Professional Documents
Culture Documents
'
Nottingham
Engineering
Investigation
Gas-Liquid
By
Mhunir
SCIEC`
October 2010
Abstract
ABSTRACT
Three different experimental campaigns had been launched to investigate the two-phase flow structures like the void waves in bubbly flow, liquid periodicity of the Taylor bubbles in slug flow, huge waves in chum flow, disturbance waves in slugs and flow and wisps in wispy-annular flow. The three experiments and data annular had been considered in increasing order of complexity. Time varying data analysis in all the three campaigns. At the simplest level of analysis statistical were acquired measureswere extracted from the time series, presented, discussed and conclusions drawn.
In the first campaign, time varying void fraction has been measured using a stack of five flush mounted ring-type conductance probes fitted with a 0.005 m internal diameter pipe, 3m long. The test section was located 390 pipe diameters from the two-phase mixer. Three liquids of different physical properties with viscosity values 1,10 and 12 mPa s were mixed with air to establish two-phase flow. Pressuredrop of by a distance of 100 measurementwas carried out using two pressuretaps separated diameters. Upstream tap was positioned 290 pipe diameters from the mixer. pipe Structure frequency was determined from Power Spectrum Density (PSD) of autofunction; structure velocity was estimated by cross-correlating two signals covariance 0.002 m apart. It is shown that flow structure becomes more periodic as separated liquid viscosity increases. Air/more viscous liquid two-phase flow exhibits higher structure frequency than air/water two-phase flow maintained at similar conditions. On the other hand, structure velocity decreases increase in liquid viscosity due to with lower liquid phasemomentum. Analyzing the time series by considering the variations
11
MB ALAMU2010
Abstract
in amplitude and frequency space to generate Probability Density Function (PDF) provides additional detail. The plot of PDF against void fraction for air/water and air/12 mPa s viscous liquid at same liquid and gas superficial velocities of 0.64 m/s and 0.27 m/s shows a marked shift between the void fractions as liquid viscosity increases. For air/12 mPa s two-phase flow, the PDF is characterized by a single, serrated, taller, narrower peak, with void fraction varies from 0.18 to 0.40, fingerprinting bubbly flow. In the case of air/water, the PDF displays a twin peak distribution typical of slug flow, with an averagevoid fraction of 0.80 and a long tail that extends to zero void fraction. Next, activity in the second campaign is summarized.
In the second campaign, dynamic drop concentration, Sauter Mean Diameter (SMD) and Mass Median Diameter (MMD) were measuredin annular two-phase flow for the first time employing a new generation instrumentation based laser diffraction technique. The measurementwas conducted with air/water flow on a 0.019 in internal diameter vertical special test section. The gas superficial velocity was varied from 13 m/s to 43 m/s at fixed liquid superficial velocities of 0.05 and systematically 0.15 m/s. Additional tests were carried out with the gas velocity fixed at 14 m/s for liquid superficial velocities range between 0.03 and 0.20 m/s. Conductance probes were used to log the void fraction and the film hold-up. Spraytec to measure the drop concentration and the characteristic drop diameters was used (SMD & MMD). Pressuredrop was investigated and measured between two pressure taps separated 1.55 m apart in the vertical direction. All instruments were linked and synchronized to achieve simultaneous data acquisitions such that one-to-one was established between the time varying film hold-up, drop correspondence
111
MB ALAMU2010
Abstract
concentration and the system pressuredrop. The time averagedvalues extracted from the dynamic signals were compared with data from previous measurements.They were found to be in good agreement. Sauter Mean Diameter (SMD) and entrained fraction display an obvious similar signature characterized by inflection points at the transition to co-current annular flow when gas superficial velocity is 21 m/s. Mass Median Diameter (MMD) on the other hand detected the transition to mist annular flow which occurs at a gas superficial velocity of 30 m/s. Additional detail is revealed when the variation in the time series of the Mass Median Diameter (MMD) is considered in amplitude and frequency spaceto generateProbability Density Function (PDF). The PDF of the MMD changed from multiple maxima/peaks to single maximum/peak signifying transition from heterogeneous/ multi-modal to
homogeneous/mono-modaldrop size distribution at a gas superficial velocity of 30 m/s. This change has been linked to transition from co-current to mist annular flow. Beyond the transition, momentum of drops produced from thinner film start decreasing as the drop size decreasesas gas superficial velocity increases.However, exact opposite trend is observed in the case of drop momentum produced from an thicker film. The entrained drops within this regime exhibit higher mass density, drop concentration/ entrained fraction as a result of greater interaction momentum and with the gas core. A relationship between pressure drop and entrained fraction is observed to follow a power law having minimum at a point where pressure drop is minimum marking transition to mist flow at gas superficial velocity of 30 m/s and entrained fraction of 0.20. Visualization studies of the high recorded during the experiment speed videos show that at a fixed gas superficial gas velocity of 14 m/s, below liquid superficial of 1.2 m/s, wisps are seen in region classified as chum flow regime by velocity
iv
MB ALAMU 2010
Abstract
existing flow pattern transition models. The observed wisps have been linked to the incomplete atomization of liquid film principally causedby dominance of gravity over drag force. Huge waves dominate with high amplitude and high standard deviation the film hold-up. At another level of analysis, fluctuation of film hold-up and drop concentration with time is considered. Both show evidence of periodicity, film thickness being more periodic than drop concentration. Thus, frequency analysis is carried out to establish the relationship between drop and wave frequency. Surprisingly, wave frequency is seento be higher than drop frequency most casesexamined. This is due to the rate of drop coalescenceand turbulent diffusion in the dispersedphase.Traditional Strouhal number-Lockhart-Martinelli parameterprovides a good correlation for the wave frequency. However, correlating drop frequency using same approach proves inadequate.Secondcampaign, therefore, concludes that the subtle changesin Sauter Mean Diameter (SMD), entrained fraction and the Mass Median Diameter (MMD) make the evidencesof flow pattern transition within annular flow more compelling.
In the third and the last campaign, flow structures were investigated in the most flow geometry used. The test section is made up of a 0.005 m internal complex diameter vertical dividing junction; located 1.3 m vertically downstream of the twophase mixer constitutes the test section. The junction splits and mal-distributes the incoming two-phase flow into the vertical run and the horizontal side arm. The fraction of gas and liquid taken off at the horizontal side arm determines the measured flow splits. Effect of liquid viscosity on partial separation of phases between the outlets of the junction has been investigated by testing liquids. The with air/water and air/viscous
V MB ALAMU2010
Abstract
liquid phase viscosity has been increased systematically from I to 36 mPa s. Test implemented varied between 0- 32 m/s and 0.003 - 1.3 m/s for gas and liquid matrix phasesuperficial velocities respectively. Plotting corresponding liquid hold-up against gas superficial velocity for each take off elucidates chum - annular transition boundary at gas superficial velocity of 15 point m/s. Liquid hold-up is seento increaseas liquid viscosity and fraction of gastaken off increasessuggesting corresponding increase in partial separation of phases.However, effect of liquid viscosity does not become significant until a threshold is exceeded when fraction of gas taken off equals 0.40. In all casesexamined, periodicity of flow structures is observedto increaseas liquid viscosity increases.
Considering the results of the three investigations carried out, it can be concluded that
VI MB ALAMU2010
List of Publications
LIST OF PUBLICATIONS
Alamu, M. B., G. P. Van der Meulen, Azzopardi, 131 2010. Dynamic Drop Size , Measurement in Vertical Annular Two-Phase Flow. 7t' International Conferenceon Multiphase Flow ICMF 2010, Tampa, FL USA, May 30-June 4,2010. Alamu, M. B., G. P. Van der Meulen,HernandezPerez,V., Azzopardi, B. J. (2010). Investigation of Liquid Loading in Gas Well: Simultaneous Measurementsof Drop Size, Film Thickness and PressureDrop. Proceedingsof the ASME 2010 29th International Conference on Ocean, Offshore and Arctic Engineering, OMAE2010, June 6-11,2010, Shanghai,China. Alamu, M. B., Khiang. N., Azzopardi, B. J.,2007:"Effects of Liquid Viscosity on TwoPhaseflow Split at a T-Junction with Vertical inlet Pipe," paper 294, Int. Conf.Multiphase Flow 2007, Leipzig, Germany, 9-13 July Alamu, M. B., G. P. Azzopardi, B.J. 20 10. Effect of Entrained Fraction, Annular , Film thickness and PressureDrop on Gas & CondensateProduction J.Energy Res. Tech. (Submitted). Alamu, M. B., Azzopardi, B. 1,2010. Dynamic Drop Concentration Measurement in Annular Two-Phase Flow. Int. J.Heat & F.Flow. (Submitted).
vii
Acknowledgements
ACKNOWLEDGEMENTS
All praises due to almighty, the beneficent, the merciful. My profound appreciation goes to Prof. B. J. Azzopardi for his supervision, guidance and patience and supportive roles in played throughout the programme. His constructive suggestions and ideas contributed immensely to the successful completion of this PhD thesis. Special gratitude to Amanda for making available some of her data used on T -junction Babatunde Oloro and Simisola Oniru are gratefully acknowledged for experiment providing air-sugar solution data. Dr. Ryuhei Kaji support is gratefully acknowledged. Supports received from the School of Chemical & Environmental Engineering and the International Office, The University of Nottingham are gratefully acknowledged.
Special acknowledgement goes to Dr. Paul Kippax and Dr. Ann Virden of Malvern Instruments, UK, for loan of the Spraytec used in the drop size measurementand for technical support throughout the experiments. Support received providing quality from Cal Gavin Ltd. (Process Intensification Engineering) is also gratefully
acknowledged.
This work has been undertaken within the Joint Project on Transient Multiphase Flows and Flow Assurance. I wish to acknowledge the contributions made to this project by the UK Engineering and Physical Sciences Research Council (EPSRC) and the following: -
Advantica; BP Exploration; CD-adapco; Chevron; ConocoPhillips; ENI; ExxonMobil; FEESA; IFP; Institutt for Energiteknikk; PDVSA (INTEVEP); Petrobras; PETRONAS;
ScandpowerPT; Shell; SINTEF; StatoilHydro and TOTAL. I sincerely express my gratitude for this support."
viii
Acknowledgements
I like to thank Prof Cern Sarica for providing the opportunity to visit Tulsa University Fluid Flow Project (TUFFP). The placementat Tulsa University Fluid Flow Project, Tulsa, Oklahoma,USA was madepossiblethrough the Business-Engineering Science and Travel Scholarship (BESTS), Roberts Money of the Graduate School, University of Nottingham. This sponsorship is gratefully acknowledged. I also like to thank Prof Ovadia Shoham, head of Tulsa University Technology Separation Projects (TUSTP) for his exciting discussion during my stay in Tulsa.
Special thank goes to the laboratory technicians Phil, Fred, Mick, Reg, Paul, Vikki and Marion for their supportive roles in the laboratory. Paul your effort in aligning the laser is gratefully appreciated.
My profound gratitude to Valente for his wonderful support. Peter, Safa, Muktar, Nasir, Lokman,Seunand Mayowahaveall beenwonderful colleagues. They haveall
contributed immensely to make this programme a success.
Finally, I like to expressmy sincere appreciation for the supportive role of Dr. Nasiru
ix
Dedications
This work is dedicated to myfamily: my dadfor visionary suggestion, my grand motherfor her passionate love and care, my mumfor her support and encouragement, brothers and sistersfor their kindnessand support and to Kudroh for her love and my support throughout the programme.
Table of contents
Table of contents
ABSTRACT .............................................................................................................. VII LIST OF PUBLICATIONS .................................................................................... VIII ACKNOWLEDGEMENTS ................................................................................... XI TABLE OF CONTENTS ................................. ........................................................ LIST OF TABLES CHAPTER 1 XXI 11
1
1
INTRODUCTION
1 1.1 Flow Assurance .................................... .......................................................... 3 1.2 Periodic Structures in Gas-Liquid Flow .......................................................... 5 1.3 Flow Patterns in Vertical Pipes ....................................................................... 6 1.3.1 Bubble Flow .................................. ..........................................................
................................. .......................................................... 1.3.5 Dispersed-Bubble Flow 10 .......................................................................... 11 1.4 Flow Pattern Maps ............................... ........................................................ 15 1.5 Aims and Objectives ..................................................................................... 16 1.6 Structure of the Thesis .................................................................................. 17 CHAPTER 2 .................................................... ........................................................
LITERATURE REVIEW 17 .............................................................................................
7 1.3.2Slug Flow ....................................... .......................................................... 1.3.3ChurnFlow 9 .................................... .......................................................... 1.3.4 Annular Flow 9
Table of contents
49 2.3.8 Sawant et al. (2009) ................................................................................ 50 2.4 Drop size correlation .................................................................................... 51 2.5 T -junction relevant literature ........................................................................ 55 2.6 Justification .................................................................................................. 59 2.7 Statementof the problem ..............................................................................
CHAPTER 3 ............................................................................................................ 61 EFFECTOF LIQUID VISCOSITYON GAS-LIQUID FLOW STRUCTURES ......................... 61
61 3.1 Introduction .................................................................................................. 62 3.2 Experimental Description ............................................................................. 64 3.3 Instrumentation ............................................................................................ 3.4 Results and Discussion 67 ................................................................................. 3.4.1 Time Resolved Measurementand Probability Density Function (PDF) 67 .. 3.4.2 Void fraction 70 .......................................................................................... 3.4.3 Structure frequency 71 ................................................................................ 3.4.4 Structure velocity 74 ................................................................................... 3.4.5 Flow Distribution Coefficient, Co 82 ........................................................... 3.4.6 Drift flux approach 82 ................................................................................. 3.4.7 Slippage between gas and liquid phases 89 ................................................. 3.4.8 Flow pattern map 91 ................................................................................... 3.4.9 PressureDrop 93 ......................................................................................... 3.4.10 Performance of structure velocity correlation 95 ....................................... 3.5 Conclusions 100 ................................................................................................ 102 CHAPTER 4 .......................................................................................................... FLOW.102 TWO-PHASE ANNULAR DYNAMICDROPSIZEMEASUREMENT VERTICAL IN 102 4.1 Introduction ................................................................................................ 104 4.2 Experimental Arrangements ....................................................................... 106 4.2.1 Flow Rig .............................................................................................. 108 4.3 Instrumentation .......................................................................................... 108 4.3.1 Spraytec ............................................................................................... 114 4.3.2 Conductance probe ............................................................................... 116 4.4 MeasurementAccuracy .............................................................................. 117 4.5 Results ....................................................................................................... 117 4.5.1 Flow Pattern Map ................................................................................. 127 4.5.2 Void Fraction ....................................................................................... 130 4.5.3 Drop size .............................................................................................. 4.5.3.1 Identification of Flow Transitions within Annular Flow basedon Drop 131 Size Measurement ..................................................................................... 139 4.5.4 Drop concentration ............................................................................... 140 4.5.5 Entrained Fraction ................................................................................ 143 4.5.6. Film hold-up ....................................................................................... 147 4.5.6.1 Film hold-up and Drop Size ........................................................... 149 4.5.7 PressureDrop ....................................................................................... 153 4.5.7.1 PressureDrop and Entrained Fraction ............................................ 158 4.5.7.2 PressureDrop and Drop Size Drop coalescenceand break-up...... -
Xll
Table of contents
4.6 Validation of Results 165 .................................................................................. 4.7 Shear Stresses 167 ............................................................................................ 4.8 Flow Structures 170 .......................................................................................... 4.9 Drop velocity 174 ............................................................................................. 4.10 Film velocity Distribution 175 ........................................................................ 4.11 Probability Density Function (PDF) 178 ......................................................... 4.12 ProposedFlow Pattern Map 182 ...................................................................... 4.13 Rosin Rammler Analysis 184 .......................................................................... 4.14. Assessmentof entrained fraction correlations 188 .......................................... 4.15 Drop size correlation 190 ................................................................................ 4.16 Discussions 192 .............................................................................................. 4.17 Conclusions 196 .............................................................................................. CHAPTER 5 200 ..........................................................................................................
........................................................... 200 WAVE DYNAMICS IN VERTICAL ANNULAR
CHAPTER 6
6.1 Introduction 237 ................................................................................................ 6.2 Interfacial shear stressdependentmodel 238 ..................................................... 6.3 Wall shear stressdependentmodel 246 ............................................................. 250 6.4 Averaged shear stresses dependentmodel .................................................. 6.7 Conclusions 253 ................................................................... ............................. CHAPTER 7 254 ................................................................................... ....... ................
SPLIT OF GAS-LIQUID TWO-PHASE FLOW AT VERTICAL DIVIDING JUNCTION......... 254
263 time series- effect of liquid physical properties ................ 7.3.2 Flow Pattern transition 266 ..................................................... .................... 7.3.3 Phasedistribution 270 ...................................................................... ........... 7.3.4 Flow pattern identification 272 ........................................................ ........... 275 7.3.5 PhaseSplit ............................................................................. .............. 7.3.6 Effect of physical properties 285 ............................... .................................. 288 7.3.7 Structure frequency .................................................... .......................... 7.4 Conclusions 295 ................................................................................................
Xlii
Table of contents
CHAPTER 8
..........................................................................................................
296
296
296 8.1 Conclusions ................................................................................................ 303 8.2 Recommendationsfor future works ............................................................ 305 REFERENCES ....................................................................................................... 318 NOMENCLATURE ............................................................................................... 321 APPENDIX A ........................................................................................................ 322 APPENDIX B ........................................................................................................
SPRAYTEC VIGNETTING DISTANCE FOR VARIOUS LENS APPLICATIONS ..................... 322
APPENDIX C APPENDIX D
........................................................................................................ ........................................................................................................
..............................................................................
323 325
325
APPENDIX E
........................................................................................................
.....................................................................
326
326
APPENDIX F
........................................................................................................ .
................................................................................... .
327
327
APPENDIX G
....................................................................................................... .
.
328
328
APPENDIX H
329 . .......................................................................................................
329
xiv
List of figures
List of figures
Figure 1.1: Schematic diagram of offshore operation with Floating Production, Storageand offloading Vessels (FPSO) and Dry Completion Unit (DCU)............ 2 6 Figure 1.2: Flow Patterns in vertical and inclined pipes, Shoham (2006) ..................... 7 Figure 1.3: Flow Patterns in horizontal and near horizontal pipes ............................... 8 Figure 1.4: Flow Patternsin entire range of inclinations, Shoham, (2006) .................. 11 Figure 1.5: Upward Vertical through a concentric annulus ........................................ 12 Figure 1.6: Flow Pattern Map Vertical Flow (Griffith and Wallis, 1961) ................... 13 Figure 1.7: Flow Pattern Map for Vertical Flow (Govier and Aziz, 1972) ................. Figure 1.8: Flow Pattern Map Vertical Flow (Taitel - Bamea - Dukler (1980 ) )...... 13 14 Figure 1.9: Flow-pattern map for horizontal flow (Mandhane et al., 1974) ............... 14 Figure 1.10: Flow-pattern map for horizontal flow (Baker, 1954) ............................ 19 Figure 2.1: Schematicsof slug unit in horizontal slug flow ...................................... Figure 2.2: Schematic Diagram of slug flow parameters,Shoham(2006) 21 ................. Figure 2.3: Schematicsof annular flow two fluid model approach, Alves et al. (1991). 37 .......................................................................................................................... Figure 2.4: Schematicsof vertical annular two-phase flow 58 ....................................... Figure 3.1: Schematic flow diagram of the 5.0 mm internal diameter rig 63 ................ Figure 3.2: Conductanceprobe calibration curves, Omebere-Iyari et al. (2006)......... 65 Figure 3.3: Top- a stack of conductanceprobes which constitute the test section 66 ...... (150 mm long). Bottom- cross section of a ring conductanceprobe 66 .......................... Figure 3.4: Comparison of time resolved void fraction data for r air/water and air/12 mPa s viscous liquid at same inlet conditions where superficial gas and liquid velocities are 0.27 m/s and 0.64 m/s respectively............................................... 68 Figure 3.5: Comparison of PDFs of void fraction for air/water and air/12 mPa s
71 liquid air/water and air-IOmPa s ......................................................................... for Figure 3.8: Effect of gas superficial velocity on periodic structure frequency 72 liquid air/water, air/10 mPa s and air/12 mPa s viscous ...................................... Figure 3.9: correlation of structure frequency using gas based Strouhal number with 73 Lockerhart-Martinelli parameter ......................................................................... Figure 3.10: Structure velocity as a function of mixture velocity for air/water and 74 liquid air/12 mPa sviscous .................................................................................. Figure 3.11: Effect of physical properties on structure velocity in the slug flow regime 75 for air/water and air-12mPa s viscous liquid ....................................................... Figure 3.12: Identification of flow regimes using structure velocity variation with 76 for air/12 mPa s mixture velocity ....................................................................... Figure 3.13: Effect of physical properties on structure velocity in the slug flow regime 78 for air-12mPa s viscous liquid ............................................................................ 80 Figure 3.14: Determination of distribution coefficient from Nicklin/Davison line..... 85 Figure 3.15: Flow parameter as function of superficial mixture velocity ................. 85 Figure 3.16: Flow parameter as function of superficial mixture velocity ................. 90 Figure 3.17: Slip ratio, S, as a function of superficial velocities ratio ........................ 91 Figure 3.18: Slip ratio, S, as a function of superficial velocities ratio,VsL = 0.4 m/s.. Figure 3.19: Top - Test matrix for air/water and air-12mPa s viscosity liquid
69 liquid for VsG= 0.27 m/s, = 0.64m/s VSL viscous ............................................... 70 Figure3.6: Void fraction asa functionof mixturevelocity at different viscosity ....... Figure 3.7: Effect of gas superficialvelocity on void fraction at VSL= 0.03 m/s for
xv
List of figures
Figure 4.8: Mean drop diameter variation with increasegas superficial velocity ..... 130 Figure 4.9: Effect of gas superficial velocity on SMD and normalized drop size..... 133 Figure 4.10: Normalized wave amplitude as function of superficial liquid and gas 138 velocities .......................................................................................................... Figure 4.11: Effect of gas and liquid superficial velocities on drop concentration 140 ... Figure 4.12: Effect of gas and liquid superficial velocities on entrained fraction..... 141 Figure 4.13: Effect of gas and liquid superficial velocities on normalized entrained fraction. Measured entrained fraction normalized by the minimum entrained fraction model of Barbosa et al. (2002). 143 ........................................................... 144 Figure 4.14: Schematic of vertical annular two-phase flow ..................................... 145 Figure 4.15 (a) Film thickness variation with increasegas superficial velocity ........ Figure 4.15 (b) Film thickness variation with increase liquid superficial velocity 145 ....
liquid superficialvelocity........128 Figure4.7 (c) Void fraction variationwith increase fraction with mechanistic Figure4.7 (d) Validation of measured model of Alves void 128 (1991) et al. ......................................................................................................
representedon flow Pattern map of Taitel et al. (1980). Bottom - PDF of void fraction corresponding to air/water and air/12 mPa sviscous liquid. Inlet condition 92 VSG= 0.27 m/s, VSL= 0.64 m/s .......................................................................... 93 Figure 3.20: Schematic of pressure-gradientbehaviour in vertical flow .................... Figure 3.21: Pressure gradient as a function of mixture velocity. Liquid phase viscosity = 12 mPa s.......................................................................................... 94 Figure 3.22: Comparison of various structure velocity correlations against experimental data............................................................................................... 96 Figure 3.23: Predictive ability of various existing correlations and the proposed correlation for air/water data.............................................................................. 97 Figure 3.23: Predictive ability of various existing correlations and the proposed correlation for bubbly flow ............................................................................... 100 Figure 4.1: Schematic flow diagram of the rig used in the present study 105 ................. Figure 4.2 (a): Schematic of the test section 106 ............................................................ Figure 4.2 (b): Incident light scattering by a single drop within the measurement volume ............................................................................................................. 108 Figure 4.2 (c): Principle of operation of laser diffraction instrument utilized in present experimental campaign .................................................................................... 109 Figure 4.3: Cross sectional view of ring-type conductance probes used to measure void fraction ..................................................................................................... 114 Figure 4.4: One of the calibration curves for the conductance probes used in the present study.................................................................................................... 115 Figure 4.5 (a): Flow pattern map for vertical two-phase flow. The open symbols representthe conditions at which presentstudy were carried out...................... 121 Figure 4.5 (b): Modified flow pattern map of Hewitt & Roberts (1969) based on new information from drop concentration measurementin vertical annular two-phase flow 124 ................................................................................................................. Figure 4.6: Left Hand Side (LHS) - image of wispy annular flow at gas superficial velocity of 14 m/s and liquid superficial velocity of 0.19 m/s. Right Hand Side (RHS) - image showing evolution of `ring' disturbance wave around transition to 124 30 m/s mist annular after gas superficial velocity of ......................................... 127 Figure 4.7(a) Void fraction variation with increasegas superficial velocity ............ Figure 4.7 (b) Measured void fraction validated with the model of Alves et al. (1991) 127 ........................................................................................................................
xvi
List of figures
gas superficial velocity ......... variation with Figure 4.17 (b) Dimensionless pressure drop variation with increase gas superficial velocity ............................................................................................................ 151 151 Figure 4.17 (c) Pressuredrop variation with increaseliquid superficial velocity ..... Figure 4.17 (d) Dimensionless pressuredrop variation with increase liquid superficial velocity ............................................................................................................ 151 152 Figure 4.17 (e): Variation of gas core density with inlet conditions ......................... 153 Figure 4.17 :(f) Normalized pressuredrop vs. gas superficial velocity .................... 153 Figure 4.17 :(g) Wave amplitude vs. gas superficial velocity .................................. Figure 4.18 (a): Pressuredrop-entrained fraction profile 154 ......................................... Figure 4.18 (b): Relationship between pressuredrop and entrained fraction............ 154 Figure 4.18 (c): Relationship between pressuredrop and entrained fraction............ 155 Figure 4.18 (d): Similarities between present study and work of Mantilla (2008).... 156 Figure 4.18 (e): Normalized wave amplitude as a function of entrained fraction 156 ..... Figure 4.20: Comparison between drop size and Kolmogorov scale of turbulence. 160 . Figure 4.21: Pressuredrop and MMD as a function of superficial liquid velocity. VsG =14 m/s........................................................................................................... 161 Figure 4.22: Comparison of presentdata with data of Azzopardi et al. (1991)......... 165 Figure 4.23: Comparison of presentdata with data of Azzopardi et al. (1991)......... 166 Figure 4.24: (a) Interfacial shear stressvariation with gas superficial velocity 167 ........ Figure 4.24: (b) Wall shear stressvariation with gas superficial velocity 167 ................ Figure 4.24: (c) Wall shear stresstime seriesVsG= 42m/s, VSL= 0.15m/s 167 .............. 167 Figure 4.24: (d) Wall shearstresstime seriesVSG= 31m/s, VSL= 0.05m/s ............. Figure 4.24 (e): Photo evidence of packets of ring disturbance waves occurring when 30 m/s. The corresponding time varying film gas superficial velocity exceeds thickness/ disturbance waves (top left) where the corresponding higher drop time = 5.85 seconds(top right) are sourced............................. 168 concentration at Figure 4.25 (a):Time series of flow variable showing flow structures ..................... 171 Figure 4.25 (b): Time series of drop concentration showing flow structures...........172 Figure 4.25 (c): Time series of Interfacial and wall shear stress.............................. 173 Figure 4.26: Comparison of drop velocity data baseand present study.................... 174 Figure 4.27: Film velocity as a function gas superficial velocity ............................. 176 Figure 4.26: Comparison of film velocity with the model of Alves et al. (1991)...... 176 Figure 4.29: Changes in drop size distribution tracked by the Probability Density Function. There is a change from multiple to single peak after gas superficial 30 m/s marked transition to annular mist flow as D50 (MMD) velocity of 179 decreases.VSL= 0.05 m/s ................................................................................. Figure 4.30: Time series and PDF of data showing bi-modality at VSG= 14.08 m/s, VSL = 0.19 m/s. Film thickness was acquired from conductance probes, drop from Spraytec and pressuredrop form differential pressurescell. 181 concentration Figure 4.31: Proposed flow pattern map showing modification to wispy annular transition boundary of Hewitt & Roberts (1969) and the new transition boundary to mist annular flow. The open symbols representthe position of the presentstudy 184 the new flow map on ........................................................................................ 187 Figure 4.32: Predicting of mean drop diameter using Rosin Rammler model.......... Figure 4.33: Prediction of measuredentrained fraction data using existing correlations.
Figure 4.16: Comparisonof StandardDeviation of Film hold-up and Drop size (MMD) showing similar trendsas gas and liquid superficialvelocities increase. 149 ........................................................................................................................ 151 increase Figure 4.17 (a) Pressuredrop
xvii
List of figures
191 Azzopardi (1997,2006) .................................................................................... Figure 4.35: Normalized flow variables vs. gas superficial velocity 194 ........................ Figure 5.1: Annular flow schematicsshowing primary and secondarywave properties, Alves et al. (1991) 202 ............................................................................................ Figure 5.2: Schematic of a unit wave including forces resolution around a droplet.. 203 Figure 5.3 (a): Time-averaged film thickness information 205 ....................................... Figure 5.3 (b): Time-averaged drop concentration information 205 ............................... Figure 5.4 (a): Autocorrelation vs. time delay for huge wave 206 .................................. Figure 5.4 (b): Autocorrelation vs. time delay for disturbance wave 206 ....................... Figure 5.5 (c): Power Spectrum Density vs. wave frequency for huge wave............ 206 Figure 5.5 (d): Power Spectrum Density vs. wave frequency for disturbance wave. 206 Figure 5.6: Wave frequency as a function of gas superficial velocity 207 ...................... Figure 5.7: Wave frequency as a function liquid superficial velocity 209 ....................... Figure 5.8: Time averaged film thickness characteristic of huge wave 210 .................... Figure 5.9: Wave amplitude-Film thickness relationship. Superficial gas velocity increasestowards decreasingfilm thickness 211 ..................................................... Figure 5.10: Effect of gas superficial velocity on drop frequency 212 ............................ Figure 5.11: Effect of gas and liquid superficial velocities on the frequencies of disturbance waves on the film interface 213 ............................................................ Figure 5.12: Similarity between drop frequency and drop size distribution 215 ............. Figure 5.13 (a) : Relationship between the dimensionless wave frequency and the Lockhart-Martinelli parameter (the square root of the ratio of the superficial 217 momentum fluxes of liquid and gas) ................................................................ Figure 5.13 (b): Relationship between the dimensionless wave frequency and the
189 ........................................................................................................................ Figure 4.34: Comparisonof predicted and measuredSMD using the model of
219 Kaji (2008) ....................................................................................................... Figure 5.14 (b): Correlation of Strouhal number with Lockhart-Martinelli parameter, 219 Mantilla (2008) ................................................................................................ Figure 5.15: Strouhal number correlation for pipes of different diameter using gas 220 based Strouhal number ..................................................................................... 224 Figure 5.16: Collision field and coalescenceof drops in annular two-phase flow.... Figure 5.17: Comparison of experimental drop frequency with modal drop and wave 226 frequency ......................................................................................................... Figure 5.18 (a): Wave velocity variation with gas superficial velocity .................... 227 227 Figure 5.18 (b): Wave velocity variation with liquid superficial velocity ................ Figure 5.19: Effect of gas and liquid superficial velocities on gas core density....... 228 230 Figure 5.20: (a) Prediction of wave velocity by various models .............................. Figure 5.20: (b) Prediction of wave velocity by various model Vso = 14 m/s.......... 230 Figure 5.21: (a) Wave spacing variation with gas superficial velocity 231 ..................... Figure 5.21: (b) Dimensionless wave spacing variation with gas superficial velocity 231 ........................................................................................................................ Figure 5.22 (a) Wave spacing variation with liquid superficial velocity VsG = 14m/s 232 ........................................................................................................................ Figure 5.22 (b) Dimensionless wave spacing variation with liquid superficial velocity
Lockhart-Martinelli parameter(the squareroot of the ratio of the superficial fluxes of liquid andgas) 218 momentum ................................................................ Figure 5.14 (a): Correlationof Strouhalnumberwith Lockhart-Martinelliparameter,
List of figures
Figure 5.23: Normalized film thickness as a function gas superficial velocity 233 ......... Figure 5.24: Normalized wave amplitude variation with gas superficial velocity 234 .... Figure 6.1 (a) Interfacial shear stressas a function of gas superficial velocity 242 ......... Figure 6.1 (b) Film velocity as a function of gas superficial velocity 242 ....................... Figure 6.2 (c) Effect of gas superficial velocity on wave amplitude 242 ........................ Figure 6.2 (d) Effect of gas superficial velocity on dimensionlesswave amplitude. 242 Figure 6.3 (a) Film thickness time-series at low gas superficial velocity 244 ................. Figure 6.3 (b) Drop concentration time-series at low gas superficial velocity 244 .......... Figure 6.3 (c) Film thickness time-series at high gas superficial velocity 244 ................ Figure 6.3 (d) Drop concentration time-series at high gas superficial velocity 244 ......... Figure 6.3: (e) Characteristic drop diameter as a function of gas superficial velocity; 245 ........................................................................................................................ Figure 6.3: (f) Dimensionless wave amplitude vs. gas superficial velocity 245 .............. Figure 6.4(a): Wall shear stressprofile 246 ................................................................... Figure 6.4(b): Film velocity profile 247 ......................................................................... Figure 6.5: (a) Mean drop diameter vs. gas superficial velocity 248 ............................... 6.5: (b) Wave amplitude vs. gas superficial velocity Figure 248 .................................... Figure 6.5: (c) Normalized drop size vs. gas superficial velocity 249 ............................. Figure 6.5: (d) Normalized wave amplitude vs. gas superficial velocity 249 .................. Figure 6.6: (a) Mean drop size vs. gas superficial velocity 251 ...................................... Figure 6.6: (b) Wave amplitude vs. gas superficial velocity 251 .................................... Figure 6.7: Comparison of measuredwave height and the model prediction 252 ........... Figure 6.8: (a) Dimensionless pressuredrop vs. gas superficial velocity 253 ................. Figure 6.8: (b) Wave amplitude vs. gas superficial velocity 253 .................................... Figure 7.1: The flow diagram with t 257 test section -junction ....................................... Figure 7.2: Picture of the T -junction with conductanceprobes (Probe 2,3 and 4)... 259 Figure 7.3: Calibration curves for the vertical conductanceprobes 260 .......................... Figure 7.4: Calibration of the horizontal conductanceprobes 261 .................................. Figure 7.5: Cross sectional view of the ring type conductanceprobe 262 ....................... Figure 7.6: Time averaged void fraction for air/water and air-glycerol solution. Gas superficial velocity = 3.2 m/s, liquid superficial velocity = 0.12 m/s, Liquid 265 36.0 mPa s, Gas Take Off = 0.68, Liquid Take Off = 0.68 viscosity = .............. Figure 7.7 (a): Detection of Chum-Annular flow transition boundary at a gas superficial velocity of 15 m/s based on liquid hold-up variation within the system 266 superficial velocity increases as gas ................................................................. Figure 7.7 (b): Further analysis of the transitions within annular flow based on established information from drop size in chapter four ..................................... 267 Figure 7.8: Representation of test conditions on the flow pattern map of Taitel et at. (1980). Chum -Annular transition boundary agrees with churn-annular transition 269 observed in Figure 7.7 (a) and Figure 7.7 (b) ....................................................flow Figure 7.9: Comparison of air/water and air-viscous liquid data at the same conditions using liquid hold-up and PDF of void fraction. Gas superficial velocity = 3.2 m/s, liquid superficial velocity = 0.12 m/s. T-Junction Pressure= 1.4 bar. 271 ...................................................................................................................... Gas Take OR'= 0.68; Liquid Take Of = 0.68 271 ......................................................... 273 Figure 7.10: Use of Froude number to identifying flow pattern ............................... Figure 7.11: Variation of film momentum, fraction with gas momentum and entrained liquid viscosity. Gas superficial velocity = 15.2 m/s, liquid superficial velocity = 0.13m/s, liquid viscosities = 1.0,12.0,18.0 277 36.0 mPa s and .............................
