhenius kinetics rule kj = aj T j exp E j RT , (6.78)
and the equilibrium constant Kc,j for the ideal gas mixture is given by Kc,j = Po RT
PN
i=1
ij
exp
Go j RT
(6.79)
6.4
Temperature evolution
Because temperature T has strong physical meaning, let us formulate our energy conservation principle as a temperature evolution equation by employing a variety of constitutive laws. Let us begin with Eq. (6.48) coupled with our constitutive rule for h, Eq. (6.65): d Yi ho o ,i + T dt i=1
N T To =hi
=h
dP cP i(T )dT + jq : u = 0, dt d dt
N
(6.80)
i=1
Yi hi + jq
dP : u = 0, dt
(6.81)
248
N
CHAPTER 6. REACTIVE NAVIER-STOKES EQUATIONS dYi dhi + hi dt dt dT dYi + hi dt dt

N
Yi
i=1 N
+ jq + jq + jq
dP : u = 0, dt dP : u = 0, dt dP : u = 0, dt dP : u = 0. dt
(6.82) (6.83) (6.84)
Yi cP i dT dt
i=1 N
Y i cP i +
i=1 =cP i=1
hi
dYi dt
=Mi i jm i N
dT cP + dt
i=1
hi (Mi i jm ) + jq i
(6.85)
Now let us dene the thermal diusion ux jT as

N
j RT
i=1
T Di Mi
yi Mi + 1 yi M
P P
(6.86)
With this, our total energy diusion ux vector jq , Eq. (6.74), becomes
N
j = kT +
jm hi + jT . i
i=1
(6.87)
Now substitute Eq. (6.87) into Eq. (6.85) and rearrange to get dT + cP dt
N N
i=1
hi (Mi i
jm ) i
kT +
jm hi + jT i
i=1
cP
dT + dt
i=1
dP + : u,(6.88) dt dP (hi Mi i hi jm + (jm hi )) = (kT ) jT + + : u,(6.89) i i dt =

N
dT + cP dt cP dT + dt
i=1 N
(hi Mi i + jm hi )) = (kT ) jT + i
dP + : u,(6.90) dt dP + : u.(6.91) dt
i=1
(hi Mi i + cP i jm T ) = (kT ) jT + i
So the equation for the evolution of a material uid particle is dT = cP dt

N
i=1
(hi Mi i + cP ijm T ) + (kT ) jT + i
dP + : u. dt
(6.92)
6.5. SHVAB-ZELDOVICH FORMULATION
249
Let us impose some more details from the reaction law. First, we recall that hi = hi Mi , so dT = cP dt
N N
i=1
hi i
i=1
cP i jm T + (kT ) jT + i
dP + : u. dt
(6.93)
Impose now Eq. (6.76) to expand i dT cP = dt

N J N
hi
i=1 J j=1 N
ij rj hi ij
i=1 N
cP i jm T + (kT ) jT + i cP i jm T + (kT ) jT + i
dP + : u, (6.94) dt dP + : u. (6.95) dt
dT = rj cP dt j=1
i=1
i=1
Now let us extend the denition of heat of reaction, Eq. (4.444) to account for multiple reactions,
N
Hj
hi ij .
i=1
(6.96)
With this, Eq. (6.95) becomes, after small rearrangement, dP dT rj Hj + = cP dt dt j=1

J N
i=1
cP i jm T + (kT ) jT + : u . i diusion eects
(6.97)
By comparing Eq. (6.97) with Eq. (4.446), it is easy to see the eects of multiple reactions, variable pressure, and diusion on how temperature evolves. Interestingly, mass, momentum, and energy diusion all inuence temperature evolution. The non-diusive terms are combinations of advection, reaction, and spatially homogeneous eects.
6.5
Shvab-Zeldovich formulation
Under some restrictive assumptions, we can simplify the energy equation considerably. Let us assume the low Mach5 number limit is applicable, which can be shown to imply that pressure changes and work work due to viscous dissipation are negligible at leading order, dP/dt 0, : u 0, the incompressible limit applies, d/dt = 0,
5
Ernst Mach, 1838-1916, Austrian physicist. CC BY-NC-ND. 12 December 2011, J. M. Powers.
250
T thermal diusion is negligible, Di 0,
all species have identical molecular masses, so that Mi = M, the multicomponent diusion coecients of all species are equal, Dij = D, all species possess identical specic heats, cP i = cP , which is itself a constant, and energy diuses at the same rate as mass so that k/cP = D. With the low Mach number limit, the energy equation, Eq. (6.48), reduces to dh + jq = 0. dt (6.98)
T With Di = 0, the diusive energy ux vector, Eq. (6.74), reduces to N
jq = kT + Substituting Eq. (6.99) into Eq. (6.98), we get dh + dt
jm hi . i
i=1
(6.99)
kT +
jm hi i
i=1
= 0.
(6.100)
With all component specic heats equal and constant, cP i = cP , Eq. (6.65) for the mixture enthalpy reduces to
N
h =
i=1
Yi ho o ,i + cP (T To ) , T
N
(6.101) (6.102)
= cP (T To ) + Similarly for a component, Eq. (6.64) reduces to
Yi ho o ,i . T
i=1
hi = ho o ,i + cP (T To ). T With Eqs. (6.102,6.103), Eq. (6.100) transforms to dT d cP + dt dt

N N
(6.103)
Yi ho o ,i T
i=1
kT
i=1
jm ho o ,i + cP (T To ) i T
= 0, (6.104)
6.5. SHVAB-ZELDOVICH FORMULATION
251
d dT + cP dt dt
Yi ho o ,i T
i=1
kT
N
jm ho o ,i i T
i=1
cP (T To )
i=1
=0 N
jm i
= 0,
(6.105) = 0, (6.106)
cP
d dT + dt dt d dt
Yiho o ,i T
i=1 N i=1
kT
jm ho o ,i i T
i=1 N
cP
T+
Yiho o ,i T cP
kT
jm ho o ,i i T
i=1
= 0, (6.107)
With M = Mi , we recover mass fractions to be identical to mole fractions, Yk = yk . T Using this along with our assumptions of negligible thermal diusion, Di = 0 and equal multicomponent diusion coecients, Dij = D, the mass diusion ux vector, Eq. (6.75), reduces to
N
jm = i
k=1,k=i N
DYk , Yk , Yk ,
(6.108)
= D
k=1,k=i N
(6.109)
= D
k=1,k=i
(6.110)
= D (Yi + 1) , = DYi .
= D Yi +
Yk , k=1
N =1
(6.111)
(6.112) (6.113)
Now substitute Eq. (6.113) into Eq. (6.107) and use our equidiusion rate assumption, D = k/cP , to get d cP dt d cP dt T+ T+
N i=1
Yi ho o ,i T cP cP
kT + D kT + k cP
i=1 N
Yi ho o ,i T Yi ho o ,i T
= 0, = 0,
(6.114) (6.115)
N o i=1 Yi hTo ,i
i=1
252 d dt Now i
N i=1
CHAPTER 6. REACTIVE NAVIER-STOKES EQUATIONS Yi ho o ,i T cP k cP

N o i=1 Yi hTo ,i
T+
T+
cP
= 0.
(6.116)
there is a spatially uniform distribution of the quantity T +(1/cP ) and there is no ux of T + (1/cP ) all time,
N i=1
N o i=1 Yi hTo ,i
at t = 0,
Yi ho o ,i at the boundary of the spatial domain for T
then, Eq. (6.116) can be satised for all time by the algebraic relation T+
N o i=1 Yi hTo ,i
cP
= T (0) +
N i=1
Yi(0)ho o ,i T , cP
(6.117)
where T (0) and Yi(0) are constants at the spatially uniform initial state. We can slightly rearrange to write
N N
cP (T To ) +
Yi ho o ,i T
i=1
= cP (T (0) To ) +
Yi(0)ho o ,i , T
i=1
(6.118)
=h(T,Yi )
=h(T (0),Yi (0))
h(T, Yi ) = h(T (0), Yi(0)).
(6.119)
This simply says the enthalpy function is a constant, which can be evaluated at the initial state.
Chapter 7 Simple solid combustion: Reaction-diusion

Here we will modify the simple thermal explosion theory of an earlier chapter which balanced unsteady evolution against reaction to include the eects of diusion. Starting from a fully unsteady formulation, we will focus on cases which are steady, resulting in a balance between reaction and diusion; however, we will briey consider a balance between all three eects. Such a theory is known after its founder1 as Frank-Kamenetskii theory. In particular, we will look for transitions from a low temperature reaction to a high temperature reaction, mainly in the context of steady state solutions, i.e. solutions with no time dependence. We draw upon the work of Buckmaster and Ludford for guidance.2
7.1
Simple planar model
Let us consider a slab of solid fuel/oxidizer mixture. The material is modelled of innite extent in the y and z directions and has length in the x direction of 2L. The temperature at each end, x = L, is held xed at T = To . The slab is initially unreacted. Exothermic conversion of solid material from reactants to products will generate an elevated temperature within the slab T > To , for x (L, L). If the thermal energy generated diuses rapidly enough, the temperature within the slab will be T To , and the reaction rate will be low. If the energy generated by local reaction diuses slowly, it will accumulate in the interior, and accelerate the local reaction rate, inducing rapid energy release and high temperature. Let us assume The material is an immobile, incompressible solid with constant specic heat, Thus is constant,
David Albertovich Frank-Kamenetskii, 1910-1970, Soviet physicist. Buckmaster, J. D., and Ludford, G. S. S., 1983, Lectures on Mathematical Combustion, SIAM, Philadelphia. See Chapter 1.
2 1
253
254
CHAPTER 7. SIMPLE SOLID COMBUSTION: REACTION-DIFFUSION Thus cP = cv is a constant, Thus, there is no advective transport of mass, momentum, or energy: u = 0.
The material has variation only with x and t, The reaction can be modelled as A B, where A and B have identical molecular masses. The reaction is irreversible, The reaction is exothermic, Initially only A is present, Let us take, as we did in thermal explosion theory, YA = 1 , YB = . (7.2) (7.3) The thermal conductivity, k is constant. (7.1)
We interpret as a reaction progress variable which has [0, 1]. For = 0, the material is all A; when = 1, the material is all B.
7.1.1
Model equations
= aeE/R/T (1 ), t e q = , t x T q = k , x e = cv T q.
Our simple model for reaction is (7.4) (7.5) (7.6) (7.7)
Equation (7.4) is our reaction kinetics law. It is the equivalent of the earlier derived Eq. (1.270) in the irreversible limit, Kc . Equation (7.5) is our energy conservation expression. It amounts to Eq. (6.42), with q playing the role of the heat ux jq . Equation (7.6) is the constitutive law for heat ux; it is the equivalent of Eq. (6.74) when mass diusion is neglected and amounts to Fouriers3 law. Equation (7.7) is our caloric equation of state; it is Eq. (1.303) with q = eo o ,A eo o ,B and YB = . Equations (7.4-7.7) are completed by T T initial and boundary conditions, which are T (L, t) = T (L, t) = To ,
3
T (x, 0) = To ,
(x, 0) = 0.
(7.8)
Jean Baptiste Joseph Fourier, 1768-1830, French mathematician.
7.1. SIMPLE PLANAR MODEL
255
7.1.2
Simple planar derivation
Let us perform a simple control volume derivation of the planar energy equation, Eq. (7.7). Consider a small volume of dimension A by x. At the left boundary x, we have heat ux q, in which we notate as qx . At the right boundary, x + x, we have heat ux out of qx+x . Recall the units of heat ux are J/m2 /s. The rst law of thermodynamics is change in total energy = heat in work out .
=0
(7.9)
There is no work for our system. But there is heat ux over system boundaries. In a combination of symbols and words, we can say total energy @ t + t total energy @ t = energy ux in energy ux out . unsteady advection and diusion
=0
(7.10)
Mathematically, we can say E|t+t E|t = (Ef lux

J/kg out
Ax e|t+t e|t
kg
= (qx+x qx ) At ,
J/m2 /s (m2 s)
Ef lux
in ) ,
(7.11) (7.12)
e|t+t e|t qx+x qx = t x
(7.13)
Now, let x 0 and t 0, and we get e qx = . t x T . x (7.14)
Now standard constitutive theory gives Fouriers law to specify the heat ux: q = k (7.15)
So Eq. (7.15) along with the thermal state equation, (7.7) when substituted into Eq. (7.14) yield 2T (cv T q) = k 2 , t x 2T T q = k 2, cv t t x T 2T qaeE/R/T (1 ) = k 2 , cv t x T 2T cv = k 2 + qaeE/R/T (1 ). t x (7.16) (7.17) (7.18) (7.19)
256
CHAPTER 7. SIMPLE SOLID COMBUSTION: REACTION-DIFFUSION
So our complete system is two equations in two unknowns with appropriate initial and boundary conditions: T 2T cv = k 2 + qaeE/R/T (1 ), t x = aeE/R/T (1 ), t T (x, 0) = 0, (x, 0) = 0. (7.20) (7.21) (7.22)
T (L, t) = T (L, t) = To ,
7.1.3
Ad hoc approximation
Let us consider an ad hoc approximation to a system much like Eqs. (7.20-7.22), but which has the advantage of being one equation and one unknown. 7.1.3.1 Planar formulation
If there were no diusion, Eq. (7.14) would yield e/t = 0 and would lead us to conclude that e(x, t) = e(x). And because we have nothing now to introduce a spatial inhomogeneity, there is no reason to take e(x) to be anything other than a constant eo . That would lead us to e(x, t) = eo , so eo = cv T q. Now at t = 0, = 0, and T = To , so eo = cv To ; thus, cv To = cv T q, cv (T To ) = . q We also get the nal temperature at = 1 to be T ( = 1) = To + q . cv (7.27) (7.25) (7.26) (7.24) (7.23)
We shall adopt Eq. (7.26) as our model for in place of Eq. (7.21). Had we admitted species diusion, we could more rigorously have arrived at a similar result, but it would be more dicult to justify treating the material as a solid. Let us use Eq. (7.26) to eliminate in Eq. (7.20) so as to get a single equation for T : cv T t = k 2T + qaeE/R/T x2 1 cv (T To ) , q (7.28)
reaction source term

7.2. NON-DIMENSIONALIZATION
257
The rst two terms in Eq. (7.28) are nothing more than the classical heat equation. The non-classical term is an algebraic source term due to chemical reaction. Note when T = To , the reaction source term is qa exp(E/R/T ) > 0, so at the initial state there is a tendency to increase the temperature. When = 1, so T = To + q/cv , the reaction source term is zero, so there is no local heat release. This eectively accounts for reactant depletion. Scaling the equation by cv and employing the well known formula for thermal diusivity = k//cP = k//cv , we get T 2T q = 2 +a (T To ) eE/R/T , t x cv T (L, t) = T (L, t) = To , T (x, 0) = To . (7.29) (7.30)
From here on, we shall consider Eqs. (7.30) and cylindrical and spherical variants to be the full problem. We shall consider solutions to it in various limits. We will not return here to the original problem without the ad hoc assumption, but that would be a straightforward exercise. 7.1.3.2 More general coordinate systems
We note that Eq. (7.30) can be extended to general coordinate systems via T t = 2 T + a q (T To ) eE/R/T . cv (7.31)
Appropriate initial and boundary conditions for the particular coordinate system would be necessary. For one-dimensional solutions in planar, cylindrical, and spherical coordinates, one can summarize the formulations as T t = 1 xm x xm T x +a q (T To ) eE/R/T . cv (7.32)
Here we have m = 0 for a planar coordinate system, m = 1 for cylindrical, and m = 2 for spherical. For cylindrical and spherical systems, the Dirichlet boundary condition at x = L would be replaced with a boundedness condition on T at x = 0.
7.2
Non-dimensionalization
Let us non-dimensionalize Eqs. (7.31). The scaling we will choose is not unique. Let us dene non-dimensional variables T = cv (T To ) , q x = x , L t = at. (7.33)
258
With these choices, Eqs. (7.31) transform to qa T q 1 m = x cv t cv L2 x m T = x t aL2 x xm T x 1 E q q T exp (7.34) , cv cv R To + (q/cv )T E 1 . (7.35) + (1 T ) exp RTo 1 + q T cv To +a E , RTo q . cv To
xm
T x
Now both sides are dimensionless. Let us dene the dimensionless parameters D= aL2 , = Q=
(7.36)
Here D is a so-called Damkhler4 number. If we recall a diusion time scale d is o d = L2 , (7.37)
and a rst estimate (which will be shown to be crude) of the reaction time scale, r is 1 r = . a (7.38)
We see that the Damkhler number is the ratio of the thermal diusion time to the reaction o time (ignoring activation energy eects!): D= L2 / d thermal diusion time = = . 1/a r reaction time (7.39)
We can think of and Q as ratios as well with = activation energy , ambient energy Q= exothermic heat release . ambient energy (7.40)
The initial and boundary conditions scale to T (1, t ) = T (1, t ) = 0, T (x , 0) = 0, if T (1, t ) = 0, T (0, t ) < , T (x , 0) = 0, m = 0, if m = 1, 2. (7.41) (7.42)
7.2.1
Diusion time discussion
As an aside, let us consider in more depth the thermal diusion time. Consider the dimensional heat equation T 2T = 2, t x
4
(7.43)
Gerhard Damkhler, 1908-1944, German chemist. o
7.2. NON-DIMENSIONALIZATION with a complex representation of a sinusoidal initial condition: T (x, 0) = Aeikx .
259
(7.44)
Here, A is a temperature amplitude, k is the wave number, and i2 = 1. If one wants, one can insist we only worry about the real part of the solution, but that will not be important for this analysis. Note that since
Aeikx = A cos(kx) + Ai sin(kx),
we can think of the initial condition as a signal with wavelength = 2/k. Note when the wave number k is large, the wavelength is small, and vice versa. Let us assume a separation of variables solution of the form T (x, t) = B(t)eikx . Thus T dB ikx = e , t dt and T x 2T x2 = ikBeikx , = k 2 Beikx .

(7.45)
(7.46)
(7.47) (7.48)
So our partial dierential equation reduces to eikx
dB = k 2 Beikx , dt dB = k 2 B, B(0) = A, dt 2 B(t) = Aek t .
(7.49) (7.50) (7.51)
Note that as t , B 0 with diusion time constant d = 1 . k 2
So, a large thermal diusivity causes modes to relax quickly. But high wave number k also induces modes to relax quickly. Note also that in terms of wavelength of the initial disturbance, 1 2 . d = 4 2 So small wavelength disturbances induce fast relaxation.
260
Importantly, recall that an arbitrary signal with have a Fourier decomposition into an innite number of modes, each with a dierent wave number. The above analysis shows that each mode will decay with its own diusion time constant. Indeed some of the modes, even at the initial state, may have negligibly small amplitude. But many will not, and we cannot know a priori which ones will be large or small. Lastly, we note that in a nite domain, we will nd a discrete spectrum of wave numbers; while in an innite domain, we will nd a continuous spectrum.
7.2.2
Final form
Let us now drop the notation and understand that all variables are dimensionless. So our equations become T T 1 m xm + (1 T ) exp , = x t D x x 1 + QT T (1, t) = T (1, t) = 0, T (x, 0) = 0, if m = 0, T (1, t) = 0, T (0, t) < , T (x, 0) = 0, if m = 1, 2.
(7.52)
Note that the initial and boundary conditions are homogeneous. The only inhomogeneity lives in the exothermic reaction source term, which is non-linear due to the exp(1/T ) term. Also, it will prove to be the case that a symmetry boundary condition at x = 0 suces, though our original formulation is more rigorous. Such equivalent boundary conditions are T (1, t) = 0, T (0, t) = 0, x T (x, 0) = 0, m = 1, 2, 3. (7.53)
7.2.3
Integral form
As an aside, let us consider the evolution of total energy within the domain. To do so we integrate a dierential volume element through the entire volume. We recall dV xm dx, for m = 0, 1, 2, (planar, cylindrical, spherical). xm
1
T 1 m m dx = x x t D x T dx = t =
1 0
xm T x
T x
dx + xm (1 T ) exp
1
1 + QT 1 + QT
dx, (7.54)
xm
0
1 D x 1 T D x
xm
dx +
0 1
xm (1 T ) exp
dx, (7.55)
xm T dx
0
+
x=1 0
xm (1 T ) exp
1 + QT
dx .
thermal energy change
boundary heat ux
internal conversion (7.56)
7.3. STEADY SOLUTIONS
261
Note that the total thermal energy in our domain changes due to two reasons: 1) diusive energy ux at the isothermal boundary, 2) internal conversion of chemical energy to thermal energy.
7.2.4
Innite Damkhler limit o
Note for D diusion becomes unimportant, and we recover a balance between unsteady eects and reaction: dT = (1 T ) exp dt 1 + QT , T (0) = 0. (7.57)
This is the problem we have already considered in thermal explosion theory. Let us here commence on a dierent route than that taken in the earlier chapter. We recall that thermal explosion theory predicts signicant acceleration of reaction when t e . Q (7.58)
7.3
Steady solutions
Let us seek solutions to the planar (m = 0) version of Eqs. (7.53) that are formally steady, so that /t = 0, and a balance between reaction and energy diusion is attained. In that limit, Eqs.(7.52) reduce to the following two point boundary value problem: 1 d2 T + (1 T ) exp 0 = D dx2 0 = T (1) = T (1). 1 + QT , (7.59) (7.60)
This problem is dicult to solve analytically because of the strong non-linearity in the reaction source term.
7.3.1
High activation energy asymptotics
Motivated by our earlier success in getting approximate solutions to a similar problem in spatially homogeneous thermal explosion theory, let us take a similar approach here. Let us seek low temperature solutions where the non-linearity may be weak. So let us dene a small parameter with 0 << 1. And now let us assume a power series expansion of T of the form T = T1 + 2 T2 + . . . (7.61)
Here we assume T1 (x) O(1), T2 (x) O(1), . . .. We will focus attention on getting an approximate solution for T1 (x).
262
With assumption Eq. (7.61), Eq. (7.59) expands to d2 T1 + . . . = D(1 T1 . . .) exp((1 QT1 + . . .)). dx2 (7.62)
Ignoring higher order terms and simplifying we get d2 T1 = D(1 T1 ) exp() exp(QT1 ). dx2 (7.63)
Now let us once again take the high activation energy limit and insist that be dened such that This gives us D d2 T1 = (1 T1 ) exp() exp(T1 ). 2 dx (7.65) 1 . Q (7.64)
We have gained an analytic advantage once again by moving the temperature into the numerator of the argument of the exponential. Now let us neglect T1 as small relative to 1 and dene a new parameter such that D exp() = DQ exp(). (7.66)
Our governing equation system then reduces to d2 T1 = eT1 , 2 dx T1 (1) = T1 (1) = 0. (7.67) (7.68)
It is still not clear how to solve this. It seems reasonable to assume that T1 should have symmetry about x = 0. If so, we might also presume that the gradient of T1 is zero at x = 0: dT1 dx = 0,
x=0
(7.69)
m which induces T1 (0) to take on an extreme value, say T1 (0) = T1 , where m could denote m maximum or minimum. Certainly T1 is unknown at this point. Let us explore the appropriate phase space solution. To aid in this, we can dene q as
q
dT1 . dx
(7.70)
263
We included the minus sign so that q has a physical interpretation of heat ux. With this assumption, our equation system can be written as two autonomous ordinary dierential equations in two unknowns: dq = eT1 , dx dT1 = q, dx q(0) = 0,
m T1 (0) = T1 .
(7.71) (7.72)
m This is slightly unsatisfying because we do not know T1 . But we presume it exists, and is a constant. Let us see if the analysis can reveal it by pressing forward. Let us scale Eq. (7.72) by Eq. (7.71) to get
dT1 q = T1 , dq e
m T1 |q=0 = T1 .
(7.73)
Now Eq. (7.73) can be solved by separating variables. Doing so and solving we get eT1 dT1 = qdq, q2 + C, eT1 = 2 applying the initial condition, m T1 e = C, 2 q m = (eT1 eT1 ), 2 q = 2(eT1 eT1 ).
m
(7.74) (7.75) (7.76) (7.77) (7.78) (7.79)
The plus square root is taken here. This will correspond to x [0, 1]. The negative square root will correspond to x [1, 0]. We can now substitute Eq. (7.79) into Eq. (7.72) to get dT1 = dx Once again separate variables to get dT1 2(eT1 eT1 )
m
2(eT1 eT1 ).
(7.80)
= dx.
(7.81)
Computer algebra reveals this can be integrated to form 2 tanh1 1 m eT1 /2

eT1 m eT1
= x + C.
(7.82)
264
m Now, when x = 0, we have T1 = T1 . The inverse hyperbolic tangent has an argument of zero there, and it evaluates to zero. Thus we must have C = 0, so
2 tanh1
1 m eT1 /2 1
eT1 m eT1
= x, = eT1
m /2
(7.83) x. 2 (7.84)
tanh1 Now recall that
eT1 m eT1
sech1 t = tanh1 so Eq. (7.84) becomes sech1 e

m T1 T1 2
1 t2 ,
= eT1
m /2
x, 2
m /2
(7.85) x , 2
m /2
m T1 T1 2
= sech eT1 = eT1

m /2
(7.86) (7.87) x 2 . (7.88)
T1 2
sech eT1
m /2
x , 2
m /2
T1 = 2 ln eT1
sech eT1
So we have an exact solution for T1 (x). This is a nice achievement, but we are not sure it m satises the boundary condition at x = 1, nor do we know the value of T1 . We must choose m T1 such that T1 (1) = 0, which we have not yet enforced. So 0 = 2 ln eT1
m /2
sech eT1
m /2
(7.89)
Only when the argument of a logarithm is unity does it map to zero. So we must demand that eT1
m /2
sech eT1
m /2
= 1, m = sech1 eT1 /2 , 2 m = cosh1 eT1 /2 , 2 m m = eT1 /2 cosh1 eT1 /2 . 2
(7.90) (7.91) (7.92) (7.93)
eT1 eT1
m /2
m /2

T1 m 8
265
0.2
0.4
0.6
0.8
m Figure 7.1: Plot of maximum temperature perturbation T1 versus reaction rate constant for steady reaction-diusion in the high activation energy limit
m m Equation (7.93) gives a direct relationship between and T1 . Given T1 , we get explicitly; thus we can easily generate a plot; see Fig 7.1. The inverse cannot be done analytically, but can be done numerically via iterative techniques. We notice a critical value of , c = m 0.878458. When = c = 0.878458, we nd T1 = 1.118684. For < c , there are two m admissible values of T1 . That is to say the solution is non-unique for < c . Moreover, both solutions are physical. For = c there is a single solution. For > c there are no steady solutions. Presumably re-introduction of neglected processes would aid in determining which solution is realized in nature. We shall see that transient stability analysis aids in selecting the correct solution when there are choices. We shall also see that reintroduction of reaction depletion into the model induces additional physical solutions, including those for > c . For = 0.4 < c we get two solutions. Both are plotted in Fig 7.2. The high temperature m m solution has T1 = 3.3079, and the low temperature solution has T1 = 0.24543. At this point we are not sure if either solution is temporally stable to small perturbations. We shall later prove that the low temperature solution is stable, and the higher temperature solution is unstable. Certainly for > c , there are no low temperature solutions. We will see that upon reintroduction of missing physics, there are stable high temperature solutions. To prevent high temperature solutions, we need < c . Now recall that
= DQe .
(7.94)
266

T1 3.0 2.5 2.0 1.5 1.0 0.5 x 1.0 0.5 0.5 1.0
Figure 7.2: Plot of T1 (x) for = 0.4 showing the two admissible steady solutions. Now to prevent the high temperature solution, we demand = DQe < 0.878458. Let us bring back our dimensional parameters to examine this criteria: DQe < 0.878458, L2 a q E E < 0.878458 exp cv To RTo RTo Thus the factors that tend to prevent thermal explosion are High thermal diusivity; this removes the thermal energy rapidly, Small length scales; the energy thus has less distance to diuse, Slow reaction rate kinetics, High activation energy. (7.96) (7.97) (7.95)
7.3.2
Method of weighted residuals
The method of the previous section was challenging. Let us approach the problem of calculating the temperature distribution with a powerful alternate method: the method of weighted residuals, using so-called Dirac5 delta functions as weighting functions. We will rst briey review the Dirac delta function, then move on to solving the Frank-Kamenetskii problem with the method of weighted residuals.
5
Paul Adrien Maurice Dirac, 1902-1984, English physicist.
7.3. STEADY SOLUTIONS The Dirac D -distribution (or generalized function, or simply function), is dened by

267
f (x)D (x a)dx =
0 if a [, ] f (a) if a [, ]
(7.98)
From this it follows by considering the special case in which f (x) = 1 that

D (x a) = 0 if x = a D (x a)dx = 1.
(7.99) (7.100)
Let us rst return to Eq. (7.67) rearranged as: d2 T1 + eT1 = 0, dx2 T1 (1) = T1 (1) = 0. We shall approximate T1 (x) by
N
(7.101) (7.102)
T1 (x) = Ta (x) =
i=1
ci fi (x).
(7.103)
At this point we do not know the so-called trial functions fi (x) nor the constants ci . 7.3.2.1 One-term collocation solution
Let us choose fi (x) to be linearly independent functions which satisfy the boundary conditions. Moreover let us consider the simplest of approximations for which N = 1. A simple function which satises the boundary conditions is a polynomial. The polynomial needs to be at least quadratic to be non-trivial. So let us take f1 (x) = 1 x2 . Ta (x) = c1 (1 x2 ). (7.104)
Note this gives f1 (1) = f1 (1) = 0. So our one-term approximate solution takes the form (7.105)
We still do not yet have a value for c1 . Let us choose c1 so as to minimize an error of our approximation. The error of our approximation e(x) will be e(x) = d2 Ta + eTa (x) , dx2 d2 c1 (1 x2 ) + exp(c1 (1 x2 )), = 2 dx = 2c1 + + exp(c1 (1 x2 )). (7.106) (7.107) (7.108)
268
If we could choose c1 in such a way that e(x) was exactly 0 for x [1, 1], we would be done. That will not happen. So let us choose c1 to drive a weighted domain-averaged error to zero. That is let us demand that
1
1 (x)e(x) dx = 0.
1
(7.109)
We have introduced here a weighting function 1 (x). Many choices exist for the weighting function. If we choose 1 (x) = Tm (x) our method is known as a Galerkin6 method. Let us choose instead another common weighting, 1 (x) = D (x). This method is known as a collocation method. We have chosen a single collocation point at x = 0. With this choice, Eq. (7.109) becomes
1 1
D (x) 2c1 + exp(c1 (1 x2 )) dx = 0.
(7.110)
The evaluation of this integral is particularly simple due to the choice of the Dirac weighting. We simply evaluate the integrand at the collocation point x = 0 and get 2c1 + + exp(c1 ) = 0. Thus = 2c1 ec1 . (7.112) (7.111)
Note here is a physical parameter with no relation the Dirac delta function D . Note that m with our approximation Ta = c1 (1 x2 ), that the maximum value of Ta is Ta = c1 . So we can say
m = 2Ta eTa .
m
(7.113)
m We can plot Ta as a function of ; see Fig 7.3. Note the predictions of Fig. 7.3 are remarkably similar to those of Fig. 7.1. For example, when = 0.4, numerical solution of Eq. (7.113) yields two roots: m Ta = 0.259171, m Ta = 2.54264.
(7.114)
Thus we get explicit approximations for the high and low temperature solutions for a oneterm collocation approximation: Ta = 0.259171(1 x2 ), Ta = 2.54264(1 x2 ). (7.115) (7.116)
Plots of the one-term collocation approximation Ta (x) for high and low temperature solutions are given in Fig 7.4.
6
Boris Gigoryevich Galerkin, 1871-1945, Russian-Soviet engineer.
269
Ta m 8
0.1
0.2
0.3
0.4
0.5
0.6
0.7
m Figure 7.3: Plot of maximum temperature perturbation Ta versus reaction rate constant for steady reaction-diusion using a one-term collocation method.
Ta 3.5 3.0 2.5 2.0 1.5 1.0 0.5 x 1.0 0.5 0.5 1.0
Figure 7.4: Plots of high and low temperature distributions Ta (x) using a one-term collocation method.
270
We can easily nd c for this approximation. Note that dierentiating Eq. (7.113) gives d m m = 2 exp T1 (1 T1 ). m dT1 (7.117)
m m A critical point exists when d/dT1 = 0. We nd this exists when T1 = 1. So c = 1 2(1)e = 0.735769.
7.3.2.2
Two-term collocation solution
We can improve our accuracy by including more basis functions. Let us take N = 2. We have a wide variety of acceptable choices for basis function that 1) satisfy the boundary conditions, and 2) are linearly independent. Let us focus on polynomials. We can select the rst as before with f1 = 1 x2 as the lowest order non-trivial polynomial that satises both boundary conditions. We could multiply this by an arbitrary constant, but that would not be of any particular use. Leaving out details, if we selected the second basis function as a cubic polynomial, we would nd the coecient on the cubic basis function to be c2 = 0. That is a consequence of the oddness of the cubic basis function and the evenness of the solution we are simulating. So it turns out the lowest order non-trivial basis function is a quartic polynomial, taken to be of the form f2 (x) = a0 + a1 x + a2 x2 + a3 x3 + a4 x4 . We can insist that f2 (1) = 0 and f2 (1) = 0. This gives a0 + a1 + a2 + a3 + a4 = 0, a0 a1 + a2 a3 + a4 = 0. (7.119) (7.120) (7.118)
We solve these for a0 and a1 to get a0 = a2 a4 , a1 = a3 . So our approximation is f2 (x) = (a2 a4 ) a3 x + a2 x2 + a3 x3 + a4 x4 . (7.121)
Motivated by the fact that a cubic approximation did not contribute to the solution, let us select a3 = 0 so to get f2 (x) = (a2 a4 ) + a2 x2 + a4 x4 .
1
(7.122)
Let us now make the choice that 1 f1 (x)f2 (x) dx = 0. This guarantees an orthogonal basis, although these basis functions are not eigenfunctions of any relevant self-adjoint linear operator for this problem. Often orthogonality of basis functions can lead to a more ecient capturing of the solution. This results in a4 = 7a2 /8. Thus, 1 7 f2 (x) = a2 + x2 x4 . 8 8
(7.123)
7.3. STEADY SOLUTIONS Let us select a2 = 8 so that f2 (x) = 1 8x2 + 7x4 = (1 x2 )(1 7x2 ). and f1 (x) = 1 x2 , f2 (x) = (1 x2 )(1 7x2 ). So now we seek approximate solutions TA (x) of the form TA (x) = c1 (1 x2 ) + c2 (1 x2 )(1 7x2 ). This leads to an error e(x) of e(x) = d2 TA + eTA (x) , dx2 d2 c1 (1 x2 ) + c2 (1 x2 )(1 7x2 ) = dx2 + exp(c1 (1 x2 ) + c2 (1 x2 )(1 7x2 )), = 2c1 + 56c2 x2 2c2 (1 7x2 ) 14c2 (1 x2 ) + exp(c1 (1 x2 ) + c2 (1 x2 )(1 7x2 )).
1
271
(7.124)
(7.125) (7.126)
(7.127)
(7.128)
(7.129) (7.130)
Now we drive two weighted errors to zero: 1 (x)e(x) dx = 0,

1 1
(7.131) (7.132)
2 (x)e(x) dx = 0.
1
Let us once again choose the weighting functions i (x) to be Dirac delta functions so that we have a two-term collocation method. Let us choose unevenly distributed collocation points so as to generate independent equations taking x = 0 and x = 1/2. Symmetric choices would lead to a linearly dependent set of equations. Other unevenly distributed choices would work as well. So we get
1
D (x)e(x) dx = 0,
1 1 1
(7.133) (7.134)
D (x 1/2)e(x) dx = 0.
or e(0) = 0, e(1/2) = 0. (7.135) (7.136)
272
Expanding, these equations are 2c1 16c2 + exp (c1 + c2 ) = 0, 9 3 c1 c2 = 0. 2c1 + 5c2 + exp 4 16 For = 0.4, we nd a high temperature solution via numerical methods: c1 = 3.07054, and a low temperature solution as well c1 = 0.243622, So the high temperature distribution is TA (x) = 3.07054(1 x2 ) + 0.683344(1 x2 )(1 7x2 ). (7.141) c2 = 0.00149141 (7.140) c2 = 0.683344, (7.139) (7.137) (7.138)
m The peak temperature of the high temperature distribution is TA = 3.75388. This is an m improvement over the one term approximation of Ta = 2.54264 and closer to the high m temperature solution found via exact methods of T1 = 3.3079. The low temperature distribution is
TA (x) = 0.243622(1 x2 ) + 0.00149141(1 x2 )(1 7x2 ).
(7.142)
m The peak temperature of the low temperature distribution is TA = 0.245113. This in m an improvement over the one term approximation of Ta = 0.259171 and compares very m favorably to the low temperature solution found via exact methods of T1 = 0.24543. Plots of the two-term collocation approximation TA (x) for high and low temperature solutions are given in Fig 7.5. While the low temperature solution is a very accurate representation, the high temperature solution exhibits a small negative portion near the boundary.
7.3.3
Steady solution with depletion
Let us return to the version of steady state reaction with depletion without resorting to the high activation energy limit, Eq.(7.60): 1 d2 T + (1 T ) exp D dx2 0 = T (1) = T (1). 0 = Rearrange to get d2 T dx2 = D (1 T ) exp = D 1 + QT , 1 + QT . (7.145) (7.146) 1 + QT , (7.143) (7.144)
Q exp() (1 T ) exp Q exp()

TA 4
273
x 1.0 0.5 0.5 1.0
Figure 7.5: Plots of high and low temperature distributions TA (x) using a two-term collocation method. Now simply adapting our earlier denition of = DQ exp(), which we note does not imply we have taken any high activation energy limits, we get exp() d2 T = (1 T ) exp 2 dx Q T (1) = T (1) = 0. 1 + QT , (7.147) (7.148)
Equations (7.147-7.148) can be solved by a numerical trial and error method where we demand that dT /dx(x = 0) = 0 and guess T (0). We keep guessing T (0) until we have satised the boundary conditions. When we do this with = 0.4, = 15, Q = 1 (so D = e /Q/ = 87173.8), we nd three steady solutions. One is at low temperature with T m = 0.016. This is a somewhat lower m than of high activation energy limit estimate of T1 = 0.24542. We nd a second intermediate m temperature solution with T = 0.417. Again this is lower than the high activation limit m value of T1 = 3.3079. And we nd a high temperature solution with T m = 0.987. There is no counterpart to this solution from the high activation energy limit analysis. Plots of T (x) for high, low, and intermediate temperature solutions are given in Fig 7.6. We can use a one-term collocation approximation to estimate the relationship between and T m . Let us estimate that Ta (x) = c1 (1 x2 ). With that choice, we get an error of e(x) = 2c1 + exp Q 1 + c1 Q(1 x2 ) (1 c1 (1 x2 )). (7.150) (7.149)
274

T 1.0
0.8
0.6
0.4
0.2
1.0
0.5
0.5
1.0
Figure 7.6: Plots of high, low, and intermediate temperature distributions T (x) for = 0.4, Q = 1, = 15. We choose a one term collocation method with 1 (x) = D (x). Then setting 0 gives e(0) = 2c1 + We solve for and get = 2c1 Q exp 1 c1 e 1 + c1 Q . (7.152) exp Q 1 + c1 Q (1 c1 ) = 0.
1 1
1 (x)e(x)dx =
(7.151)
m The maximum temperature of the approximation is given by Ta = c1 and occurs at x = 0. m A plot of Ta versus is given in Fig 7.7. For < c1 0.2, one low temperature solution exists. For c1 < < c2 0.84, three solutions exist. For > c2 , one high temperature solution exists.
7.4
Unsteady solutions
Let us know study the eects of time-dependency on our combustion problem. Let us consider the planar, m = 0, version of Eqs. (7.52): 1 2T T = + (1 T ) exp 2 t D x 1 + QT T (1, t) = T (1, t) = 0, T (x, 0) = 0.
, (7.153)
7.4. UNSTEADY SOLUTIONS

Ta m 1.0
275
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.0
1.2
1.4
m Figure 7.7: Plots of Ta versus , with Q = 1, = 15 from a one-term collocation approximate solution.
7.4.1
Linear stability
We will rst consider small deviations from the steady solutions found earlier and see if those deviations grow or decay with time. This will allow us to make a denitive statement about the linear stability of those steady solutions. 7.4.1.1 Formulation
First, recall that we have independently determined three exact numerical steady solutions to the time-independent version of Eq. (7.153). Let us call any of these Te (x). Note that by construction Te (x) satises the boundary conditions on T . Let us subject a steady solution to a small perturbation and consider that to be our initial condition for an unsteady calculation. Take then T (x, 0) = Te (x) + A(x), A(1) = A(1) = 0, A(x) = O(1), 0 < << 1. Here A(x) is some function which satises the same boundary conditions as T (x, t). Now let us assume that T (x, t) = Te (x) + T (x, t). with T (x, 0) = A(x). (7.157) (7.156) (7.154) (7.155)
276
Here T is an O(1) quantity. We then substitute our Eq. (7.156) into Eq. (7.153) to get 1 2 (Te (x) + T (x, t)) = (Te (x) + T (x, t)) t D x2 + (1 Te (x) T (x, t)) exp . 1 + QTe (x) + QT (x, t) (7.158)
From here on we will understand that Te is Te (x) and T is T (x, t). Now consider the exponential term: exp 1 + QTe + QT = exp exp exp exp So our Eq. (7.158) can be rewritten as 1 2 (Te + T ) = (Te + T ) t D x2 + (1 Te T ) exp = 1 + QTe 1+ Q T , (1 + QTe )2 1 + QTe 1 + 1 + QTe 1 + QTe 1 + QTe 1
Q T 1+QTe
(7.159) (7.160) (7.161) (7.162)
Q , T 1 + QTe Q exp T , 2 (1 + QTe ) Q 1+ T . 2 (1 + QTe ) 1
1 2 (Te + T ) 2 D x Q + exp (1 Te T ) 1 + T , 1 + QTe (1 + QTe )2 1 2 (Te + T ) = D x2 (1 Te )Q + exp (1 Te ) + T 1 + + O(2 ) 2 1 + QTe (1 + QTe ) 2 2 1 Te T (1 Te ) + + exp = 2 2 D x D x 1 + QTe
=0
+ exp
1 + QTe
T 1 +
(1 Te )Q (1 + QTe )2
+ O(2 ) . (7.163)
277
Now, we recognize the bracketed term as zero because Te (x) is constructed to satisfy the steady state equation. We also recognize that Te (x)/t = 0. So our equation reduces to, neglecting O(2 ) terms, and canceling T 1 2T = + exp t D x2 1 + QTe 1 + (1 Te )Q (1 + QTe )2 T . (7.164)
Equation (7.164) is a linear partial dierential equation for T (x, t). It is of the form 1 2T T = + B(x)T , 2 t D x with B(x) exp 7.4.1.2 1 + QTe (x) 1 + (1 Te (x))Q (1 + QTe (x))2 . (7.166) (7.165)
Separation of variables
Let us use the standard technique of separation of variables to solve Eq. (7.165). We rst assume that T (x, t) = H(x)K(t). So Eq. (7.165) becomes H(x) 1 d2 H(x) dK(t) = = K(t) + B(x)H(x)K(t), dt D dx2 1 dK(t) 1 1 d2 H(x) = + B(x) = . K(t) dt D H(x) dx2 (7.168) (7.169) (7.167)
Since the left side is a function of t and the right side is a function of x, the only way the two can be equal is if they are both the same constant. We will call that constant . Now Eq. (7.169) really contains two equations, the rst of which is dK(t) + K(t) = 0. dt This has solution K(t) = C exp(t), (7.171) (7.170)
where C is some arbitrary constant. Clearly if > 0, this solution is stable, with time constant of relaxation = 1/. The second dierential equation contained within Eq. (7.169) is 1 d2 H(x) + B(x)H(x) = H(x), D dx2 1 d2 + B(x) H(x) = H(x). D dx2 (7.172) (7.173)
278
This is of the classical eigenvalue form for a linear operator L; that is L(H(x)) = H(x). We also must have H(1) = H(1) = 0, (7.174)
to satisfy the spatially homogeneous boundary conditions on T (x, t). This eigenvalue problem is dicult to solve because of the complicated nature of B(x). Let us see how the solution would proceed in the limiting case of B as a constant. Then we will generalize later. If B is a constant, we have d2 H + (B + )DH = 0, dx2 H(1) = H(1) = 0. (7.175)
The following mapping simplies the problem somewhat: y= x+1 . 2 (7.176)
This takes our domain of x [1, 1] to y [0, 1]. By the chain rule dH dH dy 1 dH = = . dx dy dx 2 dy So 1 d2 H d2 H = . dx2 4 dy 2 So our eigenvalue problem transforms to d2 H + 4D(B + )H = 0, dy 2 This has solution H(y) = C1 cos ( 4D(B + ) )y + C2 sin ( 4D(B + )) y . At y = 0 we have then H(0) = 0 = C1 (1) + C2 (0), so C1 = 0. Thus H(y) = C2 sin ( 4D(B + )) y . (7.180) (7.179) (7.178) H(0) = H(1) = 0. (7.177)
At y = 1, we have the other boundary condition: H(1) = 0 = C2 sin ( 4D(B + )) .

(7.181)
7.4. UNSTEADY SOLUTIONS Since C2 = 0 to avoid a trivial solution, we must require that sin ( 4D(B + )) = 0.
279
(7.182)
For this to occur, the argument of the sin function must be an integer multiple of : 4D(B + ) = n, Thus = n2 2 B. 4D (7.184) n = 1, 2, 3, . . . (7.183)
We need > 0 for stability. For large n and D > 0, we have stability. Depending on the value of B, low n, which corresponds to low frequency modes, could be unstable. 7.4.1.3 Numerical eigenvalue solution
Let us return to the full problem where B = B(x). Let us solve the eigenvalue problem via the method of nite dierences. Let us take our domain x [1, 1] and discretize into N points with x = 2 , N 1 xi = (i 1)x 1. (7.185)
Note that when i = 1, xi = 1, and when i = N, xi = 1. Let us dene B(xi ) = Bi and H(xi ) = Hi . We can rewrite Eq. (7.172) as d2 H(x) + D(B(x) + )H(x) = 0, dx2 H(1) = H(1) = 0. (7.186)
Now let us apply an appropriate equation at each node. At i = 1, we must satisfy the boundary condition so H1 = 0. At i = 2, we discretize Eq. (7.186) with a second order central dierence to obtain H1 2H2 + H3 + D(B2 + )H2 = 0. x2 We get a similar equation at a general interior node i: Hi1 2Hi + Hi+1 + D(Bi + )Hi = 0. x2 (7.189) (7.188) (7.187)
280
At the i = N 1 node, we have HN 2 2HN 1 + HN + D(BN 1 + )HN 1 = 0. x2 At the i = N node, we have the boundary condition HN = 0. These represent a linear tri-diagonal system of equations of the form
2 ( Dx2 + B2 ) 1 Dx2 1 Dx2 2 ( Dx2 + 1 Dx2
(7.190)
(7.191)
0 B3 )
1 Dx2
0 ... 0
... 0
=L
... ... ...
0 0 ... ... ...
H2 H2 ... 0 H3 . . . 0 H3 . . = . (7.192) . ... ... . . . . . ... ... . . . . . . . ... ... . .

=h
This is of the classical linear algebraic eigenvalue form L h = h. All one need do is discretize and nd the eigenvalues of the matrix L. These will be good approximations to the eigenvalues of the dierential operator L. The eigenvectors of L will be good approximations of the eigenfunctions of L. To get a better approximation, one need only reduce x. Note because the matrix L is symmetric, the eigenvalues are guaranteed real, and the eigenvectors are guaranteed orthogonal. This is actually a consequence of the original problem being in Sturm-Liouville form, which is guaranteed to be self-adjoint with real eigenvalues and orthogonal eigenfunctions. 7.4.1.3.1 Low temperature transients For our case of = 0.4, Q = 1, = 15 (so D = 87173.8), we can calculate the stability of the low temperature solution. Choosing N = 101 points to discretize the domain, we nd a set of eigenvalues. They are all positive, so the solution is stable. The rst few are = 0.0000232705, 0.000108289, 0.000249682, 0.000447414, . . . . (7.193)
The rst few eigenvalues can be approximated by inert theory with B(x) = 0, see Eq. (7.184): n2 2 = 0.0000283044, 0.000113218, 0.00025474, 0.00045287, . . . . 4D (7.194)
The rst eigenvalue is associated with the longest time scale = 1/0.0000232705 = 42972.9 and a low frequency mode, whose shape is given by the associated eigenvector, plotted in Fig 7.8. This represents the fundamental mode. The rst harmonic mode is associated with the next eigenfunction, which is plotted in Fig 7.9. The second harmonic mode is associated with the next eigenfunction, which is plotted in Fig 7.10.
281
0.14 0.12 0.10 0.08 0.06 0.04 0.02 x
1.0
0.5
0.5
1.0
Figure 7.8: Plot of lowest frequency, fundamental mode, eigenfunction versus x, with = 0.4, Q = 1, = 15, low temperature steady solution Te (x).
0.10 0.05 x
1.0
0.5 0.05 0.10
0.5
1.0
Figure 7.9: Plot of rst harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, low temperature steady solution Te (x).
0.10 0.05 x
1.0
0.5 0.05 0.10
0.5
1.0
Figure 7.10: Plot of second harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, low temperature steady solution Te (x).
282
0.14 0.12 0.10 0.08 0.06 0.04 0.02 1.0 0.5 0.5 1.0 x
Figure 7.11: Plot of fundamental eigenfunction versus x, with = 0.4, Q = 1, = 15, intermediate temperature steady solution Te (x).
0.10 0.05 x
1.0
0.5 0.05 0.10
0.5
1.0
Figure 7.12: Plot of rst harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, intermediate temperature steady solution Te (x). 7.4.1.3.2 Intermediate temperature transients For the intermediate temperature solution with T m = 0.417, we nd the rst few eigenvalues to be = 0.0000383311, 0.0000668221, 0.000209943, . . . (7.195)
Except for the rst, all the eigenvalues are positive. The rst eigenvalue of = 0.0000383311 is associated with an unstable fundamental mode. All the harmonic modes are stable. We plot the rst three modes in Figs. 7.11-7.13. 7.4.1.3.3 High temperature transients For the high temperature solution with T m = 0.987, we nd the rst few eigenvalues to be = 0.000146419, 0.00014954, 0.000517724, . . . (7.196)
All the eigenvalues are positive, so all modes are stable. We plot the rst three modes in Figs. 7.14-7.16.
283
0.10 0.05 x
1.0
0.5 0.05 0.10
0.5
1.0
Figure 7.13: Plot of second harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, intermediate temperature steady solution Te (x).
1.0
0.5
0.5
1.0
0.05
0.10
0.15
Figure 7.14: Plot of fundamental eigenfunction versus x, with = 0.4, Q = 1, = 15, high temperature steady solution Te (x).
0.15 0.10 0.05 x
1.0
0.5 0.05 0.10 0.15
0.5
1.0
Figure 7.15: Plot of rst harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, high temperature steady solution Te (x).
284

0.15
0.10
0.05
1.0
0.5 0.05
0.5
1.0
0.10
Figure 7.16: Plot of second harmonic eigenfunction versus x, with = 0.4, Q = 1, = 15, high temperature steady solution Te (x).
300 000 200 000 100 000 0 0.015
0.010 0.005 0.000 1.0 0.5 0.0 0.5 1.0
Figure 7.17: Plot of T (x, t), with = 0.4, Q = 1, = 15.
7.4.2
Full transient solution
We can get a full transient solution to Eqs. (7.153) with numerical methods. We omit details of such numerical methods, which can be found in standard texts.
7.4.2.1
Low temperature solution
For our case of = 0.4, Q = 1, = 15 (so D = 87173.8), we show a plot of the full transient solution in Fig. 7.17. We see in Fig. 7.18 that the centerline temperature T (0, t) relaxes to the long time value predicted by the low temperature steady solution:
t
lim T (0, t) = 0.016.
(7.197)

T 0.015
285
0.010
0.005
50 000
100 000
150 000
200 000
250 000
t 300 000
Figure 7.18: Plot of T (0, t) along with the long time exact low temperature solution, Te (0), with = 0.4 Q = 1, = 15.
1.5 106 1. 106 500 000 0
1.0
0.5
0.0 1.0 0.5 0.0 0.5 1.0
Figure 7.19: Plot of T (x, t), with = 1.2, Q = 1, = 15. 7.4.2.2 High temperature solution
We next select a value of = 1.2 > c . This should induce transition to a high temperature solution. We maintain = 15, Q = 1. We get D = e //Q = 261521. The full transient solution is shown in Fig. 7.19 Figure 7.20 shows the centerline temperature T (0, t). We see it relaxes to the long time value predicted by the high temperature steady solution:
t
lim T (0, t) = 0.9999185.
(7.198)
It is clearly seen that there is a rapid acceleration of the reaction for t 106 . This compares with the prediction of the induction time from the innite Damkhler number, D , o thermal explosion theory of explosion to occur when t e15 e = = 2.17934 105 . Q (1)(15) (7.199)
286

T 1.0
0.8
0.6
0.4
0.2
200 000 400 000 600 000 800 000 1. 106 1.2 106 1.4 106
Figure 7.20: Plot of T (0, t) along with the long time exact high temperature exact solution with = 1.2, Q = 1, = 15. The estimate under-predicts the value by a factor of ve. This is likely due to 1) cooling of the domain due to the low temperature boundaries at x = 1, and 2) eects of nite activation energy.
Chapter 8 Laminar ames: Reaction-advection-diusion

Here we will consider premixed one-dimensional steady laminar ames. Background is available in many sources.123 This topic is broad, but we will restrict attention to the simplest cases. We will consider a simple reversible kinetics model A B, (8.1)
where A and B have identical molecular masses, MA = MB = M, and are both calorically perfect ideal gases with the same specic heats, cP A = cP B = cP . Because we are modeling the system as premixed, we consider species A to be composed of molecules which have their own fuel and oxidizer. This is not common in hydrocarbon kinetics, but is more so in the realm of explosives. More common in gas phase kinetics are situations in which cold fuel and cold oxidizer react together to form hot products. In such a situation, one can also consider non-premixed ames in which streams of fuel rst must mix with streams of oxidizer before signicant reaction can commence. Such ames will not be considered in this chapter. We shall see that the introduction of advection introduces some unusual mathematical diculties in properly modeling cold unreacting ow. This will be overcome in a way which is aesthetically unappealing, but nevertheless useful: we shall introduce an ignition temperature Tig in our reaction kinetics law to suppress all reaction for T < Tig . This will serve to render the cold boundary to be a true mathematical equilibrium point of the model; this is not a traditional chemical equilibrium, but it is an equilibrium in the formal mathematical sense nonetheless.
Buckmaster, J. D., and Ludford, G. S. S., 2008, Theory of Laminar Flames, Cambridge, Cambridge. Williams, F. A., 1985, Combustion Theory, Benjamin-Cummings, Menlo Park, California. 3 Law, C. K., 2006, Combustion Physics, Cambridge, Cambridge.
2 1
287
288
CHAPTER 8. LAMINAR FLAMES: REACTION-ADVECTION-DIFFUSION
8.1
8.1.1
8.1.1.1
Governing Equations
Evolution equations
Conservative form
Let us commence with the one-dimensional versions of Eqs. (6.1-6.3) and (6.5): + (u) = 0, t x t 1 e + u2 2 x u2 + P = 0, (u) + t x 1 2 u e + u + j q + (P )u = 0, 2 m (YB ) + (uYB + jB ) = M B . t x (8.2) (8.3) (8.4) (8.5)
8.1.1.2
Using standard reductions similar to those made to achieve Eqs. (6.34),(6.37),(6.42), (6.50), the non-conservative form of the governing equations, Eqs. (8.2-8.5) is u +u + t x x u P u + u + t x x x e e j q u u + u + +P t x x x x m YB jB YB + u + t x x 8.1.1.3 Formulation using enthalpy = 0, = 0, = 0, = M B . (8.6) (8.7) (8.8) (8.9)
It will be more useful to formulate the equations using enthalpy. First rewrite Eqs. (8.6-8.9) using the material derivative, d/dt = /t + u/x: u d + dt x du P + dt x x u u de j q +P + dt x x x m dYB jB + dt x
= 0, = 0, = 0, = M B .
(8.10) (8.11) (8.12) (8.13)
8.1. GOVERNING EQUATIONS
289
Now recall the denition for enthalpy, Eq. (3.78), h = e + P v = e + P/, to get an expression for dh: P 1 d + dP, 2 de P d 1 dP dh = + , dt dt 2 dt dt de P d dh 1 dP 2 = , dt dt dt dt dh dP de P d = . dt dt dt dt dh = de (8.14) (8.15) (8.16) (8.17)
Now using the mass equation, Eq. (8.10) to eliminate P u/x in the energy equation, Eq. (8.12), the energy equation can be rewritten as de j q P d u + = 0, dt x dt x Next use Eq. (8.17) to simplify Eq. (8.18): 8.1.1.4 dh j q dP u + = 0. dt x dt x (8.19) (8.18)
Low Mach number limit
In the limit of low Mach number, one can do a formal asymptotic expansion with the reciprocal of the Mach number squared as a perturbation parameter. All variables take the form = o + (1/M 2 )1 , where is a general variable. In this limit, the linear momentum equation can be shown to reduce at leading order to Po /x = 0, giving rise to Po = Po (t). We shall ultimately be concerned only with time-independent ows where P = Po . We adopt the constant pressure assumption now. Also in the low Mach number limit, it can be shown that viscous work is negligible, so u/x 0. Our evolution equations then in the low Mach number, constant pressure limit are u +u + = 0, t x x h j q h + u + = 0, t x x m YB YB jB + u + = M B . t x x
(8.20) (8.21) (8.22)
Note that we have eectively removed the momentum equation. It would re-appear in a non-trivial way at the next order. Equations (8.20-8.22) take on the conservative form
290
+ (u) = 0, t x (h) + (uh + j q ) = 0, t x m (YB ) + (uYB + jB ) = M B . t x
(8.23) (8.24) (8.25)
8.1.2
Constitutive models
Equations (8.20-8.22) are supplemented by simple constitutive models: Po = RT, h = cP (T To ) YB q, YB m jB = D , x YB T + Dq , j q = k x x B = aT eE/(RT ) (8.26) (8.27) (8.28) (8.29) YB 1 YB
=B /A
Kc = e
q/(RT )
1 (1 YB ) 1 M Kc
=A
H(T Tig ),
(8.30)
(8.31)
m Thus, we have nine equations for the nine unknowns, , u, h, T , j q , YB , jB , B , Kc . Many of these are obvious. Some are not. First, we note the new factor H(T Tig ) in our kinetics law, Eq. (8.30). Here H is a Heaviside unit step function. For T < Tig it takes a value of zero. For T Tig , it takes a value of unity. Next, let us see how to get Eq. (8.28) from the more general Eq. (6.75). We rst take the thermal diusion coecient to be zero, T Di = 0. We also note since MA = MB = M that yk = Yk ; that is mole fractions are the same as mass fractions. So Eq. (6.75) simplies considerably to N
jim
=
k=,1k=i
Dik
Yk . x
(8.32)
Next we take Dik = D and write for each of the two diusive mass uxes that
m jA = D m jB
YB , x YA = D . x
(8.33) (8.34)
8.1. GOVERNING EQUATIONS Since YA + YB = 1, we have also

m jA = D m jB
291
YA , x YB . = D x
(8.35) (8.36)
m m Note that jA + jB = 0, as required. Under the same assumptions, Eq. (6.74) reduces to
j q = k
T m m + jA hA + jB hB , x T m m + jA (hA,To + cP (T To )) + jB (hB,To + cP (T To )), = k x T m m m m = k + jA hA,To + jB hB,To + (jA + jB ) cP (T To ), x