xix
List offigures Figure 7.12: Variation of film thickness with liquid viscosity .................................. 277 Figure 7.13: Effect of increase in liquid viscosity on phase split at similar conditions. 279 ........................................................................................................................ Figure 7.14: Effect of increase in gas superficial velocity on the phase split for liquid 18.OmPas. Fraction of gas taken off = 0.56 ; fraction of liquid taken viscosity of 280 0.42 off = ......................................................................................................... 281 Figure 7.15 (a): Effect of liquid viscosity on phasesplit .......................................... Figure 7.15 (b): Effect of liquid viscosity on phase split. Comparison between 282 liquid air/water and air-18 mPa s viscous ......................................................... Figure 7.15(c): Effect of liquid viscosity on phase split. Comparison between air/water 283 liquids air-8 mPa s and air-18 mPa s viscous .................................................... Figure 7.16: Comparisons with prediction models for the phase split at aT -junction. Gas superficial velocity = 15.2 m/s, liquid superficial velocity = 0.13 m/s, liquid 284 18.0 mPa s viscosity = ...................................................................................... Figure 7.17 (a): Effect of physical property (Morton & Archimedes numbers) on 287 liquid hold-up .................................................................................................. Figure 7.17 (b): Effect of physical property (Morton & Archimedes numbers) on liquid hold-up 287 .................................................................................................. Figure 7.19: Variation of structure frequency correlated with Strouhal number with liquid phaseReynolds number 291 ......................................................................... Figure 7.20: structure frequency at side arm and main pipe as a function of structure frequency approaching the junction 292 .................................................................. Figure 7.21: Difference in structure frequency approaching the junction and structure frequency after separation at run arm plotted as function of difference in structure frequency approaching the junction and structure frequency after separationat the 293 side arm...........................................................................................................
xx
List of tables
List of tables
Table 2.1: coefficients for entrained fraction correlation of Oliemans et al. (1986) ...47 86 Table 3.1: Table of physical properties, Co and C1 for the test fluids ......................... 98 Table 3.2: Table of coefficients Co and C1 for different correlations ......................... 114 Table 4.1: Spraytec technical specifications ............................................................ Table 4.2 : Annular flow transition models, their respective superficial gasvelocity at the beginning of annular flow and experimental evidences of transition to co123 flow current annular ......................................................................................... Table 4.3: High speedcine movies/imagesrecorded during the experimental for conditions around transition to mist flow at gas superficial velocity of 30 m/s and 126 beyond ............................................................................................................. Table 4.4: Particle Size Distribution table for gas superficial velocity, VSG= 14.08m/s and liquid superficial velocity, VSL= 0.19 m/s................................................. 187 189 Table 4.5: Statistical results of models performance (24 data points) ...................... 190 Table 4.6: Statistical results of models performance (36 data points) ...................... Table 7.1: Fluid properties at different viscosities 276 ...................................................
xxi
Chapter 1
Introduction
Chapter 1
Introduction
1.1 Flow Assurance
With ever growing world population having improving quality of life, energy consumption is projected to increase year after year. The situation implies that hydrocarbon stocks, including both commercial and strategic reservesof lighter oils are running out. We are left therefore, with no robust alternative but to harnessheavy oil resources which present themselves as inevitable option to meet the future energy demand.Unfortunately, exploration and production of heavy oil and their associatedgas posesa lot of environmental and flow assurance challenges.
Flow is only assuredwhen the energy driving the flow is greater than the resistanceto flow. Flow assuranceis a term which encapsulates number of fluid flow, heat transfer, a issues that have important implications for the transportation production, and chemistry hydrocarbonsfrom reservoirsto processingfacilities, Pickering et al. (2001). of
In oil and gas production, deepwaterfields present similar flow assurancechallengesto those encounteredin the traditional shallow water developments.Many of the problems during the development of reservoirs located in deep waters, stem from the encountered characteristics of the reservoirs as some of them are located in turbidite sandstone formations with vast water depth. The depth of the formation below the seabedis often Usually, extraction requires intelligent completion and Floating Production, quite small. Storage and offloading Vessels (FPSO) as shown in Figure 1.1. These reservoirs are
MB ALAMU 2010
Chapter 1
Introduction
characterised by low-energy having relatively low pressures and temperatures compared to conventional reservoirs. Elf's Girassol and Dalia reservoirs in deepwater offshore Angola are good examples where development posed great flow assurance implications because not only is pressure for driving the flow limited, the low temperature implies greater difficulties with heat conservation and the avoidance of solid formation
Gos
Otl sediments
Figure 1.1: Schematic diagram of offshore operation with Floating Production, Storage and offloading Vessels (FPSO) and Dry Completion Unit (DCU).
Strand et al. (2000) reported some of the physical the British viscosity North sea production facilities.
ranging typically
The viscosity
Chapter 1
Introduction
In this hostile environment, design of production tubing and transportation systemfor the production of oil demandsspecial consideration on account of high fluid viscosity with special attention given to slug flow. This is primarily because of the liquid phase intermittency along the pipe during production. The piping system (e.g. as in Figure 1.1) must be able to withstand the fluctuating nature of the slug flow and at the sametime be able to accommodate the pressure drop associatedwith fluid movement in the pipe. Unmitigated and increasingpressuredrop may causefrequent shut-ins and inevitable loss of revenuefrom production. The increasein riser height in deepwaterdevelopment has an important implication for system stability. In particular, increasesin riser height can be destabilizing, causing multiphase slugging which in some cases might lead to shut-down of the facility. Increasing the riser height also increasesthe severity of slugging, measuredin terms of the magnitudeof pressurefluctuations and slug sizes.
MB ALAMU 2010
Chapter 1
Introduction
Taylor bubbles in slug flow, huge waves in churn flow, disturbancewaves in annular flow and wisps in wispy annular flow. The void waves, slugs, huge waves, disturbancewaves and wisps, have unique characteristic velocities and frequencies which can be used to classify them into flow patterns. Thus, information about this periodic structure is valuable to reconstructthe overall pressuredrop in any multiphase flow system.
The focus of this thesis is on these periodic structures. This thesis also examines dependence of flow structures dependence on flow rates, fluid physical properties and flow channel geometry.
In practice, phase distribution in a pipe is usually determined from periodic structure extracted from dynamic data logged by the sensor, when the sensorsc ans the flow structures during the passage of two-phase flow. From very detailed measurements, Sekoguchiand Mori (1997) have shown that more than one of these structurescan occur simultaneously. The void waves, slugs, huge waves, disturbancewaves and wisps, have characteristic velocities and frequencieswhich can be used to classify them into unique flow patterns. Flow pattern maps are generatedbased on information from two-phase flow periodic structures. Dependenceof the structure velocities and frequencieson flow link them with flow pattern transitions have been reported, rates and methods which Azzopardi (2004). This valuable information can be used to reconstruct the overall pressure drop in any multiphase flow system. Based on the uniqueness of the flow structure,flow patternsand flow pattern mapsare discussednext.
MB ALAMU2010
Chapter 1
Introduction
In vertical pipes, two-phaseflows can be classified into five major flow regimes:
"
" " " "
Bubbleflow
Slug flow Churn flow and Annular flow Dispersed-bubbleflow
Figure 1.2 shows common vertical gas-liquid two-phase flow normally encounteredin in the industry. Stratified flows do not occur in vertical gas-liquid two-phaseflow. pipes For horizontal flow, the patterns encounteredare shown in Figure 1.3 while Figure 1.4 indicates flow patternsfor the entire pipe inclinations.
5 MB ALAMU2010
lntrocluctiun
"
00
"'
0,0 0
"
" " "
00
" "
0 00 0 O p0 000 00 O O
0
0 C 0 0 "
"
0 0 O0 0 0
a) L
0
"
p0 A
Bubble
Slug
Churn
and inclined
Annular
DispersedBubble
homogeneous through the pipe cross section. Bubble flow occurs at approximately low liquid rates, with low turbulence, and is characterized by slippage between relatively liquid-phases, resulting in large values of liquid holdup. the gas-and the
Al B ALAAR' 0/U
Intrudm tion
is located in a large bullet-shaped gas pocket termed Taylor-bubble with a phase diameter almost equal to the pipe diameter. The flow consists of successive Taylor
bubbles and liquid slugs, which bridge the pipe cross section. A thin liquid film flows downward between the Taylor-bubble and the pipe wall. The film penetrates into the
following liquid slug and creates a mixing zone aerated by small gas bubbles.
StratifiedSmooth
cz
StratifiedWavy
ElongatedBubble
'..
Slug
"
""
""
""
Annular cn WavyAnnular
"
,IrOf
rr
ro
0r
rfr-p-
C'
rrrr
rr<,
noonr r
r'
ror,
rrr.
r"
r'rr po o
nnne r+
n'
DispersedBubble
Direction
of flow
pipes
AIBAL.
1%IL 2010
/rurudw tiai
-*
Stratified Slug Annular
DispersedBubble
Chum
- ---
Stratified
Slug
Falling-Film Annular
Chapter 1
Introduction
Out of the existing flow patterns,annular flow is the least understooddue to its complex Good knowledge of the conditions leading to atomization of part of liquid film nature. by gas flow is of considerablepractical importance for heat and masstransfer processes in two-phase flow systems.The mechanismsof mass, momentum, and energy transfer between the film and gas core flow is significantly altered by the inception of For example, the burnout and post-burnout heat flux in light-water cooled entrainment. the effectiveness of the emergency core cooling systems in water nuclear reactors, the performance of the film cooling of jet and rocket engines can be reactors, and
9 MB ALAMU2010
Chapter 1
Introduction
significantly
affected by the entrainment of the liquid film into the gas core flow.
Numerous other examples where the atomization and entrainment of the liquid film imposes operational and performance limitations could be found in various chemical engineering systems.
In the petroleum industry, gas well operatesin annular flow. However, liquid loading into the gas well and in the gas condensatelines has been identified as a common that lowers productivity of nearly all the gas wells in the world. Again the problem becauseof poor understandingof annular flow mechanisms.Design of problem persists knowledge of first-order parameterssuch as pressuredrop. processequipment requires Unfortunately, pressure drop can only be modeled successfully if and only if the behavior of entrained droplet in the gas core is understood. Annular flow forms the theme of this thesis. Contribution has been made to the existing annular flow central literature as detailed in ChapterFour.
10
MB ALAMU2010
Chapter 1
Introduction
homogenous no-slip. Figure 1.5 shows Dispersed-Bubble Flow as encountered in concentric annulus,among other flow patterns.
"r rr r"
it")
"
"r
'r.
Bubble Flow
DispersedBubble Flow
Churn Flow
Annular Flow
11 MB ALAMU2010
Chapter 1
Introduction
and determinations of flow pattern are carried out by visual observations. empirical Therefore, their applications are limited and they are only reliable in the range of conditions similar to those under which the data were acquired. Their extensionto other flow conditions is highly uncertain. Figures 1.6,1.7,1.8,1.9 and 1.10 show typical flow maps for vertical and horizontal flows. pattern
1.0
0.8
N
0.6
0.4 0.2 0 0 0.2 0.4 0.6 0.8 1246
vi 2
10 20 40
100
9d
Figure 1.6: Flow Pattern Map Vertical Flow (Griffith and Wallis, 1961)
12 MB ALAMU2010
Chapter 1
Introduction
loo (QPLCWA)
/4
(PA
y 10
1/3
QLG
........................................ Bubble
0.1
Churn
n V I V. ni
0.1
10 XvSG, ft/s
100
1,000
Figure 1.7: Flow Pattern Map for Vertical Flow (Govier and Aziz, 1972)
100
10
x
JQ . La
Q1
0.01 0.01
0.1 Superficial
1 Gas Velocity,
10 m/s
100
Figure 1.8: Flow Pattern Map Vertical Flow (Taitel - Barnea - Dukler (1980)).
13 MBALAMU2010
Chapter 1
Introduction
10
...........................
-"-"--
..................
Annular
0.1
-------------------
- -------------
Stratified-Smooth
0.01 n nn7
0.02
f' /s
0.1 1
VssG , rn/s
10
100
Figure 1.9: Flow-pattern map for horizontal flow (Mandhane et al., 1974) 10
i
Annular StatifedWavy
Dispersed-Bubble ................................
104
....
102
10"1
100
101
102
103
104
14
MB ALAMU2010
Chapter 1
Introduction
in annular flow. Dynamic and time-resolved measurements drop structures of entrained liquid fraction were carried out and reported for the first time size and diffraction-basedinstrument. using 3.5mm T-junction experiments,these involve partial separationof two-phaseflow T-junction. Air/water, and air-glycerol solutions of various at a vertical were used to investigate effect of liquid viscosity on the gas and concentrations the liquid take-offs.
The aims of these studiesare statedas follows: 1. To use conductanceprobe techniqueto acquire void fraction data in order to identify the effect of liquid viscosity on periodic flow structures in pipe flows ( 5mm and 19mm ID pipes) andT -junction experiments. 2. To use laser diffraction principle to obtain time-resolved drop sizes and entrained fraction in annular two-phaseflow.
MB ALAMU 2010
is
Chapter 1
Introduction
1. To investigate the effects of liquid viscosity periodic structures on gas-liquid pipe flow and phasesplitting at vertical T-junction. 2. To use characteristicsof periodic structuresto identify and classify flow patterns. 3. To establish relationship betweenwave (liquid film) and drop frequency by taking of measurements film and drop concentration. simultaneous 4. To use drop size/entrainedfraction information to identify and classify flow pattern transition boundariesin annular flow regimes 5. To establishrelationship betweenpressuredrop and entrained liquid fraction
16 MB ALAMU 2010
Chapter 2
Literature Review
Chapter 2
Literature Review
Slug and annular flow literature are predominantly reviewed in this section. Slug flow is discussedfirst and then review of relevant work on annular flow follows. As set out earlier in the objective of this study, more effort is directed to annular flow in order to better understandcomplex nature of the dispersedphaseperiodic structures.
Slug flow hydrodynamics is complex with characteristic unsteady flow behaviour, hold-up and pressuredistributions. There has been no consensus yet structure velocity, for predicting frequency of the periodic structure that occurs in on a unified correlation flow. One of the identified causesof the divergent views on slug frequency model slug is the paucity of air-viscous liquid data. Most of the existing correlations are basedon data whereas periodic structures in slug flow behave differently when air/water carried out with liquids more viscous than water. experimentsare
literature is awash with publications on disturbance waves and its Although the in gas-liquid two-phase flow the mechanism of atomization of part of the evolution liquid film to form drops in annular two-phaseflow is still not entirely understood.It has been observed that drop creation only occurs when there are large disturbance waves the film interface. Woodmanseeand Hanratty (1969) observedthat ripples on presenton these waves were a precursor to drops. Though it has been reported that drops occur in havebeen bursts Azzopardi (2006) all previous drop size or concentration measurements
17 MB ALAMU2010
Chapter 2
Literature Review
time integrated. In subsequentsubsectionsrelevant literatures are reviewed along with descriptionsof flow patternsbasedon the periodic structuresassociatedwith them.
18 MB ALAMU2010
Chapter 2
Literature Review
., m o O R O
OO
Oo O co C9
O O0O
000
Figure 2.1: Schematics of slug unit in horizontal slug flow. Xiaodong (2005) used a numerical approachto simulate the flowing field around Taylor bubble with a view to determining Taylor bubble configuration in vertical slug flow. He the mechanistic model of Zhang et al.(2003) to obtain the liquid hold-up in modified in which the average diameter of small bubbles is estimated by the wake region, Hinze (1955).The length of wake region is set at value which can give the correlation of fraction of 0.37 in the slug region. This value comes from the assumptions averagevoid highest void fraction is 0.52 and the empirical value is 0.22 when the wake that the for fully developed.He recommended further studiesthat surfacetension and region gets be included for the determination of configuration of Taylor bubble. viscosity effects Bonizzi and Issa (2003) investigated gas bubble entrainment in two-phase horizontal flow. A model of gas bubble entrainment into slug was proposed in order to slug improve the accuracyof prediction of slug characteristics.It was found that more studies for different pipe configurations and flow conditions to improve an are necessary important factor in determining characteristicsof slug flow. Issa et at. (2006) proposed improved closure models for gas entrainment and interfacial shear for slug flow in two horizontal pipes. The models were validated against experimental data. They however, detailed studieson high pressureand high gas velocities. suggested more
19 MB ALAMU 2010
Chapter 2
Literature Review
Van Hout et al. (2002) investigated experimentally the translational velocities of elongatedbubbles in continuous liquid slug for various flow rates, pipe inclinations and pipe diameters.Measurements were carried out by cross-correlatingthe output signals of optical fibre probesand by image processtechnique. consecutive The velocities of single elongated bubbles in a stagnant and a flowing liquid were for comparison with experimental results of elongated bubbles in continuous measured flow case. The results show that measuredvelocities of single elongatedbubbles slug in all casespredicted quite well by correlations while the velocities in continuous were flow, for certain cases, were considerably under predicted. This discrepancy is slug to the presenceof dispersedbubbles in the liquid slug. They therefore proposed ascribed to accountfor this discrepancy. a model
2.1.1.1 Structure Velocity, VT Slug flow in gas-liquid two-phase flow can be described by Taylor (elongated)bubble by liquid slugs which in most cases may contain dispersed bubbles. A separated for slug flow requires information about this Taylor bubble with respect complete model to their propagation or translational velocity (VT) as shown in Figure 2.2.
20 MB ALAMU 2010
Chapter 2
Literature Review
Taylor bubble
LF
Liquid slug
LS
Figure 2.2: Schematic Diagram of slug flow parameters, Shoham(2006). Nicklin et al. (1962) performed an experiment to determine translational slug velocity 25.4 mm ID using water/air as test fluid. They concluded that absolute with a pipe slug can be represented by the expression: velocity of a
(2.1)
the expression for the translational velocity of Taylor bubble as: and proposed
VTB = Co VM + Cl VD
(2.2)
Or
21
MB ALAMU 2010
Chapter 2
Literature Review
VTB =
CoVM
+ VDrift
(2.3)
The value of the coefficient Co,of the mixture velocity VM; and C1,of the drift velocity, VD can be determinedexperimentally. It is generally assumedtheoretically that by translationalvelocity is a linear function of mixture (slug) velocity as expressed Equation (2.3). For laminar flow the ratio Vmax/Vmean approaches 2.0 and indeed there is strong indication that Co increasesas the Reynolds number decreasesand reachesa value of 2.0. A more precise theory showsthat Co,for laminar flow, equals 2.27 as reported about in Taylor (1961); Collins et al. (1978), for the case where the surface tension is The exact value of Co for turbulent flow, and in particular for laminar flow, is neglected. inconclusive.
Taitel & Barnes (1990) stated that drift velocity is expected to depend on the liquid or the bubble Reynolds number. Zukoski (1966) however, shows that the viscosity, dependence the drift velocity on viscosity is negligible for Reynolds number Re > 300 of where: DVSGPA PL (2.4)
Re=
22 MB ALAMU 2010
Chapter 2
Literature Review
Where:
Re = Reynolds number, [-], Vs0 = superficial gas velocity, [m/s], D= pipe diameter, [m]
L = dynamic liquid viscosity, [kg/m-s], PL= liquid phasedensity, [kg/m3] , Davies and Taylor (1950) performed a potential flow analysis around the nose of the bubble and obtainedthe following expression: elongated VD = C, gD
(2.5)
The drift velocity of long bubbles moving in a liquid at rest dependsprimarily on the force that creates the drift, i. e. gravity. However other forces such viscous or surface tension may have secondaryeffect; therefore many authors have studied the influence of the physical properties of the operating fluids and the pipe inclination to the drift coefficient C1. The dependencyfound is expressedthrough the general relationship given by Zukoski (1996) as:
(2.6)
gtp S= PL
Fj,`U7--
(2.7)
Nf =D2
(2. 8)
23 MB ALAMU 2010
Chapter 2
Literature Review
WhereNf is the dimensionlessinverseviscosity number. L is the liquid viscosity, [mPa s or kg/m-s ], PL is the liquid phasedensity, [kg/m3].(3is the pipe inclination, and Eo is the Eotvos number, which is the ratio of gravitational to the surfacetension forces
\pgD 0 Q (2.9)
For small diameter pipes, Weber (1981) showed that the surface tension is important as its increaseconsiderably decreasethe drift velocity. He proposed a correlation basedon Zukoski (1966) data using the following relation: 76 1.76 Cl =0.54 - .
a
(2.10)
Bendiksen (1984) investigated large bubbles in inclined pipes and suggested two different regimes of bubble motion, and therefore confirmed the existence of Ferre (1979) first transition criteria in horizontal pipes. He suggestedliquid based Froude to track the transitions between these two bubble regimes. The Froude number number be expressed as: can VSL 9D (2.11) Where: JL= superficial liquid velocity, [m/s], FrL= liquid Froude number, [-], due gravity, [m/s2],D= pipe diameter, [m]. g= acceleration
FrL -
24 MB ALAMU 2010
Chapter 2
Literature Review
He suggestedfurther this transition might occur at a critical value of 3.5 and validated his model for all pipe inclinations as expressed below: ........... <3.5 FrL FrL > 3.5 (2.12)
C_0.54cos+0.35sin, 0.35sin
Theron (1989) proposed a continuous form of Equation (2.12) from Bendiksen (1984). This is particularly useful to avoid problem of discontinuity at translational Froude during numerical calculations. number Theron's idea can be expressed as:
C,=(-0.5+
cos+0.35sin
(2.13)
r=1. o+
fr-"-Cos
while:
Vu
FrM =
gD
(2.15)
2.1.1.3 Mean Bubble Motion Coefficient/Flow distribution Coefficient, Co The parameter Co in the bubble translational velocity equation is a distribution coefficient related to the velocity profiles in dispersed systems. It may be closely
25 MB ALAMU2010
Chapter 2
Literature Review
by the ratio of the maximum to the mean velocity in the liquid aheadof approximated the bubble. The exact value of Co is not clear for all flow conditions and reported experimentalvalues spreadconsiderablyfrom 1.02 to 1.35. Hale (1994) attributed this to the fact that the plot of translational velocity VT versus the mixture velocity VM is not linear, but insteadbendsvery slightly upwards,Bendiksen (1984). perfectly Nicklin et al. (1962) suggesteda constant of 1.2 for fully developed turbulent flow, but the variation of this coefficient at low Reynolds numbers (< 8000). noticed From various expressions found in the literature, the parameter Co can be expressed through the following generalexpression: Co =Co(Fr., Re, /j)
(2.16)
Fri, =
gD
(2.17)
And:
Re, =
Experimental result of Ferre (1979) suggested that the discrepanciesin the coefficient Co be due to the occurrence of a flow transition. He found two different critical might
26 MBALAMU2010
Chapter 2
Literature Review
Froude numbers at which the values of the drift velocity change, and proposed the following expression: 1.10........... 53.5 FrL Co = 1.30 2.26 < Fr. <8.28 ......... 1.02............ Fr, z 8.28
(2.19)
Similarly, Bendiksen (1984) while investigating motion of a single bubble in pipes that there was a critical Froude number at which the values suddenly changed. noticed As a result, he suggested that the following criteria should be used for all pipe inclinations:
Co -
Theron (1989) included the effect of the Froude number on the distribution coefficient Co,and presenteda simple continuous form of the correlation which is valid for pipes of inclinations. This is particularly useful to avoid problem of discontinuity at all Froude number during numerical calculations. translational 0.23 Co =1.3+0.13sin2g r
(2.21)
6 r=1.o+ Fr cos,
(2.22)
27 MB ALAMU 2010
Chapter 2
Literature Review
while:
VM
FrM =
gD
(2.23)
Recently, Petalas& Aziz (1998) usedthe Stanford Multiphase Flow Databaseto account for the effect of the slug Reynolds number on the coefficient Co,and empirically derived the following correlation: 1.64+0.12sin Co = 3' Reo. 9
(2.24)
Taitel and Barnea (1990) proposeda unified model considering the slug hydrodynamics for vertical, horizontal and inclined pipes. Several approaches for hydrodynamics of liquid film were investigated. Two methods were developed to predict pressuredrop The pressuredrop can be calculated using a force balancealong a slug acrossa slug unit. for the liquid slug zone. unit or only Zhang et al. (2003) developed a mechanistic model for slug flow based on a balance between turbulent and kinetic energy of the liquid phase and the surface free energy of the dispersed gas bubbles. The model predicts transition boundary between slug and dispersedbubble flows for all inclinations.
28 MBALAMU2010
Chapter 2
Literature Review
Bubbles with a chord length larger than the pipe diameter D are considered Taylor bubbleswhile bubbleswith smaller than this are referred to as dispersedbubbles. In similar manner,the translational velocity for dispersedbubbles in the liquid slug VB is a linear combination of the bubble drift velocity VDr; and the mixture velocity in the ft VM. slug zone VB = CoVM VDrift + (2.25)
Unlike the caseof Taylor bubbles, the translational velocity and the gas velocity are the same for small bubbles. For vertical case VDriftis the drift velocity of a bubble in the For very small bubbles the bubbles behaveas rigid spheresand the free rise is pipeline. by Stokeslaw. For larger bubble a boundary layer solution is applicable. When governed the bubble size exceedssome critical value, dcD,the rise velocity of the dispersedbubble tends to be constantand independentof the bubble diameter. The bubbles in the slug zone are usually larger than dcD For a relatively large and . deformable bubble, the equation proposed by Harmathy (1960) for the bubble rise is consideredto be sufficiently accurate: velocity (PL - Pc) a9 VDrift = 1.53 P2
0.25
(2.26) This equation shows that free rise velocity is independent of the bubble size. The free a bubble within a swarm of bubbles is lower than the free rise of a single rise velocity of
29 MBALAMU2010
Chapter 2
Literature Review
bubble. This can be viewed as the decrease buoyancy that acts on a single bubble in a of gas-liquid mixture. is This decrease correlated in the form:
Vo = VDrift (1.0 S)n
(2.27) For relatively large bubbles, Wallis (1969); and Govier & Aziz (1972) suggested use the Fernandeset al. (1983) used n=0.5. A value of n=0 was recommended by of n=1.5. Wallis (1969) for a region termed churn-turbulent. This is very close to the flow of bubbles in the slug region and, thus, the value n=0 is recommended.
The value of Co dependson the concentrationdistribution of bubbles in the liquid slug as demonstratedby the method of Zuber and Findlay (1965). Wallis (1969) points out that Co for vertical dispersed flow "usually lies between 1.0 and 1.5 with a most probable of about 1.2". value
2.1.2 Transition
Bamea (1986) suggesteda combined criterion for transition to dispersed-bubbleflow to occur by the expression:
dmax5duo, dm 5dcB
(2.28)
30 MB ALAMU 2010
Chapter 2
Literature Review
Where:
0.6
dm
(0.75 _
+ 4.15 -%1-a
(e)-0.4 PL (2.29)
a=
duo =2l
0.4Q
; - PG Jg L
(2.30)
dcB
PL
fM VA gcos9
8 (PL -Pc)
(2.31)
d,,, = the maximum stablediameter of the dispersedbubbles [m] , dCD= the critical bubble diameter,above which bubbles start to deform [m] and coalesce, dcB= critical bubble diameter, [m] The combined criterion ensuresthat neither agglomeration of bubbles nor `creaming' flow exists. The model of Barnea (1986) is only that the dispersed-bubble will occur and for gas void fraction of 0.52. The value of a=0.52 valid represent the maximum
bubbles above which bubble is not possible, yielding a transition to possible packing of slug flow.
31
MB ALAMU 2010
Chapter 2
Literature Review
In conclusion, basedon the extensiveliterature review on modeling of the velocity of the periodic structure in slug flow, no consensushas been reached on the value of flow distribution coefficient, Co, and the coefficient of drift velocity term, C1, in estimating flow structurevelocity. The gap identified in the review above forms the basis for experimental campaign described in Chapter Three using air/water, air/10 mPa s and air/12 mPa s viscous liquids with a view to determining experimentally the value of Co and C1.
Shoham (2006) has identified deep understandingof the following flow parametersas the key to modeling annular two-phaseflow. The parametersare: " " " Interfacial shearstress. Fraction of liquid entrainedas droplets in gas core. Liquid film thickness
32 MB ALAMU 2010
Chapter 2
Literature Review
"
9
In this study our investigations are limited to the first three parameters.Since the thrust this work is directed towards annular flow, it is therefore imperative to review of literature on annular flow. Fortunately, Shoham (2006) and Mantilla (2008) pertinent have published detailed and excellent literature review including past and recent efforts be referred to in the next discussion. which would
2.1.3.1 Interfacial Waves in annular flow Interfacial waves are very important in two-phase flows as they play a central role in the transfer betweenthe gas and liquid phases.Waves are recognized mass and momentum the source of entrained droplets, Woodmansee(1968). Most of the studies published as interfacial waves have beenperformed in horizontal channelswhere the curvature of on the pipes is eliminated, providing better conditions for observations and measurements. Wave studies have dealt with identification of wave types and the prediction of the between them, Andritsos (1986). Also, the predictions of the conditions at transition interfacial waves are unstable leading to slug flow or atomization have been which the considered.
Several studies on annular flow for both vertical and horizontal pipes have been The main characteristicsof interfacial waves, such as the velocity (celerity), presented. frequency and wavelength and also geometrical parameters including the height,
33
MB ALAMU2010
Chapter 2
Literature Review
base length have been presented in various studies including amplitude and wave Chopra (1982), Martin and Azzopardi (1984), Azzopardi (1986), Azzopardi (1997), Wolf et al. (2001), Fukano et al. (1983), Jayanti et al. (1990), Paras and Karabelas (1991b), Andritsos (1986), Paraset al. (1994)) etc. More recently, studieshave presented obtained under normal and microgravity conditions (Han et al. wave characteristics (2006)). Waves in stratified two-phaseflow have also been studied extensively, dealing the predictions of the conditions at which the waves become unstable mainly with through linear stability theory (Hanratty and Hershman (1961), Lin (1985), Bruno and McCready (1988), Andreussi et al. (1985), Uphold (1997), Kuru et al. (1995) among others). Models published in the literature attempt to predict wave characteristicsutilizing non linear theory especially for roll waves. These models predict the wave celerity, parametersbased on the Two-Fluid Model equations wavelength and wave geometrical (Dressier (1949), Miya et al. (1971), Watson (1989), Johnson et al. (2005)). Another is the "shallow water" theory, which aims at reducing the Two-Fluid approach used Model equations, based on the fact that the film thickness is small as comparedto the length (wavelength). Most of the models are 1-D, but a 3-D model axial characteristic has been published by Pols (1998), basedon shallow water theory for the prediction of disturbance wave characteristicsin horizontal pipes. The use of 1-D models for roll or horizontal annular two-phase flow should be based on the determination of average for the wave characteristics,since the film could be asymmetric with respect conditions In 2-inch pipes the film becomes symmetric for superficial to the pipe circumference. 50 m/s at nearly atmosphericconditions. gas velocities above
34 MB ALAMU 2010
Chapter 2
Literature Review
For lower gas velocities, the asymmetry of the film can be easily observedbecausethe disturbancewaves (rings) exhibit an angle with the vertical plane.
In the following a summary of the important publications in the literature regarding techniques, entrainment correlations, stratified, annular flow and wave experimental modeling.