=0
(8.37) (8.38) (8.39) (8.40) (8.41) (8.42) (8.43)
T m m jB hA,To + jB hB,To , x T m = k jB (hA,To hB,To ), x = k

=q
T m jB q, x T YB j q = k + Dq . x x For the equilibrium constant Kc , we recall that j q = k Kc = exp = exp = exp = exp = exp = exp Go , RT (g o g o ) B A , RT o o gA gB , RT ho T so (ho T so ) A A B B RT ho ho so so A B A exp B RT R so o ho o ho o B,T B,T A,T exp RT
(8.44) (8.45) (8.46) , , so o A,T R . (8.47) (8.48) (8.49)
Here because of constant specic heats for A and B, many terms have canceled. Let us take now so o = so o and our denition of the heat release q to get A,T B,T q Kc = exp . (8.50) RT
292
8.1.3
Alternate forms
Further analysis of the evolution equations combined with the constitutive equations can yield forms which give physical insight. 8.1.3.1 Species equation
If we combine our species evolution equation, Eq. (8.22) with the constitutive law, Eq. (8.28), we get YB YB + u t x x D YB x = M B . (8.51)
Note that for ows with no advection (u = 0), constant density ( = constant), and no reaction (B = 0), this reduces to the classical heat equation from mathematical physics YB /t = D( 2 YB /x2 ). 8.1.3.2 Energy equation
Consider the energy equation, Eq. (8.21), using Eqs. (8.27,8.29) to eliminate h and j q : (cP (T To ) YB q) +u (cP (T To ) YB q) + t x x
=h =h
T YB + Dq x x
=j q
= 0, (8.52)
T YB
q cP
+ u
x
2
T YB
q cP
k T YB q D cP x x cP D YB x
= 0, (8.53)
T k T q T + u t x cP x2 cP
YB YB + u t x x
=M B
= 0, (8.54)
We notice that the terms involving YB simplify via Eq. (8.51) to yield an equation for temperature evolution T k 2T q T + u M B = 0, 2 t x cP x cP T k 2T q T +u = + M B . 2 t x cP x cP
=
(8.55) (8.56)
8.1. GOVERNING EQUATIONS We note that thermal diusivity is = k . cP
293
(8.57)
We note here that this denition, convenient and in the combustion literature, is slightly non-traditional as it involves a variable property . So the reduced energy equation is T T +u t x = 2T q + M B . 2 x cP (8.58)
Note that for exothermic reaction, q > 0 accompanied with production of product B, B > 0, induces a temperature rise of a material particle. The temperature change is modulated by energy diusion. In the inert zero advection limit, we recover the ordinary heat equation T /t = ( 2 T /x2 ). 8.1.3.3 Shvab-Zeldovich form
Let us now adopt the assumption that mass and energy diuse at the same rate. This is not dicult to believe as both are molecular collision phenomena in gas ames. Such an assumption will allow us to write the energy equation in the Shvab-Zeldovich form. We note the dimensionless ratio of energy diusivity to mass diusivity D is known as the Lewis4 number, Le: Le = . D k . cP (8.59)
If we insist that mass and energy diuse at the same rate, we have Le = 1, which gives D== (8.60)
With this assumption, the energy equation, Eq. (8.53), takes the form t T YB T YB T YB q cP + u x T YB T YB T YB q cP x k cP x k T k YB q cP x cP x cP x x T YB T YB q cP q cP = 0, (8.61) t t
4
q cP q cP
+ u
q cP
= 0, (8.62)
+u
q cP
k cP x
= 0. (8.63)
Warren Kendall Lewis, 1882-1975, American chemical engineer. CC BY-NC-ND. 12 December 2011, J. M. Powers.
294
We rewrite in terms of the material derivative d q T YB dt cP = k 2 cP x2 T YB q cP . (8.64)
Equation (8.64) holds that for a material uid particle, the quantity T YB q/cP changes only in response to local spatial gradients. Now if we consider an initial value problem in which T and YB are initially spatially uniform and there are no gradients of either T or YB at x , then there will no tendency for any material particle to have its value of T YB q/cP change. Let us assume that at t = 0, we have T = To and no product, so YB = 0. Then we can conclude that the following relation holds for all space and time: q T YB = To . (8.65) cP Solving for YB , we get YB = cP (T To ) . q (8.66)
For this chapter, we are mainly interested in steady waves. We can imagine that our steady waves are the long time limit of a situation just described which was initially spatially uniform. Compare Eq. (8.66) to our related ad hoc assumption for reactive solids, Eq. (7.26). They are essentially equivalent, especially when one recalls that plays the same role as YB and for the reactive solid cP cv . We can use Eq. (8.66) to eliminate YB in the species equation, Eq. (8.51) to get a single equation for temperature evolution. First adopt the equal diusion assumption, Eq. (8.60), in Eq. (8.51): Next eliminate YB : t cP (T To ) q + u x cP (T To ) q x k cP x cP (T To ) q = M B . (8.68) Simplifying, we get cP T T +u t x k 2T x2 = qM B . (8.69) YB YB + u t x x k YB cP x = M B . (8.67)
Now let us use Eq. (8.30) to eliminate B : cP T T +u t x 2T k 2 x 1 YB Kc 1 Y B H(T Tig ). (8.70)
= qaT eE/(RT ) (1 YB ) 1
8.1. GOVERNING EQUATIONS Now use Eq. (8.50) to eliminate Kc and Eq. (8.66) to eliminate YB so to get cP T T +u t x = qaT e k 2T x2 cP (T To ) 1 q 1e
q/(RT )
295
E/(RT )
cP (T To ) q cP (T To ) q
H(T Tig ). (8.71)
Equation (8.71) is remarkably similar to our earlier Eq. (7.28) for temperature evolution in a heat conducting reactive solid. The dierences are as follows: We use cP instead of cv , We have included advection, We have accounted for nite , We have accounted for reversible reaction. We have imposed and ignition temperature.
8.1.4
Equilibrium conditions
We can gain insights examining when Eq. (8.71) is in equilibrium. There is one potential equilibria, the state of chemical equilibrium where Kc = YB /(1 YB ) = YB /YA . This occurs when 1e
q/(RT ) cP (T To ) q cP (T To ) q
= 0.
(8.72)
This is a transcendental equation for T . Let us examine how the equilibrium T varies for some sample parameters. For this and later calculations, we give a set of parameters in Table 8.1. Let us consider how the equilibrium temperature varies as q is varied from near zero to its maximum value of Table 8.1. The chemical equilibrium ame temperature is plotted as a function of q in Figure 8.1. For q = 0, the equilibrium temperature is the ambient temperature. As q increases, the equilibrium temperature increases monotonically. The corresponding equilibrium mass fraction as a function of q is given in Fig. 8.2. When q = 0, the equilibrium mass fraction is YB = 0.5; that is, there is no tendency for either products or reactants. As q increases, the tendency for product increases and YB 1. There is also the state of complete reaction when YB = 1, which corresponds to 1 cP (T To ) = 0, q T = To + (8.73) q . cP (8.74)
296
Parameter Value Fundamental Dimensional To 300 cP 1000 R 287 q 1.5 106 E 1.722 105 o 105 a 105 uo 1.4142 Po 105 Tig 600 Secondary Dimensional o 1.1614 k 0.0116114 Fundamental Dimensionless 1.4025 Q 5 2 D 2 TIG 0.2
Units K J/kg/K J/kg/K J/kg J/kg m2 /s 1/s m/s Pa K kg/m3 J/m/s
Table 8.1: Numerical values of parameters for simple laminar ame calculation.
8.2. STEADY BURNER-STABILIZED FLAMES

T eq K 2000 1500 1000 500 q J kg 1 500 000
297
750 000
Figure 8.1: Variation of chemical equilibrium temperature with heat release.

YB eq 1.0 0.8 0.6 0.4 0.2 0 750 000 q J kg 1 500 000
Figure 8.2: Variation of chemical equilibrium product mass fraction with heat release. Numerically, at our maximum value of q = 1.5 106 J/kg, we get complete reaction when T = 1800 K. Note this state is not an equilibrium state unless Kc . As x , we expect a state of chemical equilibrium and a corresponding high temperature. At some intermediate point, the temperature will have its value at ignition, Tig . We will ultimately transform our domain so the ignition temperature is realized at x = 0, which we can associate with a burner surface. WE say that the ame is thus anchored to the burner. Without such a prescribed anchor, our equations are such that any translation in x will yield an invariant solution, so the origin of x is arbitrary.
8.2
Steady burner-stabilized ames
Let us consider an important problem in laminar ame theory: that of a burner-stabilized ame. We shall consider a doubly-innite domain, x (, ). We will assume that as x , we have a fresh unburned stream of reactant A, YA = 1, YB = 0, with known velocity uo , density o and temperature To . The values of o and To will be consistent with a state equation so that the pressure has a value of Po .
298
8.2.1
Formulation
Let us consider our governing equations in the steady wave frame where there is no variation with t: /t = 0. This, coupled with our other reductions, yields the system d (u) = 0, dx dT cP (T To ) d2 T ucP k 2 = qaT eE/(RT ) 1 dx dx q 1e Po = RT.
q/(RT )
(8.75)
H(T Tig ),
(8.76) (8.77)
Now we have three equations for the three remaining unknowns, , u, T . For a burner-stabilized ame, we assume at x that we know the velocity, uo , the temperature, To , and the density, o . We can integrate the mass equation to get u = o uo . We then rewrite the remaining dierential equation as o uo cP d2 T dT k 2 dx dx = cP (T To ) Po qaT 1 eE/(RT ) 1 R q 1e
q/(RT )
(8.78)
H(T Tig ).
(8.79)
Our boundary conditions for this second order dierential equation are dT 0, dx x . (8.80)
We will ultimately need to integrate this equation numerically, and this poses some diculty because the boundary conditions are applied at . We can overcome this in the following way. Note that our equations are invariant under a translation in x. We will choose some large value of x to commence numerical integration. We shall initially assume this is near the chemical equilibrium point. We shall integrate backwards in x. We shall note the value of x where T = Tig . Beyond that point, the reaction rate is zero, and we can obtain an exact solution for T . We will then translate all our results so that the ignition temperature is realized at x = 0. Let us scale temperature so that T = To 1 +
q T . cP To
(8.81)
8.2. STEADY BURNER-STABILIZED FLAMES Inverting, we see that T = cP (T To ) . q cP (Tig To ) . q
299
(8.82)
We get a dimensionless ignition temperature TIG of TIG = (8.83)
Let us also dene a dimensionless heat release Q as q Q= . cP To Thus T = 1 + QT . To
(8.84)
(8.85)
Note that T = 1 corresponds to our complete reaction point where YB = 1. Let us also restrict ourselves, for convenience, to the case where = 0. With these scalings, and with = our dierential equations become o uocP To Q dT d2 T kTo Q 2 dx dx = Po qa exp R 1 + QT
Q 1
E , RTo 1 T To (1 + QT ) T 1 T
(8.86)
1 exp dT 0, x . dx Let us now scale both sides by o uo cP To to get Q
1 + QT
H(T TIG ), (8.87) (8.88)
dT k 1 d2 T Po q a Q 2 = exp dx o cP uo dx o RTo cP To uo
=1 =Q
1 + QT T 1 T
1 T 1 + QT
1 exp
Q 1
1 + QT
H(T TIG ). (8.89)
Realizing that Po = o RTo , the ambient thermal diusivity, o = k/(o cP ) and canceling Q, we get 1 d2 T a dT exp o = 2 dx uo dx uo 1 + QT 1 T 1 + QT Q T 1 1 T
1 exp
1 + QT
H(T TIG ).
(8.90)
300
Now let us scale x by a characteristic length, L = uo /a. This length scale is dictated by a balance between reaction and advection. When the reaction is fast, a is large, and the length scale is reduced. When the incoming velocity is fast, uo is large, and the length scale is increased. x ax x = = . (8.91) L uo With this choice of length scale, Eq. (8.90) becomes dT a d2 T = exp o 2 dx uo dx2 1 + QT 1 T 1 + QT Q 1 1 + QT
1 exp
T 1 T
H(T TIG ).
(8.92)
Now, similar to Eq. (7.36), we take the Damkhler number D to be o D= u2 aL2 = o . o ao (8.93)
In contrast to Eq. (7.36) the Damkhler number here includes the eects of advection. With o this choice, Eq. (8.92) becomes dT 1 d2 T = exp dx D dx2 1 + QT 1 T 1 + QT Q 1 1 + QT
1 exp
T 1 T
H(T TIG ).
(8.94)
Lastly, let us dispose with the notation and understand that all variables are dimensionless. Our dierential equation and boundary conditions become dT 1 d2 T dx D dx2 = exp 1 + QT 1T 1 + QT Q 1 1 + QT
1 exp dT dx 0.
T 1T
H(T TIG ),
(8.95) (8.96)
Equation (8.95) is remarkably similar to our Frank-Kamenetskii problem embodied in Eq. (7.59). The major dierences are reected in that Eq. (8.95) accounts for advection, variable density, reversible reaction, doubly-innite spatial domain.
301
8.2.2
Solution procedure
There are some unusual challenges in the numerical solution of the equations for laminar ames. Formally we are solving a two-point boundary value problem on a doubly-innite domain. The literature is not always coherent on solutions methods or the interpretations of results. At the heart of the issue is the cold boundary diculty. We have patched this problem via the use of an ignition temperature built into a Heaviside5 function; see Eq. (8.30). We shall see that this patch has advantages and disadvantages. Here we will not dwell on nuances, but will present a result which is mathematically sound and oer interpretations. Our result will use standard techniques of non-linear analysis from dynamic systems theory. We will pose the problem as a coupled system of rst order non-linear dierential equations, nd their equilibria, use local linear analysis to ascertain the stability of the chemical equilibrium xed point, and use numerical integration to calculate the nonlinear laminar ame structure. 8.2.2.1 Model linear system
Before commencing with the dicult Eq. (8.95), let us consider a related linear model system, given here: 3 dT d2 T 2 = 4(1 T )H(T TIG ), dx dx dT dx 0. (8.97)
The forcing is removed when either T = 1 or T < TIG . Dening q dT /dx, we rewrite our model equation as a linear system of rst order equations: dq = 3q + 4(1 T )H(T TIG ), q() = 0, (8.98) dx dT = q, q() = 0. (8.99) dx The system has an equilibrium point at q = 0 and T = 1. The system is also in equilibrium when q = 0 and T < TIG , but we will not focus on this. Taking T = T 1, the system near the equilibrium point is d dx q T = 3 4 1 0 q T . (8.100)
The local Jacobian matrix has two eigenvalues = 4 and = 1. Thus the equilibrium is a saddle. Considering the solution away from the equilibrium point, but still for T > TIG , and returning from T to T , the solution takes the form q T
5
0 1
1 + ex 5
C1 + 4C2 C1 + 4C2
1 + e4x 5
4C1 4C2 C1 + C2
(8.101)
Oliver Heaviside, 1850-1925, English engineer. CC BY-NC-ND. 12 December 2011, J. M. Powers.
302
We imagine as x that T approaches unity from below. We wish this equilibrium to be achieved. Thus, we must suppress the unstable = 4 mode; this is achieved by requiring C1 = C2 . This yields q T = 0 1 1 + ex 5 5C1 5C1 = 0 1 + C1 ex 1 1 . (8.102)
We can force T (xIG ) = TIG by taking C1 = (1 TIG )exIG . The solution for x > xIG is T = 1 (1 TIG )e(xxIG ) , q = (1 TIG )e(xxIG ) . At the interface x = xIG , we have T (xIG ) = TIG and q(xIG ) = (1 TIG ). For x < xIG , our system reduces to 3 d2 T dT 2 = 0, dx dx T (xIG ) = TIG , dT dx = 0.
x
(8.103) (8.104)
(8.105)
Solutions in this region take the form T = TIG e3(xxIG ) + C 1 e3(xxIG ) , q = 3TIG e
3(xxIG )
(8.106) (8.107)
+ 3Ce
3(xxIG )
All of them have the property that q 0 as x . One is faced with the question of how to choose the constant C. Let us choose it to match the energy ux predicted at x = xIG . At the interface x = xIG we get T = TIG and q = 3TIG + 3C. Matching values of q at the interface, we get (1 TIG ) = 3TIG + 3C. Solving for C, we get C= So we nd for x < xIG that T = TIG e3(xxIG ) + 4TIG 1 3 1 e3(xxIG ) , (8.110) (8.111) 4TIG 1 . 3 (8.109) (8.108)
q = 3TIG e3(xxIG ) + (4TIG 1) e3(xxIG ) . 8.2.2.2 System of rst order equations
Let us apply this technique to the full non-linear Eq. (8.95). First, again dene the nondimensional Fourier heat ux q as q=
dT . dx
(8.112)
303
Note this is a mathematical convenience. It is not the full diusive energy ux as it makes no account for mass diusion aects. However, it is physically intuitive. With this denition, we can rewrite Eq. (8.112) along with Eq. (8.95) as dq = Dq + D exp dx 1 + QT Q 1 1 exp 1 + QT 1T 1 + QT T 1T H(T TIG ), (8.113) (8.114)
dT = q. dx We have two boundary conditions for this problem: q() = 0.
(8.115)
We are also going to x our coordinate system so that T = TIG at x = xIG 0. We will require continuity of T and dT /dx at x = xIG = 0. Our non-linear system has the form d dx 8.2.2.3 Equilibrium q T = f (q, T ) g(q, T ) . (8.116)
Our equilibrium condition is f (q, T ) = 0, g(q, T ) = 0. By inspection, equilibrium is found when q = 0, 1 exp
Q 1
(8.117) or T < TIG . (8.118)
1 + QT
T 1T
= 0
We will focus most attention on the isolated equilibrium point which corresponds to traditional chemical equilibrium. We have considered the dimensional version of the same equilibrium condition for T in an earlier section. The solution is found via solving a transcendental equation for T . 8.2.2.4 Linear stability of equilibrium
Linear stability of the equilibrium point can be found by examining the eigenvalues of the Jacobian matrix J: J=
f q g q f T g T
=
eq
D 1
f T eq
(8.119)
304
This Jacobian has eigenvalues of = D 2 1 1 4 f D2 T .

eq
(8.120)
For D >>
f /T , the eigenvalues are approximated well by 1 D, 2 1 f D T .

eq
(8.121)
This gives rise to a stiness ratio, valid in the limit of D >> 1 D2 f . 2 T
f /T , of (8.122)
We adopt the parameters of Table 8.1. With these, we nd the dimensionless chemical equilibrium temperature at T eq = 0.953. (8.123)
This corresponds to a dimensional value of 1730.24 K. Our parameters induce a dimensional length scale of uo /a = 1.4142 105 m. This is smaller than actual ames, which actually have much larger values of D. So as to de-stien the system, we have selected a smaller than normal value for D. For these values, we get a Jacobian matrix of J= 2 0.287 1 0 . (8.124)
We get eigenvalues near the equilibrium point of = 2.134, = 0.134. (8.125)
The equilibrium is a saddle point. Note the negative value of f /T at equilibrium gives rise to the saddle character of the equilibrium. The actual stiness ratio is |2.134/(0.134)| = 15.9. This is well predicted by our simple formula which gives |22 /(0.287)| = 13.9. In principle, we could also analyze the set of equilibrium points along the cold boundary, where q = 0 and T < TIG . Near such points, linearization techniques fail because of the nature of the Heaviside function. We would have to perform a more robust analysis. As an alternative, we shall simply visually examine the results of calculations and infer the stability of this set of xed points.
8.2. STEADY BURNER-STABILIZED FLAMES 8.2.2.5 Laminar ame structure
305
8.2.2.5.1 TIG = 0.2. Let us get a numerical solution by integrating backwards in space from the equilibrium point. Here we will use the parameters of Table 8.1, in particular to contrast with a later calculation, the somewhat elevated ignition temperature of TIG = 0.2. We shall commence the solution near the isolated equilibrium point corresponding to chemical equilibrium. We could be very careful and choose the initial condition to lie just o the saddle along the eigenvector associated with the negative eigenvalue. It will work just as well to choose a point on the correct side of the equilibrium. Let us approximate x by xB , require q(xB ) = 0, T (xB ) = 1 , where 0 < << 1. That is we perturb the temperature to be just less than its equilibrium value. We integrate from the large positive x = xB back towards x = . When we nd T = TIG , we record the value of x. We translate the plots so that the ignition point is reached at x = 0 and give results. Predictions of T (x) are shown in in Figure 8.3. We see the temperature has T (0) =
T 1.0
T T eq 0.953
0.8 0.6 0.4 0.2 x
10
15
20
25
Figure 8.3: Dimensionless temperature as a function of position for a burner-stabilized premixed laminar ame, TIG = 0.2. 0.2 = TIG . As x , the temperature approaches its equilibrium value. For x < 0, the temperature continues to fall until it comes to a nal value of T 0.132. Note that this value cannot be imposed by the boundary conditions, since we enforce no other condition on T except to anchor it at the ignition temperature at x = 0. Predictions of q(x) are shown in in Figure 8.4. The heat ux is always negative. And it clearly relaxes to zero as x . This indicates that energy released in combustion makes its way back into the fresh mixture, triggering combustion of the fresh material. This is the essence of the laminar diusion ame. Predictions of (x) 1/(1 + QT (x)) are shown in in Figure 8.5. We are somewhat troubled because (x) nowhere takes on dimensionless value of unity, despite it being scaled by o . Had we anchored the ame at an ignition temperature very close to zero, we in fact would have seen approach unity at the anchor point of the ame. We chose an elevated ignition temperature so as to display the actual idiosyncrasies of the model. Even for a ame anchored at Tig To , we would nd a signicant decay of density in the cold region of the ow for x < xig .
306

q 5 0.02 0.04 0.06 0.08 0.10 0.12 5 10 15 20 25 x
Figure 8.4: Dimensionless Fourier heat ux as a function of distance for a burner-stabilized premixed laminar ame, TIG = 0.2.
0.8
0.6
0.4
0.2
10
15
20
25
30
Figure 8.5: Dimensionless density as a function of distance for a burner-stabilized premixed laminar ame, TIG = 0.2. Predictions of u(x) 1+QT (x) are shown in in Figure 8.6. Similar to , we are somewhat troubled that the dimensionless velocity does not take a value near unity. Once again, had we set Tig To , we would have found the u at x = xig would have taken a value very near unity. But it also would modulate signicantly in the cold region x < xig . We also would have discovered that had Tig To that the temperature in the cold region would drop signicantly below To , which is a curious result. Alternatively, we could iterate on the parameter Tig until we found a value for which limx T 0. For such a value, we would also nd u and to approach unity as x . We can better understand the ame structure by considering the (T, q) phase plane as shown in Fig. 8.7. Here, green denotes equilibria. We see the isolated equilibrium point at (T, q) = (0.953, 0). And we see a continuous one-dimensional set of equilibria for {(T, q)|q = 0, T (, TIG ]}. Blue lines are trajectories in the phase space. The arrows have been associated with movement in the x direction. This is because this is the direction in which it is easiest to construct the ame structure.

u 6 5 4 3 2 1 5 0 5 10 15 20 25 30 x
307
Figure 8.6: Dimensionless uid particle velocity u as a function of distance for a burnerstabilized premixed laminar lame, TIG = 0.2. We know from dynamic systems theory that special trajectories, known as heteroclinic, are those that connect one equilibrium point to another. These usually have special meaning. The heteroclinic orbit shown here in the thick blue line is that of the actual ame structure. It connects the saddle at (T, q) = (0.953, 0) to a stable point on the one-dimensional continuum of equilibria at (T, q) = (0.132, 0). It originates on the eigenvector associated with the negative eigenvalue at the equilibrium point. Moreover, it appears that the heteroclinic trajectory is an attracting trajectory, as points that begin on nearby trajectories seem to be drawn into the heteroclinic trajectory. The orange dashed line represents our chosen ignition temperature, TIG = 0.2. Had we selected a dierent ignition temperature, the heteroclinic trajectory originating from the saddle would be the same for T > TIG . However it would have turned at a dierent location and relaxed to a dierent cold equilibrium on the one-dimensional continuum of equilibria. Had we attempted to construct our solution by integrating forward in x, our task would have been more dicult. We would likely have chosen the initial temperature to be just greater than TIG . But we would have to had guessed the initial value of q. And because of the saddle nature of the equilibrium point, a guess on either side of the correct value would cause the solution to diverge as x became large. It would be possible to construct a trial and error procedure to hone the initial guess so that on one side of a critical value, the solution diverged to , while on the other it diverged to . Our procedure, however, has the clear advantage, as no guessing is required. As an aside, we note there is one additional heteroclinic orbit admitted mathematically; however, its physical relevance is far from clear. It seems there is another attracting trajectory for T > 0.953. This trajectory is associated with the same eigenvector as the physical trajectory. Along this trajectory, T increases beyond unity, at which point the mass fraction becomes greater than unity, and is thus non-physical. Mathematically this trajectory continues until it reaches an equilibrium at (T, q) (, 0). These dynamics can be revealed using the mapping of the Poincar sphere; see Fig. 8.8. Details can be found in some dynamic e
308

q
0.2
0.4
0.6
0.8
1.0
1.2
0.05
0.10
0.15
Figure 8.7: (T ,q) phase plane for a burner-stabilized premixed laminar ame, TIG = 0.2. systems texts.6 In short the mapping from (T, q) (T , q ) via T = q = T 1 + T 2 + q2 q 1 + T 2 + q2 , , (8.126) (8.127)
is introduced. Often a third variable in the mapping is introduced Z = 1/ 1 + T 2 + q2 . This induces T 2 + q2 + Z 2 = 1, the equation of a sphere, known as the Poincar sphere. e As it is not clear that physical relevance can be found for this, we leave out most of the mathematical details and briey describe the results. This mapping takes points at innity in physical space onto the unit circle. Equilibria are marked in green. Trajectories are in blue. While the plot is somewhat incomplete, we notice in Fig. 8.8 that the saddle is evident around (T , q ) (0.7, 0). We also see new equilibria on the unit circle, which corresponds to points at innity in the original space. One of these new equilibria is at (T , q ) = (1, 0). It is a sink. The other two, one in the second quadrant, the other in the fourth, are sources. The heteroclinic trajectories which connect to the saddle from the sources in the second and fourth quadrants
6
Perko, L., 2001, Dierential Equations and Dynamical Systems, Third Edition, Springer, New York.

q' 1.0
309
0.5
1.0
0.5
0.5
1.0
0.5
1.0
Figure 8.8: Projection of trajectories on Poincar sphere onto the (T , q ) phase plane for a e burner-stabilized premixed laminar ame, TIG = 0.2. form the boundaries for the basins of attraction. To the left of this boundary, trajectories are attracted to the continuous set of equilibria. To the right, they are attracted to the point (1, 0). 8.2.2.5.2 TIG = 0.076. We can modulate the ame structure by altering the ignition temperature. In particular, it is possible to iterate on the ignition temperature in such a way that as x , T 0, 1, and u 1. Thus the cold ow region takes on its ambient value. This has some aesthetic appeal. It is however unsatisfying in that one would like to think that an ignition temperature, if it truly existed, would be a physical property of the system, and not just a parameter to adjust to meet some other criterion. That duly noted, we present ame structures using all the parameters of Table 8.1 except we take Tig = 414 K, so that TIG = 0.076. Predictions of T (x) are shown in in Figure 8.9. We see the temperature has T (0) = 0.076 = TIG . As x , the temperature approaches its chemical equilibrium value. For x < 0, the temperature continues to fall until it comes to a nal value of T 0. Thus in dimensional terms, the temperature has arrived at To in the cold region. Predictions of q(x) are shown in in Figure 8.10. There is really nothing new here. Predictions of (x) 1/(1 + QT (x)) are shown in in Figure 8.11. We see that this special
310

T 1.0 T 0.8 T eq 0.953
0.6
0.4
0.2
10
15
20
25
Figure 8.9: Dimensionless temperature as a function of position for a burner-stabilized premixed laminar ame, TIG = 0.076.
q 5 0.02 0.04 0.06 0.08 0.10 0.12 0.14 10 15 20 25 x
Figure 8.10: Dimensionless Fourier heat ux as a function of distance for a burner-stabilized premixed laminar ame, TIG = 0.076.
value of TIG has allowed the dimensionless density to take on a value of unity as x . Predictions of u(x) 1+QT (x) are shown in in Figure 8.12. In contrast to our earlier result, for this special value of TIG , we have been able to allow u = 1 as x . We can better understand the ame structure by considering the (T, q) phase plane as shown in Fig. 8.13. Figure 8.13 is essentially the same as Figure 8.7 except the ignition point has been moved.
8.2.3
Detailed H2-O2 -N2 kinetics
We give brief results for a premixed laminar ame in a mixture of calorically imperfect ideal gases which obey mass action kinetics with Arrhenius reaction rates. Multi-component diusion is modelled as is thermal diusion. We consider the kinetics model of Table 1.2. The

1.2 1.0 0.8 0.6 0.4 0.2 5 0 5 10 15 20 25 30 x
311
Figure 8.11: Dimensionless density as a function of distance for a burner-stabilized premixed laminar ame, TIG = 0.076.
u 6 5 4 3 2 1 5 0 5 10 15 20 25 30 x
Figure 8.12: Dimensionless uid particle velocity u as a function of distance for a burnerstabilized premixed laminar ame, TIG = 0.076. solution method is described in detail in a series of reports from Reaction Design, Inc.78910 The methods employed in the solution here are slightly dierent. In short, a large, but nite domain dened. Then the ordinary dierential equations describing the ame structure are discretized. This leads to a large system of non-linear algebraic equations. These are solved by iterative methods, seeded with an appropriate initial guess. The temperature is
Kee, R. J., et al., 2000, Premix: A Fortran Program for Modeling Steady, Laminar, One Dimensional Premixed Flames, from the Chemkin Collection, Release 3.6, Reaction Design, San Diego, CA. 8 Kee, R. J., et al., 2000, Chemkin: A Software Package for the Analysis of Gas-Phase Chemical and Plasma Kinetics, from the Chemkin Collection, Release 3.6, Reaction Design, San Diego, CA. 9 R. J. Kee, et al., 2000, The Chemkin Thermodynamic Data Base, part of the Chemkin Collection Release 3.6, Reaction Design, San Diego, CA. 10 Kee, R. J., Dixon-Lewis, G., Warnatz, J., Coltrin, M. E., and Miller, J. A., 1998, A Fortran Computer Code Package for the Evaluation of Gas-Phase Multicomponent Transport Properties, Sandia National Laboratory, Report SAND86-8246, Livermore, CA. CC BY-NC-ND. 12 December 2011, J. M. Powers.
7
312

q
0.2
0.4
0.6
0.8
1.0
1.2
0.05
0.10
0.15
Figure 8.13: (T ,q) phase plane for a burner-stabilized premixed laminar ame, TIG = 0.076. pinned at To at one end of the domain, which is a slightly dierent boundary condition than we employed previously. At an intermediate value of x, x = xf , the temperature is pinned at T = Tf . Full details are in the report. For To = 298 K, Po = 1.01325 105 P a, and a stoichiometric unreacted mixture of 2H2 + O2 + 3.76 N2 , along with a very small amount of minor species, we give plots of Y( x) in Fig. 8.14. There is, on a log-log scale, an incubation period spanning a few orders of magnitude of length. Just past x = 2 cm, a vigorous reaction commences, and all species relax to a nal equilibrium. Temperature as a function of distance is shown in Fig. 8.15. Density and particle velocity are shown in Figs. 8.16 and 8.17, respectively.
313
10
10
Yi
10
10
H2 O2 H2O OH H O HO2 H2O2 N2

2
10
15
10
20
10
10
10 x [cm]
10
Figure 8.14: Mass fractions as a function of distance in a premixed laminar ame with detailed H2 -O2-N2 kinetics.
2500
2000
[K ] T
1500
1000
500
0 4
10
12
14
x xf [cm]
Figure 8.15: Temperature as a function of distance in a premixed laminar ame with detailed H2 -O2 -N2 kinetics.
314
x 10
9 8
[ g/cm3 ]
7 6 5 4 3 2 1 4 2 0 2 4 6 8 10 12 14
x xf [cm]
Figure 8.16: Density as a function of distance in a premixed laminar ame with detailed H2 -O2 -N2 kinetics.
1600 1400 1200
u [cm/s]
1000 800 600 400 200 4
10
12
14
x xf [cm]
Figure 8.17: Fluid particle velocity as a function of distance in a premixed laminar ame with detailed H2 -O2-N2 kinetics.
Chapter 9 Simple detonations: Reaction-advection

Plato say, my friend, that society cannot be saved until either the Professors of Greek take to making gunpowder, or else the makers of gunpowder become Professors of Greek. Undershaft to Cusins in G. B. Shaws Major Barbara, Act III. Let us consider aspects of the foundations of detonation theory. Detonation is dened as a shock-induced combustion process. It is well modeled by considering the mechanisms of advection and reaction. Diusion plays a higher order role; its neglect is not critical to the main physics. The denitive text is that of Fickett and Davis.1 The present chapter is strongly inuenced by this monograph.
9.1
9.1.1
Reactive Euler equations

One-step irreversible kinetics
Let us focus here on one-dimensional planar solutions in which all diusion processes are neglected. This is known as a reactive Euler2 model. We shall also here only be concerned with simple one-step irreversible kinetics in which A B. (9.1)
We will adopt the assumption that A and B are materials with identical properties, thus MA = MB = M,
1 2
c P A = cP B = cP ,
(9.2)
Fickett, W., and Davis, W. C., 1979, Detonation, U. California, Berkeley. Leonhard Euler, 1707-1783, Swiss mathematician.
315
316
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION
but A is endowed with chemical energy which is released as it forms B. So here we have N = 2 and J = 1. The number of elements will not be important for this analysis. Extension to multi-step kinetics is straightforward, but involves a sucient number of details to obscure many of the key elements of the analysis. At this point, we will leave the state equations relatively general, but we will soon extend them to simple ideal, calorically perfect relations. Since we have only a single reaction, our reaction rate vector rj , is a scalar, which we will call r. Our stoichiometric matrix ij is of dimension 2 1 and is ij = 1 1 . (9.3)
Here the rst entry is associated with A, and the second with B. Thus our species production rate vector, from i = J ij rj , reduces to j=1 A B = 1 1 (r) = r r . (9.4)
Now the right side of Eq. (6.5) takes the form Mi i . Let us focus on the products and see how this form expands. MB B = MB r, = MB kA , = MA kA , YA = MA k , MA = kYA , = k(1 YB ). (9.6) (9.7) (9.8) (9.9) (9.10) (9.11) (9.5)
For a reaction which commences with all A, let us dene the reaction progress variable as = YB = 1 YA . And let us dene r for this problem, such that Expanding further, we could say r = aT exp r = k(1 YB ) = k(1 ). E RT
(9.12)
(1 ) = aT exp
E RT
(1 ),
(9.13)
but we will delay introduction of temperature T . Here we have dened E = E/M. Note that the units of r for this problem will be 1/s, whereas the units for r must be mole/cm3 /s, and that r= r. (9.14) MB The use of two dierent forms of the reaction rate, r and r is unfortunate. One is nearly universal in the physical chemistry literature, r; the other is nearly universal in the one-step detonation chemistry community r.
9.1. REACTIVE EULER EQUATIONS
317
9.1.2
Thermicity
e v P,
Let us specialize the denition of frozen sound speed, Eq. (3.385), for our one-step chemistry: c=v This gives, using v = 1/, c =
2 2
P+
e P v,
(9.15)
P+
e v P,
e P v,
(9.16)
Let us, as is common in the detonation literature, dene the thermicity via the relation c =
2 e P,v . e P v,
(9.17)
We now specialize our general mathematical relation, Eq. (3.59), to get P e so that one gets
e P,v e P v,
v,
p,v
v,e
= 1,
(9.18)
.
v,e
(9.19)
Thus, using Eq. (9.19), we can rewrite Eq. (9.17) as c2 = = P ,

v,e
(9.20) .
v,e
1 P c2
(9.21)
For our one step reaction, we see that as the reaction moves forward, i.e. as increases, that is a measure of how much the pressure increases, since > 0, c2 > 0.
9.1.3
Parameters for H2-Air
Let us postulate some parameters which loosely match results of the detailed kinetics calculation of Powers and Paolucci 3 for for H2 -air detonations. Rough estimates which allow one-step kinetics models with calorically perfect ideal gas assumptions to approximate the results of detailed kinetics models with calorically imperfect ideal gas mixtures are given in Table 9.1.
Powers, J. M., and Paolucci, S., 2005, Accurate Spatial Resolution Estimates for Reactive Supersonic Flow with Detailed Chemistry, AIAA Journal, 43(5): 1088-1099. CC BY-NC-ND. 12 December 2011, J. M. Powers.
3
318
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION Parameter M R Po To o vo q E a Value 1.4 20.91 397.58 1.01325 105 298 0.85521 1.1693 1.89566 106 8.29352 106 5 109 0 Units kg/kmole J/kg/K Pa K kg/m3 m3 /kg J/kg J/kg 1/s
Table 9.1: Numerical values of parameters which roughly model H2 -air detonation.
9.1.4
Conservative form
Let us rst consider a three-dimensional conservative form of the governing equations in the inviscid limit: + (u) = 0, t t 1 e+ uu 2 (u) + (uu + P I) t 1 P + u e + u u + 2 () + (u) t e r = 0, = 0, = r, = e(P, , ), = r(P, , ). (9.22) (9.23) (9.24) (9.25) (9.26) (9.27)
For mass conservation, Eq. (9.22) is identical to the earlier Eq. (6.1). For linear momentum conservation, Eq. (9.23) is Eq. (6.2) with the viscous stress, = 0. For energy conservation, Eq. (9.24) is Eq. (6.3) with viscous stress = 0, and diusive energy transport jq = 0. For species evolution, Eq. (9.27) is Eq. (6.5) with diusive mass ux jm = 0. i Equations (9.22-9.27) form eight equations in the eight unknowns, , u, P , e, , r. Note that u has three unknowns, and Eq. (9.23) gives three equations. Note also that we assume the functional forms of e and r are given. The conservative form of the equations is the most useful and one of the more fundamental forms. It is the form which arises from the even more fundamental integral form, which admits discontinuities. We shall take advantage of this in later forming shock jump equations. The disadvantage of the conservative form is that it is unwieldy and masks simpler causal relations.
319
9.1.5
We can reveal some of the physics described by Eqs. (9.22-9.24) by writing them in what is known as a non-conservative form. 9.1.5.1 Mass
We can use the product rule to expand Eq. (9.22) as + u + u = 0. t

=d/dt
(9.28)
We recall the well known material derivative, also known as the derivative following a material particle, the total derivative, or the substantial derivative: d + u . dt t Using the material derivative, we can rewrite the mass equation as d + u = 0. (9.29) dt Often, Newtons notation for derivatives, the dot, is used to denote the material derivative. We can also recall the denition of the divergence operator, div, to be used in place of , so that the mass equation can be written compactly as + div u = 0. (9.30) The density of a material particle changes in response to the divergence of the velocity eld, which can be correlated with the rate of volume expansion of the material region. 9.1.5.2 Linear momenta
We can use the product rule to expand the linear momenta equations, Eq. (9.23), as u +u + u( (u)) + u u + P = 0, (9.31) t t u +u + (u) +u u + P = 0, (9.32) t t
=0
u + +u u + P = 0, t du + P = 0. dt Or in terms of our simplied notation with v = 1/, we have u + v grad P = 0. The uid particle accelerates in response to a pressure gradient.
(9.33) (9.34)
(9.35)
320 9.1.5.3 Energy
We can apply the product rule to the energy equation, Eq. (9.24) to get t 1 1 e+ uu + e+ uu 2 2 1 + e + u u (u) t 2 1 +u e + u u + (P u) = 0, 2 1 1 e+ uu + e+ uu + (u) 2 2 t
=0
(9.36)
1 +u e + u u + (P u) = 0, 2 1 1 e + u u + u e + u u + (P u) = 0. 2 2
(9.37) (9.38)
This is simpler, but there is more that can be done by taking advantage of the linear momenta equations. Let us continue working with the product rule once more along with some simple rearrangements e 1 1 + u e + u u + u u u + (P u) = 0, t t 2 2 u e + u (u )u + u P + P u = 0, + u e + u t t u e + u e +u + (u )u + P +P u = 0, t t
= de/dt =0
(9.39) (9.40) (9.41)
de + P u = 0. dt
(9.42)
Now from Eq. (9.29), we have u = (1/)d/dt. Eliminating the divergence of the velocity eld from Eq. (9.42), we get de P d = 0, dt dt de P d = 0. dt 2 dt (9.43) (9.44)
Recalling that = 1/v, so d/dt = (1/v 2 )dv/dt = 2 dv/dt, our energy equation becomes de dv +p = 0. dt dt Or in terms of our dot notation, we get e + pv = 0.
(9.45)
(9.46)
321
The internal energy of a uid particle changes in response to the work done by the pressure force. 9.1.5.4 Reaction
Using the product rule on Eq. (9.27) we get + + u + (u) = r, t t = r, + u + + (u) t t

=d/dt =0
(9.47) (9.48)
d = r, dt = r.
(9.49) (9.50)
The mass fraction of a uid particles product species changes according to the forward reaction rate. 9.1.5.5 Summary
In summary, our non-conservative equations are + div u u + v grad P e + Pv e r 0, 0, 0, r, e(P, , ), r(P, , ), 1 . = v = = = = = = (9.51) (9.52) (9.53) (9.54) (9.55) (9.56) (9.57)
9.1.6
One-dimensional form
Here we consider the equations to be restricted to one-dimensional planar geometries. 9.1.6.1 Conservative form
In the one-dimensional planar limit, Eqs. (9.22-9.27) reduce to + (u) = 0, t x u2 + P = 0, (u) + t x (9.58) (9.59)
322 t
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION 1 e + u2 2 x 1 P u e + u2 + 2 () + (u) t x e r
= 0, = r, = e(P, , ), = r(P, , ).
(9.60) (9.61) (9.62) (9.63)
9.1.6.2
The non-conservative Eqs. (9.51-9.57) reduce to the following in the one-dimensional planar limit: u x P u+v x e + Pv e r + = 0, = 0, = = = = (9.64) (9.65) (9.66) (9.67) (9.68) (9.69) (9.70)
0, r, e(P, , ), r(P, , ), 1 . = v
Equations (9.64-9.70) can be expanded using the denition of the material derivative: +u t x u u +u t x v e e +P +u t x t t u x P +v x v +u x +u x e r +
= 0, = 0, = 0, = r,
(9.71) (9.72) (9.73) (9.74) (9.75) (9.76) (9.77)
= e(P, , ), = r(P, , ), 1 . = v
9.1. REACTIVE EULER EQUATIONS 9.1.6.3 Reduction of energy equation
323
Let us use standard results from calculus of many variables to expand the caloric equation of state, Eq. (9.75): de = e P dP +
v,
e v
dv +
P,
d.
P,v
(9.78)
Taking the time derivative gives e= e P e P+ v v, v+

P,
.
P,v
(9.79)
Note this is simply a time derivative of the caloric state equation; it says nothing about energy conservation. Next let us use Eq. (9.79) to eliminate e in the rst law of thermodynamics, Eq. (9.66) to get e P e P+ v v, P+ P v+
P,
+P v = 0,
P,v e P,v e P v, =c2
(9.80)
=e e + v P, e P v, =2 c2
v+
= 0,
(9.81)
P + c v c2 = 0, P = 2 c2 v + c2 , P = c2 (r v) .
2 2
(9.82) (9.83) (9.84)
Now since v = (1/2 ), we get P = c2 + c2 r. (9.85)
Note that if either r = 0 or = 0, the pressure changes will be restricted to those from classical isentropic thermo-acoustics: P = c2 . If r > 0, reaction induces positive pressure changes. If r < 0, reaction induces negative pressure changes. Moreover note that Eq. (9.85) is not restricted to calorically perfect ideal gases. It is valid for general state equations.
9.1.7
Characteristic form
Let us obtain a standard form known as characteristic form for the one-dimensional unsteady equations. We follow the procedure described by Whitham (1927-).4 For this form
4
Whitham, G. B., 1974, Linear and Nonlinear Waves, Wiley, New York. CC BY-NC-ND. 12 December 2011, J. M. Powers.
324
let us use Eq. (9.85) and generalized forms for sound speed c and thermicity to recast our governing equations, Eq. (9.71-9.77) as u +u + t x x u 1 P u +u + t x x P P +u c2 +u t x t x +u t x c2 r = 0, = 0, = c2 r, = r, = c2 (P, ), = r(P, , ), = (P, , ). (9.86) (9.87) (9.88) (9.89) (9.90) (9.91) (9.92)
These take the general form
Let us write the dierential equations in matrix form: u 0 0 1 0 0 0 t x 1 0 u 0 u 1 0 0 u 0 x t 2 c 0 1 0 P + c2 u 0 u 0 P t x 0 0 0 1 0 0 0 u t x wj wj + Bij = Ci . t x
0 0 = 2 . c r r
(9.93)
Aij
(9.94)
Let us attempt to cast this the left hand side of this system in the form wj /t + wj /x. Here is a scalar which has the units of a velocity. To do so, we shall seek vectors i such that i Aij wj wj + i Bij = i Ci = mj t x wj wj + t x . (9.95)
For i to have the desired properties, we will insist that i Aij = mj , i Bij = mj . Using Eq. (9.96) to eliminate mj in Eq. (9.97), we get i Bij = i Aij , 0 = i (Aij Bij ).
(9.96) (9.97)
(9.98) (9.99)
325
Equation (9.99) has the trivial solution i = 0. For a non-trivial solution, standard linear algebra tells us that we must enforce the condition that the determinant of the coecient matrix be zero: |Aij Bij | = 0. Specializing Eq. (9.100) for Eq. (9.93), we nd u 0 0 1 0 u 0 2 c ( u) 0 u 0 0 0 0 u (9.100)
= 0.
(9.101)
Let us employ standard co-factor expansion operations to reduce the determinant: u 0 0 u 1 ( u) c2 ( u) 0 u c2 ( u) ( u) ( u) ( u)2 + 2 ( u) ( u)2 c2 There are four roots to this equation; two are repeated: = = = = u, u, u + c, u c. (9.105) (9.106) (9.107) (9.108) = 0, = 0, = 0. (9.102) (9.103) (9.104)
Let us nd the eigenvector i associated with the eigenvalues = uc. So Eq. (9.99) reduces to (u c) u 0 0 1 0 (u c) u 0 = (0 0 0 0). ( 1 2 3 4 ) 2 c ((u c) u) 0 (u c) u 0 0 0 0 (u c) u (9.109) Simplifying, c 0 4 ) 3 c 0 0 c 1 0 c 0 0 0 0 = (0 0 0 0). 0 c
( 1
(9.110)
326
We thus nd four equations, with linear dependencies: c1 c3 3 1 c2 1 2 c3 c4 Now c = 0, so Eq. (9.114) insists that 4 = 0. (9.115) = 0, = 0, = 0, = 0. (9.111) (9.112) (9.113) (9.114)
We expect a linear dependency, which implies that we are free to set at least one of the remaining i to an arbitrary value. Let us see if we can get a solution with 3 = 1. With that Eqs. (9.111-9.113) reduce to c1 c3 = 0, 1 c2 = 0, 1 2 c = 0. Solving Eq. (9.116) gives 1 = c2 . Then Eq. (9.117) becomes c2 c2 = 0. Solving gives 2 = c. (9.120) This is redundant with solving Eq. (9.118), which also gives 2 = c. So we have for = u c that i = ( c2 So Eq. (9.93) becomes 1 0 ( c2 c 1 0 ) 2 c 0 c 1 0 ) . (9.121) after multiplication by i from Eq. (9.121): u 0 0 0 0 0 t x 1 u 0 u 1 0 0 u t x P + ( c2 c 1 0 ) 0 P c2 u 0 u 0 0 1 0 t x 0 0 1 0 0 0 u t x 0 0 = ( c2 c 1 0 ) 2 c r r (9.122) (9.119) (9.116) (9.117) (9.118)
9.1. REACTIVE EULER EQUATIONS Carrying out the vector-matrix multiplication operations, we get
t t t
327
u t ( 0 c 1 0 ) P + ( 0 c(u c) u c 0 ) = ( c2
0 0 c 1 0 ) 2 . c r r
x u x P x x
(9.123)
Simplifying, c u u + (u c) t x + P P + (u c) t x = c2 r. (9.124)
Now let us conne our attention to lines in x t space on which dx = u c. dt On such lines, Eq. (9.124) can be written as c u dx u + t dt x + P dx P + t dt x = c2 r. (9.126) (9.125)
Consider now a variable, say u, which is really u(x, t). From calculus of many variables, we have u u dt + dx, t x du u dx u = + . dt t dt x du = (9.127) (9.128)
If we insist dx/dt = u c, let us call the derivative du/dt , so that Eq. (9.126) becomes c dP du + = c2 r. dt dt (9.129)
In the inert limit, after additional analysis, Eq. (9.129) reduces to the form d/dt = 0, which shows that is maintained as a constant on lines where dx/dt = u c. Thus one can say that a signal is propagated in x t space at speed u c. So we see from the characteristic analysis how the thermodynamic property c has the added signicance of inuencing the speed at which signals propagate.
328
Let us now nd i for = u. For this root, we nd 0 0 0 0 1 4 ) 0 0 0 0 0 0 0 0 = (0 0 0 0). 0 0
( 1
(9.130)
It can be seen by inspection that two independent solutions i satisfy Eq. (9.130): i = ( 0 0 1 0 ) , i = ( 0 0 0 1 ) . (9.131) (9.132)
These two eigenvectors induce the characteristic form which was already obvious from the initial form of the energy and species equations: P P +u c2 t x = c2 r, +u t x +u = r. t x (9.133) (9.134)
On lines where dx/dt = u, that is to say on material particle pathlines, these reduce to P c2 = c2 r and = r. Because we all of the eigenvalues are real, and because we were able to nd a set of four linearly independent right eigenvectors i so as to transform our four partial dierential equations into characteristic form, we can say that our system is strictly hyperbolic. Thus it is well posed for initial value problems given that initial data is provided on a noncharacteristic curve, and admits discontinuous solutions described by a set of Rankine-Hugoniot jump conditions which arise from a more primitive form of the governing equations. In summary we can write our equations in characteristic form as dP du + c dt+ dt+ du dP c dt dt d dP c2 dt dt d dt = c2 r, = c2 r, = c2 r, = r on on on on dx = u + c, dt dx = u c, dt dx = u, dt (9.135) (9.136) (9.137) (9.138)
dx = u. dt
329
9.1.8
Rankine-Hugoniot jump conditions
As described by LeVeque5 the proper way to arrive at what are known as Rankine6 -Hugoniot7 jump equations describing discontinuities is to use a more primitive form of the conservation laws, expressed in terms of integrals of conservative form quantities balanced by uxes and source terms of those quantities. If q is a set of conservative form variables, and f(q) is the ux of q (e.g. for mass conservation, is a conserved variable and u is the ux), and s(q) is the internal source term, then the primitive form of the conservation form law can be written as x2 d x2 s(q(x, t))dx. (9.139) q(x, t)dx = f(q(x1 , t)) f(q(x2 , t)) + dt x1 x1 Here we have considered ow into and out of a one-dimensional box for x [x1 , x2 ]. For our reactive Euler equations we have u 0 0 u2 + P u q= f(q) = s(q) = . (9.140) e + 1 u2 , 0 u e + 1 u2 + P , 2 2 r u If we assume there is a discontinuity in the region x [x1 , x2 ] propagating at speed U, we can break up the integral into the form d dt
x1 +U t
q(x, t)dx +
x1
d dt
x2
q(x, t)dx
x1 +U t+ x2
= f(q(x1 , t)) f(q(x2 , t)) +
s(q(x, t))dx.
x1
(9.141)
Here x1 + Ut lies just before the discontinuity and x1 + Ut+ lies just past the discontinuity. Using Leibnizs rule, we get
x1 +U t
q(x1 + Ut , t)U + 0 +
x1
q dx + 0 q(x1 + Ut+ , t)U + t

x2
x2 x1 +U t+
q dx t
(9.142)
= f(q(x1 , t)) f(q(x2 , t)) +
s(q(x, t))dx.
x1
Now if we assume that x2 x1 0 and that on either side of the discontinuity the volume of integration is suciently small so that the time and space variation of q is negligibly small, we get q(x1 )U q(x2 )U = f(q(x1 )) f(q(x2 )), U (q(x1 ) q(x2 )) = f(q(x1 )) f(q(x2 )).
5 6
(9.143) (9.144)
LeVeque, R. J., 1992, Numerical Methods for Conservation Laws, Birkhuser, Basel. a William John Macquorn Rankine, 1820-1872, Scottish engineer. 7 Pierre Henri Hugoniot, 1851-1887, French mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
330
Note that the contribution of the source term s is negligible as x2 x1 0. Dening next the notation for a jump as q(x) q(x2 ) q(x1 ), the jump conditions are rewritten as U q(x) = f(q(x)) . (9.146) (9.145)
If U = 0, as is the case when we transform to the frame where the wave is at rest, we simply recover 0 = f(q(x1 )) f(q(x2 )), f(q(x1 )) = f(q(x2 )), f(q(x)) = 0. (9.147) (9.148) (9.149)
That is the uxes on either side of the discontinuity are equal. We also get a more general result for U = 0, which is the well-known U= f(q(x2 )) f(q(x1 )) f(q(x)) = . q(x2 ) q(x1 ) q(x) (9.150) equa
The general Rankine-Hugoniot equation then for the one-dimensional reactive Euler tions across a non-stationary jump is given by 2 u2 1 u1 2 1 2 u2 + P2 1 u2 P1 2 u2 1 u1 2 1 U 2 e2 + 1 u2 1 e1 + 1 u2 = 2 u2 e2 + 1 u2 + P2 1 u1 e1 + 1 u2 + P1 2 2 2 1 2 2 2 2 1 1 2 2 1 1 2 u2 2 2 u1 1
(9.151)
Note that if there is no discontinuity, Eq. (9.139) reduces to our partial dierential equations which are the reactive Euler equations. We can rewrite Eq. (9.139) as d dt
x2 x1 x2
q(x, t)dx + (f(q(x2 , t)) f(q(x1 , t))) =
s(q(x, t))dx,
x1
(9.152)
Now if we assume continuity of all uxes and variables, we can use Taylor series expansion and Leibnizs rule to say
x2 x1
q(x, t)dx + t
f f(q(x1 , t)) + (x2 x1 ) + . . . f(q(x1 , t)) = x let x2 x1 x2 f q(x, t)dx + (x2 x1 ) = x x1 t x2 x2 f q(x, t)dx + dx = x1 t x1 x
x2
s(q(x, t))dx,
x1 x2
s(q(x, t))dx,
x1 x2
s(q(x, t))dx.
x1
(9.153)
9.1. REACTIVE EULER EQUATIONS Combining all terms under a single integral, we get
x2 x1
331
q f + s dx = 0. t x
(9.154)
Now this integral must be zero for an arbitrary x1 and x2 , so the integrand itself must be zero, and we get our partial dierential equation: q f + s = 0, t x q(x, t) + f(q(x, t)) = s(q(x, t)), t x which applies away from jumps. (9.155) (9.156)
9.1.9
Galilean transformation
We know that Newtonian8 mechanics have been constructed so as to be invariant under a socalled Galilean9 transformation which takes one from a xed laboratory frame to a constant velocity frame with respect to the xed frame. The Galilean transformation is such that our original coordinate system in the laboratory frame, (x, t), transforms to a steady wave frame, (, t), via x x = x Dt, t = t. We thus get dierentials d = x x dx + x t dt = dx + x x dt = dx Ddt, t t dt = dt. t (9.159) (9.160) (9.157) (9.158)
Scaling d by dt gives us then x dx d x = D. dt dt (9.161)
Taking as usual the particle velocity in the xed frame to be u = dx/dt and dening the particle velocity in the laboratory frame to be u = d/dt, we see that x u = u D.
8 9
(9.162)
after Isaac Newton, 1643-1727, English polymath. after Galileo Galilei, 1564-1642, Italian polymath. CC BY-NC-ND. 12 December 2011, J. M. Powers.
332
Now a dependent variable has a representation in the original space of (x, t), and in the transformed space as (, t). And they must both map to the same value of and the same x point. And there is a dierential of in both spaces which must be equal: d = dx + x t t dt =
x
= =
dt, d + x x t t x dt, (dx Ddt) + x t t x D dx + dt. x t x t t x . x t
(9.163) (9.164) (9.165)
Now consider Eq. (9.165) for constant x, thus dx = 0, and divide by dt: t t =
x
(9.166)
More generally the partial with respect to t becomes =

x
D . x t t x . x t . x t
(9.167)
Now consider Eq. (9.165) for constant t, thus dt = 0, and divide by dx: x More generally, x =
t
=
t
(9.168)
(9.169)
Let us consider how a material derivative transforms then + ( + D) u , +u = D t x x t x t x t t x = +u . x t t x
(9.170) (9.171)
Thus we can quickly write our non-conservative form, Eqs. (9.64-9.67) in the transformed frame u +u + = 0, (9.172) x x t u 1 P u +u + = 0, (9.173) x x t e v v e +u +u = 0, (9.174) P x x t t = r. (9.175) +u x t
9.2. ONE-DIMENSIONAL, STEADY SOLUTIONS
333
Moreover, we can write these equations in conservative form. Leaving out the details, which amounts to reversing our earlier steps which led to the non-conservative form, we get () u + x t 2 + P u () + u x t 1 P e + u2 + u + x 2 () + () u x t = 0, = 0, = 0, = r. (9.176) (9.177) (9.178) (9.179)
1 e + u2 2 t
9.2
One-dimensional, steady solutions
Let us consider a steadily propagating disturbance in a one dimensional ow eld. In the laboratory frame, the disturbance is observed to propagate with constant velocity D. Let us analyze such a disturbance.
9.2.1
Steady shock jumps