2.1.3.2 Relevant Literature Important studies in annular flow have documentedand presentedin Azzopardi (2006) Shoham(2006). Some of thesedetailed documentationsare repeatedhere to identify and knowledge is deficient. In their respectivebooks, Wallis (1969), Hewitt & Hallwhere Taylor (1970) presentedgeneral discussionson annular flow. The early study by Dukler (1960) attempted to develop a model for annular falling-film in vertical pipes. Later
Hewitt (1961) extended that work to cover upward vertical flow. More recently, other for vertical annular flow have beenpublished by Oliemans et al.(1986), Yao and models Sylvester(1987), Hassan and Kabir(1988), and Alves et al.(1991). Hassan and Kabir (1988) developed a comprehensivemodel for predicting two phase flow behavior in The model includes analysis of annular flow using previously established wellbores. Yao and Sylvester (1987) presenteda simple annular flow model for prediction methods. film thickness and pressure drop. The Oliemans et al. (1986) and Alves et al. of the (1991) models are basedon the two- fluid model, with closure relationship correlations for the interfacial friction factor and entrainment fraction. Finally, Caetanoet al. (1992) for annular flow in an annulusconfiguration. presenteda model
35 MB ALAMU 2010
Chapter 2
Literature Review
Extensive studies have been carried out on the physical phenomena associatedwith annular flow. Turner et al. (1969), Ilobi and Ikoku (1981) studied the minimum gas velocity required for liquid removal in vertical pipes. Taitel et al.(1980) adopted this concept to predict the transition boundary to annular flow. Several investigators developed correlations for the interfacial friction factor, including Wallis (1969), Henstock and Hanratty (1976), Whalley and Hewitt (1978), and Asali et al.(1985). Others such as Wallis (1969), Hanratty and Asali (1983), and Ishii and Mishima (1989) have studied the entrainmentprocess.Oliemans et al. (1986) presenteda comprehensive both interfacial shear and entrainment and developed their own correlations review of for three parameters. Becauseof the physical nature of Alves et al. (1991) annular flow model, it will be used Chaptersto validate experimental results of this study. frequently in subsequent
36 MB ALAMU2010
Chapter 2
Literature Review
. ,
(a)
p+dp SL
AF ptdp
.If
AG
{5
1
=wLl
rwr
1 =,
(h)
, g sin()
=,
Figure 2.3: Schematics of annular flow two fluid model approach, Alves et al. (1991).
2.2.1 Momentum
Equations
The model is derived by applying the momentum balances to the liquid film and the Note that because the flow is considered incompressible at a given core, respectively. location, the rate of change if momentum is neglected, resulting in reduction of the
37 MB ALAMU 2010
Chapter 2
Literature Review
balancesto force balances.The momentum (force) balancesfor the liquid momentum film and core are given, respectively, by:
-TwLAL +TIT,
FFF
- `J)
-PL9sin8
=0 (2.32)
And
S,
-T A - `dL)
(dPl
For fully developed flow, the pressure gradient in the film and core are equal. Thus, Equation 2.32 and Equation 2.33 can be combined by eliminating the pressuregradient, in the combined momentumequation for annular flow, given by: resulting L +rlsl(I AF +I
Fc
) -(Pc-Pc)Ssine=O. (2.34)
-z
AF =m5L(d-5L )
S, =; r(d -28L)
And SL _i. (2.35)
The hydraulic diametersof the liquid film and core are given, respectively, by:
38 MBALAMU2010
Chapter 2
Literature Review
and dc = (d - 28, )
(2.36)
(2.37)
(2.38)
The core velocity is determined in a similar manner. The flow rate in the core is given
by
qc = 4c + qLfE
= AP
(2.39)
and the core superficial velocity and actual velocity are, respectively,
+ VSLfE Vsc _ VSG (VSG +VSLfE)d (d - 25L )Z (2.40)
and v_
(2.41)
39 MB ALAMU2010
Chapter 2
Literature Review
Assuming homogeneousno slip mixture of the gas- phaseand the entrained droplets in the core, the core void fraction is determinedas
a _ vsa +VSLAE VSG (2.42)
the core density and viscosity are given, respectively, by and ac Pc = PG + Pc(1- ac) ; c = PG + L(1- ac) ac (2.43)
As the flow in the core is homogeneousno- slip flow, the gas velocity is equal to the velocity i. e. vc = vc core Thus, the total void fraction of the flow can be determined from the relationship /vG, or also from geometrical consideration, either of which yield the same aT = VSG result
L)2 aT =a(1-2
(2.44)
and
Note that the liquid - film wall shearstress,z , is determined on the basis of ReF. The Reynolds number is defined as core superficial
Re., = pcv. cd Pc
(2.46)
40 MBALAMU2010
Chapter 2
Literature Review
Where fF is the film friction factor. For a smooth pipe, the Blasius equation can be as given by: used,
fF = CF Re F
(2.48)
Where CF = 16 for laminar flow and 0.046 for turbulent flow, and the exponentn is 1 for laminar flow, and 0.2 for turbulent flow.
(2.49)
I= 1+300
dL
(2.51)
41 MB ALAMU2010
Chapter 2
Literature Review
frequent operational shut-ins and re-starts to carry out remedial or intervention assignments the well and production lines alike. This can lead to loss of revenuedue on
to production impairment.
Liquid film extraction technique based on mass balance appearsto be the commonest way to determine entrained liquid fraction in annular two-phase flow. and most reliable The practicability of this approach is rather doubtful in field application as a result of operational constraints.Therefore, producersare compelled to basethe determinationof liquid fraction in natural gas stream on the available empirical correlations. entrained Some of these correlations lack physical basis because their derivations were not Sometimestheir predictions can be misleading. mechanistic. Measurementof entrained liquid fraction employing laser diffraction techniquesusing Spraytec presents itself as a robust, cost effective alternative to close the gaps and deficiencies in knowledge caused by the traditional empirical models. Another the present arrangementis that it can be incorporated into the well head in advantageof part of intelligent completion. offshore platforms as
42 MBALAMU2010
Chapter 2
Literature Review
droplets in the core as well as drop concentration should not be detachedfrom liquid loading as the following discussionpresents.
Liquid loading is a common operating problem in gas production. The producer is forced to shut down production after the liquid content of the gas has exceeded a permissible limit as a result of liquid loading. Revenue is lost as liquid loading continues to impair The problem can be conceptualized as follows. production.
In a producing gas and gas condensatewell, as reservoir pressure decreases, entrained liquid forms an increasing restriction on gas production becauseliquid (water, oil, and is usually produced simultaneouslywith the gas production, the flow pattern condensate) inside the production tubing being annular dispersedtwo-phase flow. The liquid phase flows partly as a wavy film along the pipe circumference, and partly as entrained droplets in the turbulent gas core. At the end of the lifetime of gas wells, the gas decreases strongly. Due to this decrease,the drag force of the gas phase production rate the liquid phasemay not be sufficient anymore to bring all the liquid to the exerted on Liquid starts to drain downward (flow reversal). In such a situation, depending surface. the liquid could accumulatedown hole, block the inflow on the gasreservoir conditions, into the production tubing and gas production could cease. This phenomenon called liquid loading, occurs at a gas rate below the minimum in the pressuregradient curve, is closely related to flooding, Westende al. (2007). et and In the gas producing industry, the onset of liquid loading is commonly predicted using a developed by Turner et al. (1969). The idea behind this correlation is to correlation velocity that can elutriate the largest droplet, present in the estimate the minimum gas in the production tubing gets below this minimum gas core. When the gas velocity
43 MB ALAMU2010
Chapter 2
Literature Review
velocity, liquid loading occurs. It is, thus, implicitly assumedthat the dispersedphaseis causingliquid loading, although direct evidencefor this is not available, Westendeet al. (2007).
Belfroid et al. (2008) analyzed behavior of different natural gas wells based on both production data collected over a period of time and the numerical simulations results using both commercially available software and dedicated dynamic models. The study others confirmed many wells start liquid loading at gas rates well above the and many predicted by classic steady state models such as Turner while the liquid loading values is strongly dependent on inclination angle, flow regime transitions and the point interaction between tubing outflow behavior and the reservoir Inflow Performance Relationship (IPR).
Westende et al. (2007) performed detailed Phase Doppler Anemometry (PDA) in annular flow with air/water in a 50mm internal diameter vertical pipe measurements to understand the role of dispersion in an annular flow close to liquid loading. Probability Density Function (PDF) of the drop diameter, drop concentration, pressure of entrained fraction were determined from time varying gradient and amount in order to establish distribution of dispersed phase hydrodynamic measurements In that study, drop size and entrained fraction data were not time resolved. properties. Other differences between the presentstudy and the work of Westendeet al. (2007) are discussedas follows. Some of the liquid superficial velocities tested in the presentstudy higher than in the work of Westende al. (2007). Whilst we employ laser diffraction et are
44 MB ALAMU 2010
Chapter 2
Literature Review
technique to size liquid droplets and entrained fraction only; Westendeet al. (2007) utilized PDA to determineliquid film thickness,drop sizesand pressuredrop.
2.3.3 Entrainment
fraction correlations
Entrainment fraction in annular flow is defined as a fraction of the total liquid flow flowing in the form of droplets through the central gas core. Its prediction is important for the estimation of pressuredrop and dry-out in annular flow. In gas production it is important to determine liquid content of the natural gas stream. The most widely used fraction correlations for vertical annular two-phase flow include Wallis (1969), entrained Oliemans et al. (1986), Ishii and Mishima (1989) , Pan and Hanratty (2002a) and Sawantet al.(2009). Thesecorrelations are presentedas follows: recently
45 MBALAMU2010
Chapter 2
Literature Review
(2.52)
=104
One limitations of this correlation according to Wallis is that the correlation give errors for entrainment fractions greater than 0.5. It is also not valid for low liquid Reynolds
(2.54)
46 MB ALAMU 2010
Chapter 2
Literature Review
The values of the coefficients are given in Table 2.1 . Table 2.1: coefficients for entrained fraction correlation of Oliemans et al. (1986)
Coefficient o
Value -2.52
i
1.08
02
0.18
P3
0.27
04
0.28
P5
-1.80
P6
1.72
P7
0.70
PS
1.44
P9
0.46
(2.55)
Since the criterion for the entrainmentfraction calculation is basedon a force balanceon is limited to liquid Reynolds numbers greater than 2 for vertical a wave, the correlation
47 MB ALAMU2010
Chapter 2
Literature Review
downward flow, and 160 for vertical upward and horizontal flow for low viscosity liquids. These limits mark the initiation of roll waves. The correlation is limited to
air/water systems.
-VG")ZD a
PLPc
(2.58)
is The term VGc1 the critical gas velocity at the onset of atomization, given by
VGcr
=40
DPcPG
(2.59)
48 MB ALAMU 2010
Chapter 2
Literature Review
(2.60) According to the authorsthe correlation is only valid for mist annular flow regime. Sawantet al. (2008) is given by the following expression: ) EF = EMAX tanh(aWe'ZS Where : EMS=1Ra=2.31x10-4Rei. 35eL ; ReLMLn 250 ln(ReL) -1265 = (2.61)
o25
PcVscD OP We. Pc
ReL =
PLVsLD
PL
49 MBALAMU2010
Chapter 2
Literature Review
d32
p Rem We b =1.91 D Pc
+ 0.4E
ys` Vsc
(2.62)
Where,
We =
y2 PSG D
a
'[-];
Rem =
Pc ysc DE=
PG
MLE
PL VSL
IH.
The first term in the RHS of the Equation (4.11) is the contribution from drop break-up the secondterm represents contribution from drop coalescence. while Another simple model was suggestedby Azzopardi (2006) for vertical annular twoflow. It is expressed follows: as phase
za2
PLVSL
(2.63)
PSCVSG
PscVsc
In conclusion, a lot of work has been carried out in annular flow on entrained fraction drop size prediction including information about the disturbancewaves that prediction, drops. So far, there is no reliable information on the inter-relationship between create drop concentration fluctuation and disturbance wave frequency. This is partly due to data and non-existenceof time resolved drop data. scarcity of
MBALAMU2010
50
Chapter 2
Literature Review
2.5 T-junction
literature relevant
Works related to T-junction and its application has been reviewed by Mak et al. (2006). Based on the equation of mass, momentum and energy balance, Dionissios (2007) the dynamic separationof gas and liquid inside a T-junction with a horizontal studied branch. He proposed a mathematical model which predicts phase run and a vertical distribution and pressure drop across the T -junction of different orientations and pipe diameters taking into account the flow pattern' before the junction. The model was by good agreementbetweenmeasuredand predicted data. Adechy et al. (2004) validated CFD to simulate dispersedcore flow simultaneously with the flow of liquid film used the walls in which the core is representedas a dispersed two-phase mixture and along the liquid film modelled as a thin boundary layer. They reported predictions data which measurementsfor a certain range of phase split. Wang (2008) agreed very well with data on oil-water split across T -junction. They applied CFD and validated published data with experimental. They reported a reasonable agreementbetween the numerical two methods. Gas-liquid two-phase flow experiments where liquid phase viscosity is varied in order to study the influence of liquid viscosity acrossthe T -junction is systematically in horizontal flow and very rare in vertical flow with a tee to divert part of the scarce flow to a side-arm. Hong (1978) worked with different liquid viscosities in his ID = 0.0095 m in order to investigate flow split acrossthe experiment with aT -junction junction. He reported a distinct difference in phase split between water and liquid 5mPa s. However, he concluded that a further increase in viscosity to 10 viscosity of has a little effect on the phasesplit.These experiments are of particular relevance mPa s
51 MB ALAMU 2010
Chapter 2
Literature Review
to the current study, since the pipe diameter employed (ID = 0.0095 m) is the closest to
that utilized in the present work. Hong (1978) varied the liquid and gas flow rates as the liquid viscosity, and found that increasingthe inlet liquid superficial velocity well as decreased fraction of liquid taken off in the side arm. He also found that increasing the the gas flow rate increasedthe fraction of liquid taken off. Hong suggesteda mechanism for this splitting phenomenon,which is similar to that originally proposed by Oranje (1973). Hong argued that an abrupt change in the direction of gas entering the side arm a centripetal force, which creates an under pressure inside the 900be nd, produced drawing liquid into the side arm. When the gas intake is small, the centripetal force and hence the under pressureis small comparedwith the inertial force of the liquid stream, the liquid flows straight through the junction. Under these circumstances,liquid only so the side arm after a higher portion of the gas is diverted into the branch. Mak et al. enters (2006) used samefacility as used in this study but only tested with air/water.
Viscosity is an important property becauseof its overall implication on flow assurance in a multiphase flow system.Liquid viscosity influences flow assurancesuch that flow is transport energy is greaterthan resistanceto flow by liquid viscosity. only assuredwhen In oil and gas industry most of the existing flow models and pattern maps are largely derived based on liquid with lower viscosity, in most cases water, with viscosity of 1 Some oil wells flow oil as viscous as 10,000 mPa s at reservoir condition. This mPa s. big gap betweenlaboratory and field analysis. createsa
52 MB ALAMU 2010
Chapter 2
Literature Review
A common characteristic of two-phase flows through T-junctions is the maldistribution of the phasesbetween the outlets. The occurrenceof this maldistribution in equipment downstream of the junction can constitute a major problem in operation and control of the processand power industries as well as in oil and gas production. Stacey et al. (2000) gives a good example taken from natural gas distribution networks during winter. Observationshave shown that small quantities of the heavier components condensing out of the gas flow could arrive in appreciable amounts at any one of the delivery stations. This is causedby liquid emerging preferentially from an outlet in one the junctions used to split the flow betweendelivery stations. Another example given of by Azzopardi (1994 c) describes pipe work supplying two-phase feed to exchangers mounted in parallel. Maldistribution of the phaseswill have a operating as condensers direct effect on the performance of the heat exchangers. Those exchangersreceiving mainly liquid will underperform significantly because the in-tube heat transfer
during sub-cooling will be much lower than the correspondingcondensation coefficients coefficient. In contrast, those exchangersreceiving mainly vapor will perform slightly better than expected.However, the entire bank could operatebelow specification as this be sufficient to compensatefor poor performers. On a more positive would not normally the maldistribution of phasesthat occurs at aT -junction can be used for phase note, in the processingindustry. separation Currently, the large thick-walled vesselsbeing used as separatorsare expensiveto build the large inventory of flammable material that is contained in such vesselspresent and to safe operation. Hence, the partial separation of gas-liquid mixtures some challenges by using aT -junction will help reduce the load on the main separator,possibly possible
53 MB ALAMU2010
Chapter 2
Literature Review
leading to smaller units. Azzopardi et al. (2002) have reported the successfulapplication of aT -junction within the chemical industry. The T -junction was used to partially separatethe flashed products from a reactor. The output gas and liquid phase streams then introduced at different points of a distillation column to improve its were efficiency. A survey of the existing literature showsthat most of the previous work on dividing twoflow at aT -junction has been for sizes greater than 30mm in hydraulic diameter. phase Although Stacey et al. (2000), Lee and Lee (2001,2004), Das et al. (2005) and Wren et (2005) have worked with pipe diameters ranging from 5 to 10 mm, these al. experimental studies were performed using either rectangular channels or horizontal Tjunctions. Lee and Lee (2004) provide further motivation for researchin this areawhen that the flow distribution from a header to a parallel channel which is reporting important for predicting the transfer performance of compact heat exchangerscan be as an accumulationof small T -junctions. simulated Various models have been developed for predicting the flow behavior at aT -junction. The model by Azzopardi (1987) for predicting the phase split for annular flow T -junction was modified by Azzopardi (1988) to give better approaching a vertical by assumingthe total deposition of drops. The use of a geometrical model to predictions describe phase split was suggestedby Azzopardi and Whalley (1982). Shoham et al. (1987) later developed a flow pattern-dependent geometrical model by assuming film thickness, no entrainment and uniform mass flux distribution. The flow constant were annular and stratified flow. Under annular flow conditions, regimes considered Shohamet al. (1987) ascribes the preferential liquid take off at high gas flow rates to
54 MB ALAMU2010
Chapter 2
Literature Review
centripetal forces. This gives reasonableagreementwith their experiments.Sliwicki and Mikielewicz (1988) consideredthe local forces at the front corner of the Tjunction and ignored the effects of gravity in analyzing the diversion of the liquid film for annular flow. The fraction of drops diverted into the side arm was also calculated. Using the data of Azzopardi and Whalley (1982), they optimized the constantsin their equations. The model by Sliwicki and Mikielewicz (1988) is only valid for smooth films and is to be more accurateat lower gas velocities for which the film is lesswavy. expected The main aim of this study therefore, is to conduct an experimental investigation on the at a 5mm diametervertical Tjunction and comparethe resultswith phasesplit occurring by Azzopardi (1988), Shoham et al. (1987) and Sliwicki and relevant models Mikielewicz (1988). Furthermore, time varying, void fraction data obtained using ringtype conductanceprobes, have been analyzed to give valuable information about the flow periodic structuresaround the junction.
2.6 Justification
As reported by Belfroid et al. (2008) virtually all of the world's gas wells are either at from liquid loading and that the modeling of liquid loading behavior risk of or suffering is still quite immature. The prediction of the minimum stable gas rate using the existing be misleading becauseof their non-physical basis. For instance,almost all models could do not include entrained liquid fraction in determining the onset of existing models liquid loading. This probably may be due to lack of drop size and entrainedfraction data.
55 MB ALAMU 2010
Chapter 2
Literature Review
Vital information from time resolved drop concentrationcan help in formulating a good mechanisticmodel with a soundphysical basis. Drop size data has been measured in vertical annular two-phase flow using optical techniques.Azzopardi et al. (1980,1991) have used the angular scattering of light into small forward angles. The same approach has been used by Simmons and Hanratty (2001) and Al-Sarkhi and Hanrratty (2002) for horizontal annular flow. The instruments employed used the assumption that the scattering was dominated by Fraunhofer diffraction and time averagingwas almost inevitably employed to improve measurement The method provides average values over a finite volume. Drops size accuracy. distributions are extracted from the angular variation of scattered light. In addition, information of the time averagedconcentrationswas also determined (by integration). In practice, drop size data are usually time integrated. This integration in time may the quality of the data because of the complex mathematics and the compromise in time and space.Hence, analyzing data this way in amplitude and assumptionsmade frequency space with respect to time to yield Probability Density Functions and or to identify the dominant structurefrequency using Power Spectrum Density may often give interpretations. misleading
Azzopardi and Teixeira (1994 a) and van't Westende(2007) have used PhaseDoppler to measuredrop size in the past. This provides data at one point in space. anemometry The sampling position has to be traversed about the pipe cross section to obtain fully data. This approachalso provides drop velocity information. Azzopardi & representative
56 MBALAMU2010
Chapter 2
Literature Review
Teixeira (1994 a) have shown that the drop size distribution from the diffraction and PhaseDoppler anemometry instruments are the same if they are both convertedto the samebasis. The pertinent issue is non-existenceof time resolved drop-size data before this study. Literature is awash with time resolved void fraction, film thickness and pressuredrop data in annular two-phase flow, the whole idea of dynamic, timed averaged drop data is entirely inchoate. concentration This study presentsnew, time resolved drop size and drop concentration data for the first time. They were simultaneously acquired alongside time resolved film thickness and drop data. This work, therefore, contributes to the on-going discussions on pressure liquid loading problem in natural gas production and transportation by providing new information from experiments to better understand the behavior of the and novel in annular two-phase flow as well as the mechanisms governing periodic structures transfer of momentum between film and the gas core in vertical annular exchangeand two-phaseflow.
57 MBALAMU2010
Chapter 2
Literature Review
0o0
0
00 000
Entrained
liquid Droplet
Gas phase
-o
0 O
0o
e
o
oo 00 0
0 OoeOe
eO
000 oe eo
o00
_o0 o 0 0 Oe
0o
00
go
Wave
o"o0 e1
crest
Two-phase
flow
Woodmansee and Hanratty (1969) showed evidence that the creation of drops from the film on the channel walls did not take place from of the film but specifically from called disturbancewaves, Figure 2.4, which travel over the periodic structures, usually film at velocities of a few metresper second. Azzopardi (2006) presentsmore evidence this. Therefore, it might be expectedthat there would be an interrelationship between of the fluctuations of drop concentration and the frequency of disturbance waves. Unfortunately, hitherto there has been hardly any work presenting information about the in which the drop concentration varies with time. The most useful study is that of way Azzopardi & Whalley (1980 a) who used a camera looking axially up the pipe to record drops passing up the pipe. To ensurethat it was known which waves the drops came from, a method of injecting artificial waves was employed. The liquid flow rate was set
58 MBALAMU2010
Chapter 2
Literature Review
at the value just before disturbancewaves appeared.A small volume of liquid was then injected rapidly into the film. This produced a single wave which travelled up the
vertical pipe. The cine films taken with this technique were analyzed manually. It was found that before liquid injection there were no drops. Whilst the wave was in the pipe there were an increasingnumber of drops as the wave approachedthe cameraat the end the pipe. However, the increasewas not monotonic; drops came in bursts or waves. of Oncethe disturbancewave had exited the pipe no more drops were seen.
59 MB ALAMU 2010
Chapter 2
Literature Review
There is also supporting evidencethat drops are createdfrom periodic structures,usually disturbance waves which travel over the film at velocities of a few metres per called Therefore, it might be expectedthat there would be an interrelationship between second. the fluctuations of drop concentration and the frequency of disturbance waves where they are sourced.
Unfortunately, there has been hardly any work presenting quality information about the in which the drop concentration varies with time. The interrelationship between way fluctuations of drop concentration and the frequency of disturbance waves has never beenreported. All previous drop size and drop concentration measurementshave always been time integrated. This integration in time may compromise the quality of the data. Hence, data in amplitude and frequency spacewith respectto time to yield Probability analyzing Density Functions and or to identify the dominant structure frequency using Power Spectrum Density may often give misleading interpretations of actual dominating flow mechanism.
60 MB ALAMU 2010
Chapter 3
Chapter 3
Effect of Liquid Viscosity on Gas-Liquid Flow
Structures
3.1 Introduction
Viscosity is an important fluid property which has been found to be significant in twoflow. Flow pattern characteristics, including flow pattern transition phase gas-liquid boundaries,have been observedto behavedifferently when the viscosity of the liquid is increasedover and above that for water. Analysis is carried out with liquid of different viscosities. There are many important practical situations where effect of liquid viscosity on twoflow regime has been a source of concern to both academia and flow assurance phase Typical among these is the petroleum industry where the existing flow pattern engineers. are basedon liquid with lower viscosity, in most caseswater with maps and correlations 1.0 mPa s. Mandil (2002) reported some oil wells flowing oil up to 10 000 viscosity of Therefore, a wide gap exists between laboratory and field data. mPa s viscosity.
The approachof some authors has beento assumethat all vertical two-phase flow occurs in a highly turbulent manner which made them often assume viscous effects are This has been a logical approach, for instance, most practical oil-well flow negligible. have liquid flow ratesand gas-liquid ratios of such magnitudesthat both phases problems be in the turbulent flow. It has been noted, however, that in case where this will has been made serious discrepanciesoccur when the resulting correlation is assumption
61 MBALAMU2010
Chapter 3
applied to low production wells or wells producing very viscous crude as stated in Hagendoornet al. (1964). Both conditions suggestthat perhapsviscous effects may be the these discrepancies.In the first case, the increased energy losses have been cause of to be responsiblefor the increasedslippagebetweenthe gas and liquid phases observed as the liquid viscosity increases.This is contrary to what one might expect from Stokeslaw of friction, but the sameobservationswere made by Ros et al. (1961) that attributed this behaviour to the velocity distribution in the liquid as affected by the presenceof the pipe wall.
Air was drawn from the compressedgas mains and supplied to the mixing unit where it pumped from the liquid tank by means of a centrifugal mixes with glycerol-solution Inflow of air and glycerol/water mixture was controlled by using separatebanks pump. The mixing unit consists of an annular section surrounding a of calibrated rotameters.
62 MB ALAMU2010
Chapter 3
porous wall. Downstream of the mixer are probes 1,2 and 3 before pressure tap 1, Figure 3.1. The probes calibration curves are given in Figure 3.2. The pressure taps are separated 0.5 m apart to allow flow develop before the test section. The pressure taps connected to a Rosemount differential pressure transducer (Model 305) to measure were the pressure drop between taps 1 and 2.
Pressure
tap 2
DP cell
Pressure tap I
Is
Gai rota
Figure 3.1: Schematic flow diagram of the 5.0 - mm internal diameter rig.
The two phasestravel upward along 150 mm length, 5 mm internal diameter stainless steel,vertical pipe, where time varying, cross-section averaged void fractions were five identical conductance probes 4,5,6,7, measuredusing a stack of flow path shown in Figure 3.3. along the and 8 placed
63 MBALAMU2010
Chapter 3
Beyond the test section, the two-phase flow travels further upward and then horizontally for 1.07 m and finally vertically downwards for 3.12 m. The outlet streams were
collected in a cylindrical cyclone 1000 mm internal diameter where the phases were After separation, the liquid was collected and returned to the tank. Air was separated. to the atmosphere. Volumetric flow rate of liquid was determined by timing an vented from the cyclone separator over a lengthy period of time to minimize effluent measurement uncertainties.
Air was metered using a calibrated wet gas meter which allowed the flow rates to be of time. The liquid level in the separatorwas kept constantsuch measuredover a period that all measurements were carried out with reference to that datum. The datum served as a baselineto ensureaccuracy.
3.3 Instrumentation
The electronic circuitry employed for the conductanceprobe is similar to that of Fossa (1998) and Fossaand Guglielmini (1998). The conductancetechnique has been chosen to measurevoid fraction becausewater is an electrical conductor, while air is essentially Tap water was usedto calibrate the conductanceprobes. resistive. Glycerol in its raw form is not conductive. Sodium Chloride solution was addedto the glycerol solution to raise the conductivity of the solution to that of water. Andreussi et (1988) and Tsochatzidis et al. (1992) are some of the researchersto have used al. In this technique, a cross-sectionalaveragedvoid conductance probe successfully. fraction can be determinedonce the relationship between the electrical impedanceand
64 MB ALAMU 2010
Chapter 3
the phase distribution has been established. For the ring conductance probes used in the present experiments, the distance between the electrode plates and width of each plate are 1.7 and 0.5 mm, respectively. This gives electrode spacing to pipe diameter ratio (De/D) of 0.34 and electrode width to pipe diameter ratio (s/D) of 0.1. The probes give a is proportional to the resistance of the two-phase mixture. This voltage output which is converted to dimensionless conductance by referring to the value obtainable response when the pipe is full of liquid.
Oprobe6 Oprobe 4
dprobe2 0 probe 3
" 16
0
0
0.2
0.4
0.6
0.8
Dimensionless Conductance, Ge* [-] Figure 3.2: Conductance probe calibration curves, Omebere-lyari et al. (2006). The probe calibration equation used is the sameas that of Omebere-Iyari et al. (2006). Omebere-lyari artificially createdinstantaneous void fractions using plastic plugs with known diametersand relating this to the dimensionlessconductance. cylindrical rods of One of the calibration equationsusedin the data acquisition is shown in Figure 3.2 for Probe4.
65 MB ALAMU2010
Chapter 3
The output from the conductance probes and the differential pressure cell were fed into a PC via a National Instruments acquisition card and processed using a LabView programme.
Smm
Electrodes
Direction of flow
Figure 3.3: Top- a stack of conductance probes which constitute the test section (150 mm long). Bottom- cross section of a ring conductance probe.
66 MB ALAMU 2010
Chapter 3
and Probability
Density Function
Time varying/time resolved void fraction data for air/water and air/12 mPa s are in Figure 3.4. Time-resolved information can be examined at a number of presented levels. A great deal of information can be obtained by considering the time seriesof the fraction. An example of this for the two liquids studied cross-sectionallyaveragedvoid is shown in Figure 3.4. These are taken at gas superficial velocity of 0.27 m/s and a liquid superficial velocity of 0.64 m/s. In both casesthe results show the characteristic higher and lower void fractions which epitomise slug flow. It is alternate regions of fraction in the liquid slug part is higher for the water than for clearly seenthat the void the glycerol data. The time series data can be analysedfurther by considering the variations in amplitude frequency space. The former is examined using the Probability Density Function and (PDF), how often each value of void fraction occurs. Costigan and Whalley (1997) have density function (PDF) plots to explain the featuresof the suggested use of probability
67 MB ALAMU 2010
Chapter 3
two-phase flow. PDF plots of air/water and air/glycerol data are compared at same inlet condition as in Figure 3.4. Figure 3.5 presents the finding.
Air water
I
r. Mrel
Figure 3.4: Comparison of time resolved void fraction data for r air/water and mPa s viscous liquid at same inlet conditions where superficial gas and liquid air/12 0.27 m/s and 0.64 m/s respectively. velocities are
As can be seen in Figure 3.5, there are typical shapes or signatures which are obvious in these plots. Generally, a narrow single peak at low void fraction is typical of bubbly flow. A double peak is usually found in slug flow. The low void fraction peak to the liquid slug whilst peak at the higher void fraction is associated with corresponds the Taylor bubble region. The third type, a single peak at high void fraction with a long tail down to lower void fractions is recognised as the signature for churn flow. The last flow is characterised by a single peak at very high void fraction. The PDF one, annular solution are plotted in Figure 3.5 to elucidate any effect of of air/water and air/glycerol liquid physical properties and to identify flow pattern within the system based on the
68 Al B ALAMU 2010
Chapter 3
void fraction distribution. According to Figure 3.5, the signature displays by air/water indicates slug flow whilst the typical shape displays by air/glycerol data fits the description of bubbly flow.
10
-"Water(Slug Flow) sLiquid 13scostty(BubblyFlow)
-12mPA
a 4
2
I
0.2
0.8
Figure 3.5: Comparison of PDFs of void fraction for air/water liquid for VSG= 0.27 m/s, VSL = 0.64 m/s. viscous
There is a marked shift to the left towards decreasingvoid fraction when liquid viscosity increasedfrom 1 mPa s (water) to 12 mPa s at the same inlet conditions. This can be by time varying signal in Figure 3.4 where the average void fraction for supported be seento be around 0.30 and that of air/water 0.65. This reduction in air/glycerol can fraction, shown as lateral displacement in Figure 3.5 is a direct cross-sectionalvoid increasein liquid viscosity. effect of
69 MB AL IMU20I0
Chapter 3
000
012345 Misture `elocity, V,, [m/s] t.
0
Figure 3.6: Void fraction as a function of mixture velocity at different viscosity. In order to investigate effect of liquid viscosity further on phase distribution, liquid fixed at VSL= 0.03 m/s for mixture varied between 0 and 2.5 superficial velocity was for both air/water and air/10 mPa s sugar solutions. Results of the comparison are m/s in Figure 3.7. The plot shows significant decreasein void fraction at higher liquid shown The possible reasonfor decreasein void fraction as liquid viscosity increases viscosity. be due to decreasein liquid film velocity of the viscous phase adjacent to the wall may higher liquid hold-up and less void distribution acrossthe cross section of which causes the pipe.
70 MBALAMU2010
Chapter 3
OWVater[1mPas) "lOmPas 1
0.95
r--
U. 9
0.85
00 or
o"
0.8
0.75
0
o"
00o
cP
""
0.5 1 1.5 2 2.5
71 MBALAMU2010
Chapter 3
regime frequency data could be correlated using Strouhal number and LockhartMartinelli parameter.