2 u2 1 u1 0 2 2 u2 + P2 1 u2 P1 1 0 = 0 2 u2 e2 + 1 u2 + P2 1 u1 e1 + 1 u2 + 2 2 2 2 1 0 2 u2 2 2 u1 1
In the steady frame, our jump conditions, Eq. (9.151) have U = 0 and reduce to .
P1 1
(9.180)
The mass jump equation can be used to quickly simplify the energy and species jump equations to get a revised set 2 u2 2 u2 + P2 2 1 P2 e2 + u2 + 2 2 2 2 = 1 u1 , = 1 u2 + P1 , 1 1 P1 = e1 + u2 + , 1 2 1 = 1 .
(9.181)
9.2.2
9.2.2.1
Ordinary dierential equations of motion

Conservative form
Let us now assert that in the steady laboratory frame that no variable has dependence on t, so / t = 0, and / x = d/d. With this assumption our partial dierential equations of x motion, Eqs. (9.176-9.179), become ordinary dierential equations:
334
d () = 0, u d x d 2 + P u d x d 1 P e + u2 + u d x 2 d () u d x e r = 0, = 0, = r, = e(P, , ), = r(P, , ).
(9.182) (9.183) (9.184) (9.185) (9.186) (9.187)
We shall take the following conditions for the undisturbed uid just before the shock at x = 0 : ( = 0 ) x u( = 0 ) x P ( = 0 ) x e( = 0 ) x ( = 0 ) x = = = = = o , D, Po , eo , 0. (9.188) (9.189) (9.190) (9.191) (9.192)
Note that in the laboratory frame, this corresponds to a material at rest since u = u + D = (D) + D = 0.
9.2.2.2
Unreduced non-conservative form
We can gain insights into how the dierential equations, Eqs. (9.182-9.185) behave by writing them in a non-conservative form. Let us in fact write them, taking advantage of the reductions we used to acquire the characteristic form, Eqs. (9.86-9.89) to write those equations after transformation to the steady Galilean transformed frame as d u d +u d x d x d 1 dP u u + d d x x d dP c2 u u d x d x d u d x
= 0, = 0, = c2 r, = r.
(9.193) (9.194) (9.195) (9.196)
335
We can use Cramers10 rule to invert the coecient matrix to solve for the evolution of each state variable. This rst requires the determinant of the coecient matrix . u 0 c2 u 0 0 0 u 0 1 , 0 u 0 0 0 u
Let us attempt next to get explicit representations for the rst derivatives of each equation. In matrix form, we can say our system is d u 0 0 0 d x u 0 u 1 0 d 0 d x dP = 2 . (9.197) c2 u 0 u 0 c r d x d r 0 0 0 u d x
(9.198)
u 0 0 u , 1 = u c2 u 0 u c2 u , = u u(2 ) u = u2 (2 c2 ). u
(9.199) (9.200) (9.201)
As appears in the denominator after application of Cramers rule, we see immediately that when the waveframe velocity u becomes locally sonic ( = c) that our system of dierential u equations is potentially singular. Now by Cramers rule, d/d is found via x 0 0 u 0 1 0 u 0 0 0 u , 0 0 0 u 1 u c2 r 0 u , 2 (2 c2 ) u u u()(c2 r) , u2 (2 c2 ) u c2 r . u(2 c2 ) u 0 0 c2 r r
d = d x
(9.202)
= = =
10
(9.203) (9.204) (9.205)
Gabriel Cramer, 1704-1752, Swiss mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.
336
Similarly, solving for d/d we get u x u 0 0 0 1 0 0 0 2 2 c u c r u 0 0 r 0 u , u 0 0 1 0 0 u c2 u c2 r u , u2 (2 c2 ) u u2 (c2 r) , u2 (2 c2 ) u c2 r . 2 u c2 0 0 u 0 0 0 c2 r 0 0 r u , u 0 0 u 0 u 2 c u 0 c2 r , 2 (2 c2 ) u u uc2 r2 u , u2 (2 c2 ) u uc2 r . u 2 c2 u 0 c2 u 0
d u = d x
(9.206)
= = = For dP/d, we get x
(9.207) (9.208) (9.209)
dP d x
(9.210)
= = = Lastly, we have by inspection
(9.211) (9.212) (9.213)
r d = . d x u We can employ the local Mach number in the steady wave frame, u2 M2 2 , c to write our system of ordinary dierential equations as r d = , d x u(1 M 2 )
(9.214)
(9.215)
(9.216)
9.2. ONE-DIMENSIONAL, STEADY SOLUTIONS r d u = , d x 1 M2 dP r u = , d x 1 M2 r d = . d x u
337 (9.217) (9.218) (9.219)
We note the when the ow is locally sonic, M = 1, that our equations are singular. Recalling an analogy from compressible ow with area change, which is really an application of lHpitals rule, in order for the ow to be locally sonic, we must insist that simultaneously o the numerator must be zero; thus, we might demand that at a sonic point, r = 0. So an end state with r = 0 may in fact be a sonic point. For multi-step reactions, each with their own thermicity and reaction progress, we require the generalization r = 0 at a local sonic point. Here is the vector of thermicities and r is the vector of reaction rates. This condition will be important later when we consider so-called eigenvalue detonations. 9.2.2.3 Reduced non-conservative form
Let us use the mass equation, Eq. (9.182) to simplify the reaction equation, Eq. (9.185), then integrate our dierential equations for mass momentum and energy conservation, apply the initial conditions, and thus reduce our system of four dierential and two algebraic equations to one dierential and ve algebraic equations: u P + 2 u 1 P e + u2 + u 2 d d x e r = o D, = Po + o D 2 , 1 Po = o D eo + D 2 + 2 o 1 = r, u = e(P, , ), = r(P, , ). , (9.220) (9.221) (9.222) (9.223) (9.224) (9.225)
With some eort, we can unravel these equations to form one ordinary dierential equation in one unknown. But let us delay that analysis until after we have examined the consequences of the algebraic constraints
9.2.3
Rankine-Hugoniot analysis
Let us rst analyze our steady mass, momentum, and energy equations, Eqs. (9.220-9.222). Our analysis here will be valid both within, (0 < < 1) and at the end of the reaction zone ( = 1).
338 9.2.3.1
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION Rayleigh line
Let us get what is known as the Rayleigh11 line by considering only the mass and linear momentum equations, Eqs. (9.220-9.221). Let us rst rewrite Eq. (9.221) as P+ 2 D 2 2 u2 = Po + o . o (9.226)
Then, the mass equation, Eq. (9.22) allows us to rewrite the momentum equation, Eq. (9.226) as P+ Rearranging to solve for P , we nd P = Po 2 D 2 o 1 1 o . (9.228) 2 D 2 2 D 2 o = Po + o . o (9.227)
In terms of v = 1/, and a slight rearrangement, Eq. (9.228) can be re-stated as P = Po P Po Note: This is a line in (P, 1/) space, the Rayleigh line. The slope of the Rayleigh line is strictly negative. The magnitude of the slope of the Rayleigh line is proportional the square of the wave speed; high wave speeds induce steep slopes. The Rayleigh line passes through the ambient state (Po , 1/o ). Small volume leads to high pressure. These conclusions are a consequence of mass and momentum conservation alone. No consideration of energy has been made. The Rayleigh line equation is valid at all stages of the reaction: the inert state, a shocked stated, an intermediate reacted state, and a completely reacted state. It is always the same line. A plot of a Rayleigh line for D = 2800 m/s and the parameters of Table 9.1 is shown in Fig. 9.1. Here the point labeled O is the ambient state (1/o , Po).
11
D2 (v vo ) , 2 vo D2 v = 1 1 . Po vo vo
(9.229) (9.230)
John William Strutt, 3rd Baron Rayleigh, 1842-1919, English physicist.
339
P Pa 8. 106 6. 106 4. 106 2. 106

O
0.5
1.0
1 m3 kg
Figure 9.1: Plot of Rayleigh line for parameters of Table 9.1 and D = 2800 m/s. Slope is 2 D 2 < 0. o 9.2.3.2 Hugoniot curve
Let us next focus on the energy equation, Eq. (9.222). We shall use the mass and momentum equations (9.220,9.221) to cast the energy equation in a form which is independent of both velocities and wave speeds. From Eq. (9.220), we can easily see that Eq. (9.222) rst reduces to 1 P e + u2 + 2 1 Po = eo + D 2 + , 2 o (9.231) (9.232) Now the mass equation (9.220) also tells us that u= so Eq. (9.231) can be recast as 1 e+ 2 1 e eo + D 2 2 1 e eo + D 2 2 o
2
o D,
(9.233)
D2 + + +
1 Po = eo + D 2 + , 2 o
(9.234) (9.235) (9.236)
Po P = 0, o P Po = 0. o
(o )(o + ) 2
340
Now the Rayleigh line, Eq. (9.228) can be used to solve for D 2 , Po P D = 2 o
2
1 1 o
Po P 2 o
o o
(9.237)
Now use Eq. (9.237) to eliminate D 2 in Eq. (9.236): e eo + 1 2 Po P 2 o o o

=D 2
(o )(o + ) 2 1 2 Po P o
P Po = 0, o = 0, = 0, = 0, = 0, = 0.
(9.238)
o + P Po + o 1 P 1 1 Po e eo + (Po P ) + + 2 o o 1P P Po 1 Po 1 Po 1 P + + e eo + 2 2 o 2 2 o o 1 Po Po P P e eo + + 2 o o 1 1 1 e eo + (P + Po ) 2 o e eo + Rearranging, we get e eo = P + Po 2 -average pressure 1 1 o change in volume .
(9.239) (9.240) (9.241) (9.242) (9.243)
(9.244)
change in energy
This is the Hugoniot equation for a general material. It applies for solid, liquid, or gas. Note that This form of the Hugoniot does not depend on the state equation. The Hugoniot has no dependency on particle velocity or wave speed. The Hugoniot is valid for all e(); that is it is valid for inert, partially reacted, and totally reacted material. Dierent degrees of reaction will induce shifts in the curve. Now let us specify an equation of state. For convenience and to more easily illustrate the features of the Rankine-Hugoniot analysis, let us focus on the simplest physically-based state equation, the calorically perfect ideal gas for our simple irreversible kinetics model. With that we adopt the perfect caloric state equation studied earlier, Eq. (7.7): e = cv T q.
(9.245)
9.2. ONE-DIMENSIONAL, STEADY SOLUTIONS We also adopt an ideal gas assumption P = RT.
341
(9.246)
Solving for T , we get T = P//R. Thus cv T = (cv /R)(P/). Recalling that R = cP cv , we then get cv T = (cv /(cP cv ))(P/). And recalling that for the calorically perfect ideal gas = we get cv T = 1 P . 1 1 P q. 1 1 P v q. 1 (9.248) cP , cv (9.247)
Thus our caloric equation of state, Eq. (9.245), for our simple model becomes e(P, , ) = In terms of v, Eq. (9.249) is e(P, v, ) = (9.250) (9.249)
As an aside, we note that for this equation of state, the sound speed can be deduced from Eq. (9.16) as c =
2 2
P+
e v P,
e P v,
P+
1 P 1
1 v 1
P . v
(9.251) (9.252)
c2 = For the thermicity, we note that
P 1 P = P v = . 2 v
P = RT = P = 1 P c2 =
v,e
v,e
R R e + q T = , v v cv Rq = = ( 1)q, cv v
(9.253) (9.254) (9.255)
1 q 1 q ( 1)q = ( 1) 2 = . 2 c c P
Now at the initial state, we have = 0, and so Eq. (9.249) reduces to eo = 1 Po . 1 o (9.256)
342
We now use our caloric state relations, Eqs. (9.249,9.256) to specialize our Hugoniot relation, Eq. (9.244), to 1 1 P Po o q = P + Po 2 1 1 o . (9.257)
Employing v = 1/, we get a more compact form: 1 1 (P v Po vo ) q = (P + Po ) (v vo ) . 1 2 (9.258)
Let us next operate on Eq. (9.257) so as to see more clearly how such the Hugoniot is represented in the (P, v) = (P, 1/) plane. 1 1 (P v Po vo ) + (P v + Po v P vo Po vo ) = q, 1 2 1 1 1 1 1 1 v + v vo Po vo v + vo = q, 1 2 2 1 2 2 Po + 1 P +1 v vo vo v = q, 2 1 2 1 P = 2q + Po
+1 v 1 +1 v 1 o
(9.259) (9.260) (9.261) (9.262)
vo
Note that as v 1 vo , +1 P . (9.263)
For = 7/5, this gives v vo /6 induces and innite pressure. In fact an ideal gas cannot be compressed beyond this limit, known as the strong shock limit. Note also that as v , 1 Po < 0. +1 (9.264)
So very large volumes induce non-physical pressures. Let us continue to operate to get a more compact form for the calorically perfect ideal gas Hugoniot curve. Let us add a common term to both sides of Eq. (9.262):
+1 2q + Po 1 vo v 1 1 + P+ Po = Po , +1 +1 +1 vv 1 o
(9.265)
P+
1 Po +1
+1 v vo 1
= 2q + Po
+1 vo v 1

P Pa 8. 106
343
6. 106
N
4. 106
N
2. 106
C W
1 m3 kg
0.5
1.0
Figure 9.2: Plot of = 0, 1/2, 1 Hugoniot curves and two Rayleigh lines for D = 2800 m/s, D = 1991.1 m/s and parameters of Table 9.1. + = +1 1 Po v vo , +1 1 +1 Po vo Po v 2q + 1 1 Po vo , +Po v +1 +1 1 2q + Po vo Po vo , 1 +1 4 2q + 2 Po vo , 1 2q 4 + 2 , Po vo 1 1 2q 4 + . + 1 Po vo ( + 1)2 (9.266)
(9.267) (9.268) (9.269) (9.270) (9.271)
= = 1 P + Po + 1 1 P + Po + 1 +1 v 1 1 vo v 1 vo + 1 = =
Equation (9.271) represents a hyperbola in the (P, v) = (P, 1/) plane. As proceeds from 0 to 1 the Hugoniot moves. A plot of a series of Hugoniot curves for values of = 0, 1/2, 1 along with two Rayleigh lines for D = 2800 m/s and D = 1991.1 m/s are shown in Fig. 9.2. The D = 1991.1 m/s Rayleigh line happens to be exactly tangent to the = 1 Hugoniot curve. We will see this has special signicance. Let us dene, for convenience, the parameter 2 as 2 1 . +1 (9.272)
344
With this denition, we note that 1

4
= 1
1 +1
4 2 + 2 + 1 2 + 2 1 = . 2 ( + 1) ( + 1)2
(9.273)
With this denition, Eq. (9.271), the Hugoniot, becomes P + 2 Po v 2 vo = 22 q + 1 4 . Po vo (9.274)
Now let us seek to intersect the Rayleigh line with the Hugoniot curve and nd the points of intersection. To do so, let us use the Rayleigh line, Eq. (9.230), to eliminate pressure in the Hugoniot, Eq. (9.274): 1 D2 Po vo v 1 + 2 vo v 2 vo = 22 q + 1 4 . Po vo (9.275)
We now structure Eq. (9.275) so that it can be solved for v: 1 + 2

2
D2 D2 v + Po vo vo Po vo
v 2 vo
= 22
q + 1 4 . Po vo
(9.276)
Now, let us regroup to form v vo D2 Po vo + v vo (1 + 2 ) 1 + D2 Po vo 2 1 + 2 + 22 v vo

2
D2 Po vo
q 1 + 4 = 0, Po vo (9.277) q 1 = 0. Po vo (9.278)
D2 Po vo
v vo
(1 + 2 ) 1 +
D2 Po vo
2 1 +
D2 Po vo
22
This quadratic equation has two roots: v vo 1+ =

D2 Po vo
(1 + 2 )
2
2 PDvo o 1+
D2 Po vo 2
(1 + 2 )2 4 PDvo 1 + 1 + o
2
D2 Po vo
2 22 Pq o ov
2 PDvo o
(9.279)
For a given wave speed D, initial undisturbed conditions, Po , vo , and material properties, , q, Eq. (9.279) gives the specic volume as a function of reaction progress . Depending on these parameters, we can mathematically expect two distinct real solutions, two repeated solutions, or two complex solutions.
345
9.2.4
Shock solutions
We know from Eq. (9.181) that does not change through a shock. So if = 0 before the shock, it has the same value after. So we can get the shock state by enforcing the Rankine-Hugoniot jump conditions with = 0:
D2 Po vo
v vo
1+ =
(1 + 2 )
2
2 PDvo o 1+
D2 Po vo 2
(1 + 2 )2 4 PDvo 1 + 1 + o
2
D2 Po vo
2 . (9.280)
2 PDvo o
With considerable eort, or alternatively, by direct calculation via computational algebra, the two roots of Eq. (9.280) can be shown to reduce to: v = 1, vo Po vo v = 2 + 2 (1 + 2 ). vo D (9.281) (9.282)
The rst is the ambient solution; the second is the shocked solution. The shock solution can also be expressed as 1 2 Po vo v = + . vo + 1 + 1 D2 In the limit as Po vo /D2 0, the so-called strong shock limit, we nd v 1 . vo +1 The reciprocal gives the density ratio in the strong shock limit: +1 . o 1 (9.285) (9.284) (9.283)
With the solutions for v/vo , we can employ the Rayleigh line, Eq. (9.230) to get the pressure. Again, we nd two solutions: P = 1, Po D2 P = (1 2 ) 2 . Po Po vo (9.286) (9.287)
346
The rst is the inert solution, and the second is the shock solution. The shock solution is rewritten as P 2 D2 1 = . Po + 1 Po vo + 1 In the strong shock limit, Po vo /D2 0, the shock state reduces to 2 D2 P . Po + 1 Po vo Note, this can also be rewritten as 2 D 2 P , Po + 1 Po vo 2 D 2 . + 1 c2 o Lastly, the particle velocity can be obtained via the mass equation, Eq. (9.233): u o = , D v = , vo 1 2 Po vo = . + 1 + 1 D2 In the strong shock limit, Po vo /D2 0, we get u 1 . D +1 Note that in the laboratory frame, one gets uD 1 , D +1 1 u 1 , D +1 u 2 . D +1 (9.296) (9.297) (9.298) (9.295) (9.292) (9.293) (9.294) (9.290) (9.291) (9.289) (9.288)
9.2.5
Equilibrium solutions
Our remaining dierential equation, Eq. (9.223) is in equilibrium when r = 0, which for one-step irreversible kinetics, Eq. (9.12), occurs when = 1. So for equilibrium end states,
9.2. ONE-DIMENSIONAL, STEADY SOLUTIONS we enforce = 1 in Eq. (9.279) and get v vo 1+ =

D2 Po vo
347
(1 + 2 )
2
2 PDvo o 1+
D2 Po vo 2
(1 + 2 )2 4 PDvo 1 + 1 + o
2
D2 Po vo
2 22 Poqvo
2 PDvo o
(9.299)
This has only one free parameter, D. There are two solutions for v/vo at complete reaction. They can be distinct and real, repeated and real, or complex, depending on the value of D. We are most interested in D for which the solutions are real; these will be physically realizable. 9.2.5.1 Chapman-Jouguet solutions
Let us rst consider solutions for which the two roots of Eq. (9.299) are repeated. This is known as a Chapman12 -Jouguet13 (CJ) solution. For a CJ solution, the Rayleigh line is tangent to the Hugoniot at = 1 if the reaction is driven by one-step irreversible kinetics. We can nd values of D for which the solutions are CJ by requiring the discriminant under the square root operator in Eq. (9.299) to be zero. We label such solutions with a CJ subscript and say D2 1 + CJ Po vo
2
1 + 2
2 DCJ Po vo
1+ 1+
2 DCJ Po vo
2 22
q Po vo
= 0.
(9.300)
2 Equation (9.300) is quartic in DCJ and quadratic in DCJ . It has solutions 2 1 + 42 Poqvo 4 2 P2q o 2 (1 + 22 4 ) DCJ ov = . Po vo (2 1)2
(9.301)
The + root corresponds to a large value of DCJ . This is known as the detonation branch. For the parameter values of Table 9.1, we nd by substitution that DCJ = 1991.1 m/s for our H2 -air mixture. It corresponds to a pressure increase and a volume decrease. The - root corresponds to a small value of DCJ . It corresponds to a pressure decrease and a volume increase. It is known as the deagration branch. For our H2 -air mixture, we nd DCJ = 83.306 m/s. Here we are most concerned with the detonation branch. The deagration branch may be of interest, but neglected mechanisms, such as diusion, may be of more importance for this branch. In fact laminar ames in hydrogen move much slower than that predicted by the CJ deagration speed. We also note that for q 0, that
12 13
David Leonard Chapman, 1869-1958, English chemist. Jacques Charles Emile Jouguet, 1871-1943, French engineer. CC BY-NC-ND. 12 December 2011, J. M. Powers.
348
2 2 DCJ /(Po vo ) = (1 + 2 )/(1 2 ) = . Thus for q 0, we have DCJ Po vo , and the wave speed is the ambient sound speed. 2 Taylor series expansion of the detonation branch in the strong shock limit, Po vo /DCJ 0 shows that 2 DCJ 2q( 2 1), vo , vCJ +1 q PCJ 2( 1) , vo 2q( 2 1), uCJ +1 2q( 2 1) uCJ . +1
(9.302) (9.303) (9.304) (9.305) (9.306)
Importantly the Mach number in the wave frame at the CJ state is u2 CJ 2 MCJ = 2 , cCJ u2 CJ = , PCJ vCJ = = =
2 2q( 2 1) (+1)2 , 2q vo ( 1)vo +1 2 ( 2 (+1)2 ( 1) +1
(9.307) (9.308) (9.309) (9.310) , (9.311) (9.312)
1)
2 ( (+1)2
= 1.
( 1) +1
+ 1)( 1)
In the strong shock limit, the local Mach number in the wave frame is sonic at the end of the reaction zone. This can be shown to hold away from the strong shock limit as well. 9.2.5.2 Weak and strong solutions
For D < DCJ there are no real solutions. For D > DCJ , there are two real solutions. These are known as the weak and strong solution. These solutions are represent the intersection of the Rayleigh line with the complete reaction Hugoniot at two points. The higher pressure solution is known as the strong solution. The lower pressure solution is known as the weak solution. Equilibrium end state analysis cannot determine which of the solutions, strong or weak, is preferred if D > DCJ .

D ms 6000 5000 4000 3000 2000
DCJ
349
1000 0 1000 2000 3000 4000 uf m s
Figure 9.3: Plot of detonation wave speed D versus piston velocity uf ; parameters from Table 9.1 loosely model H2 -air detonation. We can understand much about detonations, weak, strong, and CJ, by considering how they behave as the nal velocity in the laboratory frame is changed. We can think of the nal velocity in the laboratory frame as that of a piston which is pushing the detonation. While we could analyze this on the basis of the theory we have already developed, the algebra is complicated. Let us instead return to a more primitive form. Consider the RankineHugoniot jump equations, Eqs. (9.220-9.222) with caloric state equation, Eq. (9.249) with the nal state, denoted by the subscript f , being = 1 and the initial state being = 0, Eq. (9.220) being used to simplify Eq. (9.222), and the laboratory frame velocity u used in place of u: (uf D) = o D, Pf + (uf D)2 = Po + o D 2 , 1 Pf 1 Pf 1 Po 1 2 Po + D + . q + (uf D)2 + = 1 f 2 f 1 o 2 o
ef =eo
(9.313) (9.314) (9.315)
Let us consider the unknowns to be Pf , f , and D. Computer algebra solution of these three equations yields two sets of solutions. The relevant physical branch has a solution for D of + 1 1 q + + D = uf 4 2 u2 f Po + o u2 f +1 1 q + 4 2 u2 f
2
We give a plot of D as a function of the supporting piston velocity uf in Figure 9.3. We notice on Fig. 9.3 that there is a clear minimum D. This value of D is the CJ value of DCJ = 1991.1 m/s. It corresponds to a piston velocity of uf = 794.9 m/s. A piston driving at uf = 794.9 m/s will just drive the wave at the CJ speed. At this piston speed, all of the energy to drive the wave is supplied by the combustion process itself. As uf increases beyond 794.9 m/s, the wave speed D increases. For such piston velocities,
(9.316)
350
the piston itself is supplying energy to drive the wave. For uf < 794.9 m/s, our formula predicts an increase in D, but that is not what is observed in experiment. Instead, a wave propagating at DCJ is observed. We last note that in the inert, q = 0, small piston velocity, uf 0, limit that Eq. (9.316) reduces to D = Po /o . That is, the wave speed is the ambient sound speed. 9.2.5.3 Summary of solution properties
Here is a summary of the properties of solutions for various values of D that can be obtained by equilibrium end state analysis: D < DCJ : No Rayleigh line intersects a complete reaction Hugoniot in real space. There is no real equilibrium solution. D = DCJ : There is a two repeated solutions at a single point, which we will call C, the Chapman-Jouguet point. At C, the Rayleigh line is tangent to the complete reaction Hugoniot. Some properties of this solution are uCJ /cCJ = 1; the ow is sonic in the wave frame at complete reaction. This is the unique speed of propagation of a wave without piston support if the reaction is one-step irreversible. At this wave speed, D = DCJ , all the energy from the reaction is just sucient to drive the wave forward. Because the end of the reaction zone is sonic, downstream acoustic disturbances cannot overtake the reaction zone. D > DCJ ; Two solutions are admitted at the equilibrium end state, the strong solution at point S, and the weak solution at point W . The strong solution S has u/c < 1, subsonic. piston support is required to drive the wave forward some energy to drive the wave comes from the reaction, some comes from the piston. if the piston support is withdrawn, acoustic disturbances will overtake and weaken the wave. The weak solution W has u/c > 1, supersonic often thought to be non-physical, at least for one-step irreversible kinetics because of no initiation mechanism. exceptions exist.
351
9.2.6
ZND solutions: One-step irreversible kinetics
We next consider the structure of the reaction zone. Structure was rst considered contemporaneously and independently by Zeldovich, von Neumann14 and Dring15 Equation (9.279) o gives v() for either the strong or weak branches of the solution. Knowing v() and thus (), since v = 1/, we can use the integrated mass equation, Eq. (9.220) to write an explicit equation for u(). Our Rankine-Hugoniot analysis also give T (). These can be employed in the reaction kinetics equation, Eq. (9.223, 9.13) to form a single ordinary dierential equation for the evolution of of the form a(T ()) exp RTE() (1 ) d = , d x u() (0) = 0. (9.317)
We consider the initial unshocked state to be labeled O. We label the point after the shock N for the Neumann point, named after von Neumann, one of the pioneers of detonation theory. Recall = 0 both at O and at N. But the state variables, e.g. P , , u, change from O to N. Before we actually solve the dierential equations, we can learn much by considering how P varies with in the reaction zone by using Rankine-Hugoniot analysis. Consider the Rankine-Hugoniot equations, Eqs. (9.220-9.222) with caloric state equation, Eq. (9.249), Eq. (9.220) being used to simplify Eq. (9.222): = o D, u P + 2 = Po + o D 2 , u P 1 Po 1 2 Po 1 P 1 q + u2 + = + D + . 1 2 1 o 2 o
e =eo
(9.318) (9.319) (9.320)
In Fig. (9.4), we plot P versus for three dierent values of D: D = 2800 m/s > DCJ , D = 1991.1 m/s = DCJ , D = 1800 m/s < DCJ . We can also, via detailed algebraic analysis get an algebraic expression for M as a function of . We omit that here, but do get a plot for our system. In Fig. (9.5), we plot M versus for three dierent values of D: D = 2800 m/s > DCJ , D = 1991.1 m/s = DCJ , D = 1800 m/s < DCJ . The plot of M versus is log-log so that the sonic condition may be clearly exhibited.
14 15
Computer algebra reveals the solution for P () to be 1 Po P () = Po + o D 2 1 1 +1 o D 2
2( 2
1)q . D2
(9.321)
John von Neumann, 1903-1957, Hungarian-American genius. Werner Dring, 1911-2006, German physicist. o CC BY-NC-ND. 12 December 2011, J. M. Powers.
352
P Pa 5. 106 4. 106
N
DCJ S
3. 106 N
N D D DCJ DCJ
2. 106 1. 106
O
C W
DCJ
0.2
0.4
0.6
0.8
1.0
Figure 9.4: P versus from Rankine-Hugoniot analysis for one-step irreversible reaction for D = 2800 m/s > DCJ , D = 1991.1 m/s = DCJ , D = 1800 m/s < DCJ for H2 /air-based parameters of Table 9.1.
M 5.0 3.0
D DCJ D DCJ O O O W
DCJ
2.0 1.5 1.0

D N DCJ D DCJ C
0.01
0.02
0.05
0.10
0.20
0.50
1.00
Figure 9.5: M versus from Rankine-Hugoniot analysis for one-step irreversible reaction for D = 2800 m/s > DCJ , D = 1991.1 m/s = DCJ , D = 1800 m/s < DCJ for H2 /air-based parameters of Table 9.1.
353
For D = 2800 m/s > DCJ , there are two branches, the strong and the weak. The strong branch commences at N where = 0 and proceeds to decrease to = 1 where the equilibrium point S is encountered at a subsonic state. There other branch commences at O and pressure increases until the supersonic W is reached at = 1. For D = 1991.1 m/s = DCJ , the behavior is similar, except that the branches commencing at N and O both reach complete reaction at the same point C. The point C can be shown to be sonic with M = 1. We recall from our earlier discussion regarding Eqs. (9.216-9.219) that sonic points are admitted only if r = 0. For one-step irreversible reaction, r = 0 when = 1, so a sonic condition is admissible. For D = 1800 m/s < DCJ , the strong and weak branches merge at a point of incomplete reaction. At the point of merger, near = 0.8, the ow is locally sonic; however, this is not a point of complete reaction, so there can be no real-valued detonation structure for this value of D. Finally, we write an alternate dierential-algebraic equations which can be integrated for the detonation structure: = o D, u P + 2 = Po + o D 2 , u 1 2 P 1 Po e+ u + = eo + D 2 + , 2 2 o 1 P q, e = 1 P = RT, d 1 E = a exp d x u RT (9.322) (9.323) (9.324) (9.325) (9.326) . (9.327)
We need the condition (0) = 0. These form six equations for the six unknowns , u, P , e, T , and . 9.2.6.1 CJ ZND structures
We now x D = DCJ = 1991.1 m/s and integrate Eq. (9.317) from the shocked state N to a complete reaction point, C, the Chapman-Jouguet detonation state. We could also integrate from O to C, but this is not observed in nature. After obtaining (), we can use x our Rankine-Hugoniot analysis results to plot all state variables as functions of x. In Fig. (9.6), we plot the density versus x. The density rst jumps discontinuously from O to its shocked value at N. From there it slowly drops through the reaction zone until it relaxes near x 0.01 m to its equilibrium value at complete reaction. Thus the reaction zone has a thickness of roughly 1 cm. Similar behavior is seen for the pressure in Fig. 9.7. The wave frame velocity is shown in Fig. 9.8. Since the unshocked uid is at rest in the laboratory frame with u = 0 m/s, the uid in the wave frame has velocity u = 0 1991.1 m/s = 1991.1 m/s. It is shocked to a lower velocity and then relaxes to its equilibrium value.
354
kg m3 5 4 3 2 1 m x
0.015
0.010
0.005
0.000
Figure 9.6: ZND structure of () for D = DCJ = 1991.1 m/s for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1.
P Pa 2.5 106 2. 106 1.5 106 1. 106 500 000 0.010 0.005 0.000 m x
Figure 9.7: ZND structure of P () for D = DCJ = 1991.1 m/s for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1.
ms u 0.010 500 0.005 0.000 m x
1000
1500
2000
Figure 9.8: ZND structure of wave frame-based uid particle velocity u() for D = DCJ = x 1991.1 m/s for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.

u ms 1500
355
1000
500
0 0.010 0.005 0.000 m x
Figure 9.9: ZND structure of laboratory frame-based uid particle velocity u() for D = x DCJ = 1991.1 m/s for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
T K 2500 2000 1500 1000 500 m x
0.010
0.005
0.000
Figure 9.10: ZND structure of T () for D = DCJ = 1991.1 m/s for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1. The structure of the velocity proles is easier to understand in the laboratory frame, as shown in Fig. 9.9. Here we see the unshocked uid velocity of 0 m/s. The uid is shocked to a high velocity, which then decreases to a value at the end of the reaction zone. The nal velocity can be associated with that of a supporting piston, uf = 794.9 m/s. The temperature is plotted in Fig. 9.10. It is shocked from 298 K to a high value, then continues to mainly increase through the reaction zone. Near the end of the reaction zone, there is a nal decrease as it reaches its equilibrium value. The Mach number calculated in the wave frame, M is shown in Fig. 9.11. It goes from an initial value of M = 4.88887, which we call the CJ detonation Mach number, MCJ = 4.88887, to a post-shock value of M = 0.41687. Note this result conrms a standard result from compressible ow that a standing normal shock must bring a ow from a supersonic state to a subsonic state. At equilibrium it relaxes to M = 1. This relaxation to a sonic state when = 1 is what denes the CJ state. We recall that this result is similar to that obtained in so-called Rayleigh ow of one-dimensional gas dynamics. Rayleigh ow admits heat transfer to a one-dimensional channel, and it is well known that the addition of heat always induces the ow to move to a sonic (or choked) state. So we can think of the CJ detonation
356

M 5 4 3 2 1 m x
0.015
0.010
0.005
0.000
x Figure 9.11: ZND structure of wave frame-based Mach number M () for D = DCJ = 1991.1 m/s for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
1.0 0.8 0.6 0.4 0.2 0.0 0.015 0.010 0.005 0.000 m x
Figure 9.12: ZND structure of () for D = DCJ = 1991.1 m/s for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1. wave as a thermally choked ow. The reaction progress variable is plotted in Fig. 9.12. Note that it undergoes no shock jump and simply relaxes to its equilibrium value of = 1 near x = 0.01 m. Lastly, we plot T () on a log-log scale in Fig. (9.13). The sign of x is reversed so as to x avoid the plotting of the logarithms of negative numbers. We notice on this scale that the temperature is roughly that of the shock until = 0.001 m, at which point a steep rise x begins. We call this length the induction length, ind . When we compare this gure to Fig. 2 of Powers and Paolucci, 2005, we see a somewhat similar behavior. However the detailed kinetics model shows ind 0.0001 m. The overall reaction zone length rxn is predicted well by the simple model. Its value of rxn 0.01 m is also predicted by the detailed model. Some of the nal values at the end state are dierent as well. This could be due to a variety of factors, especially dierences in the state equations. Comparisons between values predicted by the detailed model of Powers and Paolucci, 2005, against those of the simple model here are given in Table 9.2.
357
T K 2600. 2400.
2200.
2000.
1800. m x 0.001 0.01
10 5
10 4
Figure 9.13: ZND structure on a log-log scale of T () for D = DCJ = 1991.1 m/s for x one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
parameter rxn ind DCJ Ps PCJ Ts TCJ s CJ Mo Ms MCJ
simple detailed 2 10 m 102 m 103 m 104 m 1991.1 m/s 1979.7 m/s 6 2.80849 10 P a 2.8323 106 P a 1.4553 106 P a 1.6483 106 P a 1664.4 K 1542.7 K 2570.86 K 2982.1 K 4.244 kg/m3 4.618 kg/m3 1.424 kg/m3 1.5882 kg/m3 4.88887 4.8594 0.41687 0.40779 1 0.93823
Table 9.2: Numerical values of parameters which roughly model CJ H2 -air detonation.
358

kg m3 5 4 3 2 1 m x
0.0002
0.0001
Figure 9.14: ZND structure of () for strong D = 2800 m/s > DCJ for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1.
P Pa 5. 106 4. 106 3. 106 2. 106 1. 106 0.0001 0 m x
Figure 9.15: ZND structure of P () for strong D = 2800 m/s > DCJ for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1. 9.2.6.2 Strong ZND structures
We increase the detonation velocity D to D > DCJ and obtain strong detonation structures. These have a path from O to N to the equilibrium point S. These detonations require piston support to propagate, as the energy supplied by heat release alone is insucient to maintain their steady speed. Similar to our plots of the CJ structures, we give plots of the strong, D = 2800 m/s structures of (), P (), u(), u(), T (), M(), () in Figs. 9.14-9.20, respectively. x x x x x x x The behavior of the plots is qualitatively similar to that for CJ detonations. We see however that the reaction zone has become signicantly thinner, rxn 0.0001 m. This is because the higher temperatures associated with the stronger shock induce faster reactions, thus thinning the reaction zone. Comparison with Fig. 9.2 reveals that the shocked and nal values of pressure agree with those of the Rankine-Hugoniot jump analysis. We also note the nal value of M is subsonic. This allows information to propagate from the supporting piston all the way to the shock front.
359
ms u 0.0001 500 1000 1500 2000 2500 0 m x
Figure 9.16: ZND structure of wave frame uid particle velocity u() for strong D = x 2800 m/s > DCJ for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
u ms 2000 1500 1000 500 0 0.0001 0 m x
Figure 9.17: ZND structure of laboratory frame uid particle velocity u() for strong x D = 2800 m/s > DCJ for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
T K 4000
3000
2000
1000
0.0001
m x
Figure 9.18: ZND structure of T () for strong D = 2800 m/s > DCJ for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1.
360
M 7 6 5 4 3 2 1 0.0002 0.0001 0 m x
x Figure 9.19: ZND structure of wave frame Mach number M () for strong D = 2800 m/s > DCJ for one-step irreversible reaction for H2 /air-based parameters of Table 9.1.
1.0 0.8 0.6 0.4 0.2 0.0 0.0002 0.0001 0 m x
Figure 9.20: ZND structure of () for strong D = 2800 m/s > DCJ for one-step irreversible x reaction for H2 /air-based parameters of Table 9.1.

0.08
361
0.06
0.04
0.02
3.2249 1021
1.61245 1021
m x
Figure 9.21: ZND structure of () for unshocked, weak D = 2800 m/s > DCJ for one-step x irreversible reaction for H2 /air-based parameters of Table 9.1. The galactic distance scales are far too large to be realistic representations of reality! 9.2.6.3 Weak ZND structures
For the simple one-step irreversible kinetics model, there is no path from O through the shocked state N to the weak solution W . There is a direct path from O to W ; however, it is physically unrealistic. For D = 2800 m/s, we plot versus x in Fig. 9.21. Numerical solution was available only until 0.02. Numerical precision issues arose at this point. Note, importantly, that rxn 1021 m is unrealistically large! Note the distance from Earth to the Large Magellanic Cloud, a dwarf galaxy orbiting the Milky Way, is 1021 m. Our combustion model is not well calibrated to those distances, so it is entirely unreliable to predict this class of weak detonation! 9.2.6.4 Piston problem
We can understand the physics of the one-step kinetics problem better in the context of a piston problem, where the supporting piston connects to the nal laboratory frame uid velocity up = u( ). Let us consider pistons with high velocity and then lower them x and examine the changes of structure. up > up,CJ . This high velocity piston will drive a strong shock into the uid at a speed D > DCJ . The solution will proceed from O to N to S and be subsonic throughout. Therefore changes at the piston face will be able to be communicated all the way to the shock front. The energy to drive the wave comes from a combination of energy released during combustion and energy supplied by the piston support. up = up,CJ At a critical value of piston velocity, up,CJ , the solution will go from O to
362
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION N to C, and where it is locally sonic.
up < up,CJ For such ows, the detonation wave is self-supporting. There is no means to communicate with the supporting piston. The detonation wave proceeds at D = DCJ . We note that DCJ is not a function of the specic kinetic mechanism. So for a one-step irreversible kinetics model, the conclusion the DCJ is the unique speed of propagation of an unsupported wave is veried. It should be noted that nearly any complication added to the model, e.g. reversibility, multi-step kinetics, multi-dimensionality, diusion, etc., will alter this conclusion.
9.2.7
Detonation structure: Two-step irreversible kinetics
Let us consider a small change to the one-step model of the previous sections. We will now consider a two-step irreversible kinetics model. The rst reaction will be exothermic and the second endothermic. Both reactions will be driven to completion, and when they are complete, the global heat release will be identical to that of the one-step reaction. All other parameters will remain the same from the one-step model. This model is discussed in detail by Fickett and Davis, and in a two-dimensional extension by Powers and Gonthier.16 We will see that this simple modication has profound eects on what is a preferred detonation structure. In particular, we will see that for such a two-step model the CJ structure is no longer the preferred state of an unsupported detonation wave, the steady speed of the unsupported detonation wave is unique and greater than the CJ speed, there is a path from the unshocked state O to the shocked state N through a sonic incomplete reaction pathological point P to the weak equilibrium end state W , there is a strong analog to steady compressible one-dimensional inert ow with area change, i.e. rocket nozzle ow. Let us pose the two step irreversible kinetics model of 1 : A B, 2 : B C. (9.328) (9.329)
Let us insist that A, B, and C each have the same molecular mass, MA = MB = MC = M, and the same constant specic heats, cP A = cP B = cP C = cP . Let us also insist that both reactions have the same kinetic parameters, E1 = E2 = E, a1 = a2 = a, 1 = 2 = 0. Therefore the reaction rates are such that k1 = k2 = k.
16
(9.330)
Powers, J. M., and Gonthier, K. A., 1992, Reaction Zone Structure for Strong, Weak Overdriven, and Weak Underdriven Oblique Detonations, Physics of Fluids A, 4(9): 2082-2089. CC BY-NC-ND. 12 December 2011, J. M. Powers.
363
Let us assume at the initial state that we have all A and no B or C: A (0) = A , B (0) = 0, C (0) = 0. Since J = 2, we have a reaction vector rj of length 2: rj = r1 r2 . (9.331)
Our stoichiometric matrix ij has dimension 3 2 since N = 3 and J = 2: 1 0 ij = 1 1 . 0 1
(9.332)
Elementary row operations gives us the row echelon form YA 1 0 d M YA + YB = 0 1 dt YA + YB + YC 0 0 YA + YB + YC = 1. This can be thought of as an unusual matrix equation: YA ( 1 1 1 ) YB = ( 1 ) . YC
We recall here that d/dt denotes the material derivative following a uid particle, d/dt = /t + u/ x. For our steady waves, we will have d/dt = ud/d. We next recall that x i / = Yi /Mi, which for us is Yi/M, since the molecular masses are constant. So we have YA A 1 0 r1 M M M d r1 YB = B = 1 1 = r1 r2 . (9.334) r2 dt YC C 0 1 r2 r1 r2
For species production rates, from i = J ij rj we have j=1 A / A 1 0 r1 1 d 1 1 r1 B / = B = 1 1 = r1 r2 . r2 dt C / C 0 1 r2
(9.333)
(9.335)
We can integrate the homogeneous third equation and apply the initial condition to get (9.336)
(9.337)
We can perform an analogous exercise to nding the form = + D and get YA 1 1 0 YB = 0 + 1 1 1 . 2 YC 0 0 1

=F
(9.338)
364
The column vectors of F are linearly independent and lie in the right null space of the coecient matrix (1, 1, 1). The choices for F are not unique, but are convenient. We can think of the independent variables 1 , 2 as reaction progress variables. Thus, for reaction 1, we have 1 , and for reaction 2, we have 2 . Both 1 (0) = 0 and 2 (0) = 0. The mass fraction of each species can be related to the reaction progress via Y A = 1 1 , Y B = 1 2 , Y C = 2 . (9.339) (9.340) (9.341)
When the reaction is complete, we have 1 1, 2 1, and YA 0, YB 0, YC 1. Now our reaction law is YA r1 kA YA d M M YB = r1 r2 = kA kB = k YA YB . (9.342) dt YC r2 kB YB Eliminating YA , YB and YC in favor of 1 and 2 , we get 1 1 (1 1 ) d 1 2 = k (1 1 ) (1 2 ) . dt 2 1 2 This reduces to 1 1 1 d 1 2 = k 1 21 + 2 . dt 2 1 2
(9.343)
(9.344)
The second of these equations is the dierence of the rst and the third, so it is redundant and we need only consider d dt 1 2 =k 1 1 1 2 . (9.345)
In the steady wave frame, this is written as d1 = (1 1 )k, d x d2 u = (1 2 )k. d x u (9.346) (9.347)
Because the rates k1 = k2 = k have been taken identical, we can actually get 2 (1 ). Dividing our two kinetic equations gives 1 2 d2 = . d1 1 1
(9.348)
365
Because 1 (0) = 0 and 2 (0) = 0, we can say that 2 (1 = 0) = 0. We rearrange this dierential equation to get 1 1 d2 + 2 = . d1 1 1 1 1 This equation is rst order and linear. It has an integrating factor of exp d1 1 1 = exp ( ln(1 1 )) = 1 . 1 1 (9.349)
Multiplying both sides by the integrating factor, we get 1 d2 + 1 1 d1 Using the product rule, we then get d d1 2 1 = , 1 1 (1 1 )2 1 2 = d1 . 1 1 (1 1 )2 (9.351) (9.352) 1 1 1
2
2 =
1 . (1 1 )2
(9.350)
Taking u = 1 and dv = d1 /(1 1 )2 and integrating the right side by parts, we get 1 d1 2 = , 1 1 1 1 1 1 1 = + ln(1 1 ) + C, 1 1 2 = 1 + (1 1 ) ln(1 1 ) + C(1 1 ). Now since 2 (1 = 0) = 0, we get C = 0, so 2 () = 1 () + (1 1 ()) ln(1 1 ()). x x x x Leaving out details of the derivation, our state equation becomes e(T, 1 , 2 ) = cv (T To ) 1 q1 2 q2 . We nd it convenient to dene Q(1 , 2 ) as Q(1 , 2 ) 1 q1 + 2 q2 . So the equation of state can be written as e(T, 1 , 2 ) = cv (T To ) Q(1 , 2 ). (9.359) (9.358) (9.357) (9.356) (9.353) (9.354) (9.355)
366
CHAPTER 9. SIMPLE DETONATIONS: REACTION-ADVECTION Parameter M R Po To o vo q1 q2 E a Value 1.4 20.91 397.58 1.01325 105 298 0.85521 1.1693 7.58265 106 5.68698 106 8.29352 106 5 109 0 Units kg/kmole J/kg/K Pa K kg/m3 m3 /kg J/kg J/kg J/kg 1/s
Table 9.3: Numerical values of parameters for two-step irreversible kinetics. The frozen sound speed remains c2 = P v = There are now two thermicities: 1 = 2 = 1 P c2 1 1 P c2 2 =
v,e,2
P .
(9.360)
q1 , 1 P q2 . 1 P
(9.361) (9.362)
=
v,e,1
Parameters for our two step model are identical to those of our one step model, except for the heat releases. The parameters are listed in Table 9.3. Note that at complete reaction Q(1 , 2 ) = Q(1, 1) = q1 + q2 = 1.89566 106 J/kg. Thus, the overall heat release at complete reaction 1 = 2 = 1 is identical to our earlier one-step kinetic model. Let us do some new Rankine-Hugoniot analysis. We can write a set of mass, momentum, energy, and state equations as = o D, u P + 2 = Po + o D 2 , u 1 2 P 1 Po e+ u + = eo + D 2 + , 2 2 o 1 P 1 q1 2 q2 , e = 1 2 = 1 + (1 1 ) ln(1 1 ).
(9.363) (9.364) (9.365) (9.366) (9.367)

P Pa
N
367
5. 106 N
D D
4. 106
N D D D P D D D D D D D
3. 106
D D
2. 106
1. 106
O
0.2
0.4
0.6
0.8
1.0
Figure 9.22: Pressure versus 1 for two-step kinetics problem for three dierent values of D: D = 2800 m/s > D, D = 2616.5 m/s = D, D = 2500 m/s < D; parameters are from Table 9.3. Let us consider D and 1 to be unspecied but known parameters for this analysis. These equations are ve equations for the ve unknowns, , u, P , e, and 2 . They can be solved for (D, 1 ), u(D, 1 ), P (D, 1), e(D, 1 ), and 2 (1 ). The solution is lengthy, but the plot is revealing. For three dierent values of D, pressure as a function of 1 is shown in in Fig. 9.22. There are three important classes of D, each shown in Fig. 9.22, depending on how D compares to a critical value we call D. D > D. There are two potentially paths here. The important physical branch starts at point O, and is immediately shocked to state N, the Neumann point. From N the pressure rst decreases as 1 increases. Near 1 = 0.75, the pressure reaches a local minimum, and then increases to the complete reaction point at S. This is a strong solution. There is a second branch which commences at O and is unshocked. On this branch the pressure increases to a maximum, then decreases to the end state at W . While this branch is admissible mathematically, its length scales are unphysically long, and this branch is discarded. D = D. Let us only consider branches which are shocked from O to N. The unshocked branches are again non-physical. On this branch, the pressure decreases from N to the pathological point P . At P , the ow is locally sonic, with M = 1. Here the pressure can take two distinct paths. The one chosen will depend on the velocity of the supporting piston at the end state. On one path the pressure increases to its nal value at the strong point S. On the other the pressure decreases to its nal value at the weak point W .
368
M
O
5.0 O 3.0
D D W D D D D P
2.0 1.5 1.0

D D D D N D D
0.01
0.02
0.05
0.10
0.20
0.50
1.00 1
Figure 9.23: Mach number M versus 1 for two-step kinetics problem for three dierent values of D: D = 2800 m/s > D, D = 2616.5 m/s = D, D = 2500 m/s < D; parameters are from Table 9.3. D < D. For such values of D, there is no physical structure for the entire reaction zone 0 < 1 < 1. This branch is discarded. Mach number in the wave frame, M as a function of 1 is shown in in Fig. 9.23. The results here are similar to those in Fig. 9.22. Note the ambient point O is always supersonic, and the Neumann point N is always subsonic. For ows originating at N, if D > D, the ow the ow can undergo remains subsonic throughout until its termination at S. For D = D, a subsonic to supersonic transition at the pathological point P . The weak point W is a supersonic end state. The important Fig. 9.22 bears remarkable similarity to curves of P (x) in compressible inert ow in a converging-diverging nozzle. We recall that for such ows, a subsonic to supersonic transition is only realized at an area minimum. This can be explained because the equation for evolution of pressure for such ows takes the form dP/dx (dA/dx)/(1 M 2 ). So if the ow is locally sonic, it must encounter a critical point in area, dA/dx = 0 in order to avoid innite pressure gradients. This is what is realized in actual nozzles. For us the analogous equation is the two-step version of Eq. (9.218), which can be shown to be dP (1 r1 + 2 r2 ) u = . (9.368) d x 1 M2
Because the rst reaction is exothermic, we have 1 > 0, and because the second reaction is endothermic, we have 2 < 0. With r1 > 0, r2 > 0, this gives rise to the possibility that
369
1 r1 + 2 r2 = 0 at a point where the reaction is incomplete. The point P is just such a point; it is realized when D = D. Finally, we write our dierential-algebraic equations which are integrated for the detonation structure: = o D, u P + 2 = Po + o D 2 , u 1 P 1 Po e + u2 + = eo + D 2 + , 2 2 o 1 P e = 1 q1 2 q2 , 1 P = RT, 2 = 1 + (1 1 ) ln(1 1 ), 1 1 E d1 . = a exp d x u RT (9.369) (9.370) (9.371) (9.372) (9.373) (9.374) (9.375)
We need the condition 1 (0) = 0. These form seven equations for the seven unknowns , u, P , e, T , 1 , and 2 . We also realize that the algebraic solutions are multi-valued and must take special care to be on the proper branch. This becomes particularly important for solutions which pass through P . 9.2.7.1 Strong structures
Here we consider strong structures for two cases: D > D and D = D. All of these will proceed from O to N through a pressure minimum, and nish at the strong point S. 9.2.7.1.1 D > D Structures for a strong detonation with D = 2800 m/s > D are given in Figs. 9.24-9.31. The structure of all of these can be compared directly to those of the onestep kinetics model at the same D = 2800 m/s, Figs. 9.14-9.20. Note the shock values are identical. The reaction zone thicknesses are quite similar as well at rxn 0.0001 m. The structures themselves have some dierences; most notably, the two-step model structures display interior critical points before complete reaction. We take special note of the pressure plot of Fig. 9.25, which can be compared with Fig. 9.22. We see in both gures the shock from O to N, followed by a drop of pressure to a minimum, followed by a nal relaxation to an equilibrium value at S. Note that the two curves have the opposite sense of direction as 1 commences at 0 and goes to 1, while x commences at 0 and goes to 0.0002 m. x We can also compare the M() plot of Fig. 9.29 with that of M (1 ) of Fig. 9.23. In both the supersonic O is shocked to a subsonic N. The Mach number rises slightly then falls in the reaction zone to its equilibrium value at S. It never returns to a supersonic state.
370
kg m3 5 4 3 2 1 m x
0.0002
0.0001
Figure 9.24: ZND structure of () for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
P Pa 5. 106 4. 106 3. 106 2. 106 1. 106 0.0001 0 m x
Figure 9.25: ZND structure of P () for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
ms u 0.0001 500 1000 1500 2000 2500 0 m x
Figure 9.26: ZND structure of u() for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
371
u ms 2000 1500 1000 500 0 0.0001 0 m x
Figure 9.27: ZND structure of u() for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
T K 5000 4000 3000 2000 1000 m x
0.0001
Figure 9.28: ZND structure of T () for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
M 7 6 5 4 3 2 1 0.0002 0.0001 0 m x
x Figure 9.29: ZND structure of M() for strong D = 2800 m/s > D for two-step irreversible reaction with parameters of Table 9.3.
372
1 1.0 0.8 0.6 0.4 0.2 0.0 0.0002 0.0001 0 m x
Figure 9.30: ZND structure of 1 () for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.
2 1.0 0.8 0.6 0.4 0.2 0.0 0.0002 0.0001 0 m x
Figure 9.31: ZND structure of 2 () for strong D = 2800 m/s > D for two-step irreversible x reaction with parameters of Table 9.3.