Oair-water pair-lOmPas Oair-12mPas 30 24 7
Q1
is
4-
a) 12
a L
46
V r
(o
-6
Figure 3.8: Effect of gas superficial velocity on periodic structure frequency for air/water, air/10 mPa s and air/12 mPa s viscous liquid. The frequency of periodic structure in this case, void wave in all flow patterns has been correlated using gas based Strouhal number and Lockhartencountered Martinelli parameter. Strouhal number and Lockerhart-Martinelli parameter are
(3.1)
= -
EPLvG] [_l
2 PL VSL
O.S'
(3.2)
72 MBALAMU2010
Chapter 3
The result is shown in Figure 3.9. The figure shows Strouhal number increaseswith increase in Lockhart-Martinelli parameter. The scatter noticed in the figure may be as a result of
significant variation in gas density. pair-water
10 1 I u)
pair-lOmPas
pair-12mPas
0
00
0.1
0.01
Ad% oo
04,
13
E
= 0.001 0.0001
0.00001
0.000001
0 00 ooo 0
0.01
Figure 3.9: correlation of structure frequency using gas based Strouhal number with Lockerhart-Martinelli parameter.
In this analysis, a constantand unique value was used in each of the casesconsidered.These data follow sametrend as the trend observedin Figure 5.15 and will fall on the sameline if together. The dependenceof structure frequency on superficial gas and liquid plotted is clearly demonstratedin Figure 3.9. velocities The best linear equation fitting the experimentaldata points in Figure 3.9 takesthe form: T-G StG = 0.009 VSG L
(3.3)
Combining Equations (3.1) and (3.2) yields the following expression for slug frequency:
73 MB ALAMU 2010
Chapter 3
fslug
0.009 V SG -
TPG L
(3.4)
'
3.5 a L
3 3 L IA
00 o vv 0 00 00 OCC, o
'A 10%
0 0.1
oa o0
1
<o 0 00 0000
10
0
100
Figure 3.10: Structure velocity as a function of mixture velocity for air/water and liquid. air/12 mPa s viscous
Figure 3.10 shows how structure velocity varies with mixture velocity for air/water and liquid respectively. The figure shows that at low superficial air/12 mPa s viscous difference in structure velocity between velocities, there exists a noticeable air/water and
74 MB ALAMU 2010
Chapter 3
liquid. However, at higher mixture velocities, both structure air/12 mPa s viscous velocities becomeindistinguishableFigure 3.10. Further analysis to determinethe drift velocity of the flow structure is carried out on data in Figure 3.10. The data were re-plotted on a log-normal plot as shown in Figure 3.11. The intercept on the vertical in Figure 3.10 and Figure 3.11 gives an indication of drift velocity of the flow structurevelocity accordingNicklin et al. (1962).
Flow structure in air-glycerol mixture exhibits higher drift velocity as shown in Figure 3.10 and Figure 3. The reasonfor this behavior can be attributed to the effect of surface tension on structure velocity. Water has a surface tension of 0.073 N/m that of glycerol being 0.046 N/m at the test condition. Surface tension forces hold the flow structure together. The higher the surface tension acting between the interface, the lesser the frequency Figure 3.8. structure
10 Oair-12mPas Oair-water
00
E
00 d GpO 0
00
d L
44
uu N
O
0.1
p12345 Mixture velocity, VM [m/s]
Figure 3.11: Effect of physical properties on structure velocity in the slug flow for air/water and air-12mPa s viscous liquid. regime
75 MBALAMU2010
Chapter 3
6 0 air-12mPa s i i
4 U O U
01
00
w L
L
80 0
bubble, slug
i i
r N
flow
0
churnflow 10
annularflow 15 20
05
Mixture velocity,VM[m/s]
Figure 3.12: Identification of flow regimes using structure velocity variation with mixture velocity for air/12 mPa s. Analysing the data further, structure velocity variation with mixture velocity for air/12 liquid is presentedin Figure 3.12 to extend the analysis to cover churn mPa s viscous flow regimes. Figure 3.12 reveals that in the bubble and slug flow regimes, and annular velocity increasesmonotonically with mixture velocity. However, a point of structure inflection is reached around VM =5 m/s. This point marks beginning of churn flow After this point, the structure velocity remains constant with increasein mixture regime. Another transition is boundary is crossed around VM = 14 m/s where the velocity. becomeshigher than in churn flow. This point marks the beginning of structure velocity flow. annular In bubble/slug flow dominated regime, bubble coalescencetakes place. In the slug flow, to form bigger bubbles until a critical bubble diameter is Taylor bubbles may coalesce shear overcomesthe surface tension and the gravity forces. reachedwhen gas stream
76 MB ALAMU2010
Chapter 3
Churning of liquid film, flow reversal, flooding and liquid loading may occurs. At this bubble collapse and break-up occurs.Net momentum transfer acrossthe phasesis point, and collapse compensateeach other. Hence, the existenceof zero as bubble coalescence
in Figure 3.12. In chum flow, huge waves dominate the surface of the plateau as noticed
liquid film. The liquid flows as film around the pipe wall whilst the continuous phase flows in the centre transporting the entraineddroplets as core flow. gas
Transition to annular flow occurs around VM = 14 m/s according to Figure 3.12. This is interesting as the value agreeswith Vs0 = 15 m/s at the transition to annular particularly in a pipe of similar diameter with T -junction to divert part of the flow reported in Chapter7. Although the controlling mechanismis different for T -junction and pipe flow, this link in transition to annular flow between the two systems suggests there is interrelationship betweenthe flow pattern transitions even when geometry is different. At the beginning of annular flow the gas velocity is sufficient to lift the entrainedliquid the drag from the gas phaseovercomesthe gravity force acting downward forcing up as the droplet down.
In conclusion, the mechanism governing the flow from one flow pattern to another is different as discussedusing information from dynamics of flow structure in two-phase flow. A common area exists in flow pattern transition between pipe flow and dividing junction as demonstratedby transition to annular flow.
77
MB ALAMU2010
Chapter 3
5 O air-12mPas
N4 E
3
U O C,
y=0.5475x + 0.935
.o
o-.
vu 1
bubbly,slugflow
0
012345 Mixture velocity, VM [m/s] Figure 3.13: Effect of physical properties on structure velocity in the slug now for air-12mPa s viscous liquid. regime
Flow distribution coefficient, Co and drift velocity, VD can be determined experimentally in slug flow regime when structurevelocity is plotted against mixture velocity especially in Figure 3.13 for air-glycerol solution based on the idea of Nicklin et al. as shown (1962). According to Azzopardi (2004), one interesting finding is that the velocities of void Taylor bubbles and wisps all lay on the curve proposed by Nicklin et al. (1962) waves, for slug flow. Following this, it is seenthat a deviation from the Nicklin's curve occurs the flow changesto chum. According to Figure 3.13, basedon the idea of Nicklin, when it is observedthat Co = 0.5475, the slope of the straight line representingthe data.This is lower than the value of 1.2 reported by Nicklin et al. (1962) from air/water much The possible reason may be that the bubbles entrained in the liquid are experiments. It is well known that small bubbles tend to flow in the near wall region wall-peaking. profile of the gas holdup, whereas large bubbles tend to and present a wall-peaking
78 MB ALAMU2010
Chapter 3
to the core region forming a core-peakingprofile. Based on the definition of Co, migrate velocity of the bubble travelling at the centre-line is lower than that at the wall for a core-peaking behaviour. The likely cause of this phenomenon may be the presenceof tiny bubbles dispersedin the liquid slug. This was observedduring the experiment.What as slug flow turned bubbly flow when colour of the liquid turned milky actually started few minutes of start-up. This can be supported by the PDF plot in Figure 3.7. after Therefore, it is not surprising that Nicklin's equation fails to capture Co as expectedfor air/12 mPa s viscous liquid flow becauseit was formulated based on on slug flow mechanism. The Co value of 0.5475 is too low compared to Co = 1.2 for air/water. This is the limitation of this method. The Ca should clearly be more than or equal to 1.2 at least. Again, it is also evident from time resolved void fraction plot Figure 3.5 that air/12 mPa liquid displays a lower value of void fraction when it is compared with s viscous data at sameinlet condition. air/water
79 MBALAMU2010
Chapter 3
Oair-12mPa s y
r"
E
>4 t .u
O 7
00 00 OO
IL u
L
O OO
00
bubble,slugflow
0 012345
Figure 3.14 shows a good agreementbetween air/water data and Nicklin et al. (1962) line. The slope of the air/water line, that is the flow distribution coefficient, Co equals 1.20 and the drift velocity, VD. which is the intercept on the structure velocity line, being 0.0775 m/s. The value of Co agrees with Co =1.17 reported by Omebere-Iyari equal (2006). The reasonfor this good agreementmay be due to the nature of the formulation Nicklin's equation. The derivation was basedon potential flow analysis according of the Davies and Taylor (1950). Effect of surfacetension and liquid viscosity was not taken to Therefore, it is not surprising it works well for air/water. into account.
liquid shows a departure from Nicklin's line before VM =1 m/s Air/12 mPa s viscous VM =2 m/s. This may be as a result of the effect of physical properties, surface and after
80 MBALAMU2010
Chapter 3
tension and liquid viscosity. It seems liquid viscosity has a stronger effect on flow distribution coefficient, Cowhile drift velocity is surfacetension dependent. As earlier reviewed in Chapter2 sub-sections2.1.2,2.1.3 and 2.1.4, for laminar flow the ratio Vmax/mean 2.0 and indeed there is strong indication that Co increases approaches
the Reynolds number decreasesand reachesa value of about 2.0. A more precise as theory shows that Cofor laminar flow equals 2.27 as reported by Taylor (1961); Collins (1978) for the casewhere the surface tension is neglected. The exact value of Co et al. for both turbulent flow and laminar flows is inconclusive. Hale (1994) has attributed the non-uniqueness value of Co to the nature of the plot of of or bubble translational velocity against mixture velocity plot. The plot structure velocity is not perfectly linear, but instead bends very slightly upwards (Bendiksen, 1984). The Co is not clear for all flow conditions and reported experimental values exact value of from 1.02 to 1.35,Xiaodong (2005). Nicklin et al. (1962) suggested spreadconsiderably 1.2 for fully developed turbulent flow, but noticed the variation of this a constant of low Reynolds numbers(< 8000). coefficient at Again, refer to the classical work of Weber (1981) for small diameter pipes, Weber that the surface tension is important as its increase considerably decreasethe showed drift velocity. This statement agrees with recent observations in Figure 3.13. Using air/12 mPa s glycerol solution has a lower surface tension of air/water as a reference, 0.046 N/m and a higher VD = 0.935 m/s compares to air/water with VD = 0.0775 m/s from Figure 3.14.
81 MB ALAMU2010
Chapter 3
Next, calculating Co from measured void fraction using drift flux approach is considered.
0.35 gD
Drift velocity
(3.5)
Where: Drift velocity = rise velocity of Taylor bubble in liquid slug, m/s. Recently, Taitel & Barnea (1990) has emphasisedthat drift velocity is expected to dependon liquid viscosity basedon observationfrom their experimental work. Next, using similar approachflow distribution coefficient and coefficient of secondterm in Equation (3.5) will be determinedfrom measuredvoid fraction.
V-=CpVM+Vp Eg
LSG
(3.6)
Woldesemayatand Ghajar (2007) carried out comparison of the performance of 68 void basedon unbiaseddata set of 2845 data points covering wide range fraction correlations made. The duo engagedon comprehensive assessments of parametersthan any previous
82 MB ALAMU2010
Chapter 3
literature search for the available void fraction correlations and experimental void fraction data. After systematically refining the data, the performance of the correlations in correctly predicting the diverse data sets was evaluated. The analysis showed that most of the correlations developedare very restricted in terms of handling a wide variety data sets. Based on the observations made, an improved void fraction correlation of handle all data sets regardlessof flow patterns and inclination which could acceptably developed. The inclination angle and the system pressure effect on the angles was measured void fraction data were closely analyzeda nd accounted for in their n ew correlation. Bankoff (1960) has shown that the non-uniformity in phase distribution is a function of Zuber and Findlay (1965) emphasizedthe importance drift velocity. The pressurewhile latter claimed drift velocity is a function of the concentration profile and that it is also dependenton momentum transfer betweenthe phases.The concentration profiles across the pipe for a given inclination are generally assumedconstant for flows without mass transfer along the pipe length.
Integrating these observations and noting that this effect is normal to the pipe
factor of the form [1 + cosh] 0.25 inclination, a correction was introduced. The functional dependency of drift velocity on the momentum transfer between the phases assuming dimensional force interaction would be a maximum one for pipes with very high
inclination angles operating near atmospheric pressure. This effect could conveniently
Patin /P'YS
83 MB ALAMU2010
Chapter 3
Thesetwo correction factors have been introduced into the drift velocity expressionof Dix (Coddington and Macian, 2002) correlation. Hence, the modified correlation of Woldesemayatand Ghajar (2007) becomes:
VSG sg=
CO VSG + VDrrJ!
; Co=1+()fl.
PL
(3.8)
VD = drift velocity, m/s; V. = superficial mixture velocity, m/s; VG = gas velocity, m/s, Co= flow distribution coefficient, [-]
Therefore, Co can be estimated from the measured void fraction by calculating drift from Equation (3.8) and substitute the value in Equation (3.7). The results are velocity in Table 3.1 Figure 3.15 and Figure 3.16 respectively. presented , From Figure 3.15 and Figure 3.16, Cois varying with mixture velocity (liquid superficial because as liquid superficial velocity goes to zero, the Reynolds velocity) probably pys`d) becomes for the liquid (ReL = more laminar. For instance, if VSL= 0.8 number PPL Re = 400. Nicklin et al.(1962) reported Co can go to 2.0. This explains why the m/s, the Co for air/12 mPa s in Figure 3.15 clusters around 2.0. Basically, the distribution of figure shows that air/12 mPa s glycerol solution Co concentratedaround 2.0 while that of air/10 1.60 Figure 3.16. mPa s concentratedaround
84 MB ALAMU2010
Chapter 3
4 0water (1mPas)
0
012mPas
ko 0
as
u2
o o ] g, V., (: C
.
1
"r r "r 7
'IhD
8)
13
600 ^n 00
bubblyflow
rI' rl
05
20
VS
"r
uq
u3
"r
TO it
oz 1
000o bubble
05
OCP@ flow
00 (0o
20
10 is elocity,A'm,[nJs] Mixture A
85 MB ALAMU2010
Chapter 3
This implies that flow structure, Taylor bubble in case of slug flow, travel faster at the centre-line with air/12 mPa s glycerol solution two-phase flow than in air/water at the sameinlet condition.
Table 3.1: Table of physical properties, Co and Cl for the test fluids
Fluid Liquid density Surface tension Drift velocity C. C1
lkg/m I
ImN/ml
Im/sl
I-I
I-I
Omebere-Iyari (2006) using air/water with the same experimental facility reported a 1.17 for Co and -0.16 m/s for drift velocity when he plotted structure velocity value of under intermittent flow conditions (Slug & Chum). This is in against mixture velocity with the averagevalue of 1.20 for Co obtained using present approach. good agreement Also, the drift velocity calculated using Equation (3.8) yields VD = 0.16 m/s which 100% with the Omebere-Iyari (2006) value of VD = -0.16m/s in slug and chum agrees flow regimes. The only difference is in the flow direction as indicated by the negative flow possibly due to dominance of chum flow. sign which signified reverse
Flow distribution coefficient appearsto be higher for the larger liquid viscosities used as indicated in Figure 3.15 and Figure 3.16 respectively. Both figures have same Co of 1.60 if the statistical average is determined from the distribution Co. This is not about difference in viscosity is very small about 2 mPa s. In the case of 10 surprising as the
86
MB ALAMU2010
Chapter 3
Co decreasesfrom a value of about 5.0 with increasing mixture velocity to a mPa s, from 4.0 to a value of about 1.2. This ties in 1.6; for air/water, Co decreases value of with the findings of Schmidt et al. (2008).
The bifurcation along the average Co line is another interesting feature noticed in the trend displayed by Co for Bubbly flow as highlighted by Figure 3.15 and Figure 3.16 The bifurcation delineateschum, slug and annular flow from bubbly flow. respectively. This can also be supported by visual observation during the experiment. Based on this division, flow pattern above the average Co line is characterizedby periodic structures travelling faster than the mixture velocity at the centre-line. This is characteristicof slug, flow. They are often dominated by huge waves which travel faster chum and annular than disturbance.Another possible reasonto support the former statementmay be corewhereby the flow structurestravel along the centre-line couple with the peaking effect influence of centripetal accelerationon the flow structure.
In the casewhere Co is less than unity for instance the flow was observedto be bubbly dominates,with bubble velocity in the liquid streamless than where wall-peaking effect the mixture velocity. All values of Co below 1.0 for air/water, air/lOmPa s and air/12 liquids may be a direct result of wall-peaking effect in bubbly flow mPa s viscous is less than the total superficial velocities. regime where the centre-line velocity if Nicklin's expressionis suitable for slug flow then the expressionhas to be Therefore, in Co as liquid physical properties change. Before the modified to reflect changes Nicklin's expression,drift velocity will be discussedfirst. modification to
87 MB ALAMU2010
Chapter 3
Drift velocity is the velocity of gas phase relative to the gas-liquid mixture velocity. Drift velocity of 0.07752 m/s is obtained if it is calculated from expressionof Nicklin's expression (0.35 gD) for the present casewhere pipe internal diameter equals 0.005
m. However, a different value of drift velocity, VD = -0.16 m/s was reported by Omebere-lyari (2006) which agreesvery well with the value VD = 0.16 m/s using the drift velocity correlation by Woldesemayat and Ghajar (2007) except in the proposed flow direction. These values are an order of magnitude higher than the value obtained from Equation (3.5). The difference in value obtained is due to the inclusion of physical in the calculation of drift velocity using Equation (3.8). Using the value of properties drift velocity to determine the drift velocity coefficient, C1(Equation 2.5), the results are in Table 3.1. The value of C1= 0.7455 obtained for air/water is more than twice reported of C1= 0.35 using Nicklin's equation. a value in Modification to the Nicklin's expressionto reflect new C. and C1 is necessary order to with upmost accuracy in the slug flow regime. Therefore, an predict structure velocity improved version of Nicklin `s expressionis proposed for slug translational velocity for liquid two-phaseflows as follows: air/water and air viscous
VSLug = COVM + Cl
gD
(3.9)
Where
VM = VSG +VSL
The coefficients of the LHS of Equation (3.9) are available in Table 3.2 for different correlations.
88 MB ALAMU2010
Chapter 3
pl sg) S_
8
g ;E=S.
`
S +
(3.10
Where: rp=VSG VSL S= Slip velocity ratio, [-] In order to investigate effect of liquid viscosity on slip ratio Equation (3.10) and Equation (3.11) were utilized using the measuredvalue of void waves obtained during the experiment with air and liquid of different liquid viscosities. The results are shown in Figure 3.17. (3.11)
89 MB ALAMU 2010
Chapter 3
100
Owater [lmPa s] 010mPa s
O 12mPasO
mi C/9
O r r
10
13 0O 1
C CtA
00OLT
i slug, churn , annular flow
0001
bubbly flow
0.1
0.01 0.1 1 10 100
Another comparison was carried out for VSL= 0.4 m/s for air/water and air-lOmPa s as in Figure 3.18. Again, higher slippage is experienced by the fluid because of the liquid phase. The lateral separation distance between the water and resistanceof the is a direct effect of liquid viscosity as the flow experiencehigher lOmPas sugar solution liquid phaseswhen liquid viscosity increasesfrom 1 mPa s slippage between gas and (water) to 10 mPa s for sugarsolution.
90 MB ALAMU 2010
Chapter 3
O
i Cl)
-
iw
GI X
00
c9
1
0.1 1 10 100
91 MB ALAMU2010
Chapter 3
presence of micro-bubbles
in the liquid
to remove without
liquid
M 10
10
' z
>
1
Dhparssd bubble Now Int"If flow bubble flow Wdwdu churn flow
0 p0
s0 0
:
0
g 0.1 00
00
04
00 o
0.1
00
slus%w
000e000
0.1 1
10
100
100 T
10
10
8 e. U. 06 IL
-12aPh"LIqud
%K-*WllD6bbhllu-
06
4
s
df aoffy.
1 iA[an
I' f
0
0 0.2 0.6 0.4 Voidfraction1-1 0.8
0.2
0.8 k
3.19: Top - Test matrix for air/water and air-12mPa s viscosity liquid Figure flow Pattern map of Taitel et at. (1980). Bottom - PDF of void represented on fraction corresponding to air/water and air/12 mPa sviscous liquid. Inlet condition Vsc = 0.27 m/s, Vs,, = 0.64 m/s.
92 MB ALAMU 2010
Chapter 3
-j
dpj\ -. 7 LL
dL,F
VSG
Figure 3.20: Schematic of pressure-gradient behaviour in vertical flow According to the Figure 3.20, the gravitational pressure drop decreaseswhile frictional drop increaseswith gas or mixture superficial velocities. Total pressuredrop on pressure hand, decreaseswith increase in gas superficial velocities, exhibiting a the other then increasesas flow regime changes. minimum and data reproducethis behaviour as can be seen in Figure 3.21. Figure 3.21 is Experimental pressure drop plot as a function of mixture velocity. The the experimental measured drop was measuredacrosstwo pressuretaps separateda distance 0.50 m apart. pressure is noticed to decrease from 12410 [Pa/ml from Bubble/Slug flow Pressure drop
93 MB ALAMU2010
Chapter 3
characterisedby high liquid hold-up associatedwith high gravitational pressuregradient during slug/Intermittent flow at 12330 [Palm]. exhibiting a minimum 12420
12400
0
7
12380 CO
a i
12360 12340
12320
s
012345 1\lixture velocity, V. [m/s]
Figure 3.21: Pressure gradient as a function of mixture velocity. Liquid phase viscosity = 12 mPa s. This is becausethe superficial gas velocity is not high enough to cause high frictional drop. Gravitational pressure drop dominates because of high liquid hold-up pressure the pipe cross-section.As a result, the total pressuregradient decreases exhibiting across The pressuredrop behaviour reported here agreeswith the result of Nan Da a minimum. Hlaing et al (2007) which examined effect of liquid viscosity on flow regimes and Nan Da Hlaing et al. (2007) reported that in the Bubble, the slug and pressuregradients. flow regimes, the pressuregradients decreasedwith increasing mixture the slug-churn Reynolds number while in annular and mist flow regimes, pressure gradients increased with mixture Reynolds number.
94 MB ALAMU 2010
Chapter 3
In order to minimise pressure-dropand losses during two-phase flow operators more than not will like to operatein the Intermittent/Slug flow regime. This practice is often typical and is preferred in the production of oil and gas in vertical wells, Shoham(2006).
'Measured
i
YMeasured
(3.12)
AAPE
1100
i=1
('Predicted
YMeasured) i
11
YMeasured
(3.13)
The smaller the AAPE the better the correlation. For similar value of AAPE, the lowest defines the best correlation. standarddeviation value
95 MBALAMU2010
Chapter 3
E//ect of Liquid
Viscosity on Gas-Liquid
Flow Structures
YMeasured
(3.14)
The Standard Deviation STDEV can be calculated using: inn Zj_0 RE, DREI -
STDEV =
n2 (3.15)
Figure 3.22 presents the results when the predicted structure velocity by various and the experimental data are plotted against mixture velocity. correlations
*air-%iscousliquid Oair-Hater
ANeN correlation (air-%iscous liquid)
3
H 4
0 0
U3
a >
q2
U i-1 N
A" 0
o0 o 0% 00
00110 ,
0
1
Figure 3.22: Comparison of various structure velocity correlations against experimental data.
Chapter 3
The new correlations for air/water and air/viscous liquid are the results of changes in the of Equation (3.9). Equation (3.9) is the Nicklin's expression for translational coefficients Taylor bubble in slug flow. velocity of Bediksen (1984) correlation for structure velocity in slug flow is given as: VT
Structure velocity
0.54 gD
Drift velocity
(3.15)
The coefficients and the results of statistical assessment carried out on the existing correlations and the proposed new correlations are tabulated in Table 3.2.
* Bendiksen(1984)
0 correlation B, C1 = 0.10
ONicklin
E
43
N
.0
cJ
.1
U
L
u1 a) aa
u
010,
x1'
0 L
01234
least AAPE (Absolute Average Percentage Relative Error). The performance with the correlation B can predict the experimental data within 15 % Figure 3.23.
97
MB ALAMU 010
Chapter
Effect of Liquid
Viscosity on Gas-Liquid
Flow Structures
Unfortunately, the proposed correlation A, whose coefficients were obtained from the experimental data based on Nicklin's idea, could not re-produce original experimental
data. However, correlation gives better prediction of structure velocity when the drift velocity was tuned until a better match was achieved between coefficient of experimental and predicted structure velocity. In this case, the predicted structure
velocity was accurate within 15 % when the coefficient of drift velocity term was reduced 0.10.
Air/Water
Correlation Nicklin al.(1962) et Co 1.20 C, 0.35 APE AAPE STDEV 15.10 15.10 8.16
Bediksen (1984)
A (air-water) Proposed correlation B Proposed (air-water) correlation
1.20
1.20 1.20
0.54
0.75 0.10
Co 1.20 1.20
AAPE STDEV
12.07 42.10 44.25 14.55 41.82 43.22 53.47 68.64 56.94 1.77 29.4 33.26
ideas 1.60 based NicklinBedkisen Modified corelation on flow 1.00 forbubbly Proposed correlation
It is clearly seen in Table 33.2that predictive ability of the various correlation increases drift velocity (or drift velocity) decreases. However, this is not the as the coefficient of in air/12 mPa s viscous liquid analysis, Table 3.2. In the case of air/12 mPa s case viscous liquid, Nicklin's/Bediksen idea was used by adjusting the value of distribution
the drift velocity in the structure velocity Equation coefficient and the coefficient of
98 Al B ALAMU 20/0
Chapter 3
(3.9). The modified correlation basedon Nicklin/Bediksen ideas producesdrift velocity 0.64 and a Co of 1.60 was used in the calculation basedon observationin coefficient of Figure 3.15. This correlation performs gives worst performance with highest AAPE of 68.64. Nicklin's correlation with the drift velocity coefficient of 0.35 performs better than the modified correlation with a lesserAAPE of 42.10. Bediksen (1984) correlation better than the modified correlation and the Nicklin's with lesser AAPE of performs 41.82. The reasonwhy the modified correlation fails to predict experimental data may be that the modification of Nicklin/Bediksen's expressionis not suitable to model bubbly flow. It may be that the integrity of the liquid slug has been affected by numeroustiny bubble in the liquid as observedduring the experiments. Therefore, another model is entrained for estimating structure velocity in the bubbly flow regime. Going by the proposed in Figure 3.15 and Figure 3.16, it is reasonable to assumethat in bubbly observation flow, the flow distribution coefficient does not exceed 1.0. Harmathy (1960) suggested Equation 2.26 for calculating drift velocity in bubbly. Again, the Equation is expressed as: (Pc - Pc) 179 1.53 PL
0.25
(3.16)
Replacing drift velocity term in Equation (3.9) with the expression in Equation (3.16) 1.0 for Co in the sameequation yields Equation (3.17). Equation (3.17) and substituting is the proposedcorrelation for structurevelocity in bubbly flow regime.
99 MB ALAMU 2010
Chuptci
E%fec o%Liquid t
iiscositt'
on Gus-Liquid
Flow Structures
VT
Structure velocity
0"q (PL
PL
Drift
0.2s
- Pc)
velocity
(3.17)
O Bendiksen (1984) 0Nicklin et al. (1962) O'Modified correlation based on Nicklin/Bediksen ideas 0 Proposed Correlation for bubbly flow
f
5
41
Ii r.. r
4
3
;. I
J L rr u L r z v J rr
2
1
u
A
Figure 3.23: Predictive ability of various existing correlations and the proposed correlation for bubbly flow.
The proposed correlation for bubbly flow gives best performance with the least AAPE _ 29.40, Table 3.2 and Figure 3.23.
3.5 Conclusions
be drawn from the study of effect The following conclusions can of liquid viscosity and in vertical two-phase flow: surface tension on periodic structures
Chapter 3
increasein liquid viscosity. 1. Void fraction decreases with 2. Structurefrequency increases with increasedliquid viscosity. 3. Both liquid viscosity and surfacetension causesshift to transition boundary. 4. Higher viscosity liquid phase induces higher slippage between gas and liquid phasein two-phasegas-liquid flow. 5. Flow distribution coefficient, Co has a strong dependence liquid viscosity. on , 6. Structurevelocity is strongly dependenton viscosity and surfacetension. 7. Flow structure travels faster at the centre-line in medium of higher viscosity and lower surfacetension. 8. Structures are held together by surface tension forces. Lower surface tension fluid tends to produce more periodic structures than their higher surface tension Hence, fluid with lower surface tension has higher structure frequency counterparts. lower drift velocity. and 9. Structurevelocity has beenusedto characteriseflow into various regimes. 10. Slip between gas and liquid phaseshas been used to dermacate bubbly, churn, flow. slug and annular 11. Increase in liquid phase viscosity shifts transition boundary to left on flow pattern map. 12.A new model has been proposed to calculate structure velocity in bubbly flow. It the experimentalstructurevelocity within 15 %. predict
Chapter 4
Chapter 4
Dynamic Drop Size Measurement in Vertical
Annular Two-Phase Flow
4.1 Introduction
Drop size data has been measured in vertical annular two-phase flow using optical techniques. Azzopardi et al. (1980,1991) have used the small forward angle light The sameapproachhas beenused by Simmons and Hanratty (2001) and Alscattering. Sarhki and Hanratty (2002) for horizontal annular flow. The instruments employed used the assumption that the scattering was dominated by Fraunhofer diffraction and time inevitably employed to improve measurement accuracy. The averaging was almost average values over a finite volume. Drop size distributions were method provides extracted from the angular variation of scatteredlight. In addition, information on the time averaged concentration was also determined. The other approach used utilized PhaseDoppler Anemometry; Azzopardi and Teixeira (1994 a), van't Westende(2007). This provides data at one point in space.The sampling position has to be traversedabout the pipe cross section to obtain fully representativedata. This approach also provides information about drop velocity. Azzopardi & Teixeira (1994 a) have shown that the drop size distribution from the diffraction and Phase Doppler anemometry instruments the sameif they are both convertedto the samebasis. are Though it has been reported that drops occur in bursts Azzopardi (2006), almost all drop size or concentration measurementshave been time integrated. This previous integration over time may compromise the quality of the data becauseof the complex
Chapter 4
the assumptionsmade in time and space. Hence, analyzing data this mathematicsand in amplitude and frequency spacewith respectto time to yield Probability Density way Functions and or to identify the dominant structure frequency using Power Spectrum
Density may often give misleading interpretations.
Dynamic time averaged drop-size measurementsare reported for the first time for two-phase flow. They were carried out on a 19 mm internal diameter vertical annular as fluids. A laser light scattering technique was employed to pipe with air and water the drop size and concentrationvariations in time. Simultaneously, time-resolved obtain of film thickness using conductanceprobes employing a pair were made measurements of flush mounted rings as electrodes and of pressure gradient. The gas superficial from 13-43 m/s at liquid superficial velocities of 0.05 and 0.15 m/s. velocities ranged Additional tests were carried out at a fixed gas velocity of 14 m/s for liquid superficial 0.03-0.18 m/s. Though structures are not clearly visible in the signals velocities of they have been analyzed in amplitude and frequency spaceto yield Probability acquired Density Function (PDF) and to identify the dominant frequencies. Cross-correlation between two film thickness probes provides the wave velocities. Characteristic frequencies for both waves on the film interface and drop concentration have been reported. Therefore, this work contributes experimentally to provide information on mechanisms for liquid entrainment in annular two-phase flow by providing new data responsible
Chapter 4
taken simultaneously for film thickness (waves), drop sizes/concentrationand pressure drop measurements.
For the sake of clarity, drop size is presentedhere in Chapter Four while and results and discussionson wave measurements deliberately deferred till Chapter Five. Chapter are Six extends Chapter Five by discussingwave amplitude and modeling in greater detail. Because of the link between these consecutive chapters, sometimes presentation and discussionsof results are carried in way that referencesare inevitably madeto them.
Spraytec instrument. Film thickness was measured using conductance probe employing flush mounted rings as electrodes. Pressure drop across the system was a pair of differential pressure cell. The output from the conductance probes and monitored with a differential pressure cell were fed into a PC via a National Instruments acquisition the a LabView programme. card and processed using
In the first series of experimentsthe gas superficial velocity was varied between 13 and
of 0.05 and 0.15 m/s. The second series kept the 43 m/s at liquid superficial velocities
104
MB ALAMU2010
Chapter 4
gas velocity at 14 m/s for liquid superficial velocities of 0.03-0.18 m/s. Pressure difference over a length of the pipe has also beenmeasured.
Figure 4.1: Schematic flow diagram of the rig used in the present study
105
MB ALAMU2010
Chapter 4
Purge air
Purge air
cD
0. 00
Doe
. 00 DUG
0- ring
Liquid film
extractor
00
D0D O 00
DD000 DD000
DDDOO
00 DD 00 000 D410 000
Conductance probes
t
Differential pressure cell
t
Conductance probes
d
N R
i a
4
Figure 4.2 (a): Schematic of the test section the new drop size unit, liquid film extractor and a The special test section comprises Figure 4.2 displayed the schematics stack of conductance probes. of the combined the optical access to the drop size unit from the laser beam. special test section showing
106 MBALAMU2010
Chapter 4
The fluids used were water and air. Water is taken from a storage tank and pumped through a bank of calibrated rotametersto monitor the flow rate into mixer. Bronkhorst EL-Flow, a dynamic massflow controller was used for some inlet conditions to regulate gas flux. The schematicof instrument is available in Appendix A. A bank of calibrated air rotameters was used in other category of experiments to monitor air into two-phasemixer. The mixer consistedof an annular section into which air was introduced. Water emerged into the annulus through a seriesof 5 mm holes on the wall of the cappedcentral pipe. This mixer was mounted at the bottom of the pipe 310 pipe diameters from section the conductanceprobes are located. Two pressuretaps connectedto a Rosemount where differential pressure cell, located 230 pipe diameters from the mixer, separatedby a distanceof 82 pipe diameterswere usedto monitor the pressuredrop in the system.The liquid film is extractedvia a 19mm internal diameter, 350mm long, 0.80 porosity acrylic Beyond the liquid extractor, only the droplet-laden gas core flows through the pipe. laser beam to illuminate the core flow. The pipe outlet is chamber which admits to a separator, the air being connected to the compressor, vented into connected the liquid being returnedto the storagetank. atmosphere,
107 MBALAMU2010
Chapter 4L
4.3 Instrumentation
The instrumentation in this study consists of the Spraytec instrument, conductance probes and differential pressure cell.