P Pa 5. 106 4. 106 3. 106 2. 106 1. 106 m x
373
0.0001
Figure 9.32: ZND structure of P () for strong D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3.
M 7 6 5 4 3 2 1 0.0002 0.0001 0 m x
x Figure 9.33: ZND structure of M () for strong D = 2616.5 m/s = D for two-step irreversible reaction with parameters of Table 9.3. 9.2.7.1.2 D = D For D = D = 2616.5 m/s, we can nd a strong structure with a path from O to N to S. Pressure P and Mach number M are plotted in Figures 9.32-9.33. Note that at an interior point in the structure, a cusp in the P and M prole is seen. At this = 1. point, the ow is locally sonic with M 9.2.7.2 Weak, eigenvalue structures
Let us now consider weak structures with D = D = 2616.5 m/s. Special care must be taken in integrating the governing equations. In general one must integrate to very near the pathological point P , then halt. While there are more sophisticated techniques involving further coordinate transformations, one can record the values near P on its approach from N. Then one can perturb slightly all state variables so that the M is just greater than unity and recommence the integration. Figures of the structures which commence at O, are shocked to N, pass through sonic point P , and nish at the supersonic W are shown in Figs. 9.34-9.41. One may again compare the pressure and Mach number plots of Figs. 9.35,9.39 with those of Figs. 9.22,9.23.
374

kg m3 5 4 3 2 1 m x
0.000292911
Figure 9.34: ZND structure of () for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3.
P Pa 5. 106 4. 106 3. 106 2. 106 1. 106 m x
0.000585822
0.000292911
Figure 9.35: ZND structure of P () for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3. 9.2.7.3 Piston problem
We can understand the physics of the two-step kinetics problem better in the context of a piston problem, where the supporting piston connects to the nal laboratory frame uid velocity up = u( ). Let us consider pistons with high velocity and then lower them x and examine the changes of structure. up > ups . This high velocity piston will drive a strong shock into the uid at a speed D > D. The solution will proceed from O to N to S and be subsonic throughout. Therefore changes at the piston face will be able to be communicated all the way to the shock front. The energy to drive the wave comes from a combination of energy released during combustion and energy supplied by the piston support. up = ups . At a critical value of piston velocity, ups , the solution will go from O to N to P to S, and be locally sonic. This is analogous to the subsonic design condition for a converging-diverging nozzle. up [ps , upw ]. Here the ow can be complicated. Analogous to ow in a nozzle, there u

ms u 0.000585822 0.000292911 500 1000 1500 2000 2500 0 m x
375
Figure 9.36: ZND structure of u() for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3.
u ms 2000 1500 1000 500 0 0.000585822 0.000292911 0 m x
Figure 9.37: ZND structure of u() for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3. can be standing shock waves in the supersonic portion of the ow which decelerate the ow so as to match the piston velocity at the end of the reaction. Such ows will proceed from O to N through P , and then are shocked back onto the subsonic branch to terminate at S. up = upw . This state is analogous to the supersonic design condition of ow in a converging-diverging nozzle. The uid proceeds from O to N through P and terminates at W . All of the energy to propagate the wave comes from the reaction. up < upw . For such ows, the detonation wave is self-supporting. There is no means to communicate with the supporting piston. The detonation wave proceeds at D = D. We note that D is a function of the specic kinetic mechanism. This non-classical result contradicts the conclusion from CJ theory with simpler kinetics in which the wave speed of an unsupported detonation is independent of the kinetics. Note for our problem that D = 2616.5 m/s. This stands in contrast to the CJ velocity, independently computed of DCJ = 1991.1 m/s for the same mixture.
376

T K 4000 3000 2000 1000 m x
0.000585822
0.000292911
Figure 9.38: ZND structure of T () for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3.
M 7 6 5 4 3 2 1 0.000585822 0.000292911 0 m x
x Figure 9.39: ZND structure of M () for weak D = 2616.5 m/s = D for two-step irreversible reaction with parameters of Table 9.3.
9.2.8
Detonation structure: Detailed H2 O2 N2 kinetics
These same notions for detonation with simple kinetics and state equations can easily be extended to more complex models. Let us consider a one-dimensional steady detonation in a stoichiometric hydrogen air mixture with the detailed kinetics model of Table 1.2. We shall consider a case almost identical to that studied by Powers and Paolucci.17 The mixture is a stoichiometric hydrogen-air mixture of 2H2 + O2 + 3.76N2 . As we did in our modeling of the same mixture under spatially homogeneous isochoric, adiabatic conditions in an earlier chapter, we will take the number of moles of each of the minor species to be a small number near machine precision. This has the eect of removing some numerical roundo errors in the very early stages of reaction. Our model will be the steady one-dimensional reactive Euler equations, obtained by considering the reactive Navier-Stokes equations in the limit as , jm , jq all go to zero. i Our model then is 1) the integrated mass, momentum, and energy equations, Eqs. (9.220Powers, J. M., and Paolucci, S., 2005, Accurate Spatial Resolution Estimates for Reactive Supersonic Flow with Detailed Chemistry, AIAA Journal, 43(5): 1088-1099. CC BY-NC-ND. 12 December 2011, J. M. Powers.
17

1 1.0 0.8 0.6 0.4 0.2 0.0 0.000585822 0.000292911 0 m x
377
Figure 9.40: ZND structure of 1 () for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3.
2 1.0 0.8 0.6 0.4 0.2 0.0 m x
Figure 9.41: ZND structure of 2 () for weak D = 2616.5 m/s = D for two-step irreversible x reaction with parameters of Table 9.3. 9.222) with an opposite sign on D to account for the left-running wave, 2) the one-dimensional steady diusion-free version of species evolution, Eq. (6.50), 3) the calorically imperfect ideal gas state equations of Eqs. (6.63), (6.71), and 4) the law of mass action with Arrhenius kinetics of Eqs. (6.76-6.79).
u P + 2 u 1 P e + u2 + u 2 dYi u d x
= o D, = Po + o D 2 , 1 Po = o D eo + D 2 + 2 o = Mi i ,
N
(9.376) (9.377) , (9.378) (9.379)
P = RT
i=1
Yi , Mi
(9.380)
378

N T
e =
i=1 J
Yi eo o ,i + T ij rj ,
j=1 N kkj
cvi (T )dT ,
(9.381) (9.382)
To
i = rj = k j
k=1
1 1 Kc,j E j RT
ij
kkj . ,
k=1
(9.383) (9.384)
kj = aj T j exp Kc,j = Po RT
PN
i=1
, Go j RT .
exp
(9.385)
In order to cleanly plot the results on a log scale, we will, in contrast to the previous right-running detonations, consider left running waves with detonation velocity D. This dierential-algebraic system must be solved numerically. One can use standard dierentialalgebraic methods to achieve this. Alternatively, and with signicant eort, one can remove all of the algebraic constraints. Part of this requires a numerical iteration to nd certain roots. After this eort, one can in principle write a set of N L dierential equations for evolution of the N L independent species. Here, we chose D DCJ = 1979.70 m/s. Because this model is not a simple one-step model, we cannot expect to nd the equilibrium state to be exactly sonic. However, it is possible to slightly overdrive the wave and achieve a nearly sonic state at the equilibrium state. Here, our nal Mach number was M = 0.9382. Had we weakened the overdrive further, we would have encountered an interior sonic point at a non-equilibrium point, thus inducing a non-physical sonic singularity. Numerical solution for species mass fraction is given in Fig. (9.42). The gure is plotted on a log-log scale because of the wide range of length scales and mass fraction scales encountered. Note that the minor species begin to change at a very small length scale. At a value of x 2.6 104 m, a signicant event occurs, known as a thermal explosion. This length is known as the induction length, ind = 2.6104 m. We get a rough estimate of the induction time by the formula tind ind /s = 7.9 107 s. Here us is the post shock velocity in the u wave frame. Its value is us = 330.54 m/s. All species contribute to the reaction dynamics here. This is followed by a relaxation to chemical equilibrium, achieved around 0.1 m. The pressure prole is given in Fig. (9.2.8). We articially located the shock just away from x = 0, so as to ease the log-log plot. The pressure is shocked from it atmospheric value to 2.83280 106 P a (see Table 9.2). After the shock, the pressure holds nearly constant for several decades of distance. Once the thermal explosion commences, the pressure relaxes to its equilibrium value. This gure can be compared with its one-step equivalent of Fig. 9.15. Similar behavior is seen in the temperature plot of Fig. 9.44. The temperature is shocked to Ts = 1542.7 K, stays constant in the induction zone, and then increases to its equilibrium value after the thermal explosion. This gure can be compared with its one-step equivalent
379
10
10 Yi
10
10
10 x (m)
Figure 9.42: Detailed kinetics ZND structure of species mass fractions for near CJ detonation, DCJ D = 1979.70 m/s, in 2H2 + O2 + 3.76N2 , Po = 1.01325 105 P a, To = 298 K, Ts = 1542.7 K,Ps = 2.83280 106 P a. of Fig. 9.18. It is very interesting to compare these results to those obtained in an earlier chapter. Earlier, an isochoric, adiabatic combustion of precisely the same stoichiometric hydrogen-air mixture with precisely the same kinetics was conducted. The adiabatic/isochoric mixture had an initial temperature and pressure identical to the post-shock pressure and temperature here. We compare the induction time of the spatially homogeneous problem, tind = 6.6 107 s, Eq. (1.362) to our estimate from the detonation found earlier, tind ind /s = u 7 7.9 10 s. The two are remarkably similar! We compare some other relevant values in Table 9.4. Note that these mixtures are identical at the onset of the calculation. The detonating mixture has reached the same initial state after the shock. And a uid particle advecting through the detonation reaction zone with undergo a thermal explosion at nearly the same time a particle that was stationary in the closed vessel will. After that the two fates are dierent. This is because there is no kinetic energy in the spatially homogeneous problem. Thus all the chemical energy is transformed into thermal energy. This is reected in the higher nal temperature and pressure of the spatially homogeneous problem relative to the detonating ow. Because the nal temperature is dierent, the two systems relax to a dierent chemical equilibrium, as reected in the dierent mass fractions. For example, note that the cooler detonating ow
OH H 2O H2 O2 H O HO 2 H 2O2 N2 10
0
380
Parameter To Ts Po Ps tind T eq P eq eq ueq eq YO2 eq YH eq YOH eq YO eq Y H2 eq Y H2 O eq YHO2 eq Y H2 O 2 eq Y N2
Spatially Homogeneous 1542.7 K 2.83280 106 P a 6.6 107 s 3382.3 K 5.53 106 P a 4.62 kg/m3 0 m/s 1.85 102 5.41 104 2.45 102 3.88 103 3.75 103 2.04 101 6.84 105 1.04 105 7.45 101
Detonation 298 K 1542.7 K 1.01325 105 P a 2.83280 106 P a 7.9 107 s 2982.1 K 1.65 106 P a 1.59 kg/m3 1066 m/s 1.38 102 2.71 104 1.48 102 1.78 103 2.57 103 2.22 101 2.23 105 3.08 106 7.45 101
Table 9.4: Comparison of relevant predictions of a spatially homogeneous model with those of a near CJ detonation in the same mixture.
381
10 P (Pa) 10
10
10
10 x (m)
10
Figure 9.43: Detailed kinetics ZND structure of pressure for near CJ detonation, DCJ D = 1979.70 m/s, in 2H2 + O2 + 3.76N2 , Po = 1.01325 105 P a, To = 298 K, Ts = 1542.7 KPs = 2.83280 106 P a. has a higher nal mass fraction of H2 O.
382
3500 3000 2500 T (K) 2000 1500 1000 500

10
10 x (m)
10
Figure 9.44: Detailed kinetics ZND structure of temperature for near CJ, DCJ D = 1979.70 m/s, detonation in 2H2 + O2 + 3.76N2 , Po = 1.01325 105 P a, To = 298 K, Ts = 1542.7 K, Ps = 2.83280 106 P a.
Chapter 10 Blast waves

Here we will study the Taylor1 -Sedov2 blast wave solution. We will follow most closely two papers of Taylor34 from 1950. Taylor notes that the rst of these was actually written in 1941, but was classied. Sedovs complementary study5 is also of interest. One may also consult other articles by Taylor for background.67 Consider a point source of energy which at t = 0 is released into a calorically perfect ideal gas. The point source could be the combustion products of an intense reaction event. We shall follow Taylors analysis and obtain what is known as self-similar solutions. Though there are more general approaches which may in fact expose more details of how self-similar solutions are obtained, we will conne ourselves to Taylors approach and use his notation. The self-similar solution will be enabled by studying the Euler equations in what is known as the strong shock limit for a spherical shock wave. Now a shock wave will raise both the internal and kinetic energy of the ambient uid into which it is propagating. We would like to consider a scenario in which the total energy, kinetic and internal, enclosed by the strong spherical shock wave is a constant. The ambient uid is initially at rest, and a point source of energy exists at r = 0. For t > 0, this point source of energy is distributed to the mechanical and thermal energy of the surrounding uid.
Georey Ingram Taylor, 1886-1975, English physicist. Leonid Ivanovitch Sedov, 1907-1999, Soviet physicist. 3 Taylor, G. I., 1950, The Formation of a Blast Wave by a Very Intense Explosion. I. Theoretical Discussion, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences 201(1065): 159-174. 4 Taylor, G. I., 1950, The Formation of a Blast Wave by a Very Intense Explosion. II. The Atomic Explosion of 1945, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 201(1065): 175-186. 5 Sedov, L. I., 1946, Rasprostraneniya Silnykh Vzryvnykh Voln, Prikladnaya Matematika i Mekhanika 10 6 Taylor, G. I., 1950, The Dynamics of the Combustion Products Behind Plane and Spherical Detonation Fronts in Explosives, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 200(1061): 235-247. 7 Taylor, G. I., 1946, The Air Wave Surrounding an Expanding Sphere, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 186(1006): 273-292.
2 1
383
384
CHAPTER 10. BLAST WAVES
Let us follow now Taylors analysis from his 1950 Part I Theoretical Discussion paper. We shall write the governing inert one-dimensional unsteady Euler equations in spherical coordinates, reduce the partial dierential equations in r and t to ordinary dierential equations in an appropriate similarity variable, solve the ordinary dierential equations numerically, and show our transformation guarantees constant total energy in the region r [0, R(t)], where R(t) is the locus of the moving shock wave. We shall also refer to specic equations in Taylors rst 1950 paper.
10.1
Governing equations
u 2u +u + = , t r r r u u 1 P +u + = 0, t r r P 2 = 0, +u t r 1 P , e = 1 P = RT.
The non-conservative formulation of the governing equations is as follows: (10.1) (10.2) (10.3) (10.4) (10.5)
e e +u t r
For review, lets look at the energy equation in a little more detail. Recall the material derivative is d/dt = /t + u/r, so the energy equation is de P d = 0. dt 2 dt (10.6)
Let us now substitute the thermal energy equation, Eq. (10.5) into the energy equation, Eq. (10.6): P d 1 d P 2 = 0, 1 dt dt 1 P d 1 1 dP P d + 2 = 0, 2 dt 1 1 dt dt P d P d 1 dP + ( 1) 2 = 0, 2 dt dt dt
(10.7) (10.8) (10.9)
10.2. SIMILARITY TRANSFORMATION P d 1 dP 2 dt dt P d dP dt dt 1 dP P d +1 dt dt d P dt P P +u r = 0, = 0, = 0, = 0, = 0.
385 (10.10) (10.11) (10.12) (10.13) (10.14)
10.2
Similarity transformation
We shall next make some non-intuitive and non-obvious choices for a transformed coordinate system and transformed dependent variables. These choices can be systematically studied with the techniques of group theory, not discussed here.
10.2.1
Independent variables
r , R(t) = t. =
So, let us transform the independent variables (r, t) (, ) with (10.15) (10.16)
We will seek solutions such that the dependent variables are functions of , the distance relative to the time-dependent shock, only. We will have little need for the transformed time since it is equivalent to the original time t.
10.2.2
Dependent variables
P = y = R3 f1 (), Po = (), o u = R3/2 1 ().
Let us also dene new dependent variables as (10.17) (10.18) (10.19)
These amount to denitions of a scaled pressure, f1 , a scaled density and a scaled velocity 1 , with the assumption that each is a function of only. Here Po , and o are constant ambient values of pressure and density, respectively.
386 We also assume the shock velocity to be of the form U= The constant A is to be determined. dR = AR3/2 . dt
(10.20)
10.2.3
Derivative transformations
= + . t t t
By the chain rule we have (10.21)
Now by Eq. (10.15) we get r dR = 2 , t R dt dR = , R(t) dt = AR3/2 , R A = 5/2 . R From Eq. (10.16) we simply get = 1. t Thus the chain rule, Eq. (10.21), can be written as A = 5/2 + . t R As we are insisting the / = 0, we get A d = 5/2 . t R d In the same way, we get = + , r r r
=0
(10.22) (10.23) (10.24) (10.25)
(10.26)
(10.27)
(10.28)
(10.29)
1 d . = R d
(10.30)
10.3. TRANSFORMED EQUATIONS
387
10.3
Transformed equations
Let us now apply our rules for derivative transformation, Eqs. (10.25,10.30), and our transformed dependent variables, Eqs. (10.17-10.19) to the governing equations.
10.3.1
Mass
First, we shall consider the mass equation, Eq. (10.1). We get 1 d 1 d 2 A d (o ) + R3/2 1 (o ) + o R3/2 1 = o R3/2 1 . (10.31) 5/2 d R R d R d r
= =/t =u = =/r = =/r =u = =u
Realizing the R(t) = R( ) is not a function of , canceling the common factor of o , and eliminating r with Eq. (10.15), we can write A d 1 d d1 2 1 + 5/2 + 5/2 = , 5/2 d R R d R d R5/2 d d1 2 d + 1 + = 1 , A d d d d d d1 2 A + 1 + + 1 d d d (10.32) (10.33) = 0, mass. (10.34)
Equation (10.34) is number 9 in Taylors paper, which we will call here Eq. T(9).
10.3.2
Linear momentum
Now consider the linear momentum equation, Eq. (10.2), and apply the same transformations: 1 R3/2 1 + R3/2 1 R3/2 1 + Po R3 f1 t r o r
=u =u =u =1/ =P
= 0, (10.35)
R3/2
1 3 5/2 dR 1 R3/2 1 + Po R3 f1 R 1 +R3/2 1 t 2 dt r o r

=u/t
= 0, (10.36)
R3/2
d1 3 5/2 R3/2 1 R AR3/2 1 + R3/2 1 d 2 r 1 + Po R3 f1 o r 1 A d1 3 A R3/2 1 + Po R3 f1 + R3/2 1 4 4 1 R d 2R r o r A R5/2
= 0, (10.37) = 0, (10.38)
388
CHAPTER 10. BLAST WAVES 1 d 1 1 d A d1 3 A R3/2 1 + Po R3 f1 = 0, (10.39) + R3/2 1 4 d 4 1 R 2R R d o R d 1 d1 Po 1 df1 A d1 3 A + 4 + = 0, (10.40) 4 4 1 R d 2R R d o R4 d d1 3 d1 Po df1 A A1 + 1 + = 0. (10.41) d 2 d o d 3 d1 1 + 2 d d1 Po 1 df1 + = 0, d o d
Our nal form is A + 1 linear momentum. (10.42)
Equation (10.42) is T(7).
10.3.3
Energy
Let us now consider the energy equation. It is best to begin with a form in which the equation of state has already been imposed. So we will start by expanding Eq. (10.11) in terms of partial derivatives: P P P +u t r
=dP/dt
+u t r
=d/dt
= 0,
(10.43)
R3
A R5/2
Po R3 f1 + R3/2 1 Po R3 f1 t r Po R3 f1 (o ) + R3/2 1 (o ) o t r R3 f1 + R3/2 1 R3 f1 t r R3 f1 + R3/2 1 t r f1 dR R3 R3 f1 3R4 f1 + R3/2 1 t dt r R3 f1 + R3/2 1 t r df1 R3 f1 3R4 (AR3/2 )f1 + R3/2 1 d r R3 f1 + R3/2 1 t r
= 0,
(10.44)
= 0,
(10.45)
= 0,
(10.46)
= 0.
(10.47)
Carrying on, we have A 1 df1 A df1 3 11/2 f1 + R3/2 1 R3 11/2 d R R R d
10.4. DIMENSIONLESS EQUATIONS A d 1 d R3 f1 5/2 + R3/2 1 R d R d A df1 A 1 df1 f1 d d 11/2 3 11/2 f1 + 11/2 A + 1 11/2 R d R R d R d d d df1 f1 d df1 A 3Af1 + 1 + 1 A d d d d Our nal form is A 3f1 + df1 d + f1 d df1 (A + 1 ) 1 = 0, d d energy.
389 = 0, = 0, = 0. (10.48) (10.49) (10.50)
(10.51)
Equation (10.51) is T(11), correcting for a typographical error replacing a r with .
10.4
Dimensionless equations
Let us now write our conservation principles in dimensionless form. We take the constant ambient sound speed co to be dened for the calorically perfect ideal gas as c2 o Po . o (10.52)
Note, we have used our notation for sound speed here; Taylor uses a instead. Let us also dene f co A 1 . A
2
f1 ,
(10.53) (10.54)
10.4.1
Mass
d d d 2 + A + A + A = 0, d d d d d 2 d + + + = 0, d d d d ( ) = d 1 d d =
With these denitions, the mass equation, Eq. (10.34) becomes A (10.55) (10.56)
d d
d 2 + , d
2
(10.57) (10.58)
mass.
Equation (10.58) is T(9a).

390
10.4.2
Linear momentum
d Po 1 A2 df d 3 + A2 A + A + 2 d d o c2 d o 3 d 1 1 df d + + + 2 d d d 3 1 df d ( ) + d 2 d d ( ) d
With the same denitions, the momentum equation, Eq. (10.42) becomes A = 0, = 0, = 0, = (10.59) (10.60) (10.61)
1 df 3 , momentum. (10.62) d 2
10.4.3
Energy
A2 df A2 f + 2 c2 co d o f A2 d A2 df (A + A) A 2 = 0, (10.63) c2 d co d o df f d df 3f + + ( + ) = 0, (10.64) d d d 1 d df df f ( + ) = 0, energy. (10.65) 3f + + d d d
The energy equation, Eq. (10.51) becomes A 3 +
10.5
Reduction to non-autonomous form

d + 2 df df d ( + ) + f = 0, d d 1 df 3 d 2 + 2 df df 3f + + f ( + ) = 0, d d
Let us eliminate d/d and d/d from Eq. (10.65) with use of Eqs. (10.58,10.62). 3f + (10.66)
(10.67)
df 3f + ( ) f d df f d
1 df d
3f ( ) + ( )2
3 2 1 df + ( ) d 2
3 2
= 0, = 0,
(10.68) (10.69)
10.5. REDUCTION TO NON-AUTONOMOUS FORM ( )2 df 3 2 f 3( ) + ( ) d 2 f df 3 22 ( )2 + f 3 3 + 2 + d 2 f df 1 22 ( )2 + f 3 3 + + d 2 f
391 = 0, = 0, = 0. (10.70) (10.71) (10.72)
Rearranging, we get ( )2 f df =f d 1 3 + 3 + 2 22 . (10.73) (10.74) Equation (10.73) is T(14). We can thus write an explicit non-autonomous ordinary dierential equation for the evolution of f in terms of the state variables f , , and , as well as the independent variable .
1 f 3 + 3 + 2 df = f d ( )2 22
(10.75)
Eq. (10.75) can be directly substituted into the momentum equation, Eq. (10.62) to get d = d
1 df d
3 2
(10.76)
Then Eq. (10.76) can be substituted into Eq. (10.58) to get

d + 2 d d = . d
(10.77)
Equations (10.75-10.77) form a non-autonomous system of rst order dierential equations of the form df = g1 (f, , , ), d d = g2 (f, , , ), d d = g3 (f, , , ). d (10.78) (10.79) (10.80)
They can be integrated with standard numerical software. One must of course provide conditions of all state variables at a particular point. We apply conditions not at = 0, but
392
at = 1, the locus of the shock front. Following Taylor, the conditions are taken from the Rankine-Hugoniot equations applied in the limit of a strong shock. We take the subscript s to denote the shock state at = 1. For the density, one applies Eq. (9.285) and nds s +1 = , o 1 o s +1 = , o 1 +1 . s = ( = 1) = 1 (10.81) (10.82) (10.83)
For the pressure, one nds, by slight modication of Eq. (9.291) (taking D 2 = (dR/dt)2 ), that
dR 2 dt c2 o 2 3
AR = c2 o A2 R3 = c2 o 1 = fs = f ( = 1) = For the velocity, using Eq. (9.298), one nds us

dR dt
+ 1 Ps , 2 Po + 1 3 R f1s , 2 + 1 3 A2 fs , R 2 c2 o +1 fs , 2 2 . +1
(10.84) (10.85) (10.86) (10.87) (10.88)
2 , +1
(10.89) (10.90) (10.91) (10.92)
2 R3/2 1s = , 3/2 AR +1 R3/2 As 2 = , AR3/2 +1 2 . s = ( = 1) = +1
Equations (10.83, 10.88,10.92) form the appropriate set of initial conditions for the integration of Eqs. (10.75-10.77).
10.6
Numerical solution
Solutions for f (), () and () are shown for = 7/5 in Figs. 10.1-10.3, respectively.
10.6. NUMERICAL SOLUTION

f 1.2 1.0 0.8 0.6 0.4 0.2
393
0.0
0.2
0.4
0.6
0.8
1.0
Figure 10.1: Scaled pressure f versus similarity variable for = 7/5 in Taylor-Sedov blast wave.
1.0
0.8
0.6
0.4
0.2
0.0
0.2
0.4
0.6
0.8
1.0
Figure 10.2: Scaled velocity versus similarity variable for = 7/5 in Taylor-Sedov blast wave.
7 6 5 4 3 2 1
0.0
0.2
0.4
0.6
0.8
1.0
Figure 10.3: Scaled density versus similarity variable for = 7/5 in Taylor-Sedov blast wave.
394
So we now have a similarity solution for the scaled variables. We need to relate this to physical dimensional quantities. Let us assign some initial conditions for t = 0, r > 0; that is, away from the point source. Take u(r, 0) = 0, We also have from the ideal gas law T (r, 0) = Po = To . o R (10.94) (r, 0) = o , P (r, 0) = Po . (10.93)
For the calorically perfect gas we further have e(r, 0) = 1 Po = eo . 1 o (10.95)
10.6.1
Calculation of total energy
Now as the point source expands, it will generate a strong shock wave. Material which has not been shocked is oblivious to the presence of the shock. Material which the shock wave has reached has been inuenced by it. It stands to reason from energy conservation principles that we want the total energy, internal plus kinetic, to be constant in the shocked domain, r (0, R(t)], where R(t) is the shock front location. Let us recall some spherical geometry so this energy conservation principle can be properly formulated. Consider a thin dierential spherical shell of thickness dr located somewhere in the shocked region: r (0, R(t)]. The volume of the thin shell is dV = 4r 2 dr (10.96) (surface area) (thickness) The dierential mass dm of this shell is dm = dV, = 4r 2 dr. (10.97) (10.98)
Now recall the mass-specic internal energy is e and the mass-specic kinetic energy is u2 /2. So the total dierential energy, internal plus kinetic, in the dierential shell is dE = 1 e + u2 dm, 2 1 = 4 e + u2 r 2 dr. 2 (10.99) (10.100)
10.6. NUMERICAL SOLUTION Now, the total energy E within the shock is the integral through the entire sphere,
R(t) R(t)
395
E=
0
dE =
0 R(t)
1 4 e + u2 r 2 dr, 2 4 1 P 1 + u2 r 2 dr, 1 2
R(t)
(10.101) (10.102) (10.103)
=
0
4 1
R(t) 0
P r 2dr + 2
0
u2 r 2 dr .
thermal energy
kinetic energy
We introduce variables from our similarity transformations next: E = 4 1