4.3.1 Spraytec
Laser diffraction instrument such as Spraytec measure changes in light scattering as a function of drop concentration and characteristic drop diameter. Spraytec provides time drop size distribution from the analysis of a diffraction pattern resolved volume-based, from the interaction between a drop and a laser beam. Power from Heliumresulting Neon laser illuminates drop saturated gas core. The measurement volume is the
intersection of a droplet field and the laser beam. Entrained droplets scatter the beam
by diffraction. Figure 4.2 (b). Un-diffracted light are recorded as obscuration principally the measurement. obscuration [%], being 100% minus transmission [%]. of
Figure 4.2 (b): Incident light scattering by, a single drop within the measurement volume.
beam and beyond the drops, the receiving optics is placed which uses a In line with the Fourier transform lens to collect the diffracted light in the forward direction. The Fourier
108
MB ALAMU 2010
Chapter 4
lens converts the scattered light into a far field diffraction pattern. This diffraction pattern is then focused onto a multi-element photoelectric detector composed of: one main detector ring (ring 0) and thirty one other concentric annular rings which produce
analogue signal proportional diffraction to the incident light intensity, Figure 4.2 (c). Once the and size distribution is
then derived employing a non-linear least square analysis which gives the most closely fitted diffraction pattern.
s ede r mmiadukU mi, Bt
I
se grits
jt 1, rkb H EMMT
(Finr3mlase marl
Lk e
w
u 02daedl' jt
tretrural
d,+aricsg
rtrned
qdm
Figure 4.2 (c): Principle present experimental The instrument mathematical improvement reconstruct
of laser diffraction
instrument
utilized
in
with
of 1% Lorenz-Mie
The
instrument (an to
Fraunholers profile
and a multiple
scattering
light scattered
improved
accuracy. Lorenz
Chapter 4
based on the Fraunhofer diffraction theory interpreted this portion of light energy as diffracted light and overestimatedthe small particle population. These features enable the instrument to provide accurate size distribution information in environments that other laser-diffraction systems can not. Corcoran et al., (2000) reported use of the Lorenz-Mie theory considerablyimproves the instrument performanceswhen measuring drops. Time resolved drop-size measurementis possible with the special invery small built Insitec card. The Insitec card records cycle-to-cycle temporal variations of the drop gas core as the instrument scansthe detector. - rich
Spraytec includes features designed to enhance the measurement of short duration including Flash Mode data acquisition software (required to achieve the full 2.5 events, kHz data acquisition rate), and a trigger input allowing measurementto be triggered by devices. The system is able to measure particles ranging from 0.5 to 1000 external microns depending on lens configuration at measurementrates of up to 2.5 kHz. The latest version of the instrument can handle measurements rate up to 10 kHz, Dumuochel et al. (2009). Dumuochel et al. recommend a comprehensive experimental protocol to follow when laser light diffraction measurementsare performed in severe operating Dumuochel et al. (2009) presentedbeam steering, the vignetting and the light conditions. multiple scattering effects as the measurement phenomena which can affect laser diffraction data quality. Although their work is not hydro dynamically related to the was used in this work as a guide to check the quality presentstudy the recommendation of our data.
Chapter 4
Beam steering effect is the manifestation of light scattered because of a refractive index in the gas phase. A refractive index gradient in the surrounded gas flow can be gradient by temperature gradients or by the presence of liquid vapor. The mathematical caused inversion procedure mistaken and interprets this supplementary scattered light to the presence of drops and calculates the drop size distribution accordingly. Beam steering deviates light at small angles and mainly affects the proportion of light detected by the first inner diodes, i. e., those sensitive to the big drops. In consequence, the drop-size distribution overestimates the big drop population and may exhibit a supplementary peak in this range of droplets. A characteristic feature of the presence of beam steering is a light intensity detected by the first diode Dumuochel et al. (2009). The data were peak of the bias of beam steering effect, beam steering does not affect carefully checked against the quality of the present study.
Vignetting is avoided by setting the separationdistancebetween the emitter and receiver to 300 mm i. e. 1.5 times focal length of lens used which was 200 mm. Drop-size test in Figure 4.2 (a) was designed such that the width of the box is not up to sections 300mm.Recommended vignetting distance from the lens to the test section by the instrument manufacturer is available in appendix B. Vignetting is a phenomenonwhere from the collection angle i. e. the drop do not diffract light at such light escapes scattered high angle that the light does not illuminate the Fourier lens. The effect is a drop population because they have greatest diffracted light underestimation of small angle.
Chapter 4
Multiple light scattering occurs when the measuringvolume contains a high number of drops either becausethe drop density is high or becausethe measuring volume is large. The reduction of the mean diameter is a consequence the overestimation of the small of drop population. It has widely reported that light multiple scattering affects the the transmissionis less than 40% and introduces a bias that depends measurement when on the transmissionand on drop characteristics.
The ratio of the diffracted light intensity to the incident light intensity provides the obscuration of the measurement. Unscattered light intensity is measured by the instrument using obscuration detector. The experiments were carefully controlled such that light obscuration did not greater than 20% i. e. transmission = 100%-obscuration. The Spraytecmultiple light algorithms helped correct multiple scattering effects.
Finally, background checks are carried out at no flow condition in order to check instrument responsequality. The background is a measure of stray light in the optical importantly, particulate contamination of the windows. A good system and, most background measurement is critical to ensure accurate and precise particle size The raw detector signal should have a beam power reading distribution measurements. (ring 0) of at least 1000 and a smooth, inverse bell-shaped curve for the remaining rings. The data reduction verification settingscontain threshold values for the transmissionand levels allowed for a valid background. If an error threshold is reached,the averagesignal messageand the background measurement is terminated. system produces an error
112 MB ALAMU2010
Chapter 4
RTSizer uses the backgroundto compensatefor any particulate contamination by using the background as a baseline that is subtracted from all particle size distribution
It is therefore, a good practice to carry out background scattering measurements.
before each test. Background measurementwas a routine experimental measurements protocol throughout the experimentalcampaign.
Spraytec can be synchronized with other instrumentation using a trigger system which data acquisition with other instrumentation. In the present allows simultaneous Spraytec was operated in external trigger mode. This mode allows the arrangement, instrument to link electronically with other instrumentation like conductanceprobesand differential pressurecell such that data acquisition can be carried out in a simultaneous In order to achieve simultaneousdata acquisition, a pulse generatorsubroutine manner. is written using LabView commandto control the execution of the three data acquisition the three different electronics. with programmesassociated
Spraytec was operated in flash mode in order to impose a sampling frequency and time respectively for the measurement. The dynamic range of the instrument is sampling to size drops between 0.5 - 460 microns based on the focal length used for capable the scattered light. The instrument model is RTS 5114, with maximum collection of 1.004 mW and characteristic wavelength range between 400 670 nm. Table 4.1 power information about the Spraytec. gives some technical
113 MB ALAMU2010
Chapter 4
:rode;
Direction
of flow
Figure 4.3: Cross sectional view of ring-type conductance probes used to measure void fraction.
114 MBALAMU2010
Chapter 4
0.8 r.,
0
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
The pressure difference between a pair of tappings 1.55 m apart was measuredby a Kent Taylor Smart Deltapi K series. In order to ensure that differential pressure cell, is only one phase in the tapping lines a liquid purging technique was used. This there bubbles from entering the tapping lines. Before the measurement a high prevented
115
MB ALAMU 2010
Chapter 4
to clear any bubbles from the lines. It was then reducedto an purging rate was applied flow rate which is low enough that the measured pressure gradient is not optimum affected by the purged flow.
The data acquisition pad consists of two data acquisition boxes. One for Spraytec and the other for conductanceprobesand the differential pressuremeter. The measurements from the probes and the differential pressurecell were acquired using a PC installed with NI DAQ card with nine channels. Eight channels for conductanceprobes and one for differential pressure meter. Calibration procedures for the differential meter and conductanceprobe are detailed in Kaji (2008). Sampling frequency was set at 1000Hz to achieve 6 secondsdata acquisition time for 6000 data points for each element of the test matrix investigated LabView programme data acquisition for conductanceprobe and DP cell whilst RTSizer servesas the controls interface for Spraytec. graphic user's
116 MB ALAMU2010
Chapter 4
Conductanceprobesmeasures void fraction within the accuracy of 10 percent.The range of the differential pressurecell was 0 -70 kPa of which the resolution was 45 Pa.
4.5 Results
First, flow pattern map is discussedwith respect to transition to annular flow. Various Next, statistical measures transition models are consideredand discussed. were extracted from the time varying signals acquired simultaneously during the experimental campaigns and measurements.Statistical average of principal flow variables were generatedfrom their respectivetime seriesand plotted against superficial gas velocity in a systematicmanner. Results are then considered,presentedand discussedin increasing level of complexity.
Chapter 4
boundary shows two limiting conditions, a critical gas velocity below which no entrainment is possible at any liquid flow rate and a critical liquid velocity below which no entrainmentis possibleat any gas velocity. In a transition region between these two limiting conditions, the critical gas velocity at the in increaseswith the decrease liquid velocity. Ishii and Grolmes (1975) onset of entrainment developed this criterion based on a mechanism of shearing off the crest of disturbance waves. They assumedthat the onsetof entrainmenttakes place when the drag force on the tip disturbancewaves exceedsthe retaining force of surface tension. Also it can be observed of that the gas velocity required for the onset of entrainment increases very sharply at the critical liquid velocity.
Based on mechanism governing vertical annular flow, Taitel et al. (1980) proposed a (red dashedline) from churn to annular flow Figure S. The idea was based transition criterion Turner et al. (1969) droplet's model. In order to establish co-current vertical annular flow on the drag force on gas phasemust overcomethe resultant effect of gravity and surfacetension forces such that a liquid droplet entrainedin the gas core can be sufficiently levitated without falling back. Thus the criterion for the transition boundary is given by the expression in Equation (4.1): FD= FG Where FD= drag force, FG= gravity force; (4.1)
118 MB ALAMU2010
Chapter 4
In Equation (4.1), FD and FG can be substitutedby defining the terms on the left hand side hand side of the equationrespectivelyas follows: and right 69(PL-Pc)
Dn4vPctc
(4.2) CD= drag coefficient, [- ] do = drop diameter, [m] [m/s2] g= gravitational acceleration, Solving for VG from above expressionyields the gas velocity on the transition boundary, to the minimum gas velocity required to lift the droplets in the core and which corresponds maintain annular flow. The droplet diameter is determined from a balance of surface tension forces that promote larger droplets and the impact of forces of the gas-phasethat tend to break the droplets into Taitel et al. quantified this phenomenon by Weber number given by the smaller ones. following expression:
"DPGVG We = Q (4.3) Very small particles (droplets or bubbles) correspondto conditions where We = 8.0, Shoham to annular flow, the larger droplets occur, where it is assumedthat (2008). At the transition
Chapter 4
the Weber number is between 20 and 30. Thus, once the We is given, the droplet diameter can be determinedas:
dD _
aWe PG-VG'
(4.4)
Turner at al. (1969) suggesteda value of CD = 0.44 for fully developed turbulent flow and We = 30 for large droplets. Basedon the fact that annular flow is associated with small liquid hold-up they approximatedVG = Vsc When various substitutions are made final expression for transition to annular flow is in Table 4.2. It can also be expressedin dimensionlessform as in Equation obtained as given (5). The dimensionless number, 3.1, on right side of Equation (4.5) is called Kutateladze number.
VSG PG 5
3.1-
169(PL
PG)I0.25
(4.5) When the experimental data are substituted into this transition model, a value of gas 12.67 m/s is obtained.This is different from a value of gas superficial superficial velocity of 21 m/s observedbasedon observationof subtle changesin entrainedfraction and velocity of liquid and gas superficial velocities increase in the present study. The SMD measured as in result of the transition model of Taitel et al. (1980) and present study may be as difference in arriving at the final form of the transition model. a result of assumptionmade
Chapter 4
--- Ishii and Grolmes (1975) Entrainment onset criterion Hewitt and Roberts (1969) wispy annular criterion
r
Z
10
-Mishima and Ishii (1984)flow transition boudaries - -Taitel. Barneaand Dukler (1980)annular flow boudarv
)h
I %ispy annular
I
00008E
------------------
0.01
t
Churn
Annulagllow
0.001
I 10 100
Superficial gas velocity, Vs. [m/s] Figure 4.5 (a): Flow pattern map for vertical two-phase flow. The open symbols represent the conditions at which present study were carried out.
Recently, Pan and Hanratty (2002) proposed an empirical correlation for the critical gas based on the experimental data. The authors presented a correlation for vertical velocity flow based on the balance of the rate of atomization of the liquid film and the rate of annular deposition of drops. The development of the correlation considers data bases with pipe diameters of 1.06 to 5.72 cm, superficial gas velocities of 20 to 119 m/s, superficial liquid 0.012 to 1.35 m/s, gas densities of 0.27 to 35 kg/m3 and surface tensions of 0.01 velocities of to 0.073 N/m. The critical superficial velocity marks the onset of atomization of liquid film
in to liquid droplet in the gas stream. The correlation droplets, as liquid velocity In annular with entrained is given in Table 4.2.
flow regime transition from annular mist to annular wispy flow is observed (Hawkes et al.,
121
MBALAML'2010
Chapter 4
(2000)). An empirical transition criteria proposed by Hewitt and Roberts (1969) for the prediction of this transition is shown in Figure 4.5 (a). The experimental observationshave shown the existence of wisps or agglomeratedliquid structure in a gas core of wispy annular flow (Bennett et al. (1965)). However, very limited information is available on the nature of interfacial waves in annular wispy flow regime as in a transition regime betweenthe annular mist and annular wispy flow. well as Sekoguchi and Takeishi (1989) reported the existenceof huge waves in theseregimeswhich they described as larger and faster waves comparedto the regular disturbancewaves. If the flow physic is considered, it is expected that in annular wispy flow and in the transition can be quite different. regimes, mechanismof entrainment
In the present study, efforts were to visualize the flow. Flow visualization studies were by studying the cine movies recordedby high speedcamerain slow motion mode carried out to study the interaction of entrained structuresin the gas core. Vision Research,Phantom, high speedcameraversion V7.1, equippedwith SR-CMOS sensorswas used to acquire cine information which was fed into a PC installed with Phantom software version 606. The PC 10/100 Ethernet (1Gbit 2Q03). The recording was carried controls the video processorvia 1000 frames per minute with a resolution of 800 x 600 pixels. In order to enhancethe out at image quality, a Nikon 24 - 85 mm F2.8 macro zoom lens was used to direct the sensor focus on to the pipe section where the cine movies were recorded.
Chapter 4
Table 4.2 : Annular flow transition models, their respective superficial gas velocity the beginning of annular flow and experimental evidences of transition to coat current annular flow.
Author
Equation
VSG[m/s]
Remarks This model shows that transition to annular flow is independentof liquid superficial velocity VG,, is the critical gas velocity at the onset of atomization Commonly referred to as flow reversalpoint criterion. Transition independentof liquid superficial velocity Investigatedchurn -transition in annular 31.8 mm vertical pipe Reportedchurn -transition in 50 annular mm vertical pipe Evidencebasedon fraction entrained
12.67
jDPLPG VQ = 40 GCr
12.39
Wallis parameter
V. _ G-
VscPc'S Fag(dP)
1.0
10.78
Experimental
20.00
Experimental
20.00
Mantilla (2008)
Experimental
20.00
Present study
Experimental
21.00
123 MB ALAMU20I0
Chapter
Dynamic
in Vertical
Annular
Two-Phase
Flow
OVSL=0.05m
Q\'SL=0.1Sm.
s
:n
0 \'SG=
14m, s
10 rM E
Bubbly -
'
1m., '
l 3 1 d e^I M1
y . O J
Annular
mist
Slug 0.1
<1<0xx$
:F
u
L
0.01
Churn wispy annular
region
L '.
U
Q f
0.001
10
100
Superficial gas velocit`, ''SG [m/s] Figure 4.5 (b): Modified flow pattern map of Hewitt & Roberts (1969) based on new from drop concentration information measurement in vertical annular two-phase flow.
Figure 4.6: Left l land Side (LIIS) - image of wispy annular flow at gas superficial 14 m/s and liquid superficial velocity of 0.19 m/s. Right Hand Side (RHS) velocity of image showing evolution of `ring' disturbance wave around transition to mist velocity of 30 m/s. annular after gas superficial
Wisps or fractal of liquid structures entrained in the gas core were observed at conditions they are not expected. This is supported by the high speed images recorded during the where In Figure 4.5 (b), one of the images extracted from the eine movie recorded experiments. during the flow Figure 4.6 LHS , where wisps were seen at gas superficial velocity Vs6 = , ,
14 m/s and at liquid superficial velocity Vs1,= 0.19 m/s, was attached to the flow map in the (wispy annular regime). This image demonstrates incompatibility of the present observation
Chapter 4
with the popular flow map, Figure 4.5 (b). This visualization evidence, however, conforms
the recent finding by Azzopardi et al. (2008). Azzopardi et al. observed wisps in the with region usually identified as chum flow regime by published flow transition models. According to Hewitt and Roberts (1969) transition criterion to wispy annular, the critical superficial liquid velocity value, VSL= 1.2 m/s must be exceededfor wispy annular flow to occur. However, wisps are seen in the present study at conditions below the critical liquid velocity value, VSL= 1.2 m/s. Further evidencesin support of existenceof superficial in region where they are not expected in annular flow will be provided in the wisps discussionusing the Probability Density Function (PDF). subsequent Other video footagesrecordedfor flow around the transition to mist annular flow and beyond in Table 4.3 column (a) and (c) respectively. A similar image recorded by are presented Mantilla (2008) was presentedin column (b) Table 4.3 to show similarity of the entrainment transition to mist annular flow. mechanismafter At relatively low gas superficial velocity, fractal of liquid film resemblessausagein shape in the gas core description of which fits into wispy annular describedby Sekoguchi are seen Takeishi (1989). The possible explanation for presence of wisps at these conditions of and lower gas and liquid superficial velocities can be attributed to incomplete relatively liquid torn off the wave crest by the shear influence of the gas phase not atomization of break down the bag of liquid due to gravity pull in the opposite enough to completely direction. Another interesting structure was observed around gas superficial velocity of 30 m/s and beyond.The wave structure capturedunder the high speedcine movie was shown in column
Chapter 4
(c) Table 4.3. The wave was referred to as ring disturbance wave according to Mantilla (2008) who observed similar wave (Table 4.3 column (b)) in his experiment with a pipe diameter of 152.4mm, horizontal annular flow. The wave was characterized by higher amplitude and frequency. Wave properties associated with the various entrainment processes is discussed in detailed in Chapter Five.
Table 4.3: High speed cine movies/images recorded during the experimental for conditions around transition to mist flow at gas superficial velocity of 30 m/s and beyond. Present study Mantilla (2008) Present study
(ei)
W)
(C)
126 MBALAAfU2010
Chapter 4
1.00
0.99 0.98 w C 0.97 0.96
1.00
-0.99
0 0
C
13
C 20-98 . L
-Measured=Predicted
00 090
0 13
" 0.91 79
> 0.96 0.95
0 0
30 40 50
I-0.94
0.93 U
20
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1.00
(b)
velocity.
4.7 (b) Measured void fraction validated with the model of Alves et al. (1991). Figure
127 MB ALAMU2010
Chapter 4
1.00
0.99
0.98 r
1.00
OVSG=14m/s
0.99 0.98 OVSG =14m/s
" -Measured=Predicted
0.97 c 0 0.96
0.95 0.94 0.93
0.97
0 0
O
0.96
4,0.95
000 000 00
0.92 0 0.05 0.1 0.15 0.2 0.25 Liquid superficial velocity, Vu[m/s]
0.92
0.94
0.96
0.98
100
Figure 4.7 (c) Void fraction variations with increase liquid superficial velocity. Figure 4.7 (d) Validation of measured void fraction with mechanistic model of Alves et al. (1991).
The measuredvoid fraction was validated with the mechanistic model of Alves et al. (1991) as shown in Figure 4.7 (b) and Figure 4.7 (d) respectively. Excellent match is observedbetweenthe measuredand the predicted values. The possible reason for the unexpected behavior noticed in Figure 4.7 (a) can be to the dispersed phase turbulence intensity. Below VSG = 21m/s, gravity attributed influence dominates (this statementwill be supported with facts from other data in the In the case of VSL = 0.05 m/s, thinner film produces relatively subsequentchapters). bigger drops. Bigger drops move at relatively slower velocity and do not follow gas Increase concentration of such bigger droplets enhances phase turbulence closely. intensity, Sawant et al. (2009), therefore, the fluctuation in the void fraction turbulence fluctuating gradient that is responsible for the higher voidage noticed gradient. It is this After this boundary, void fraction increases asymptotically with before Vso = 21 m/s. increasein gas superficial velocity.
128
MB ALAMU 2010
Chapter 4
The distribution in the caseof VSL= 0.05 m/s is bi-modal according to the PDF of D50 which is characterized by droplet diameters of 100m and 165m up to 30 m/s superficial gas velocity, Figure 4.29. Beyond VsG = 30 m/s, the distribution becomes to mono-modal with droplets size decreases around 100 gm. It may be that turbulent intensity decreases Another reasonmay with after VSG= 30 m/s as drop size decreases. be that smaller droplets dominatesthe gas core becausethey follow gas phaseturbulence Sawant et al. (2009). This allows for bigger voidage as picked up by all more closely instrumentation used in the data acquisition. Weihong et al. (2001) first reported the behavior while during their experimental study low-liquid-loading gas-liquid flow in near horizontal pipes. The phenomenon was of noticed when liquid superficial velocity was increased from VSL= 0.03 m/s to VSL = 0.05m/s at constant VSG = 25m/s, the liquid-film flow rate, hold-up, and pressure decreased,and the droplet entrained fraction increased significantly. They gradient all did not give any reason nor explain what was responsible for the behavior but rather the significance this will have on the development of predictive models for emphasized low-liquid-loading wet gas pipe lines. Their experiment was conducted on 50.1mm internal diameter pipe with inclination angles from horizontal of -2, -1, 0, 10and 2. They reported that this unexpected behavior occurred at gas superficial velocity of 25m/s which falls within the range of VsG = 21m/s suggesting similar mechanisms horizontal and vertical annular two-phaseflow. governing
Chapter 4
basedon the specific surfaceper unit volume. Or the diameter of a drop having the same the entire drop. volume/surfaceratio as
E
160
a) N a)
13 0
AA 120
E
C. O
12
13
13
80
a)
2
40 0 10 20 30
00
40 50
4.8, Sauter Mean Diameter (D32)and Mass Median Diameter (D50)decrease In Figure increasing gas superficial velocity at constant liquid superficial asymptotically with
130 MBALAMU2010
Chapter 4
velocity. D50 is seen to be greater than D32 in all cases tested.D50 for VSL = 0.15 m/s is
higher than D50for VSL= 0.05 m/s before the transition boundary at VSG= 21 m/s. After the transition, the trend reverses itself. This trend continues until another transition boundary is crossed at VsG = 30 m/s. After the transition to mist flow at VsG = 30 m/s D50for VSL= 0.15 m/s again peaksup crossing-overD50for VSL= 0.05 m/s at VsG= 30
M/S.
4.5.3.1 Identification
Measurement Subtle changes at transitions within annular flow have been elucidated by the information from characteristic mean diameters measured during the experiment. The these changes occurred marked flow regime transitions within conditions at which flow. The drop measurements were logged at constant liquid superficial annular 0.05 m/s and 0.15 m/s respectively with gas superficial velocities velocities of increased from 13 m/s to 43 m/s. Several fluctuations were seen in the systematically MMD. In order to averageout effect of this fluctuation MMD was normalized results of
by SMD, Figure4.9.
Two flow regimes transitions have elucidatedin Figure 4.9 using subtle changesin SMD MMD. First transition occurs at VsG = 21 m/s This is marked as dotted and discontinuous line in Figure 4.9. This transition is captured by SMD profile as drop drop from thicker film as gas superficial velocity from thinner film crosses-over created Gravity influence on thinner film is suppressedas drop from thinner film increases.
131 MB ALAMU2010
Chapter 4
the transition boundary at Vs0 = 21 m/s. This transition is picked up as point of crosses inflection according to MMD profile in Figure 4.9. The transition marks the beginning of co-current annular two-phase flow. In general,coflow occurs when drag force on the gas phase is sufficient enough to current annular levitate the liquid drops in the gas core. In terms of waves traversing the gas-liquid interface, before this transition, huge wave dominates although it co-exists with the disturbancewave. After this transition and before gas superficial velocity of VSG= 30 disturbancewave dominatesthe gas-liquid interface. However, after gas superficial m/s, Vso = 30 m/s huge wave vanishes. velocity Mass of a drop is directly proportional to its size according to this relationship, Ceylan et al. (2003):
lr pd a3
m=
6 (4.6)
Where, a= drop size (SMD or MMD). Therefore, drop mass densities increasesas film becomes thinner after the transition. This transition is only detectedby SMD profile Figure 4.9. The phenomenapresented here by this information from SMD will have significant effect in design of process in the formulation of mechanistic model to predict drop size distribution equipment and in terms of SMD.
4.9, MMD was moralized by SMD (secondaryaxis) and a plot of MMD/SMD In Figure velocity reveals two additional details. First, the mean indices of against gas superficial increasesindicate drop size distributions are selfMMD/SMD as superficial gas velocity
132 MB ALAMU2010
Chapter 4
is similar. Self-similarity nature of the profile suggestsexplicitly that drop coalescence dominant mechanismalthough it is believed drop break-up occurs simultaneouslybut on be tacitly neglectedin the analysis of result. a scalewhich can Secondrevelation was the detection of transition boundary to mist annular flow around 30 m/s based on the profile of MMD/SMD. This transition gas superficial velocity of boundary is marked by dashed discontinuous line in Figure 11 at Vs0 = 30 m/s. According to Figure 4.9, the drop mass density increaseswith thickness of the liquid film which producesthe droplet. This behavior is an exact opposite of the trend observed with the SMD profile. Before this transition, thinner film produces drop of higher mass density in terms of MMD.
*VSL= VSL=
2.4
i
1 113 dpQQ iil Q
oho
1.8
100
Q v A_ v
1.2 r.
50
C4
0.6
0.0
0 10 20 30
VSG
40
II11/si
50
drop size.
Chapter 4
The occurrencesat VSG= 21 m/s and VSG= 30 m/s can be linked with changeof liquid entrainment mechanisms.Evidencesfrom video footage of the flow recorded using high that although bag break-up and ligament break-up may co-exist speed camera suggest bag break-up dominates before gas superficial velocity VsG = 30 m/s while ligament break-up mechanismpresentsdominant mechanismafter gas superficial velocity Vsc = 30 m/s.
Generally, the droplet creation from film can be explained as follows. As gas velocity Wave traversing the gas-liquid interface traps gas increasesthe flow structuresincreases. bubbles into the liquid film. Density difference causes the bubbles to migrate and the wave crest causing increase in interfacial shear stress. Gravity agglomerate near drainage, gas-stream shear and bubble expansion causes liquid membrane to gets Eventually, bubble bursts forming numerousdroplets and developsvortices. The thinner. formed droplets causethe gas core density to increase and expand. This process newly liquid film lacks momentumto shed further droplet. continuesuntil
First transition describe transition to co-current upward annular flow at VsG = 21 m/s. For fully developed annular flow to occur, the drag or shear force being exerted by the the gravity force in order to keep the entrained drop flow cogas phasemust overcome the gas phase. Before the transition, the mechanism of entrainment is currently with be dominated by bag break-upfrom the video footage of the flow. to suggested
From measurements drop sizes and from flow visualization experiments,Azzopardi of identified two mechanismsby which droplet entrainment takes place. In the first (1983)
Chapter 4
mechanism (bag break-up), the gas apparently `undercuts' a large wave forming and open-endedbubble with a thick filament rim. When this bubble bursts, a rapid transient the gas phase acceleratesthe droplets. In the second mechanism, the acceleration of ligament break-up, the wave crests are pulled forward in the gas core in the form of ligaments. Oscillation within continue to grow until a point is reached when the frequency of oscillation within the ligament becomesequal to the turbulent frequency from the gas phase.At this point resonanteffect breaks up the ligament into numerous liquid droplets noticed in the gas core. The secondtransition marks the flow transition from annular to mist flow at VSG= 30 m/s. This probably reflects a change in the liquid from the `bag break-up' mechanism to the `ligament tearing' entrainment mechanism mechanism.
Bag break-up is characterizedby shearingof large liquid packets from the film to the gas At low gas velocities the droplets are generatedfrom the large flooding-type waves core. (huge wave) through the undercutting (bag break-up) mechanism. The driving force is the turbulent fluctuation in the gas phase. As the gas flow rate is increased and the large flooding-type waves subsides,disturbance waves start to dominates occurrenceof both mechanismscoexist (VsG= 21 m/s). With further and continuous increase although flow rate (after VsG = 30 m/s) the ligament break-up mechanism becoming of the gas dominant atomization mechanism. Based on the foregoing analysis of atomization or is controlled by bag break up before the transition co-current entrainment mechanism flow at VsG = 21 m/s. The dominant entrainment mechanism becomesligament annular break-upafter VSG= 21 m/s. .
Chapter 4
The usual expectation is that under low gas velocity condition, droplets are relatively large. As the gas velocity increases,more and more liquid gets entrained into the gas in decreaseof disturbance liquid film flow rate gradually decreases resulting core and Consequently the average size of droplets also decreaseswith the wave amplitude. increasein gas velocity. This is obviously not the case from the observation of present VSG= 30 m/s and seemsto be only valid for D32. study after
Wispy annular shown in Figure 4.9 are characterizedby incomplete atomization of the liquid film. It may be that droplet inertial is the driving force of droplet dispersionwhen film undergoescomplete atomization. The normalized MMD in Figure 4.9 for VSL _ 0.15 m/s, crossed-over the normalized VSL = 0.05 m/s at VsG = 30 m/s. Additional for this behavior in the mist annular regime is offered as follows. possible explanation
Ring disturbancewaves were identified in this study around VsG 30 m/s. The evolution this wave influences drop size (MMD) distribution at this transition boundary where of VSG = 30m/s and beyond. This is becausering disturbance wave appearsto dominate the creation and dispersion of entrained liquid droplets in the gas phases and control VsG > 30m/s. If this wave could be thought of as an elastic material then after to why D5oof VSL= 0.15m/s is greater than D50at VSL= 0.05m/s can be explanation as Then the degree of springinessseemsto be higher the higher the liquid flow provided. flow rate is maintainedat a constant value. Therefore, more elastic D50of rate when gas VSL= 0.15m/s tends to offer less resistanceto the driving force of the dispersedphase
Chapter 4
in greateramplitude, greaterdrop size than in the caseof D50of VSL= 0.05m/s resulting
to be more rigid. which appears
Another possible reasongoes as follows. Thinner film produces bigger drop becauseof the greater interaction between it and the turbulent gas core. Thicker film on the other hand experiencesless interaction with the gas core due to conservation of momentum. However, in mist flow (VsG >_ 30 m/s) where the droplets size are smaller and homogeneously distributed, atomization of liquid film is complete. approximately Dispersedphasedistribution is inertial-driven. Flow at higher liquid superficial velocity more structuresat constant gas superficial velocity. This may explain why will produce drop-size is higher for VSL= 0.15 m/s after VSG 30 m/s. >_ normalized
Another reasonwhy drop size of VSL= 0.15m/s is greater than that VSL= 0.05m/s may be explained by expansion of the core as liquid superficial velocity increases.Increases in liquid superficial velocity increasesfilm burst which in turn increasesdrop size and
dropconcentration.
137 MB ALAMU2010
Chapter 4
0.025 $ Q 0.02
W of 0.015
VSL= 0.05m/s
Q. E o)
0.01
'"
0.005
El a
0
0 10 20 30 40 50
When D50 is normalized by D32 as shown in Figure 4.9, an exception to the rule of that thinner film producesbigger drops is observed.This exception occurs at VSG thumb 30 m/s according to Figure 4.9 as thicker film producesbigger drops. >_
When wave amplitude is normalized by wave spacing and plotted against gas superficial trend observed in the drop size profile is also noticed, Figure 4.10. velocity, a similar The wave amplitude data use in Figure 4.10 and their detailed discussionsare available in ChapterSix. Therefore, this observationconfirms that droplet size (D50)distribution is dependenton the disturbancewave amplitude.
138 MB ALAMU2010
Chapter 4
Chapter 4
0.25
O VSL = 0.05 m/s "VSL= 0.15m/s "
0.2
"r rr
0 f 0.15
"
rr ii 40
rP 4
0.1
0.05
1)
0 10
00000
20 Gas superficial
Figure 4.11: Effect of gas and liquid superficial velocities on drop concentration. The trend shown by liquid superficial velocity, VSL= 0.15 m/s reveals additional details. The transitions at 21 m/s and 30 m/s to co-current annular and mist annular flow Thesetransitions, however, appearas point of inflections. Though, it can respectively. be concluded in terms of drop concentrationthat thicker film has greater interaction with the gas core.
(4.7) Where VD is the drop velocity, PLis the liquid density; c is the volumetric drop concentration
Chapter 4
mLE is the total liquid massflux. and Azzopardi & Teixeira (1994 b), Fore & Dukler (1995), From the available measurements, Zaidi et al. (1998), van't Westendeet al. (2007), it can be seenthat the gas superficial velocity is a good approximation to the drop velocity at centre-line while the product of liquid density and superficial liquid velocity approximatesthe total liquid massflux. The gas therefore be usedin place of drop velocity in Equation (6). Hence, superficial velocity can fraction can be estimatedfrom inlet condition by multiplying drop concentration entrained by superficial velocities ratio (VsQIVsL)"
0.7
^, 0.6
OVSL- 0.05 m/s fVSL=0.15 m/s
W 0.1
'
0
0 10 20 30 40
VSG [III/SI
50
Gassuperficial velocity,
Figure 4.12: Effect of gas and liquid superficial velocities on entrained fraction.
Therefore, Figure 4.12 is obtained when drop concentration is converted to entrained fraction. The figure relatesentrainedfraction variation with gas superficial velocity at liquid superficial velocities. The generaltrend shows that entrained fraction is constant higher for thinner film where VSL= 0.05 m/s than the caseof thicker film where liquid VSL= 0.15 m/s before Vso z 21 m/s. After VsG 2 21 m/s, the trend superficial velocity,
141 MB ALAMU2010
Chapter 4
Entrained fraction becomeshigher the higher the liquid superficial velocity. reverses. At the next level of analysis entrainmentfraction was normalized with minimum entrained fraction Equation (4.8) by Barbosaet al. (2002) and plotted against increasing superficial gas velocity, Figure 4.13.
EFMIN
0.95
+ 342.55
PL7L
PG mG
d2
(4.8) Where EFMIN minimum entrainedfraction, [%]; dp = pipe diameter, [m], PLand po are = liquid and gas densities; [kg/m3] and nip and mG are the liquid and gas massfluxes in [kg/m-s2]. respectively Figure 4.13 sharesimilar featureswith Figure 4.12. This shows good agreementbetween the model of Barbosaet al. (2002). Figure 4.13 shows entrainedfraction presentstudy and three distinct flow transitions: one, wispy annular, characterizedby dominance profile with huge waves; two, transition to annular flow characterizedby co-existenceof huge and of disturbancewave; and three, mist annular flow where disturbancewave dominates.This is supportedby measuredwave properties (seeChapter Five). Wave frequency, classification length and wave velocity have beentaken into consideration in categorizingwaves into wave huge and disturbancewaves.
Chapter 4
QQ 2 o
W
QoQo
10
Ui a IH f
U.
2
1
> O! oQ'
o
Q
0 10 20 30 40 50 Gas superficial velocity V$G(m/s] ,
Figure 4.13: Effect of gas and liquid superficial velocities on normalized entrained fraction. Measured entrained fraction normalized by the minimum entrained fraction model of Barbosa et al. (2002).
Figure 4.13 indicates that entrainedfraction for the casewhere VSL= 0.05 m/s is greater VSL= 0.15 m/s before VSGz 21 m/s. After VSG 21 m/s, the trend than the casewhere fraction increases with increased liquid superficial velocity at reverses, entrained constantgas superficial velocity. it could be concludedthat the subtle changesthat occurs in entrainedfraction Therefore, velocity of 21 m/s is another evidence of transition to coprofile around gas superficial flow. current annular
143 MBALAMU2010
Chapter 4
Therefore, using the schematic in Figure 4.14, liquid film hold-up on the wall can be estimated from the cross-sectional averaged void fraction logged with the conductance probes using Equation (4.9).
Void fraction,
Entrained
liquid
Droplet
Gas phase
Gou
ll
L1 Q pp
taU
!OO pn oa
oO
t'
en rJ
Liquid
film
thickness
VG
Wave Fg
4y
crest
,t.
.o
Two-phase
flow
Figure 4.14: Schematic of vertical annular two-phase flow. si .z Where: 6F = film thickness, [mm]; d= pipe diameter; Eg= cross sectional average void fraction, (4.9)
[-]
4.15 (a) shows that if the liquid superficial velocity is kept constant and the gas Figure is increased, the average film thickness decreases monotonically. On superficial velocity
Chapter 4
the other hand, Figure 4.15 (b) shows that when gas superficial velocity is kept constant and liquid superficial velocity is varied, the averagefilm thickness increases.
0.3 a OVSL=0.05m/s
0.25
OVSG=14m/s
I o. is E
b a o.2
0 13 0
C
'"
OVSLx0.10mis
0.20 0.15
Q
sQ 00 0 E3 E3
N N 41 C
015
LL 0.1 t 0.05
p q3 0O :0009,, 13
if 0.10
L N
0.05
0
0.00
0 10 20 30 40 s0
0
0.05
oo
0
Gassuperficial velocity, Vx (m/s]
0.1
0.15
0.2
(a)
(b)
Figure 4.15 (a) Film thickness variation with increase gas superficial velocity. Figure 4.15 (b) Film thickness variation with increase liquid superficial velocity.
An interesting hydrodynamic phenomenon occurs around VsG = 21m/s. The hydrodynamic phenomenon appearsto be transition boundary between wispy annular
flow regime. This transition appears as point of inflection around and co-current annular
VsG = 21 m/s in Figure 4.15 (a). After VSG= 21 m/s, the film is believed to become This boundary was picked up by all instrumentation more stable and axi-symmetrical. in the data acquisition. This will be discussedfurther in subsequentanalysis. used The following discussionattemptsto offer explanation for this occurrence.It is believed influences annular flow regime Geraci (2005). The that gravitational acceleration influence of gravity increasesdisturbancewaves due to the acceleration of the liquid in the opposite direction to the fluid motion McGillivray et al. (2002). The phase influence of gravity createsan unstable and chaotic film. As noticed previously in the
145 MB ALAMU2010
Chapter 4
void fraction profile Figure 4.15: (a), the dominant effect of gravity ceasesafter VsG= 21 m/s. Under the influence of gravity film is less stable and chaotic. Gravity effect is for VSL = 0.05 m/s becausethe film is thinner, less stable and more more pronounced chaotic. Thinner film of VSL= 0.05 m/s promotes greater interaction with the gas core and therefore experiencesmore turbulence than in the case of VSL= 0.10 and 0.15m/s. Values for VsL= 0.10 m/s was interpolatedbetweenVSL= 0.05 m/s and 0.15 m/s. In the case of VSL = 0.10 m/s and 0.15 rn/s the film are less affected by turbulence. Film thickness decreases the transition boundary around VsG= 21 m/s. After the transition, at it recoversbehaving linearly increasewith increasing in gas superficial velocity for both VSL=0.10m/sand 0.15m/s. The trend in the average film thickness behavior with increasing gas and liquid velocity is in good agreement with previous studies on air/water annular superficial flow. Kaji (2008) reported the chaotic nature of the film in the gravity dominatedregime to be as a result of change in film thickness with axial distance. Kaji claimed for VsG> 12 m/s, film thickness increaseswith axial distance up to about 100 pipe diametersthen decreases asymptotically. Film hold-up data, according to the foregoing, therefore, seemsto capture the transition to co-current annular flow which occursat a superficial gas velocity of 21 m/s.
146 MBALAMU2010
Chapter 4
4.5.6.1 Film hold-up and Drop Size In this sub-section,film hold-up and drop size data are analyzedto establish a relationship between film hold-up (wave) which produces the droplets and the entrained droplets. Standard deviation of the film hold-up presents the most interesting finding of all the statistical analysis carried out. The standarddeviation of the film thickness when plotted superficial gas velocity as shown in Figure 4.16 shows that Mass Median against Diameter (MMD) is directly proportional to the standard deviation of the film hold-up. There is a strong relationship betweenFigure 4.8 and Figure 4.16. The two figures display trend for VSL = 0.05 m/s and VSL = 0.15 m/s. This is another interesting similar becausedata in Figure 4.8 was acquired from Spraytec whilst film thickness observation data in Figure 4.16 was logged using conductanceprobes.
It is interesting to seein Figure 4.16 that drop size distribution is directly proportional to the deviation of film hold-up from which the drops are produced. By employing the standard film thicknesscan be determined from drop size instantaneous right statistical model distribution (MMD). Thus; dynamic film hold-up can be generatedfrom drop size information and vice-versa.In nuclear engineeringand multiphase Computational Fluid Dynamics, this valuable information can be usedto model and quantify the amount of liquid in the gas core by carrying out massbalanceanalysis. present
level of analysis, standarddeviation has been used to distinguish betweenhuge and At next disturbancewaves in annular two-phaseflow. In Figure 4.16, the standarddeviation of the increasinggas superficial film thickness is seento decrease with McGillivray & velocity.
147 MB ALAMU2010
Chapter 4
Gabriel (2002) used standarddeviation of the liquid film thickness to categorizewaves in annular two-phase flow into huge and disturbancewaves. They characterizedhuge wave as having large standarddeviation. In the presentdata the standarddeviation decreases film as increasegas superficial velocity. Figure 4.16 shows fluctuations thicknessdecreases with transition boundariesat gas superficial velocities of 21 m/s and 30m/s respectively. around After gas superficial velocity of 30 m/s, the thicker film displays a higher standarddeviation to the trend observein drop size distribution Figure 4.16. similar McGillivray & Gabriel (2002) did not specify boundary conditions for the huge and disturbancewaves basedon deviation of the film thickness from the mean.Therefore,using findings in this study we can concludethat huge wave transversesthe gas liquid interface the influence and this dominanceceases after gas superficial velocity = 21 m/s. under gravity This is reasonableas the wave is characterizedby large amplitude typical of huge wave before transition to co-current annular regime after gas superficial velocity= 21 m/s. After 21 m/s, wave amplitude becomessmaller, decreasingwith gas superficial velocity = increasinggas superficial velocity.
148 MB ALAMU2010
Chapter 4
0.12
r
2.5
2.0
0.1 0.08
rw
00
0.06 Ger
Q41 o0 o oA
1.5
0.04 0.02
.o
1.0
0.5 0.0
0 10 20 30 40 50
0
Gassuperficial VSG [m/s] velocit`;
Figure 4.16: Comparison of Standard Deviation of Film hold-up and Drop size (MMD) showing similar trends as gas and liquid superficial velocities increase.
Chapter 4
21 m/s. In the caseof VSL= 0.05 m/s, unexpectedbehavior was noticed. After VSG= 21 further the pressuredrop is expectedto recover, it did not rather it decreases m/s when It VSG= 36 m/s where it startsto recover. is not very clear if frictional pressuredrop until has over-taken gravity at this point. Result of another experimentalcampaignwhere gas superficial velocity was fixed at VSG = 14 m/s whilst the liquid superficial velocity was varied systematically between 0.03 0.19 m/s is shown in Figure 4.17 (c). The pressure drop is observed to increase with
increaseliquid superficial velocity. It reachesa maximum then decreases. Pressuredrop to maximum drop size recorded,Figure 4.21. maximum corresponds Normalized pressure drop for this case is plotted against liquid superficial velocity as in Figure 4.17 (d). It is observed that pressure continues to decline with shown liquid velocity. superficial
150
MB ALAMU 2010
Chapter 4
60
O VSL 0.05m/s
10000 E am
a
6000 d
4000 6
.LSO ; a a 40
00 00000
o000 o
CL 0
"' In N el 20 a 10
L.. "
30
`o
1`OV
a
O
2000 [0
50
10
20
30 velocity,
Gas superficial
40 Vu [m/s]
50
(b)
35 O VSL 14m/s 30 25 20 15 oNi p, 10 0bQ y=-135.8x+30.674 R=x0.9501 0, 0
4000
3000
O000
0000
00
CL 0
00
s
2D00
1000 N
0 0
0 0.05 0.15 0.1 Liquid superficial velocity, Va[m/s] 0.2
0
0 0.05 0.1 0.15 0.2 Liquid superficial velocity, V51[m/sj
(c)
(a)
Figure 4.17 (a) Pressure drop variation with Figure 4.17 (b) Dimensionless pressure drop velocity. Figure 4.17 (c) Pressure drop variation with Figure 4.17 (d) Dimensionless pressure drop superficial velocity.
increase gas superficial velocity. variation with increase gas superficial increase liquid superficial velocity. variation with increase liquid
Becausethe pressure drop data were acquired simultaneously with drop concentration and liquid hold-up a direct comparison is possible among the three data set. Liquid hold-up has been utilized in calculating gas core density. Core density is plotted against gas superficial in Figure 4.17 (e). velocity as shown
Chapter 4
E
tlo 4
V
13
13
Q Q Q 13 oap000000 QQ 00 O
i
cJ V
C u In f0 1 w 0
10
20
30
40
50
Figure 4.17 (e): Variation of gas core density with inlet conditions. It can therefore be concluded that entrained liquid fraction comes from the film burst. Film increasein burst produces droplets which causethe expansionof the gas core and subsequent density. This continuesuntil film becomesvery thin and can no longer sheddroplet. gas core The physic can also be explained as follows. An increase in gas superficial velocity at liquid superficial velocity results in an increase in the interfacial velocity at the constant liquid-gas interface. Momentum is transferred from the highly energetic gas core to the liquid film, resulting in an increase in the film velocity in the base area of the film. This increases interfacial shear stress and the friction factor which cause an increase in the pressuredrop.
Direct similarity has been found between pressure drop and wave amplitude as shown in Figure 4.17 (0 and Figure 4.17 (g) when pressuredata are qualitatively comparedwith wave amplitude profile.
152
MB ALAMU2010
Chapter 4
t
a i
c3 40
CL
30
a1$ d
00 oV
ti
Q"/`0, 13 C]
Is
CL 1 E
d
20 936 10
o oodrPo
op
w14e
30 40 50
3 os
0
0 0 10 20
10
20
30
40
50
(g)
Figure 4.17 : (f) Normalized pressure drop vs. gas superficial velocity. Figure 4.17 : (g) Wave amplitude vs. gas superficial velocity.
4.5.7.1 Pressure Drop and Entrained Fraction One of the questions that have not been addressedsatisfactorily in annular two-phase flow is whether or not the pressuredrop increaseswith entrained fraction. An attempt is in this subsection to provide original observations from the experiments which made this question. Since pressuredrop, film thickness and entrained fraction might answer data were acquired simultaneously,comparisonbetween one variable and anothercan be
made.
Figure 4.18 (a) shows how the system pressuredrop performs with respect to the entrained fraction. It is clear from the plot that pressuredrop in the system increasesmonotonically fraction and is higher with increased liquid superficial velocity. The with entrained was observedto follow power law. relationship was curve-fitted and
Chapter 4
12000
10000
E
8000
1.00
13 0.80
Yil
10
6000
a L
a L
13 Q
wisps
dY3
Aso
0.60
u 7
4000 2000
ooooso
Ik
,001
0.40
T
0.20
AQ
mist
0.00
FC
0
0 10 20 30 40 50 Gas superficial velocity VSG[m/s] ,
10000
r..
(--.
6-1
CL
cpe ,
..----. A JK
-..
t3,13
2000 o0.0
0.2
0.4
0.6
0.8
1.0
154 MB ALAMU2010
Chapter 4
AP =10887(EF>0.163
; VSL
_ O. 15m/S
Normalized pressure drop variation with entrained fraction is presented in Figure 4.18(c). In Figure 4.18(d), present data is compared with the data of Mantilla (2008) in a bigger diameter pipe. Mantilla calculated entrained who carried out similar work fraction from mass balance approach using film extraction technique. The work was horizontal and inclined pipe. carried out on
Gravity dominated regime 5000 QVSL= 0.05m/s
^ O
I 4000 I CL 3000
mist now
VSG 30m/s
OVSL=0.15m/s
o
0
2
0 7
2000
dp
1000
0
00
0 ,.
01
<>08
E3
'Vsc =30m/s
0. 00
0.20
0.40
0.60
0.80
1. 00
Vsc=21m/s
4.18 (c): Relationship between pressure drop and entrained fraction. Figure
155 MBALAMU2010
Chapter 4
gras it.
OV'SL
dominated
Qoss "VSI. N
SVSI
- 0.15w/s (present study) O N'SI. = 0.004 - 0.018m is( 6-Inch V'SI. = 0.034m' s (2-Inch ID)
ID)
ID)
5000
100000
F"
4000
& 'a,
a -01
e 00 10000
01.
3000
2000
do
1000
ob ooQ co
0.20
'
.. a a " r
mict tloiN SG ? 30m/s
C ie, pol
0
0.00
N'sc =Amis
EF = 0.05
Entrained
Fraction, EFI-]
(2008).
r-.
0< 0.005
4$
Q 0.004 ai M 3
0.003
CL E y 0.002
3 i 0.001 0 Z
0
0 10 20 30 velocity Gas superficial Figure 4.18 (e): Normalized wave amplitude 40 VSG[m/s] 50
as a function
of entrained
fraction
Chapter 4
In Figure 4.18 (c), pressuredrop falls as more droplets are torn off the wave crest into the fast moving gas core until a situation is reached where EF = 0.20. After this transition, pressure drop starts to recover and increasesmonotonically with entrained fraction. This transition point correspondsto VSG= 30 m/s. At this point and beyond, it is believed that film will undergocompleteatomization to promote mist annular flow. For gas producer the best operating practice would be to produce natural gas within the transition region with the following boundary conditions: 21 m/s _VsG < 30 m/s
0.05 EF< 0.20. _ This region is characterizedby minimum entrained fraction and minimum pressuredrop meansless shut-downsand more revenue. which In Figure 4.18(d), the trend in the present study and that of Mantilla is very similar they are acquired from different pipe diameter and flow orientation. The although Mantilla data at conditions of VSL= 0.0035 m/s and VSL= 0.018 m/s both show a slight drop at the beginning of mist flow. Mist flow occurs when there is no minimum pressure to drop size as gas velocity is increasing. Droplets are observedto be significant change homogeneouslydistributed in the gas phase. approximately Another good agreementbetween this work and that of Mantilla is that under gravity dominated flow, entrained fraction is less than 0.05. This is shown in Figure 4.18 (d) for data Mantilla took on 152.4 mm (6-Inch ID pipe) flow facility below superficial gas 21 m/s. Mantilla reported transition to annular flow occurs at a superficial velocity of 20 m/s. This is an excellent agreement with present study value of 21 gas velocity of
Chapter 4
m/s. At Vs0 = 21 m/s, drag from the gasphaseis just sufficient enoughto lift the droplet after it overcomesthe force of gravity. Another interesting observation is in Figure 4.18(e) where normalized wave amplitude (discuss in Chapter Six) was plotted against entrained fraction. Figure 4.18(c) and Figure 4.18(e) show that pressuredrop and wave amplitude behavior are analogousand The trend is similar in both cases. similar. 4.5.7.2 Pressure Drop and Drop Size - Drop coalescenceand break-up Majority of particle transportationessentiallytakes place due to the turbulent diffusion in the turbulent flow. In the caseof viscous liquids or in viscous sub-layer, on the other hand, when the length scaleof turbulenceis less than the kolmogorov scale of turbulence, i. e. AD< Ak,and molecular diffusion dominates.Typically, drops with a size comparableto the length scale of turbulencemay collide. The collision occur, in general, the length scale of turbulence is greaterthan the Kolmogorov scale of turbulence when (i. e. Ap> Ak) which is observedin the fully developedturbulent flow. Where:
3 0.25 Ak ER
(4.11) /P, [m2/s] kinematic viscosity of the medium, t7, = = dynamic viscosity of the medium, [kg/m-s] = density of the medium, [kg/hn3] p= dissipation energy per unit mass,[m2/s3] ER =
Chapter 4
Dissipation energy per unit mass,cR,for turbulent flow in a pipe can be calculated from the following expression, Shoham(2006): pressuregradient using OR , IdPI VM dL pm (4.12)
In pipe flows the turbulence length scalecan be estimatedfrom the hydraulic diameter. In fully developedpipe flow the turbulencelength scale is 7% of the hydraulic diameter i. e. 4=0.07D, D is the pipe hydraulic diameter. In this study length scaleof where
turbulence is estimatedto be 1.3 mm. Kolmogorov scale of turbulencewas estimatedusing Equation (4.12) from experimental data and plotted against superficial gas velocity and comparedwith SMD as shown in Figure 4.20. SMD is used in Figure 4.20 becauseit is smaller comparedto MMD at Figure 4.20 revealsthat the scaleof turbulence is higher for the thinner samecondition. liquid film; however, it decaysgenerally with increasing gas superficial velocity. the
Since length scale of turbulence and the drop diameter (SMD or MMD) are both greater Kolmogorov length scaleof turbulence, it is expectedthat drop coalescence will than dominate over break-up. Therefore, Kolmogorov scale of turbulence in Figure 4.20 will be used in the model to estimatedrop collision frequency in Chapter Five.
159 MBALAMU2010
Chapter 4
,., 8
130
110 90
12
" 00
&-go (3 0
10
W [3 000 13
0 fte
8 u 6
"2 70 .
L CL)
ooob3%(bo
50
30
1313
IL 4
L O
2
0
0 10 20 30 40 s0
H
O
10
Gassuperficial velocity VSG[m/s] , Figure 4.20: Comparison between drop size and Kolmogorov scale of turbulence.
An experimental campaign was conductedto investigate how pressuredrop varies with drop size at constant gas superficial velocity. The results of the revels additional details For drop break-up and coalescence. a constant gas superficial velocity of Vso = 14 about liquid superficial velocity were varied systematically between 0.03 m/s and 0.19 m/s, drop monitored by differential pressure cell drop is m/s corresponding pressure displayed in Figure 4.21. The figure shows an interesting feature with a maximum drop at VSL = 0.13 m/s after which the pressure drop start to decline with pressure further increasein liquid superficial velocity. distinct events can be deduced from plot - drop coalescenceand drop break-up. Two Drop break-up occurs when shear stresses (disruptive force) acting on the drop force (surface tension) When two drops collide, they interact overcomesthe restoring forming dump-bell. The liquid film between the drops starts to drain. If for some time
160 MB ALAMU2010
Chapter 4
drainage continues to the critical film thickness within the interaction time, coalescence will take place.
250
5000
E
A Kz ei 200
00
150
i0
'
O MMD
OOO
Pressure drop
4000
3000 01 Ma 2000
W h
E
100
50
oz
1000 0
0.0 0.1 0.2 0.3 liquid velocity.
Liquid
Figure 4.21: Pressure drop and MMD as a function of superficial VSG=14 m/s.
The drops with a size larger than a characteristic maximum stable drop size have a tendency to breakup, but smaller drops show a tendency to coalesce. Both coalescence break-up of drops seems to occur simultaneously. Since the data were acquired and drop size and pressure drop profiles by superimposing simultaneously, comparison of information the two profiles reveals additional as shown in Figure 4.21. The plot
displays one-to-one correspondence between the two data as Figure 4.21 depicts an link between the system pressure drop and the size of the droplet entrained unbreakable The positive pressure drop gradient represents drop coalescence events in the gas core. drop continue to increase in size as a result collision with another drop. However, the as indicates dominance of drop break-up events. negative gradient
161 MB ALAMU2010
Chapter 4
dominatedevent,drop size grows with liquid superficial velocity. A Under coalescence critical drop diameter (MMD = 248 m) is attainedwhere further increasein liquid flow in superficial velocity leadsto decrease drop size and the entrained rate at constant gas fraction. This diameter can also be referred to as the maximum stable drop size. Drop break-up occurs when the characteristicmaximum stable diameter is exceeded.After the inflection, the event in the dispersedphasebecomesdrop break-up dominated. point of in A correspondingdecrease entrainedfraction was also observedfor this condition. It is to seethat SMD do not behavein a similar manner to MMD. SMD continue noteworthy increasewith increasing liquid superficial velocity.
James et al. (1980) carried out experimental and analytical studies on the motion of droplets in two-phase flow. As a result of their visualization work (using a laser axial technique), Jameset al. found that larger droplets (>200 m), formed mainly from view bag break-up, tend to travel in straight lines at velocities similar to those at the time of their formation. Droplets formed by ligament break-up tend to stay in the core for much longer due to turbulent eddy interactionswithin the gas core.
Various explanations have been sought for pressure drop maxima behavior. Owen and Hewitt (1987) attributed it to suppressionof turbulence in the gas core by very large flux droplets which they observed under this condition. According to them, of entrained increasing the entrained droplet concentration reduces the turbulence, and hence the interfacial friction factor and the pressuregradient decreaseswith increasing gas mass flux. Another reason given by Owen and Hewitt (1987) responsible for this behavior is the changein interfacial structure.
Chapter 4
However, the augmentationof turbulence intensity reported by Azzopardi and Teixeira (1994 b) appearsto disagreewith the suppressionof turbulence reported by Owen and Hewitt (1987). It is noted that Hewitt and Owen based their statementpurely on the velocity profiles for the gas as measuredby, e.g., Gill et al. (1964). analysis of mean Theseprofiles they fitted to a log law Equation. Using a friction velocity obtained from the two-phase pressuredrop, they determineda 'two-phase von Karman constant'which they found differed from the classical single-phasevalue and which correlated with the momentum to that for the gas-drop mixture. They related the ratio of gas superficial in the von Karman constantto suppressionof turbulence. Azzopardi (1999) later change that the value of turbulence intensity increases with rate of entrainment and reported therefore entrained fraction. Hence, Azzopardi further proved that increasein turbulence intensity was due to newly createddrops.
Summarizing opinions by previous workers, Sawant et al. (2009) concluded that in flow smaller droplets which can follow the gas phase turbulence more closely annular have small slip velocity and increase of concentration of such droplets attenuatethe intensity. On the other hand larger droplets move at relatively slow velocity turbulence do not follow the turbulence very closely. Accordingly the increaseof concentration and the droplets enhances turbulenceintensity. of such
In conclusion of this aspect,the onset of suppressionof turbulence at maximum pressure by Owen and Hewitt (1987) and the claim by Azzopardi and Teixeira drop suggested droplet increasesturbulence intensity are both (1994 b) that newly created correct and
Chapter 4
in line with the observationof the presentstudy according to Figure 4.21. In Figure 4.21, the positive slope correspondcoalescence event with resultant effect of creation of more drops. The newly formed drop generates turbulence eddy causing increase in new interfacial shear stressand hence increasein pressuredrop. This trend continues until a is reached.At the maximum point, Figure 4.21, turbulence intensity is maximum point to be maximum. Therefore, the claim by Azzopardi and Teixeira (1994 b) that observed drops aggravateturbulence intensity is justified. newly created
In the same vein, at the maximum point in Figure 4.21, the pressure drop is also Beyond this point suppression of turbulence begins because bigger drops maximum. for turbulence generation start undergoing atomizing break-up. As the breakresponsible drop size and entrained fraction decreases and hence decrease in pressure up continues, drop. Again, the claim of Owen and Hewitt (1987) is justified.
Therefore, it can be concluded from the foregoing discussion that dispersed phase increasesas new drops are createdinto the gas stream core of annular flow. turbulence to The new drops may undergo coalescence form bigger drops. As the concentrationof drops increases, the entrained fraction systematically increases and therefore these drop increases.On the other hand, dispersed phase turbulence subsideswhen pressure dominating event is controlled by drop break-up causing drop size, entrained liquid the fraction and pressuredrop to decrease.
Chapter 4
liquid superficial velocity Vsi = 0.05 m/s. The data shows excellent agreement. same
u.u5
f
o0
f f
v 0.06 c O .p v
v C O u a 0.02
0.04
f_
0 0
0 10 20 30 40 50 60 70
Figure 4.22: Comparison of present data with data of Azzopardi et al. (1991).
Measured mean diameters are compared with the result of less advance instrumentation flow conditions as in Figure 4.23. From the results it looks (particle sizer) at very similar like the systems agree at high gas superficial velocities, but not at low flow rate. When 30m/s the value of D32and D50are almost identical for the two systems. Vsci > drop sizes (MMD & SMD) , in Figure 4.23, generally both data agree very In terms of
Chapter 4
well. However, excellent agreement was observed after gas superficial velocity, VSG> 30 m/s. Assuming the two systems have similar dynamic ranges measurements, then it appears that Spraytec under-estimates the particle size when it encounters a very big drop which implies that there were particles outside of the dynamic range of the system. The dynamic range of this version of Spraytec is: 0.5 m - 460 m.
200
1
AD50 : Particle Sizer A D32: Particle Sizer 0D32 : Spraytec- VSL = 0.05m/s f D32 : Spraytec V'SL= 0.15m/s D50 : Spraytec VSL = 0.05m/s 13 D50: Spraytec V'SL= 0.15m/s 11
v v E
140
EI 0
a 0 "
C
4)
80
wetpi> oe
10 20 30 40 50
: 00
60 70
20 0
Gassuperficialvelocity,VSG [m/s]
Figure 4.23: Comparison of present data with data of Azzopardi et al. (1991). point to While in Figure 4.23 is the distance from injection in the two experiments.
distance from the probe volume was set at 6.25 metres in the present study, it was 3.0 in the work with particle sizer, Azzopardi meter et al. (1991). It may be that flow was
the point where droplet sizes etc. was measured for the work with not fully developed at It may be that drop coalescence and drop break-up events are functions of particle sizer.
166
MB ALAMU 010
Chapter 4
It the distancesfrom injection points to the probe volumes in the two measurements. has been establishedfrom this study that drop coalescence and drop break-up contributes to hugely in determining size of the drop.
(b)"
40 q 35 u N 30 25 20 d '^ 15
Z
O O O O
o
1
Q00
a
N 12
d
0
0 000
0
C0
000
00
0 00o
00000 o 0
10 5
3"
0
10 20 Gas superficial 30 40 Vx Im/s] velocity, 50
0
0 40 30 io 10 [m/s] Gas superficial velocvity, VSG so
(a)
(b)
Figure 4.24: (a) Interfacial shear stress variation with gas superficial velocity. Figure 4.24: (b) Wall shear stress variation with gas superficial velocity. 70
Goo 400
60 s0 V
a N
40 30
4;; 20 I- 10
O
9.10
(c)
(d)
Figure 4.24: (c) Wall shear stress time series VSG= 42m/s, VSL= 0.15m/s. Figure 4.24: (d) Wall shear stress time series VSG= 31m/s, VSL= 0.05m/s.
167 MB ALAMU2010
Chapter 4
The interfacial shear stress generally increases with increase in liquid and gas superficial Figure 4.24 (a). Wall shear stress on the other hand, Figure 4.24 (b), displays a velocities
different characteristic from interfacial shear stress.
I' 0.25
Thine. ' 1. rertwl N tN dh4rn r n i11
-10000
02
Nt Y ois
n a_ sm c O
6000 d cC O 40()0 U O Rf A . ( lau o(rotranrd lprnd dropkl% ........................,
C, 2000
VSG: 42m/s VSL 0.1Sm/s =
0.05
S. 5 5.6 57 Time fsl 5.6 59
6 555.6 53 me (s)5 8
5.9
Figure 4.24 (c): Photo e idence of packets of ring disturbance waves occurring time varying film exceeds 30 m/s. The corresponding gas superficial velocity higher drop disturbance waves (top left) where the corresponding thickness/ time = 5.85 seconds (top right) are sourced. at concentration
when
is seen to increase with liquid and gas superficial velocities. However, Wall shear stress annular flow, it is peaked and then decreases. For thicker at the transition to co-current
Chapter 4
film it continues to decreases while for thinner film it recovers and continues to increase with increasing gas superficial velocity.
Close to VsG = 30 m/s, time-resolved wall shearstressbegins to fluctuate from positive to negative value. The negativity becomes stronger with increasing gas superficial Figure 4.24 (c) and Figure 4.24 (d) display the time series of the wall shear velocity. for some of the conditions tested. The reason for this behavior is due to the stress of packets of waves referred to as `ring disturbance waves' by Mantilla emergence (2008) which covered thinner part of the liquid film around the pipe circumference Figure 4.24 (e). These waves were observed to form close to the transition to mist V5G= 30 m/s. Mantilla reported occurrenceof these waves in 2-inch (50.8mm) around diameter horizontal pipe within 30 m/ s- 50 m/s superficial gas velocity. This is mechanism governing dispersed phase flow distribution in evidence suggesting same annular two-phaseflow. Sawant et al. (2008) attributed this behavior to instability of film thickness. In their flow was visualized and film was observed to be highly unstable. A theoretical experiment Moalem Maron and Dukler (1984) and experimental data obtained by Zabaraset al. study of (1986) showed that in this region wall shear stress fluctuates around zero and takes both which results in an internal recycle in the liquid film. The negative and positive values at low gas velocity and close to the churn-annular flow regime experiment was carried out boundary. However, the conditions at which these behaviors were observed in transition further away from transition boundary and well into annular flow regime as presentstudy are In fact, the negativity/fluctuation becomesstronger by existing transition models. predicted increases. as gas superficial velocity
Chapter 4
Another possible explanation for this retrogradeflow behavior may be as a result of internal liquid droplets in the gas core as explained by Sleicher (1962).When drops are vibration of them travel in axial direction. Some pinch off and rebounce on hitting the created some of liquid film createsdisturbancewaves. In some casesthe disruption causesinternal recycling liquid film or retrograde flow. The retrograde flow lowers the interfacial shearstressand of hence pressuredrop. Sleicher's argumentcontradicts observation in the present study. In the pressure increasesas gas superficial velocity. Therefore, Sleicher's argument present study doesnot explain our experimentalobservation.
by disturbancewaves around the pipe circumference produces a corresponding covered drop concentration according to the characteristic signature displayed by drop higher time series in Figure 4.24 (e). The time series of drop concentration concentration further shows a chain of smaller droplets almost of same sizes corresponding to the liquid film where film is uniformly fairly distributed around the pipe section of It is therefore reasonableto suggest that thinner film produces higher circumference. drop concentration and hence higher entrained fraction. Structures of other flow in the dispersedare considerednext. variables
Chapter 4
Figure 4.25 (a) displays typical time varying plots of principal flow variables in annular flow. Flow structures are not clear in the film thickness signal. However, when
film thickness is used to generateinterfacial shear stress the flow structures measured become very obvious as indicated in Figure 4.25(a). The time series of drop concentration, Mass Median Diameter all display similar signatures with similar peaks troughs at the sametime interval in space. and
0.1$
0.1$ F E 0.14
0.12 0.1 0.0$ IE
2000 1800
1600 1400
1200 1000
$00 600 400 200
0
2.2 2.4
2.6
2.8
Timet: 1
24
2.6
2.6
70
0 60
250
200
N s0
V
N 40
y150
300
w
H
30
cY
50
20 c
0L
10 L 2
2,2
2.4
2.6
2.8
2.2
2.8
Time [s]
Figure 4.25 (a): Time series of flow variable showing flow structures.