1 1
Po R3 f1 R2 2 Rd +2
0 1 P r2 dr 1 0
o R3 2 R2 2 Rd , 1
u2 r2 dr
(10.104)
1 Po A2 2 f d + 2 o A2 2 2 d, 2 co 0 0 1 o 1 2 2 Po f 2 d + d , = 4A2 2 co ( 1) 0 2 0 1 1 1 1 2 2 = 4A2 o f 2 d + d . ( 1) 0 2 0
4 1 4 = 1
Po f1 2 d + 2
0 1 0
o 2 2 d, 1
(10.105) (10.106) (10.107) (10.108)
dependent on only The term inside the parentheses is dependent on only. So, if we consider air with = 7/5, we can, using our knowledge of f (), (), and (), which only depend on , to calculate once and for all the value of the integrals. For = 7/5, we obtain via numerical quadrature E = 4A2 o 1 1 (0.185194) + (0.185168) , (7/5)(2/5) 2 2 = 5.3192oA . (10.109) (10.110)
Now from Eq. (10.17, 10.52,10.53, 10.110) with = 7/5, we get P = Po R3 f A2 , c2 o o 2 = Po R3 f A , Po 1 = R3 f o A2 , (10.111) (10.112) (10.113)
396 = R3 f 1
7 5
CHAPTER 10. BLAST WAVES E , 5.3192 (10.114) (10.115) . (10.116)
= 0.1343R3Ef, E r P (r, t) = 0.1343 3 f R (t) R(t) The peak pressure occurs at = 1, where r = R, and where f ( = 1) = So at = 1, where r = R, we have 2(1.4) 2 = = 1.167. +1 1.4 + 1
(10.117)
E . (10.118) R3 The peak pressure decays at a rate proportional to 1/R3 in the strong shock limit. Now from Eqs. (10.19,10.54,10.110) we get for u: P = (0.1343)(1.167)R3E = 0.1567 u = R3/2 A, = R3/2 u(r, t) = E , 5.319o . (10.119) (10.120) (10.121)
E r 1 3/2 (t) 5.319o R R(t)
Let us now explicitly solve for the shock position R(t) and the shock velocity dR/dt. We have from Eqs. (10.20, 10.110) that dR = AR3/2 , dt = R3/2 dR = 2 5/2 R = 5 Now since R(0) = 0, we get C = 0, so 2 5/2 R = 5 t = E t, 5.319o (10.126) (10.127) (10.128) 1 E , 3/2 (t) 5.319o R E dt, 5.319o E t + C. 5.319o (10.122) (10.123) (10.124) (10.125)
2 5/2 R 5.319o E 1/2 , 5 = 0.9225R5/2 1/2 E 1/2 . o
10.6. NUMERICAL SOLUTION Equation (10.128) is T(38). Solving for R, we get 1 t1/2 E 1/2 , 0.9225 o R(t) = 1.032791/5 E 1/5 t2/5 . o R5/2 =
397
(10.129) (10.130)
Thus, we have a prediction for the shock location as a function of time t, as well as point source energy E. If we know the position as a function of time, we can easily get the shock velocity be direct dierentiation: dR = 0.41311/5 E 1/5 t3/5 . o dt (10.131)
If we can make a measurement of the blast wave location R at a given known time t, and we know the ambient density o , we can estimate the point source energy E. Let us invert Eq. (10.130) to solve for E and get E = o R5 , (1.03279)5t2 o R5 = 0.85102 2 . t (10.132) (10.133)
10.6.2
Comparison with experimental data
Now Taylors Part II paper from 1950 gives data for the 19 July 1945 atomic explosion at the Trinity site in New Mexico. We choose one point from the photographic record which nds the shock from the blast to be located at R = 185 m when t = 62 ms. Let us assume the ambient air has a density of o = 1.161 kg/m3 . Then we can estimate the energy of the device by Eq. (10.133) as E = 0.85102
kg 1.161 m2 (185 m)5 , (0.062 s)2 = 55.7 1012 J.
(10.134) (10.135)
Now 1 ton of the high explosive TNT is know to contain 4.25 109 J of chemical energy. So the estimated energy of the Trinity site device in terms of a TNT equivalent is T NTequivalent = 55.7 1012 J = 13.1 103 ton. J 4.25 109 ton (10.136)
In common parlance, the Trinity site device was a 13 kiloton bomb by this simple estimate. Taylor provides some nuanced corrections to this estimate. Modern estimates are now around 20 kiloton.
398
10.7
Contrast with acoustic limit
We saw in Eq. (10.118) that in the expansion associated with a strong shock, the pressure decays as 1/R3 . Let us see how that compares with the decay of pressure in the limit of a weak shock. Let us rst rewrite the governing equations. Here we 1) rewrite Eq. (10.1) in a conservative form, using the chain rule to absorb the source term inside the derivative, 2) repeat the linear momentum equation, Eq. (10.2), and 3) re-cast the energy equation for a calorically perfect ideal gas, Eq. (10.11) in terms of the full partial derivatives: 1 2 r u + 2 t r r u 1 P u +u + t r r P P P +u +u t r t r = 0, = 0, = 0. (10.137) (10.138) (10.139)
Now let us consider the acoustic limit, which corresponds to perturbations of a uid at rest. Taking 0 < << 1, we recast the dependent variables , P , and u as = o + 1 + . . . , P = Po + P1 + . . . , u = uo +u1 + . . . .
=0
(10.140) (10.141) (10.142)
Here o and Po are taken to be constants. The ambient velocity uo = 0. Strictly speaking, we should non-dimensionalize the equations before we introduce an asymptotic expansion. However, so doing would not change the essence of the argument to be made. We next introduce our expansions into the governing equations: 1 2 r (o + 1 ) (u1 ) = 0, (o + 1 ) + 2 t r r 1 (u1 ) + (u1 ) (u1 ) + (Po + P1 ) = 0, t r o + 1 r (Po + P1 ) + (u1 ) (Po + P1 ) t r Po + P1 (o + 1 ) + (u1 ) (o + 1 ) = 0. o + 1 t r (10.143) (10.144)
(10.145)
Now derivatives of constants are all zero, and so at leading order the constant state satises the governing equations. At O(), the equations reduce to 1 1 + 2 (r 2 o u1 ) = 0, t r r
(10.146)
10.7. CONTRAST WITH ACOUSTIC LIMIT 1 P1 u1 + = 0, t o r Po 1 P1 = 0. t o t Now, adopt as before c2 = Po /o , so the energy equation, Eq. (10.148), becomes o P1 1 = c2 . o t t Now substitute Eq. (10.149) into the mass equation, Eq. (10.146), to get 1 P1 1 + 2 (r 2 o u1 ) = 0. c2 t r r o We take the time derivative of Eq. (10.150) to get 1 2 1 2 P1 r o u1 + 2 t2 co t r 2 r u1 1 1 2 P1 r 2 o + 2 2 t2 co r r t = 0, = 0.
399 (10.147) (10.148)
(10.149)
(10.150)
(10.151) (10.152)
We next use the momentum equation, Eq. (10.147), to eliminate u1 /t in Eq. (10.152): 1 2 P1 1 P1 1 r 2 o = 0, + 2 c2 t2 r r o r o P1 1 2 P1 1 r2 = 0, 2 2 t2 co r r r 1 2 P1 1 = 2 2 t2 co r r (10.153) (10.154) r2 P1 r . (10.155)
This second-order linear partial dierential equation has a well-known solution of the DAlembert form: 1 r P1 = g t r co 1 r + h t+ r co . (10.156)
Here g and h are arbitrary functions which are chosen to match the initial conditions. Let us check this solution for g; the procedure can easily be repeated for h. If P1 = (1/r)g(t r/co), then r 1 P1 , = g t t r co 2 P1 r 1 , = g t 2 t r co (10.157) (10.158)
400 and P1 r 11 = g t r co r co
1 r g t 2 r co
(10.159)
With these results, let us substitute into Eq. (10.155) to see if it is satised: r 1 1 g t 2r co co = = = = = 11 1 1 r r r2 2g t , g t 2 r r co r co r co 1 r r r 2 g t +g t , r r co co co r 1 r 1 r r 1 + g t g t 2 2 g t r co co co co co co 1 r r g t , 2 2 r co co r 1 1 . g t 2r co co (10.160) (10.161) (10.162) , (10.163) (10.164)
Indeed, our form of P1 (r, t) satises the governing partial dierential equation. Moreover we can see by inspection of Eq. (10.156) that the pressure decays as does 1/r in the limit of acoustic disturbances. This is a much slower rate of decay than for the blast wave, which goes as the inverse cube of radius.
Bibliography
Here we give a summary of some of the major books relevant to these lecture notes. Specic journal articles are not cited, though the main text of the lecture notes occasionally draws on such articles where relevant. The emphasis is on combustion, physical chemistry, and thermodynamics, with some mathematics. Some general works of historic importance are also included. The list is by no means comprehensive. M. M. Abbott and H. C. van Ness, 1972, Thermodynamics, Schaums Outline Series in Engineering, McGraw-Hill, New York.
This is written in the style of all the Schaums series. It has extensive solved problems and a crisp rigorous style that is readable by undergraduate engineers. It has a chemical engineering emphasis, but is also useful for all engineers.
R. Aris, 1962, Vectors, Tensors, and the Basic Equations of Fluid Mechanics, Dover, New York.
This book is a jewel of continuum mechanics theory. It is brief and incisive. While the focus is on the inert Navier-Stokes equations, there is a short, excellent chapter on the reactive Navier-Stokes equations. Highly recommended.
J. Bebernes and D. Eberly, 1989, Mathematical Problems from Combustion Theory, Springer, Berlin.
This monograph focuses on mathematical analysis of combustion systems, focusing on blowup phenomena.
A. Bejan, 2006, Advanced Engineering Thermodynamics, Third Edition, John Wiley, Hoboken, New Jersey.
This is an advanced undergraduate text. It gives a modern treatment of the science of classical thermodynamics. It does not conne its attention to traditional engineering problems, and considers applications across biology and earth sciences as well. The thermodynamics of irreversible processes are discussed in detail.
R. S. Berry, S. A. Rice, and J. Ross,, 2000, Physical Chemistry, Second Edition, Oxford University Press, Oxford.
This is a rigorous general text in physical chemistry at a senior or rst year graduate level. It has a full treatment of classical and statistical thermodynamics as well as quantum mechanics.
401
R. B. Bird, W. E. Stewart, and E. N. Lightfoot, 2006, Transport Phenomena, Second Edition, Wiley, New York.
This acknowledged classic of the chemical engineering literature does not focus on reaction, but has an exemplary treatment of mass, momentum, and energy advection-diusion phenomena.
L. Boltzmann, 1995, Lectures on Gas Theory, Dover, New York.

This is a detailed monograph by the founding father of statistical thermodynamics.
C. Borgnakke, and R. E. Sonntag, 2009, Fundamentals of Thermodynamics, Seventh Edition, John Wiley, New York.
This classic and popular undergraduate mechanical engineering text has stood the test of time and has a full treatment of most classical problems.
R. Boyle, 2003, The Sceptical Chymist, Dover, New York.

This is the original work of the famous early gure of the scientic revolution.
J. D. Buckmaster and G. S. S. Ludford, 1983, Lectures on Mathematical Combustion, SIAM, Philadelphia.

This is set of lecture notes which serves as an excellent graduate-level introduction to some of the mathematical challenges of combustion.
J. D. Buckmaster and G. S. S. Ludford, 2008, Theory of Laminar Flames, Cambridge, Cambridge.

This seminal monograph, rst published in 1982, highlights rigorous mathematical methods applied to laminar ames.
H. B. Callen, 1985, Thermodynamics and an Introduction to Thermostatistics, Second Edition, John Wiley, New York.
This advanced undergraduate text has an emphasis on classical physics applied to thermodynamics, with a few chapters devoted to quantum and statistical foundations.
S. Carnot, 2005, Reections on the Motive Power of Fire, Dover, New York.
This is the original source of Carnots foundational work on the heat engines. Also included is a paper of Clausius.
N. Chigier, 1981, Energy, Combustion, and Environment, McGraw-Hill, New York.

This advanced undergraduate text gives an integrated science-based discussion of the topic matter stated in its title.
R. Courant and K. O. Friedrichs, 1999, Supersonic Flow and Shock Waves, Springer, New York. 402
This seminal mathematical treatment of inviscid compressible ow gives an outstanding and rigorous discussion of the application of partial dierential equations to uid mechanics. There is only a small discussion of combustion, but the methods are of high relevance in high speed reactive ow.
S. R. de Groot and P. Mazur, 1984, Non-Equilibrium Thermodynamics, Dover, New York.

This is an inuential monograph which summarizes much of the work of the famous Belgian school of thermodynamics. It is written at a graduate level and has a strong link to uid mechanics and chemical reactions.
E. Fermi, 1936, Thermodynamics, Dover, New York.

This short classic clearly and eciently summarizes the fundamentals of thermodynamics. It is based on a series of lectures given by this Nobel prize winner.
W. Fickett and W. C. Davis, 1979, Detonation, U. California, Berkeley.

This distinctive monograph has a rich and focused discussion on detonation science. It opens with the relevant physical chemistry, then gives detailed exposition on the theory of detonation. Careful discussion of experiment is also included.
I. Glassman and R. Yetter, 2008, Combustion, Fourth Edition, Academic Press, New York.
This well known monograph give a good description of important combustion phenomena.
J. F. Griths and J. A. Barnard, 1995, Flame and Combustion, Third Edition, Chapman and Hall, Glasgow.
This is a graduate level text in combustion.
E. P. Gyftopoulos and G. P. Beretta, 1991, Thermodynamics: Foundations and Applications, Macmillan, New York.
This beginning graduate text has a rigorous development of classical thermodynamics.
J. O. Hirschfelder, C. F. Curtis, and R. B. Bird, 1954, Molecular Theory of Gases and Liquids, Wiley, New York.
This comprehensive tome is a valuable addition to any library of thermal science. Its wide ranging text covers equations of state, molecular collision theory, reactive hydrodynamics, reaction kinetics, and many other topics all from the point of view of careful physical chemistry. Much of the work remains original.
W. Hundsdorfer and J. G. Verwer, 2003, Numerical Solution of Time-Dependent AdvectionDiusion-Reaction Equations, Springer, Berlin.
This nice graduate level monograph focuses on modern numerical techniques for paradigm reactive ow systems. The systems themselves are easy to pose which allows eective exposition of the key challenges in numerical modeling.
403
A. M. Kanury, 1975, Introduction to Combustion Phenomena, Gordon and Breach, Amsterdam.

This introductory graduate level text gives practical problems and good discussion of combustion phenomena.
A. K. Kapila, 1983, Asymptotic Treatment of Chemically Reacting Systems, Pitman, Boston.

This short introductory monograph gives lecture notes on asymptotic analysis as applied to combustion.
E. L. Keating, 2007, Applied Combustion, Second Edition, CRC Press, Boca Raton, FL.
This book is an engineering text for seniors or rst year graduate students with a general coverage of topics.
J. Kestin, 1966, A Course in Thermodynamics, Blaisdell, Waltham, Massachusetts.

This is a foundational textbook that can be read on many levels. All rst principles are reported in a readable fashion. In addition the author makes a great eort to expose the underlying mathematical foundations of thermodynamics.
R. J. Key, M. E. Coltrin, and P. Glarborg, 2003, Chemically Reacting Flow: Theory and Practice, John Wiley, New York.
This comprehensive text gives an introductory graduate level discussion of uid mechanics, thermochemistry, and nite rate chemical kinetics. The focus is on low Mach number ows, and there is signicant discussion of how to achieve computational solutions.
C. Kittel and H. Kroemer, 1980, Thermal Physics, Second Edition, Freeman, San Francisco.
This is a classic undergraduate text for physics students. It has a good introduction to statistical thermodynamics and a short eective chapter on classical thermodynamics.
D. Kondepudi and I. Prigogine, 1998, Modern Thermodynamics: From Heat Engines to Dissipative Structures, John Wiley, New York.
Detailed modern exposition which exploits the authors unique vision of thermodynamics with both a science and engineering avor.
V. P. Korobeinikov, 1989, Unsteady Interaction of Shock and Detonation Waves in Gases, Taylor and Francis, New York.
This is a graduate level monograph.
K. K. Kuo, 2005, Principles of Combustion, Second Edition, John Wiley, New York.
This is a readable graduate level engineering text for combustion fundamentals. It includes a full treatment of reacting thermodynamics as well as discussion of links to uid mechanics.
404
K. J. Laidler, 1965, Chemical Kinetics, McGraw-Hill, New York.

This is a standard advanced undergraduate chemistry text on the dynamics of chemical reactions.
L. D. Landau and E. M. Lifshitz, 1987, Fluid Mechanics, Second Edition, ButterworhHeinemann, Oxford.
This classic comprehensive text of uid physics contains a small discussion of reactive ow.
B. H. Lavenda, 1978, Thermodynamics of Irreversible Processes, John Wiley, New York.

This is a lively and opinionated monograph describing and commenting on irreversible thermodynamics. The author is especially critical of the Prigogine school of thought on entropy production rate minimization.
A. Lavoisier, 1984, Elements of Chemistry, Dover, New York.

This is the classic treatise which gives the rst explicit statement of mass conservation in chemical reactions.
H. W. Liepmann and A. Roshko, 1957, Elements of Gasdynamics, John Wiley, New York.
This is an inuential text in compressible aerodynamics that is appropriate for seniors or beginning graduate students. It has a strong treatment of the physics and thermodynamics of compressible ow along with elegant and ecient text. Its treatment of both experiment and the underlying theory is outstanding.
C. K. Law, 2006, Combustion Physics, Cambridge, Cambridge.

This modern text is a comprehensive graduate level discussion of combustion theory.
J. H. S. Lee, 2008, The Detonation Phenomenon, Cambridge, Cambridge.

This is a graduate level monograph focusing on detonation.
J. C. Maxwell, 2001, Theory of Heat, Dover, New York.

This is a short readable book by the 19th century master. Here the mathematics is minimized in favor of more words of explanation.
B. Lewis and G. von Elbe, 1961, Combustion, Flames and Explosions of Gases, Academic Press, New York.
For many years this was the graduate text in combustion. It has a ne description of scientic combustion experiments and good discussion of the key mechanisms of combustion.
M. J. Moran and H. N. Shapiro, 2003, Fundamentals of Engineering Thermodynamics, Fifth Edition, John Wiley, New York.
This is a standard undergraduate thermodynamics text, and one of the more popular. Note: some examples in this set of notes have been adopted from the 1992 Second Edition of this text.
405
I. M ller and T. Ruggeri, 1998, Rational Extended Thermodynamics, Springer-Verlag, New u York.
This modern monograph gives a rigorous treatment of some of the key issues in modern continuum mechanics and classical thermodynamics.
I. M ller and W. Weiss, 2005, Entropy and Energy, Springer-Verlag, New York. u
This is a unique treatise on fundamental concepts in thermodynamics. The author provide mathematical rigor, historical perspective, and examples from a diverse set of scientic elds.
I. M ller, 2007, A History of Thermodynamics: The Doctrine of Energy and Entropy, u Springer-Verlag, New York.
The author gives a readable text at an undergraduate level which highlights some of the many controversies of thermodynamics, both ancient and modern.
E. S. Oran and J. P. Boris, 2001, Numerical Simulation of Reactive Flow, Second Edition, Cambridge, Cambridge.
This monograph by two of the leading gures of computational combustion gives a good discussion of numerical challenges as well as physically relevant problems.
W. Pauli, 2000, Thermodynamics and the Kinetic Theory of Gases, Dover, New York.
This is a monograph on thermodynamics by a well known physicist.
L. Perko, 2001, Dierential Equations and Dynamical Systems, Third Edition, Springer, New York.
This monograph gives no explicit discussion of combustion but does give an outstanding treatment of dynamic systems theory.
M. Planck, 1945, Treatise on Thermodynamics, Dover, New York.

This brief book gives many unique insights from a great physicist.
T. Poinsot and D. Veynante, 2005, Theoretical and Numerical Combustion, Second Edition, Edwards, Flourtown, PA.
This is a useful graduate level introduction to combustion.
I. Prigogine, 1967, Introduction to Thermodynamics of Irreversible Processes, Third Edition, Interscience, New York.
This is a famous book that summarizes the essence of the work of the Belgian school for which the author was awarded the Nobel prize.
K. W. Ragland and K. M. Bryden, 2011, Combustion Engineering, Second Edition, CRC Press, Boca Raton.
This is a beginning graduate text in combustion engineering.
406
L. E. Reichl, 1998, A Modern Course in Statistical Physics, John Wiley, New York.
This full service graduate text has a good summary of key concepts of classical thermodynamics and a strong development of modern statistical thermodynamics.
W. C. Reynolds, 1968, Thermodynamics, Second Edition, McGraw-Hill, New York.

This is an unusually good undergraduate text written for mechanical engineers. The author has wonderful qualitative problems in addition to the usual topics in such texts. A good introduction to statistical mechanics is included as well. Highly recommended.
D. E. Rosner, 2000, Transport Processes in Chemically Reacting Flow Systems, Dover, New York.
This graduate level monograph gives a chemical engineering perspective on reaction engineering in uid systems.
S. I. Sandler, 1998, Chemical and Engineering Thermodynamics, Third Edition, John Wiley, New York.
Advanced undergraduate text in thermodynamics from a chemical engineering perspective with a good mathematical treatment.
E. Schrodinger, 1989, Statistical Thermodynamics, Dover, New York.

This is a short monograph written by the one of the pioneers of quantum physics.
A. H. Shapiro, 1953, The Dynamics and Thermodynamics of Compressible Fluid Flow, Vols. I and II, John Wiley, New York.
This classic two volume set has a comprehensive treatment of the subject of its title. It has numerous worked example problems, and is written from an engineers perspective.
J. M. Smith, H. C. van Ness, and M. Abbott, 2004, Introduction to Chemical Engineering Thermodynamics, Seventh Edition, McGraw-Hill, New York.
This is probably the most common undergraduate text in thermodynamics in chemical engineering. It is rigorous and has went through many revisions.
R. A. Strehlow, 1984, Combustion Fundamentals, McGraw-Hill, New York.

This graduate level text has a nice discussion of physical chemistry followed by good descriptions of the authors seminal work in detonation experiments, among other standard combustion topics.
J. W. Tester and M. Modell, 1997, Thermodynamics and Its Applications, Third Edition, Prentice Hall, Upper Saddle River, New Jersey.
This entry level graduate text in thermodynamics is written from a chemical engineers perspective. It has a strong mathematical development for both classical and statistical thermodynamics.
407
K. Terao, 2007, Irreversible Phenomena: Ignitions, Combustion, and Detonation Waves, Springer, Berlin.
This unique monograph links irreversible thermodynamics with phase transition, ignition and detonation.
T.-Y. Toong, 1982, Combustion Dynamics: The Dynamics of Chemically Reacting Fluids, McGraw-Hill, New York.
This is an entry level graduate text on combustion science.
C. A. Truesdell, 1980, The Tragicomic History of Thermodynamics, 1822-1854, Springer Verlag, New York.
This idiosyncratic monograph has a lucid description of the history of nineteenth century thermal science. It is written in an erudite fashion and the reader who is willing to dive into a dicult subject will be rewarded for diligence by many new insights.
C. A. Truesdell, 1984, Rational Thermodynamics, Springer-Verlag, New York.

This is a modern update on the evolution of classical thermodynamics in the twentieth century. It is at a graduate level and contains several articles by some of the present leaders of the eld.
S. R. Turns, 2011, An Introduction to Combustion, Third Edition, McGraw-Hill, Boston.

This is a popular senior-level undergraduate text on combustion which uses many notions from thermodynamics of mixtures.
H. C. van Ness, 1983, Understanding Thermodynamics, Dover, New York.

This is a short readable monograph from a chemical engineering perspective.
W. G. Vincenti and C. H. Kruger, 1965, Introduction to Physical Gas Dynamics, John Wiley, New York.
This graduate text on high speed non-equilibrium ows contains some nice descriptions of the interplay of classical and statistical mechanics. There is an emphasis on aerospace applications.
J. Warnatz, U. Maas, and R. W. Dibble, 2006, Combustion, Fourth Edition, Springer, New York.
This short graduate level text is nonetheless comprehensive and gives a good introduction to combustion science.
F. A. Williams, 1985, Combustion Theory, Benjamin-Cummings, Menlo Park, California.

This inuential monograph was rst published in 1965. It was one of the rst texts to give a comprehensive theoretical treatment of reactive uid mechanics for model and practical systems.
L. C. Woods, 1975, The Thermodynamics of Fluid Systems, Clarendon, Oxford.

This graduate text gives a good, detailed survey of the thermodynamics of irreversible processes, especially related to uid systems in which convection and diusion play important roles.
408
I. B. Zeldovich and A. S. Kompaneets, 1960, Theory of Detonation, Academic Press, New York.
This is a readable monograph whose rst author is one of the founding fathers of modern detonation theory.
Ya. B. Zeldovich and Yu. P. Raizer, 2002, Physics of Shock Waves and High-Temperature Hydrodynamic Phenomena, Dover, New York.
This classic tome of the Russian school is a tour de force of reactive ow physics.
409

Notes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Notes

Uploaded by

Copyright:

Available Formats

LECTURE NOTES ON FUNDAMENTALS OF COMBUSTION

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.2.2 2 Gas 2.1 2.2 2.3

CC BY-NC-ND. 12 December 2011, J. M. Powers.

6.2 6.3 6.4 6.5

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CC BY-NC-ND. 12 December 2011, J. M. Powers.

Chapter 1 Introduction to kinetics

CHAPTER 1. INTRODUCTION TO KINETICS

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CHAPTER 1. INTRODUCTION TO KINETICS

Isothermal, isochoric kinetics

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.1. ISOTHERMAL, ISOCHORIC KINETICS

CC BY-NC-ND. 12 December 2011, J. M. Powers.

18 Here we have taken = = r =

CHAPTER 1. INTRODUCTION TO KINETICS

(1.15) (1.16) (1.17)

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.1. ISOTHERMAL, ISOCHORIC KINETICS

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CHAPTER 1. INTRODUCTION TO KINETICS

At high temperature, we have lim exp E j RT = 1. (1.39)

And lastly, at intermediate temperature, we have exp E j RT O(1) when T =O Ej R . (1.40)

A sketch of this modulating factor is given in Figure 1.1. Note

1.1. ISOTHERMAL, ISOCHORIC KINETICS

exp(- E j /(R T))

CC BY-NC-ND. 12 December 2011, J. M. Powers.

CHAPTER 1. INTRODUCTION TO KINETICS

= 2 2.9 10 = 1.16 1014

mole cm3 mole cm3 mole cm3

(5000 K)1 exp(0), (1.46) (5000 K)1 , (1.47) (1.48)

= 2 6.81 1018 exp

erg 4.96 1012 mole erg 8.31441 107 mole K (5000 K)

= 1.77548 1010 a13 T 13 exp a14 T

= 5.80 1013 = 8.8774 10

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.1. ISOTHERMAL, ISOCHORIC KINETICS

CHAPTER 1. INTRODUCTION TO KINETICS

Note that by our denition of molecular species concentration that i =

Substituting from Eq. (1.60) into Eq. (1.61), we get

CHAPTER 1. INTRODUCTION TO KINETICS

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.1. ISOTHERMAL, ISOCHORIC KINETICS a13 T 13 exp E 13 RT e e e a14 T 14 exp O O M E 14 RT e 2 e = 0. O M

CHAPTER 1. INTRODUCTION TO KINETICS

0.001 100 000 200 000

1.1. ISOTHERMAL, ISOCHORIC KINETICS

Brook Taylor, 1685-1731, English mathematician. CC BY-NC-ND. 12 December 2011, J. M. Powers.

CHAPTER 1. INTRODUCTION TO KINETICS

1.1. ISOTHERMAL, ISOCHORIC KINETICS 1.1.1.2 Single reversible reaction

CC BY-NC-ND. 12 December 2011, J. M. Powers.

32 In matrix form, we can simplify to d dt O O2

CHAPTER 1. INTRODUCTION TO KINETICS

Here the N J or 2 1 matrix is = 2 1 . (1.108)

So here the N N or 2 2 matrix L1 is L1 = 1 0 1 2 . (1.111)

We rearrange Eq. (1.123) to nd the familiar Kc,13 =

CHAPTER 1. INTRODUCTION TO KINETICS

1 1 1 1 (O2 + (O O )) (O2 + (O O ) + O ) O O (O2 + (O O ) + O ) . 2 2 Kc,13 2

CC BY-NC-ND. 12 December 2011, J. M. Powers.

1.1. ISOTHERMAL, ISOCHORIC KINETICS

0.00100 0.00070 0.00050