Figure 4.25 (b) the thinner film covered by disturbance waves around the pipe In the film thicknessis close to zero produced a correspondingbigger circumferencewhere
171 MBALAMU2010
Chapter 4
liquid droplet entrained in the gas core as shown in the drop concentration structure of time series.
2
a a 0 Iz
c ai
6000
4000
CL 0 L
0 2000 VSG 42m/s VSL=0.15m/s =
0
5.5 5.6 5.7 5.8 Time [s] 5.9 6
Figure 4.25 (b): Time series of drop concentration showing flow structures The time series further shows a chain of smaller droplets almost of samesize where the film is uniformly distributed around the pipe circumference. Figure 4.25 (c) shows that this structure created from disturbance waves generatea positive interfacial shearstressand a negative wall shearstress.
Chapter 4
500 450 to vn 400 CL N 350 IA 40 300 250 a v 200 150 "-, 100 50 VSG = 42m/s VSL = 0.15m/s
0
5.5 5.6
5.8
5.9
5.5
5.6
5.8
5.9
Figure 4.25 (c): Time series of Interfacial and wall shear stress
173 MB ALAMU2010
Chapter 4
MLE
= P1V
(4.13)
Where VD is the drop velocity, pi is the liquid density; c is the volumetric drop concentration M LE is the entrained mass flux. Because film thickness and film velocity is known, and film flow rate can be estimated using Equation (2.37).
100
80 E 0 > 60
O
O8 o0.
o
41
10
do,
"
F.
40 0
OAzzopardi et al. (1991) & Fore & Dukler (1995) Q Azzopardi & Teixera (1994) Zaidi et al. (1998) A Westende et aI.(2007) QVSL = 0.05m/s : present study QVSL = 0.15m/s : present study
20
0
0
20
40
60
80
100
Multiplying
Equation (2.37) by film density and divide by area of the pipe produces film
4gF. OL flux, i. c. 7kF = where qF = ApVsL(1 - fE). Total liquid mass flux can be mass pz , from the product of liquid phase density and the superficial liquid velocity estimated i. e. rim. = PLVSL. Therefore, entrained mass flux, Equation (4.13) can be estimated by
Chapter 4
rLE
= mL -
drop velocity can be expressedas VD =m pL LF,where VD is the drop velocity, PL is the liquid density; c is the volumetric drop concentrationand thLE is the entrainedmassflux.
Alternatively, entrained mass flux can be obtained from film withdrawal technique by balance. Drop velocity at the centre-line can therefore, be approximated carrying out mass from Equation (4.13) by dividing entrained mass flux by liquid density and drop
concentration.
Results obtained are shown in Figure 4.26. There is excellent agreementbetween present in the drop velocity obtainedby various researchers the past. study and
Chuptcr 4
4.0 E
3.0 a"
Q VSL = 0.15m/s
QQo
13 QQ
0
2.0
7 C LL
0 o0 00000
1.0
0
0 10 20 30 velocity 40 VSG[m/s] , velocity. 50
0.0
Gas superficial Figure 4.27: Film velocity
as a function
gas superficial
Predicted film velocity using the mechanistic model of Alves et al. (1991) is in excellent the measured film velocity as shown in Figure 4.28. agreement well with
6.0 OVSL= 0.05m/s QVSL= 0.15m/s 0 VSG = 14m/s
4.0
5.0 E
u 0
3.0
2.0
1.0
0.0
p123456 Measured film velocity [m/s]
4.26: Comparison of film velocity with the model of Alves et al. (1991). Figure
176
Al B ALAa1C' 20/0
Chapter 4
Figure 4.27 shows that film velocity increases as gas rate increases, reaches a maximum
VSG= 21 m/s, at the beginning of annular flow when liquid film seemsto start around undergoing complete atomization. Due to changeof controlling mechanism (huge wave to disturbance), film velocity then drops as the gas rate increasesfurther for VSL= 0.15 Film velocity increases,reachesa maximum close to transition to annular flow, m/s. then drops around VsG = 21 m/s and recoversagain, increasing linearly with increasein for the caseof VSL= 0.05 m/s. gas superficial velocity The most interesting part of the film behaviour is the changesobservedat the transition VSG= 21 m/s. This is not surprising becausefilm velocity is directly proportional around to entrained fraction according to Equation (4.14). According to Figure 4.27, the liquid film at VSL = 0.15 m/s film has a higher velocity the liquid film at VSL= 0.05 m/s. This implies that the film has more energy and is than in momentum and will exhibit more flow structures as liquid superficial velocity richer increasesat constant gas rate constant. Therefore, more momentum will be transferred from the highly energetic gas core to the liquid film, resulting in an increasein the film in the base area of the film. This explains why liquid film velocity is higher in velocity the caseof VSL= 0.15 m/s. drop velocities relationship will follow similar trend Figure 4.27 since gas Film and drop velocity at the centre-line are approximately equal. velocity and
Chapter 4
4.11 Probability
Time varying drop size can be examinedfurther in both amplitude and frequency space. The amplitude variation can be carried out using the Probability Density Function (PDF). This is the probability of occurrenceof each drop size as instrument scansthe
detector. In the present study, drop size distribution was tracked using Probability for a series of gas
superficial velocities at a liquid superficial velocity, VSL = 0.05m/s as shown in Figure 4.29. The shape of the distribution function describes the extent of the homogeneity or heterogeneity of the particle size in the distribution. The Probability Density Function to the right with increasing time, an indication of dominance of shows marked shifts The PDF is characterized by multiple peaks and series of maximums coalescence events. finger-prints multi-modal distribution before gas superficial velocity of 30 which clearly m/s.
An interesting feature is however, noticed after VsG = 30m/s when the drop size distribution changesfrom multiple peaksto single peak with only one maximum. The possible explanation for this behaviour is that after superficial velocity of 30 m/s, the drop size distribution becomesuniform as pressure drop starts to recover as flow from annular to mist annular flow. The peak becomes narrower and regime changes increase suggesting more uniform distribution. taller as superficial gas velocities Analysis carried out using modified Rosin-Rammler (1963) model fails to predict MMD before VSG= 30m/s. This is an indication that the model in its original form performs best for mono-modal distribution, and the experimental data was multimodal. Rosin is treatedin the subsequent sub-section. Rammler analysis
Chapter 4
12 10
!
8 001
ii
il
' II. II
II
iI II
6
4 2 0 0
I'
iI II II
i i
I
1I r
40
30 20 300 10
Figure 4.29: Changes in drop size distribution tracked by the Probability Density Function. There is a change from multiple to single peak after gas superficial 30 m/s marked transition to annular mist flow as D50(MMD) decreases. velocity of VSG= 0.05 m/s.
Figure 4.30 shows the time series and the corresponding PDF of dynamic data taken to the beginning of annular flow as predicted by the transition model of Turner et close (1969). The figures show how PDF can be used to determine the modality of the drop al. distribution and the pressure drop in the system. The multiple peakconcentration distribution noticed at this inlet condition VSG= 14.08 m/s and VSL= 0.19 m/s , which is to be in fully developedannular flow according to the droplet model of Turner supposed the presence , in the gas core, of partially atomized fractal of liquid film, suggests to as wisps or agglomerateof liquid body carried by huge wave. otherwise referred for the presence of wisps in this region is the incomplete The only explanation by chaotic and unstable nature of liquid film. The atomization process promoted in chum region has been reported by Azzopardi et al. (2008). The presenceof wisps
Chapter 4
PDF of the drop concentration, Mass Median Diameter (MMD) and the pressure drop all
show twin peak suggesting bi-modality and heterogeneousnature of the distribution, Figure 4.30. Each peak signifies a different event. The surface of the liquid film can be coveredby disturbanceand huge waves when they co-exist in annular flow. Depending on the mechanismof drop entrainment, either huge disturbance wave could dominate. Surprisingly, the PDF of the wall film which or the drops displays a single peak implying homogenousand dominancenature produced of huge wave over disturbance wave. According to Figure 4.30, the PDF of the time film thickness is characterizedby a broader and serratedpeak. Thus, the wave averaged the liquid film appearsto be a product of meeting or merger of two independent on traversing the gas-liquid interface. The resultant wave has higher amplitude and waves henceproduceshigher drop size Figure 4.30, Appendix D.
180 MB ALAMU2010
Chapter 4
0.1
a2s"ss Time Isi Liquid 0 0.1 0.2 03 0.4 0.5 (mm) 0.6 0.7 0.2 Film Thickness
7 F
I::
o.s
". ]s ". g
15 o. 1
2 a
". os "
A1]11ii 'Ilme (i1
300 E i
"
280
260 240
220
L
B 61
0 100 150 Mass Median 200 Diameter, 250 Do (Nm) 300
i2s4ss Time(s]
2600
23
2500
20
2400
is
a 2300
10 2200
2100.
2000 1 012eese
0
2000 2100 2200 2200 Pressure 2400 Drop 2500 [Pa/mJ 2600 1700 2600
Time [s]
Figure 4.30: Time series and PDF of data showing bi-modality at VSG=14.08 m/s, VSL= 0.19 m/s. Film thickness was acquired from conductance probes, drop from Spraytec and pressure drop form differential pressures cell. concentration
Chapter 4
The new wave is characterizedby velocity observedto be greaterthan the velocity of the thus, appearsto fit description of huge wave reported by Sekoguchiand parent wave and Takeishi (1989). The inlet condition at which the PDF in Figure 4.30 was generatedfalls the wispy annular regime in Figure 4.31, sameflow regime where Sekoguchiand within Takeishi spotted wisps in the gascore. The classification of mist annular flow region on flow pattern map Figure 4.31 is based drop size collected after VSG= 30m/s. Mist flow is characterizedby gas phase partly on a continuous phase where liquid condensate, oil or water exists as very small as homogeneouslydistributed droplets. approximately
Chapter 4
(1969) and Sawant et al. (2009) statementsthat the transition to annular wispy flow takesplace at higher liquid phaseReynolds number.
According the presentfindings, wispy annular prevails in annular flow until gas 21 m/s is exceeded. This agreeswith previous studies i.e. Mantilla superficial velocity of (2008), where measurements drop size and drop concentrationwere not carried out. of Also, the existing mechanisticmodels, Table 4.3, have establishedthe non-dependence this transition on superficial liquid velocity. It is therefore, proposedthat this of transition in the new map should be independentof liquid superficial velocity. The by line is represented dotted line in Figure 4.31 at superficial gasvelocity of proposed 21 m/s. This marks the new transition to co-current annular flow. Wave characteristic this transition will be detailed in Chapter Five. associatedwith Transition to mist annular flow occurs when gas superficial velocity exceeds30 m/s. There are convincing evidencefrom PDF, MMD and flow visualization studies.This is also not dependenton liquid superficial velocity. Therefore, the transition the discontinuous line in Figure 4.31 at gas superficial velocity, VsG= 30 m/s represents transition boundary to mist annular flow. new
Chapter 4
OV SL-0.05mIs
13VSL-0.15m/s
:n
OVSG-14m/s
10 E
Ind
Bnbbh
: r r' 1'1
l ya
.0d.
JCX8E3t 0%%
_
---------------
W 0.001 1
10 Superficial gasvelocity, V5 [m/s) Figure 4.31: Proposed flow pattern map showing modification to wispy annular transition boundary of Hewitt & Roberts (1969) and the new transition boundary to flow. The open symbols represent the position of the present study on mist annular the new flow map.
100
R= e-GN)n
(4.15)
R= 1-
(4.16)
Chapter 4
R=n
n (n)n e-n),
(4.17)
R is the retained weight fraction of particles with a diameter greater than D, D is where the particle size and N is the mean particle size, and n is a measure of the spreadof particle sizes. Re-writing Equation (4.16) yields:
1-R
(4.18)
log, (1-
R) = loge e-G)
(4.19)
0-
loge(1-
R) =
(4.20)
Substituting for 0, on LHS and expandingEquation (4.20) generates: 1oge[O loge(1Recall, 0= loge 1 R)] =n loge D-n loge N (4.21) (4.22)
Equation (4.22) into Equation (4.21) gives: Putting loge[loge 1- loge(l - R)] =n log, D-n loge N (4.23)
Further simplification of Equation (4.23) yields the final form of Rosin Rammler:
185 MB ALAMU2010
Chapter 4
rt loge D-n =
log,, N
(4.24)
Equation (4.24) is then in a straight-line form: y= mx +c Where: [loge (11R)J loge (4.25)
y=
(4.26)
X= loge D
(4.27) (4.28)
(4.29)
N=
el-(Dl
(4.30)
D and n are the fitting parameters. Therefore, plotting Equation (4.26) againstEquation (4.27) will produce a straight line, Figure 4.32, with a slope m and intercept C on the vertical axis. The approachabove has beenusedto representthe present data. A good straight line intercept on the vertical axis. However, it was found that the was generatedwith a clear the drop sizeswas poor especially at low superficial gasvelocity model prediction of by distribution are characterized several maximums or multiple peaks. where the
186 MB ALAMU2010
Chapter 4
4.4: Particle Size Distribution table for gas superficial velocity, VSG = 14.08 m/s and liquid superficial velocity, VSL = 0.19 m/s.
Size 0.001 0.002 0.002 %< 0.00 0.00 0.00 % In 0.00 0.00 0.00 Size 0.126 0.158 0.200 %< 0.00 0.00 0.00 % In 0.00 0.00 0.00 Size 12.59 15.85 1995 %< 0.58 0.87 1.25 % In 0.17 0.29 0.38
Table
0.003
0.003 0.004
0.00
0.00 0.00
0.00
0.00 0.00
0.251
0.316 0.398
0.00
0.00 0.00
0.00
0.00 0.00
25.12
3152 39.81
1.65
2.50 4.48
0.40
0.85 1.97
0.005 0.006
0.008
0.00 0.00
0.00
0.00 0.00
0.00
0.501 0.631
0.794
0.00 0.01
0.01
0.00 0.00
0.00
50.12 63.10
79.43
5.24 5.27
5.27
0.76 0.03
0.00
0.010
0.013 0.016
0.00
0.00 0.00
0.00
0.00 0.00
1.00
1.26 1.58
0.02
0.02 0.03
0.01
0.01 0.01
100.00
125.89 158.49
5.85
2157 32.63
0.58
1621 10.77
0.020 0.025
0.032
0.00 0.00
0.00
0.00 0.00
0.00
2.00 2.51
3.16
0.04 0.06
0.08
0.01 0.02
0.02
199.53 251.19
316.23
3222 33.93
8024
0.18 1.12
4691
0.040 0.050
0.063
0.00 0.00
0.00
0.00 0.00
0.00
3.98 5.01
6.31
0.12 0.17
0.23
0.03 0.05
0.06
398.11 501.19
630.96
100.00 100.00
100.00
19.16 0.00
0.00
0.079
0.00
0.00
7.94
0.31
0.08
794.33
100.00
0.00
Fitting Rosin"RammlerParameters
U1. V't9L{0-1JJ. 06
10 8 6 "'4 oc
zJ
..
02468 LN[D]
10
Figure 4.32: Predicting of mean drop diameter using Rosin Rammler model.
187 MBALAMU2010
Chapter 4
The particle size distribution in Table 4.4 was generatedfrom RTSizer, Spraytec The predicted MMD by Rosin Rammler approachgave MMD = 133.82 m software. while the measuredMMD = 241.7m, Appendix D. This is an indication that the model in its original form cannot satisfactorily predict drop size characterizeby multiple peak distribution. Going by the presentargumentof this study, Rosin-Rammler (1963) model fail to predict MMD before gas superficial velocity of 30 m/s but will perform better will in the single distribution, mist annular flow regime.
188 MB ALAMU2010
Chapter 4
i
O %\allis (1969)
Measured = Predicted (2002a) OPan and Hanratty
0.8
-,
r
0.6 z.
vI
OOliemans
et al. (1986)
- 0.4 ii W s 0.2
y_
0o 0
0
0
0.2 0.4 0.6 Fraction 0.8 EFI-I
rElp 19 'o
0
\Ie: isimetl Entrained
Figure 4.33: Prediction of measured entrained fraction data using existing correlations. It observable from Figure 4.33 that all models under predict entrained fraction after fraction of 0.4. This is an inadequacy of the existing correlations in predicting entrained fraction accurately in vertical co-current annular two-phase flow. Therefore, a entrained mechanistic model with sound physical basis is advocated for predicting entrained fraction in vertical annular two-phase flow. Table 4.5: Statistical results of models performance (24 data points)
M ode is
APE
AAPE
STDEV
Sawant et al. (2009) Wallis (1969) Ishii & Mishima (1989) Pan & Hanratty (2002a) Oliemans et al.(1986)
189 A4 ALAA.fU20 10
Chapter 4
M ode is
APE
AAPE
STDEV
36
-17.38 -11.85
54.57
70.19 48.92
153.77
122.3 82.5
-47.99 352.56
54.98 361.48
43.93 442.97
The statistical parameters used in Tables 4.5 and 4.6 respectively are defined as follows: APE = Average Percentage Error AAPE = Absolute Average Percentage Error STDEV = Standard Deviation
d32
ML P VL
The first term in the RHS of the Equation (4.31) is the contribution from drop break-up term represents contribution from drop coalescence. while the second
Chapter 4
200
600
Azzopardi (2006)
120 0 so CL CL
013 , 00
Q VSLa 0.15m/s
aoo
M
9 d
300 O
200
C .
40
OVSL 0.15m/s pVSL=O. OSm/s O VSG. 34m/s 0 100 200 Measured 300 400
a
0 40 80 Measured 120 D321 pm] 160 200
0
500 600 D32[ pm]
(a)
Azzopardi (1997,2006).
(b)
Figure 4.34: Comparison of predicted and measured SMD using the model of
Another simple model was suggested by Azzopardi (2006) for vertical annular twoflow. It is expressed as follows: phase zaZ
PLVSL
Psc vsc
jpsGvsc
(4.32)
The model prediction against the measureddata is shown in Figure 4.34 (b). The model poor comparedto that of Equation (4.31) as shown in Figure 4.34 performance was very (a). Tuning parameter E in the coalescence term of Equation (4.16) was varied between 1 and 3 at various flow regimes within annular flow to determine systematically influence of drop coalescence the predictive nature of the model. For instance,in the on flow in the case of VSL= 0.05 m/s and VS0> 30 m/s, best fit was obtained mist annular when E=2.5 (MLE = 125 kg/m2-s) with maximum relative error of 20%. Between 21
Vso 5 30 m/s, using E=2.5 over-predicts drop sizes and gives values which are m/s
191 MB ALAMU2010
Chapter 4
coalescence is almost zero. This obviously contradicts the physical reasoning of the In the caseof VSL= 0.15 m/s, best match with measureddata was got when E= model. 0.3 (MLE = 45 kg/m2-s) in the mist flow regime where VSG> 30 m/s. The maximum is less than -10%. relative error recorded Between 21 m/s 2 VSG:5 30 m/s, using E=0.025 (MLE = 3.75 kg/m2-s) the model
predicts drop size with maximum relative error of -30%. For the case of VSG< 21 m/s, E=0.0000125 (MLE= 0.01875 kg/m2-s) gives best result with maximum relative using error of - 48%. Overall the model does not give good result at low superficial gas velocity. For the case VSG= 14 m/s, the model over-predicteddrop sizes and the values are unrealistic. of
4.16 Discussions
Pressuredrop was normalized with wall shearstresscalculated using the model of Alves (1991). The result was shown in Figure 4.35. For the thicker film where superficial et al. liquid velocity, VSL= 0.15 m/s, pressureminimum occurs at gas superficial velocity, VsG = 21 m/s. This agrees very well with previous studies on churn - annular transition boundary of Barbosa et al. (2002) and van Westendeet al. (2007). Both reported gas 21 m/s for transition from churn to annular in vertical pipe, Table superficial velocity of film, the minimum is not clear, Figure 4.35. 4.2. For the thinner
192 MB ALAMU2010
Chapter 4
In term of drop size, thicker film produceshigher SauterMean Diameter (SMD) before transition to annular flow. At the transition boundary and beyond, thinner film overtakes thicker film curve producing higher Sauter Mean Diameter (SMD) as gas superficial velocity continues to increase,Figures 4.8,4.9 and 4.35.
In Figure 4.9, normalized MMD curve for thicker film (VSL = 0.15 m/s), crossed-over thinner film (VSL= 0.05 m/s) curve at VSG= 30 m/s. The cross-over point marks the transition to mist annular flow. Beyond this transition, thinner film producesbigger drop becauseof the greater interaction it enjoys with the gas core. Thicker film on the other hand experiencesless interaction with the gas core due to conservation of momentum. However, in mist flow (Vso >_ 30 m/s) where the droplets size are smaller and homogeneouslydistributed, flow at higher liquid rate will produce more approximately constant gas rate. This explains why normalized drop-size is higher for VSL structuresat 30 0.15 m/s after VsG>_ m/s. =
In relation to entrained fraction, pressure continues to drop as more droplets are in the gas core until a minimum is reached where EF = 0.20. After this entrained transition, pressure drop starts to recover and increases monotonically with entrained fraction. This transition and point of recovery correspondsto VSG= 30 m/s. At this point beyond, it is believed that liquid film has undergone complete atomization. Liquid and into the gas core and droplet deposition on the liquid film is steady, entrainment flow pattern being mist annular. The drag force of the gas phase must resulting tension and gravity forces for a fully developed annular flow to overcome the surface influence of gravity film is less stable and chaotic. occur.Under the
193 MB ALAMU2010
Chapter 4
Figure 4.35 shows combined normalized flow variables as a function of superficial gas liquid rates. From the present data gravity dominates flow before velocity at constant transition to co-current annular flow at gas superficial velocity, VsG= 21 m/s.
00 0 "*
bc dpo
0*0
;1O o. * f o* o* o K"
I gravitydominated
nowrgIme
1 G4
10 v
0.1 0.01
IA,
w 0.001
10 20 30 40 50 Gassuperficial velocity,VSG [m/s]
Figure 4.35: Normalized flow variables vs. gas superficial velocity. At lower gas superficial velocity and before transition to co-current annular flow, thinner film produces higher entrained fraction. However, the entrained fraction curve for film (VSL= 0.15 m/s) overtakesentrained fraction curve for thinner film (VSL_ thicker 0.05 m/s) at transition to co-current annular flow at gas superficial velocity, Vsc = 21 the transition boundary, the thicker film produces higher entrained m/s. After crossing
194 MBALAMU2010
Chapter 4
fraction than the thinner film, both continue to increasemonotonically as gas superficial increases.Liquid film before the transition boundary is believed to be covered velocity by huge waves although it co-exists disturbance wave. The dominance of principally huge wave has been establishedbasedon the standarddeviation of the film thickness. However, Disturbance waves dominatesthe flow after VsG = 21 m/s and coexistedwith huge wave until gas superficial velocity of 30 m/s. After gas superficial velocity of 30 m/s annular mist flow begins, huge wave disappears;disturbance wave then co-exists with ripple waves.
As gas superficial velocity increasesat fixed liquid superficial velocity, the slip within the gas-liquid interface increases,leading to a higher interfacial shear, and hence the total pressure-gradient increases. However, when pressure drop is normalized the behavior changes as in Figure 4.35. Pressure drop is characterized by a minimum the dominanceof gravity forces. It recovers after transition to annular flow as suggesting to increase.The amount of entrained liquid fraction gas superficial velocity continues increasesas gas flux increases.
Westendeet al. (2007) working with air/water at inlet conditions similar to the present in a 50mm diameter pipe also reported that pressure-gradient and amount of study fraction experienceda minimum around Vs(; = 20 m/s corresponding with a entrained densimetric Froude number = 1.When gas superficial velocity is less than 20 m/s i. e. Frg 1, they said liquid film thickness and the wave height become much larger. At this < however, according to Zabaras et al. (1986), wall shear stress fluctuates, transition, directed upward, and film churning occurs. Below the transition point, they occasionally
195 MB ALAMU2010
Chapter 4
also reported the presenceof large interfacial waves which are propelled fast with the gas flow, which was recognized as huge waves reported by Sekoguchi and Takeishi (1989). Zabaras et al. (1986) referred to the onset of liquid down-flow as flow reversal point and in general occur at a gas velocity below the minimum in the pressuregradient
curve.
An interesting phenomenon occurs at a superficial velocity of 21 m/s for thinner film (liquid superficial velocity of 0.05 m/s). The liquid film velocity, pressuregradient, wall shear stress, liquid hold-up, wave frequency and wave length all decreasewhile drop (entrained fraction) D32(Sauter Mean Diameter) and wave amplitude all concentration , increase significantly. This behavior was first reported by Weihong et al. (2001) who the impact this behavior might have on the development of predictive emphasized for low-liquid-loading wet gas pipelines. Weihong et al. did not give any reason models the mechanism that was responsible for this behavior. They carried out nor explained their experimental investigation on 50.1mm internal diameter pipe with inclination from horizontal of -2, -1, 00,1 and 20.The superficial gasvelocity at which this angles occurredwas reported as 25 m/s. unusual phenomenon
4.17 Conclusions
From the results and discussions presented , dynamic drop size including drop fraction/film hold-up and pressuredrop information have been used concentration,void in vertical annular two-phaseflow to accomplish the following:
Chapter 4
1. New, time-resolved drop size and drop concentration data have been obtained film hold-up and pressuredrop information. All these parameters simultaneously with show fluctuations with time. Some, such as film hold-up, are more obviously periodic than others. 2. The time averagedvalues are in agreementwith prior data. Both measuredSMD and entrained fraction captured transition to co-current annular flow at gas superficial velocity of 21 m/s. MMD profile on the other hand detected transition to mist annular flow at a gas superficial velocity of 30 m/s. 3. Considering MMD (Dsa), thinner film produces bigger drops only at this boundary 21 m/s 2 Vso S 30 m/s. Before Vso = 21 m/s, and after VsG= 30 m/s, thicker condition: film producesbigger droplets. 4. In terms of SMD (D32),thicker film produces bigger drop up to Vs0 = 21 m/s. After V50 = 21 m/s, thinner film producesbigger drop. 5. Below VSG= 21 m/s, thinner film (VSL= 0.05 m/s) produces higher entrainedfraction. After transition to co-current annular flow at gas superficial velocity, Vso = 21 m/s, fraction becomeshigher the higher the film thickness or film hold-up (i. e. VSL entrained = 0.15 m/s). 6. Below entrained fraction of 0.05 (EF= 0.05) and VSG< 21 m/s droplet flow is inertialdriven and gravity appearsto have significant effect on entrained fraction. The flow is huge waves dominate the gas-liquid interface. chaotic and
Chapter 4
Drop size or MMD is large under gravity-dominated flow. Effects of gravity diminish beyond entrained fraction of 0.05 (i. e. EF > 0.05). After EF= 0.05 pressure drops fraction until a minimum is reachedwhere EF= 0.20, VsG= 30 uniformly with entrained
MIS.
7. Pressuredrop recovers after EF = 0.20, increaseswith increase entrained fraction as velocity increases.Mist flow occurs as a result of complete atomization gas superficial of liquid film. 8. In mist flow regime droplet size do not change significantly with increase gas
superficialvelocity.
9. Examination of the time series in amplitude and frequency space reveals interesting features. PDF of MMD displays multiple peaks (multi-modal distributions) before gas VsG = 30m/s, suggesting heterogeneousnature of the drop sizes. superficial velocity, This marks transition to mist annular flow earlier detected by MMD profile. In mist flow, after VsG = 30m/s, PDF changesto single peak (mono-modal distribution) annular by homogenousuniformly distributed drop size. characterize 10. Wispy annular was observedwhere they were not expected. This is confirmed with footage of the flow. Transition from wispy to mist annular has been modified to video this observation as well as transition to mist annular flow on the proposed,new reflect flow pattern map. fraction, transition to mist annular when this boundary 11. In terms of entrained is satisfied: VSG? 30m/s, EF? 0.20. After this transition, normalized pressure condition fraction as gas superficial velocity increases. drop increaseswith entrained
198
MB ALAMU 2010
Chapter 4
12. For gas producer the best operating practice would be to produce natural gas within the transition region with the following boundary conditions: 21 m/s 30m/s <_ _VSG
0.05 >_EF 0.20 < This region is characterized by minimum entrained fraction and pressure drop which meansless shut-downsand more revenue. 13. Entrained fraction is a strong function of gas superficial velocity in two-phase vertical annular flow. 14. A new flow pattern map is proposed based on subtle changes in drop sizes and entrained fraction measurement. 15. The standarddeviation of averagefilm thickness is directly proportional to the drop distribution. size 16. At superficial gas velocity of 30 m/s, the following interesting phenomenaoccur: (i) Standarddeviation of mean film thicknessbecomesinertial dependent (ii) Mean drop size (MMD) becomesdependenton superficial liquid velocity. (iii) PDF of drop size distribution changes from heterogeneous to homogeneous distribution.
199 MB ALAMU2010
Chapter 5
Chapter 5
Wave Dynamics in Vertical Annular
Two-Phase Flow
5.1 Introduction
The mechanismsof atomization of part of the liquid film to form drops in annular twophase flow are not entirely understood. It has been observed that drop creation only occurs when there are large disturbance waves present on the film interface. Woodmanseeand Hanratty (1969) observedthat ripples on thesewaves were a precursor to drops. Woodmansee& Hanratty (1969) showed evidence that the creation of drops from the film on the channel walls did not take place from of the film but specifically from called disturbance waves which travel over the film at periodic structures, usually few meters per second.Azzopardi (2006) presentsmore evidenceof this. velocities of a Therefore, it might be expected that there would be an interrelationship between the fluctuations of drop concentration and the frequency of disturbance waves. Unfortunately, hitherto there has been hardly any work presenting information about the in which the drop concentrationvaries with time. The most useful study is that of way Azzopardi & Whalley (1980 a) who used a camera looking axially up the pipe to record drops passingup the pipe. deeper level there is need to understandinterfacial hydrodynamics, exchangeand At a betweenthe liquid film and the gas core in annular two-phase transfer of momentum
200 MB ALAMU2010
Chapter 5
flow becausea lot of topical issuesassociated with the flow are still unresolved. One of the ways in which the problem can be addressedis to measure and report simultaneously the wave properties, film hold-up, drop concentration and the system pressuredrop.
Understanding wave properties is a starting point to developing a physically based,
mechanistic model from which entrained liquid fraction can be predicted in vertical annular two-phaseflow. Results from present study shows that characteristic drop sizes, entrained fraction and the waves that created them are closely related. In this Chapter, measured wave
properties are presented.
Wave properties can be classified into two: - primary and secondaryproperties,basedon the findings of this work. First, the primary properties are the properties inherent in the fundamental to its description. Primary wave properties are dynamic wave and be usedto describethe wave. Theseproperties are statedbelow: properties which can " " " " Wave frequency Wave velocity Wave spacing and Wave amplitude
The secondaryproperties are regardedas those properties that are not directly used describe the wave but are important to its evolution and dynamics. Secondary to listed below: wave properties are
201 MB ALAMU2010
Chapter
Wave Dynamics
in Vertical
Annular
Two-Phase
Flow
" "
Some of these properties are schematically described in Figure 5.1 and Figure 5.2 respectively. Next. some of the measured wave properties are discussed.
a : , ~s
L!
(a)
p+dp
5t
Ac
pc gsinO
tp
Figure 5.1: Annular flo'tt' schematics showing primary Alves et al. (1991). properties,
Chapter 5
" "
Some of these properties are schematically described in Figure 5.1 and Figure 5.2 respectively. Next, some of the measured wave properties are discussed.
aB
(a)
p+dp
$ Ac
Tr
TI
pc gsing
tp
showing
primary
and secondary
wave
Chapter 5
6Fb
A5W (1)
FD
w 21.1
F
Droplet shedding
Vw
Fg
Wave height, SH
Figure 5.2: Schematic of a unit wave including forces resolution around a droplet.
x(t)
- xl
[x
(t + kA T) - x]dt
; zz T
(5.1)
Where: T= sampling time, s KAi = time delay, s interrogating time delay and; i=
Chapter 5
x=_
JT x(t)dt
0
Po(f)=
Or(2
R(0)+
2; rjkAr) (5.2)
Where w(kAr) is a windowing function. In this analysis basic Cosine windowing function was used as used in Azzopardi et al. (2008). It is expressedas:
(zker w(kLr)=Cos 2r )
(5.3)
In order to obtain frequency of two-phase flow structure we adopted the procedure to the time-resolved drop concentrationand wave (film) data shown in Figures 5.2 above (a) and 5.2 (b) in calculating autocorrelationsand Power Spectrum Densities displayed in Figures 5.3 (a) and Figures 5.3 (b).
204 MBALAMU2010
Chapter
Two-Phase
Flow
0.75
-VSG =14m/s VSL = 0.15m/s -VSG = 32m/s VSL= 0.15m/s VSG = 13m/s VSL = 0.05m/s -VSG = 33m/s VSL = 0.05m/s VSL=O. OSm/s --VSG=42 VSL=0.15m/s -VSG=40m/s
E E
4A a, c
0.5
I'
E 0.25
LL
o
5
5.1
5.3
5.4
5.5
Figure 10000
film
thickness
information
a a 1000
0
c
u
C
v
a 0 O
100
10
Time [s]
Figure 5.3 (b): Time averaged drop concentration information. drop concentration was logged during the experiments with Malvern I Mme-ri' ved Spra`q,:c equipped with Insitec card and time-averaged film thickness was acquired pair of flush-mounted ring conductance probes. err.ploving a
205 MB i IAfU2010
Chapter 5
1
VSG = 14m/s VSL= 0.1Sm/s
1
VSG = 32m/s VSL uqnid = 0.15m/s
0.8 c c
Lq dream
-D o (I'll
0.8 c .V C
Fnm Con[
0
0.6
0.4
0.6
0.4 0.2
-Drop
0.2 O
V
O Q
-0.2 -0.4 L 0
0.2 0.4 0.6 0.8 100.2
(a):
Figure Figure 5.4 (a): Autocorrelation 5.4 (b): Autocorrelation
I 0.1 0.01 0.001 0.0001 0.00001 0.000001 a 0.0000001 1E-08 1E-09 1E-10 1E-11 1F. 12 0 100 200 300 f[Hzj 400 500 600 I I VSG= 14m/s VSL= 0.15m/s liquid film
(b)
vs. time delay for huge wave. vs. time delay for disturbance
10 1 0.1 0.01 0.001 0.0001 0.00001 0.000001 0.0000001 if-08 1E-09 1E-10 1E11 1E-12 0.1 1 10 Frequency, f[Hz] 100 1000 VSG= 32m/s VSL =0.15m/s F [ __ "
wave.
Drop Conc.
uwann, o. op CoM
Frequency,
(c)
Figure 5.5 (c): Power Spectrum Figure 5.5 (d): Power Spectrum Density vs. wave frequency Density vs. wave frequency
(d)
for huge wave. for disturbance
wave.
The data in Figure 5.4 (a) and Figure 5.4 (b) show how the autocorrelation decays with There is less correlation at finite time than at time, t=0s. time. the peak at delay time =0s after The position of a trough
time varying drop concentration and time varying film hold-up in the case of case of disturbance. The inverse of the time delay at the trough indicates approximately half the dominant frequency.
Chapter 5
Power Spectrum Densities of these autocorrelationsare displayed in Figure 5.5 (a) and Figure 5.5 (b) respectively. Power Spectrum Densities Figure 5.5 (a) and Figure 5.5 (b) indicate the strength of occurrence of each frequency. The peak indicates the most dominant frequency. Values of dominant frequencies obtained for the waves traversing the gas-liquid interface for the test matrix implementedin this study are plotted against gas superficial in Figure 5.6. velocity as shown
"VSL= 0.05m/s 100.0 gravity dominated now
N
+
a ?eQ
dQQ
10.0
a
QpQ ppO
Hug. waves
1p p
'01CIO
a'
ea
Huge waves dominated
1.0
0 20 10 30 40 Gas superficial velocity, Vsc [m/s] 50
Figure 5.6: Wave frequency as a function of gas superficial velocity. Observations from Figure 5.6 show that at constant superficial gas velocity, the increaseswith increasein superficial liquid velocity. The possible reasonfor frequency behaviour may be due to the fact that at lower liquid superficial velocity, less energy this is required from the gas to keep the waves moving. However, at higher liquid flow rate is required from the energetic gas core to keep the wave in motion. more energy
207 MB ALAMU2010
Chapter 5
Therefore, more flow structuresare createdas a result of the higher momentum transfer which translates to higher structure frequency as observed in Figure 5.6 as liquid superficial velocity increasesas in the caseof VSL= 0.15 m/s. Frequency attributes of the wave can be explained further in terms of wave velocity. Wave frequency is directly proportional to wave velocity (i. e. f= v/?). Since the wave velocity is also higher as liquid superficial velocity increases,Figure 5.4 (a). Therefore, it is not surprising that frequency is higher as liquid superficial velocity increases. Again, for a fixed gas rate, as liquid flow rate increasesmore waves occur due to energy creation as a result of exchangeand transfer of momentum between the film and the gas core, resulting in higher frequency. In Figure 5.6 three waves have been identified namely: huge, disturbance and ripple The waves traversing the liquid film interface in annular flow have been waves. into two main categories by several researchersSawant et al. (2008); ripple classified disturbancewaves (Hewitt and Hall-Taylor, 1970; Azzopardi, 1997). and Ripple waves are relatively small wavelength and small amplitude dynamic waves traveling at much lower velocity comparedto the velocity of disturbancewaves and gas These waves are also non-coherentand lose their identity after traveling a short phase. distance. On the other hand, disturbancewaves are large amplitude kinematic waves and higher than the velocity of liquid film and ripple waves. travel at velocity much They are coherent and dominate the interfacial transfer of mass, momentum and energy. Disturbance waves also act as a roughnessto the central gas flow and contribute to the frictional pressuredrop in annular flow.
208 MB ALAMU2010
Chapter 5
Ring disturbance waves were identified in this study around VsG>_ m/s. The evolution 30 of this wave influences drop size (MMD) distribution at this transition boundary where VsG = 30 m/s and beyond. This is becausering disturbance wave appearsto dominate and control the creation and dispersion of entrained liquid droplets in the gas phases after VsG 2 30 m/s. If this wave could be thought of as an elastic material then explanation as to why D50of VSL= 0.15 m/s is greater than D50of VSL= 0.05 m/s can be provided. Then the degree of springiness seemsto be higher as the superficial liquid increasesat fixed superficial gas velocity. Therefore, more elastic D50of VSL= velocity 0.15 m/s tends to offer less resistance to the driving force of the dispersed phase in greater amplitude, greater drop size than in the case of D50of VSL = 0.05 resulting m/s which appearsto be more rigid.
100.0 O VSG=14m/s
N =
10.0 D'
000 0000000000
C)
3
1.0
0 0.05 0.1 0.15 0.2
209 MB ALAMU2010
Chapter 5
Figure 5.7 shows fluctuation in wave frequency around 10 Hz. The fluctuation is caused by chaotic nature of the liquid film. The Figure gives an indication that wave frequency is a
strong function superficial gas velocity. The wave frequency remains constant because gas
flow rate is fixed. The liquid film is covered by huge wave and the entrained fraction is less than 0.05. Wave frequency reaches a maximum and then drops as liquid superficial velocity increases from
0.03 to 0.20 m/s. This is an attribute of huge wave with frequency attaining a maximum and
then drops as the wave propagates. Huge waves are characterized by large amplitude waves as shown on Figure 5.8 and hence high energy. In this case, wave amplitude reduces as liquid flow rate increases at nearly similar gas mass flux, Figure 5.9.
0.5
E E 0.4
N N
0.3
E 0.2 U.
0.1 0
5 5.1 5.2 Time [s] 5.3 5.4 5.5
of huge waves.
210 MBALAMU'010
Chapter 5
r-. E2
'0 4
a
1.5
O O0 ,q O
00
ai
.r
CL ro 1
E
0.5
o cp
3
0 0 0.1 0.2 0.3 0.4
5.10, it can be seen that there is a great deal of scatter at the lower velocity at lower gas superficial velocities.
superficial
since the drops are far away from each is very low
the coalescence may not occur. Hence, the fluctuation and liquid superficial velocity
However, on the other hand, if the drop concentration potential outliers. in the case for higher liquid superficial velocity, the interaction
211
MB ALAMU 2010
Chapter 5
Nevertheless, the characteristic frequenciesof the fluctuations in drop concentration are to increasewith both gasand liquid velocities, Figure 5.10. seen
30
"
N
VSL= 0.15ui/s
20
C
C1
L
a 10 0 L.
0 1; 1' i
0
0
llama
10 20 30 40 50
frequency values extractedfrom the film hold-up data, Figure 5.6, Referring to the wave it is more obvious that this frequency is that of disturbance waves travelling on the film Again, this wave frequency increases with increase in both gas and liquid surface. superficial velocities.
frequency and wave frequenciesas a function of gas and liquid superficial Plotting drop in Figure 5.11, it is seenthat the wave frequencies are higher than the drop velocities as This ties in with the limited prior datum reported by frequencies in most cases.
Chapter 5
Azzopardi and Whalley (1980 a). Howe ver, at some low drop concentrationsdrop frequencies are seento be higher than their correspondingwave frequencies. Generally, that in every atomization event, that the drop concentration frequency one would expect be greater than the disturbancewave frequency. would
30 I
w. 20
aJ
131Q
0
y"" INui
Cr (U
4)
10
10
20
30
40
50
frequency of fluctuations in drop concentration,fD, is less than the frequency of That the disturbance waves,fw, is surprising as one would expect the contrary. Examination of high speed cine/video footage taken looking axially down the pipe would lead one to drop creation eventsper wave, i. e.,fD>fw. A possible explanation for the expect several frequency of drops is that drops from one atomization event become lower than expected diffused into those from other events as they pass along the pipe. Each event probably
Chapter 5
distribution of sizes. Small drops are acceleratedmore rapidly and so arrive produces a at the measuring position earlier that the larger drops that are harder to accelerate. Therefore, the faster drops could catch up with slower drops from a previous event and then undergo coalescencein a different event to form new drops. It is also possible that there is dynamic equilibrium betweendrop formation and coalescence.
It is noted that there is another type of event which might be being picked up in the frequency analysis. It had been reported by Hall Taylor et al. (1963), Hall Taylor and Nedderman (1968), Wilkes et al. (1983) that the frequency of disturbance waves falls from an initially high value to a lower one higher up the vertical pipe. This was Though the waves had a characteristic attributed to the meeting and merging of waves. (mean) velocity, there was a distribution about this mean and faster waves could catch in front of them. The combined wave would be much bigger than up with slower waves likely to be broken up to create drops. Wilkes et al. (1983) normal and so more from these wave coalescenceevents could be estimated that the amount of entrainment half the total entrainment. Perhapsit is the frequency of coalescenceevents that is being drop concentrationfrequency,fD. reported as
214 MB ALAMU2010
Chapter 5
different instrument. The collision frequency increases with the increasing drop concentration and dissipation energies.
Ju
2.5
MN
nQ,
20
2.0
r-, i
u
1100
1.5
zr
1.0
1o += p 0.5
0
A
0
0 10 20
0.0
30 40 50
Gassuperficialvelocity [ni/s]
Figure 5.12: Similarity between drop frequency and drop size distribution.
215 MBALAMU2010
Chapter 5
diameter divided by the liquid superficial velocity) has been found to be inversely proportional to the Lockhart-Martinelli parameter. This implies that the frequency is directly proportional to the gas superficial velocity but that the effect of liquid superficial velocity is a minor effect. Data from the present experiments, when plotted in this matter shows the self-same trend. Strouhal number is a dimensionlessnumber given by: fWd
Sty = VsL
(5.4)
Azzopardi (2004) showed that bubbly and slug flow regime frequency data could be Strouhal number and Lockhart-Martinelli parameter. Because of the correlated using behavior of the flow structuresbetween annular, slug and bubbly flows, Strouhal analogous Lockhart-Martinelli parameter are utilized to correlate frequency data of the number and in Figure 5.13 (a) for wave frequency and Figure 5.13 (b) for wave present study as shown and drop frequenciesrespectively.
An empirical correlation for wave frequency in terms of Strouhal number was generated by curve-fitting based on the present data. The correlation shows that Strouhal number decays with Lockhart-Martinelli parameter.However, a better correlation was obtained liquid based Strouhal number was plotted against Lockhart-Martinelli when the log-log chart as suggestedby Azzopardi (2004). The data displayed in parameter on a Figure 5.13 (a) shows a linear relationship with negative slope. The best fit obtained
216 MB ALAMU2010
Chapter 5
followed a power law. The resulting correlation is given by the expression by Equation 5.5 as follows:
St
L=0.4292
X -0.908
(5.5)
Where:
St
f.,, d
=
PLV
0.5
SG
VSL
PG VSG
10
Nf a)
.C 3
C
E
"L
1 0.01 0.1
Chapter 5
inadequacy of the Strouhal number and Lockhart-Martinelli parameter in correlation drop frequency.
100
00 10 a+
.a
..
:1
7. 10
9 r
(l U. 1
0.01
0.01 0.1 Lockhart-Martinelli par, aineter, X [-] 1
Figure 5.13 (b): Relationship between the dimensionless wave frequency and the (the square root of the ratio of the superficial Lockhart-Martinelli parameter fluxes of liquid and gas). momentum
The trend observed in Figure 5.13 (a) is similar to the trend reported by Kaji (2008) and Mantilla (2008) in Figure 5.14(a) and Figure 5.14(b) respectively. Both correlatedwave frequency using liquid based Strouhal number. Kaji used same flow facility as used in Kaji did not carry out drop size measurement. this study.
218
MB ALAMU 2010
Chapter 5
25
30
40
50 st
60 0
11)
0.1
0.01 0.01
0.1
10
100
Lockhart-Martinelli
parameter
Figure 5.14 (a): Correlation of Strouhal number with Lockhart-Martinelli parameter, Kaji (2008).
I 000
" X. 100
24nch, Water Water-uranal A 2-inch, " Nnch.Water-Gycerin o -6ArxK Walen " Gerad2005 Frkaro (1983) fo vag Shearer (1964)
-voa
10
St=025X':Za S
A
00001
0.001
0.01 x (-)
0.1
5.14 (b): Correlation of Strouhal number with Lockhart-Martinelli Figure Mantilla (2008). parameter,
Chapter 5
Kaji's data was mostly taken in slug, bubbly and chum flow regimes. Kaji did not give an empirical correlation for his data. However, the trend in Kaji's data and that of
present study is similar velocities. with present data extending the trend to higher gas superficial 1.0 >_ Sti. <_10 and 0.01 >_ X51.0 on Kaji's line in
Figure 5.14(a). Therefore, both data can be represented empirically by same Equation as proposed by Equation 5.5. Mantilla (2008) also included data from other sources to correlate his data as shown in Figure 5.14 (b). Again, the trend looks similar.
Q 50 mm 0 67 mm IV"SL= 0.25]
,-
10000
diameter
using gas
is observed when gas based Strouhal number is used to correlate A different trend information as reported by Azzopardi et al. (2008) in Figure 5.15. This shows frequency
Chapter 5
Strouhal number to increase linearly with increase in Lockhart-Martinelli parameter. The present data agreesvery well other data from different pipe diameter as shown in Figure 5.15.
221 MB ALAMU2010
Chapter 5
These models derive their namesfrom collision theory on which they were built, namely constant efficiency, dynamic deformation, static deformation and kinetic theories. Thus discussions. the theories are summarizedin the subsequent The constant efficiency model was derived by considering the collision frequency between drops in a turbulent flow field in a manneranalogousto that usedto derive kinetic theory of gases,Delichatsios and Probstein,(1975). Writing the collision frequency as the product of the collision cross section and the root mean squarerelative fluctuations. In order to facilitate the final expressionof the model, assumption velocity has to be made that a constantfraction of all collisions result in drop coalescence, the frequency From this view of coalescence coalescence the constant efficiency expression. be written as the product of collision frequency and a coalescence efficiency or may probability.
dynamic and static deformation theories,two different models were derived. Based on First, when the inertia of the colliding drops is neglectedand the characteristictime of force fluctuations is much smaller than the drop deformation, the deformation turbulent drops may be determinedan averageturbulence force squeezingthe drops. This of the force given by the product of pressurefluctuations acrossthe drop pair and the average becauseof the absenceof turbulence drops is assumed positive projected area of the between the contacting drops. The resulting model, referred to as the static deformation increasein the coalescence efficiency with increasing drop size. On model, predicts an if the drop shapesrespondto every fluctuation in the field, their the other hand force squeezingthe drop pair. This gives is determinedby the instantaneous deformation the dynamic deformation model, which predicts a to the model referred to as rise
222 MB ALAMU2010
Chapter 5
In violent collisions, the inertia of the drops is the governing factor and the effectiveness of the collision dependson the adequacyof the collision kinetic energy to accomplish film drainage work. By performing a dimensional analysis on this processone may the fourth coalescence expression,termed kinetic collision model. This model arrive at frequency similar to static deformation of yields a drop size dependence coalescence though it was obtained from a very different form of analysis. model,
The frequencies derived from the kinetic collision model and the static deformation drop size dependence, predict the evolving transients model, which show very similar best. These frequenciesare in particular, able to track the drift in the peak of the drop distribution and also at least qualitatively reproducethe bimodal feature of the size frequency expressionderived from in their experiment study. The coalescence spectra dynamic deformation model predicts oppositetrends in the displacementof the peak the fails to predict the bimodality in the experiment. The quantitative agreementis also and frequency function obtained by assumingthe Interestingly, the coalescence poor. to be a constant(collision frequency) predicts the transients coalescenceefficiency the frequenciesfrom the kinetic collision and the satisfactorily although not as well as deformation models. This is explained from the fact that the latter models predict static for the larger drops. In other words, these models reduceto an efficiency of nearly unity for the larger drops. These results show that the the constant efficiency model frequency is increasing functions of well as the coalescence efficiency as coalescence The transient coalescenceexperiments the drop sizes for the situations considered.
223 MB ALAMU2010
Chapter 5
frequencyis an increasing function of the drop pair sizes. indicate that the coalescence
Recently, a more pragmatic approachhas beensought to model collision frequency in turbulent two-phase flow. Ceylan et al. (2003) derived a mechanistic model drops in a collision field, Figure 8. The authors utilized 1-D undergoing coalescence diffusivity Equations in radial coordinatewith inclusion of eddy diffusion coefficient to describe the flow. The turbulent diffusion coefficient of the medium is related to the flow and therefore, it may be statedin terms of the micro scaleof the characteristics of turbulence which is different for both turbulent and molecular diffusion. Empirical introduced to facilitate description of turbulent coefficient in terms of scale constant was turbulence with a tacit assumptionthat drops are equally sized. of
Collision field
ei
Two-phase
flow
of 5.16: Collision field and coalescence drops in annular two-phase flow. Figure therefore can provide solution to transient drop final Equation is time dependent, The in two-phase flow. For fully developedturbulent flow Ceylan proposed size distribution determinecollision frequency in turbulent dispersion: the following equation to
Chapter S
ti = 16 Where:
(Po R a o
(QZ) R
1/3 (5.6)
co= collision frequency, Hz cpo= drop concentration, R: transport degreeof particles by turbulent flow (0 < R < 1.0 ), -a= empirical constant,a= drop size, m ER= turbulent dissipation rate, m2/S3
The tuning parameterin Equation (5.6) is IRa which was empirically determinedto be 0.72. When this value is used in Equation (5.6), the value of collision frequency was However, the bestmatch was achievedwhen the tuning parameter, pRa over-estimated. 0.1875. Substituting this value into Equation (5.6) producesexpressionbelow: =
Chapter
Wave Dynamics
in Vertical
Annular
Two-Phase
Flow
0 VSL = 0.05m/s ImprovedmodelQ V'SL= 0.15m/s Improvedmodel 'SL= 0.05m/s Measured N'SL=0.15m/s Measured -
30
0
r. Cj IM C
20
Q
10
0
G
a
0 0 10 20 30
AM
40 50
Gassuperficialvelocity fin/s]
Figure 5.17: Comparison frequency. wave of experimental drop frequency with modal drop and
Chapter 5
3.5
3.0
OVSL=o. 05m/s
3.0
E 2.5 2.0
0 O oa OO 00
O VSL=14m/s
OO Oo O
0 E
2.5
0
000
Ooo0 00
2.o
0
IS
ar
1.5
1.0 0.s 00
0 10
Gas superficial
40 velocity, VG [m/s]
30
so
0.1 velocity,
0.15 V$L[m/s]
0.2
(a)
(b)
Figure 5.18 (a): Wave velocity variation with gas superficial velocity. Figure 5.18 (b): Wave velocity variation with liquid superficial velocity.
It is observed that the wave velocity becomes higher as liquid superficial velocity increases.This can be as explained in terms of gas core density Figure 5.19. A lower gas density decreasesthe momentum and energy of the gas being transferred to the liquid film thus generating a lower velocity wave. It has also been observed that wave velocity changesaround the transition boundaries. While it is very obvious around VSG= 30 m/s, the change in velocity gradient is not clear around VSG= 21 m/s.
227 MBALAMU2010
Chapter 5
5 p VSL= 0.05m/s
M4
DVSL=0.15m/s
Q0 13 13 0 13 0 r, 000000
w V lu 1
pp 44P0000
10
20
30
40
50
KV _
F+VSPc K+ Pc
Where:
(5.9)
228 MB ALAMU2010
Chapter 5
D (1-EF)
4SF
(5.10)
where:
L f/g
Lil l
(5.12)
Results obtained are shown in Figure 5.20 (a) and Figure 5.20 (b) for the two cases investigated. Pearce (1979) model shows superiority to Kumar et al. (2002) in the first set of experimental campaign. Pearce(1979) model gives maximum relative error of +14.79 % % for K=0.30. Kumar's model on the other hand predicted and a minimum of -40.90 the wave velocity with maximum relative error of +68.52% and a minimum of -23.92 % for the casesof VSL= 0.05 m/s and VSL= 0.15 m/s respectively. In the case of VSG= 14 m/s, Pearce(1979) predicted huge wave velocity within relative +5 % with K=0.65. It is not surprising that K is very high in this case. error range of Pearcecoefficient, K, is a function of wave amplitude and amplitude of huge wave being larger than that of regular disturbancewave. This is another proof that in the caseof VSG 14 m/s the gas-liquid interface is dominatedby huge wave. =
229 MB ALAMU2010
Chapter S
Kumar model produces +79.13 % maximum relative error and +71.12 % minimum relative error in the caseof VSG= 14 m/s.
6.0
OP. a . . td419791k. DP.. O5 /s 0.90: VSl. O. a .t4. (1979). K. OAO: VSt. O.ISm/s r. t Il. (200: ): vsl. .tr. oasw/s
6.0 0 Pearce at W. (1979), k 0.65: 5.0 VSG a 14m/s Q Kumar at al. (2002) : VSG z 14m/s
5.0 4.0 d
3.0
"km.. okumr
0
d
2.0
0
a
1.0
0.0 L 0
0O
123456 Measured wave velocity [m/s]
0.0
0123456
ad3O3 13
Measured wave velocity (m/s]
(a)
(b)
Figure 5.20: (b) Prediction of wave velocity by various model VSG=14 m/s.
It can be concluded that K, the Pearcecoefficient, is not only a function of pipe diameter but also a function of inlet condition. This is becauseit is a function of wave amplitude is a function of gas and liquid superficial velocities. and wave amplitude
230 MB ALAMU2010
Chapter 5
(5.13)
Wave spacing obtained from Equation (5.13) is plotted against gas superficial velocity as displayed in Figure 5.21 (a). The figure shows that disturbancewave spacing changes transition boundaries where VSG= 21 m/s, 30 m/s and 40 m/s (for VSL = 0.05 around Wave spacingtransition for VSL= 0.15 m/s at VSG= 40 m/s is unclear. m/s) respectively.
0.3
4000
O VSL+ 0.05m/s OVSL O. lSm/s
0.2
o
0.2
00
3000 J
opO00
1d' o 0130
G
00 00 o0 0
eZ
C
C
0.1 v
m 2000
13
0.1
0.0
V
1000
00 Q ,pQ OQ EID
0
10 20 30 40 50 0 10 20 Gas superficial velocity, VsG[m/s] Gas superficial 40 velocity, VsG[m/s) 30 50
(a)
(b)
Figure 5.21: (a) Wave spacing variation with gas superficial velocity. Figure 5.21: (b) Dimensionless wave spacing variation with gas superficial velocity.
VSL = 0.05 m/s and VSL = 0.15 m/s in Figure 5.21 (a), wave spacing is For the case of indistinguishable before VSG= 21m/s. After VsG = 21m/s, wave spacing becomeshigher as The reason responsible for this observationmay be liquid velocity decreases. the superficial inertial and mass density possessby the liquid film when VSL= 0.15 m/s. As a result higher
Chapter 5
liquid film corresponding to VSL= 0.15 m/s tend to resist the force exerted by the dispersed gas phase. This increase in resistancecausesreduction in wave amplitude and hence wave
spacing.
When wavelength is normalized by film thickness result shows that in transition boundaries wavelengthis seento be higher for VSL= 0.05 m/s. are still very obvious and In the case of Vso = 14 m/s wavelength increaseswith increasein liquid superficial velocity Figure 5.22 (a). However, when the normalized wave spacing is plotted against liquid in Figure 5.22 (b), the wave spacing is observed to decreasewith superficial velocity as increase in liquid superficial velocity Figure 5.22 (b)
0.3 0.3 E 0.2
c ' 0.2 CL N
1400
p VSL=14m/s O VSLz 14m/s
13 Q0Q
13
(3 13
to 1200
E3 Qa
.lP
m N
e 1000
Q Q O QQa C3
13
800 600
13
400 tu 200 0
1313
0.1
0.15
0.2
0.1 velocity
velocity, V 1[m/sJ
0.15 V5jm/s] ,
0.2
(a)
(b)
Figure 5.22 (a) Wave spacing variation with liquid superficial velocity VsG = 14 m/s. Figure 5.22 (b) Dimensionless wave spacing variation with liquid superficial velocity VSG =14m/s.
232 MB ALAMU2010
Chapter 5
Wave spacing has been observedto be a revealing wave property according to this study. For instance when film thickness is normalized by wave spacing as shown in Figure 5.23, the various transition boundaries become readily visible. Again, when the wave amplitude is
normalized by wave spacing the result is shown in Figure 5.24. The figure reveals a
signature similar to the measurednormalized drop size profile Figure 4.9 (a). The similarity of Figure 4.9 (a) and Figure 5.24 shows that the proposed wave amplitude prediction (Chapter Six) is very reliable.
pVSL= 0.15m/s
0.0015 vi U, a c v
Qp Q DQ
00 Q0 0 0 00 Q
0.001 E
S 0.0005 Z
O <>
0
0 10 20 30 40 50
233 MB ALAMU2010
Chapter 5
4
1
1 1 1 1 1 1
1 1
1
1 1 1
CJ
13 j9, b-bPP' b
0 0 10 20 30 40 50
5.9 Conclusions
From the results and discussions presented above the following conclusions can be drawn from dynamic measurements wave properties: of 1. Dynamic properties of the wave changesas wave identity changes.Wave properties transition boundarieswithin annular flow sub-regimes. Flow pattern transition change at two-phaseflow occurs at VSG= 21 m/s, 30 m/s and 40 m/s respectively. within annular Wave frequency attains a maxima and then drops within wispy annular i. e. Om/s> Vs0 5 21 m/s for VSL= 0.05 m/s and 0 m/s > VSG:530 m/s for the caseof VSL= 0.15 m/s. dominance of huge or disturbance wave depends on liquid superficial velocity 2. The is maintainedat a constant rate. when gas superficial velocity
234 MB ALAMU2010
Chapter 5
3. Wave frequency: " " " is a strong function of gas superficial velocity increaseswith increasein gas superficial velocity increaseswith increasein liquid superficial velocity
4. Wave velocity: " " increaseswith increasein gas superficial velocity in increaseswith decrease liquid superficial velocity
5. Wave spacing: " " 6. increaseswith increasein gas superficial velocity in increaseswith decrease liquid superficial velocity Wave fluctuation with time is higher when compared with than drop concentration
fluctuation with respectto time. 7. Drop frequency is directly proportional to the standard deviation of the film
thickness. 8. Correlating wave and drop frequency using Strouhal number and Lockharta good correlation for wave frequency. However, the parameter give
Martinelli
inadequateto model drop frequency. correlating parameterproves 9. After gas superficial velocity, Vs0 = 30 m/s, the dominating mechanismbecomes
ligament break-up as huge wave disappears.The liquid film becomes predominantly by disturbancewave. covered 10. to co-current annular flow, gas superficial velocity, VSG= 21 m/s, At transition
huge and disturbancewaves co-exist. Disturbance wave dominates over huge waves.
235 MB ALAMU2010
Chapter 5
11.
Before gas superficial velocity VSG = 21 m/s, huge wave and ephemeral ,
disturbance wave covered the liquid film interface huge wave dominates over , disturbance waves. 12. 13. Drop collision frequency is directly proportional to drop size distribution. At low drop concentration (usually at low gas velocity), drop frequency, in some
to be higher than wave frequency, the controlling mechanism being cases, are seen ligament break-up. Meeting and merger of waves occur, wave bursts into droplet, and because the newly formed droplets are far apart coalescence hardly takes place drop concentrationis low under this condition. suggesting why
236 MB ALAMU2010
Chapter 6
Chapter 6
Wave Amplitude and Dynamics in Annular Two-Phase Flow
6.1 Introduction
Wave amplitude is defined as the difference between the wave peak height and wave base height. While the importance of studying the features of the waves was realized and by many investigators, few studies focused on such aspects.The wave amplitude suggested has not been satisfactorily predicted in the literature. This could be largely attributed to the the phenomena with a number of interrelated factors influencing the wave complexity of Such factors include condensation,evaporation and the balance of forces that amplitude. influence the wave structure. Such forces are exerted by, for example, simultaneously and interfacial shear stress exerted by the gas stream on gravitational acceleration, pressure the wave front and back and by the surface tension forces. The inter-dependenceand theseforces on the evolution of the wave shapeare very complex. coupling effects of Past work on wave amplitude modeling is usually based on cut-off criterion developed by DeJong (1999) defined as the sum of the averagefilm thickness plus one standarddeviation film thickness. Sawant et al. (2008) stated that accurate measurement of wave of the is not possible using the current conductanceprobe technique. This is becausethe amplitude has finite length in axial direction. The measuredfilm thickness is an averageover the probe the measuredwave heights are lower than the actual wave probe width and heights referring to the classical work of Hall Taylor (1966) and Martin(1983). However, Sawant et al. (2008) suggestedthis limitation should not affect the statistical characteristics
Chapter 6
disturbance waves such as frequency, velocity and wavelength but warned that the values of the measured amplitude presentedin that study did not representthe absolute amplitude of therefore should only be used to study the trends in variation of disturbance wave and the change in boundary conditions. Next, we discuss modeling of wave amplitude with in annular two-phaseflow. amplitude
TI = CwLAX
(6.1)
Where: Cw is a factor that accounts for the effect of the surface tension on the internal flow or dissipation flow in the wave. the circulation and liquid phaseviscosity, kg/m-s L= VLF= liquid film velocity, m/s ab,y= wave amplitude, m
238 MB ALAMU2010
Chapter 6
Since the hydrodynamics inside the wave crest can be describedin terms of viscous and tension force C,, is defined in terms of the liquid viscosity number, N. It is the surface the viscous force induced by an internal flow to the surface tension force as ratio of given by the following expression:
(iv) and;
Cw =c
ItL
N=
[PL sap
1/a 1
(6.2) Ishii and Grolmes (1975) correlated the value of cwbasedon data from several sources, as follows:
1= 3c.
11 78 N08 .
N, for 15
(6.3)
(6.4)
VF_
_
4sL
(6.5)
film thickness and entrained fraction measured from experiment and are timeSince it is possible from the Equation (6.5) to time-resolve film velocity becausefilm resolved,
Chapter 6
velocity is proportional to the entrained liquid fraction. From this film velocity distribution profile can be calculated,the result is presentedin Figure 6.1 (b). The Alves et al. (1991) mechanistic model for annular flow in vertical and off-vertical pipes is a 1-D two-fluid model. This annular flow model is formulated basedon different flow geometry and different physical phenomena.The model enablesdetailed prediction the annular flow- pattern characteristics,including the velocity distribution, liquidof film thickness, gas void fraction, and pressure gradient. Therefore, all other flow directly in this study will be calculated utilizing basedAlves. et variables not measured al. (1991) models. The model expressessome of the flow parametersas follows in the next discussion. Interfacial shearstressis expressed as:
Pc(vc -vF)2 Z_ fj 2 ,
(6.6)
in Where the interfacial friction factor is expressed terms of a dimensionlessparameter I, being expressedas: 1i fsc
I=
(6.7)
I =1+300
(6.8)
Where:
240 MB ALAMU2010
Chapter 6
dF =4SF(d-SF)ld
(6.9)
Rec =
; dc = (d -28F)
Pc = Pc ac + PL(1- ac) ;
PC = PGaC + PL (1- ac )
d= pipe diameter, [m]; SF= film thickness, [mm]; fsc = gas core friction factor, [-]
density, [kg/m3]; uc =gas core viscosity, [kg/m-s]; ac = gas core pc = gas core liquid hold-up, [-] [kg/m-s] u, = gas phaseviscosity; , Calculated interfacial stressand the film velocity are then plotted against VsG as shown Figure 6.1(a) and Figure 6.1(b) respectively to establish a link between the two variable.
40
U
5.0
* VSLs 0.05m/s QVSLs 0.15m/s 0 Q
N
In
N
X 30
N Ny M
O O OO Do
% 20 r N io d 4C
n 0 10
2.0 S
0
0 10
0a ooo
O0 00
3.0
s 0
20 30 40 50 Gas superficial velocity, VG [m/s]
(a)
(b)
241 MB ALAMU2010
Chapter 6
Figure 6.1 (a) Interfacial shear stress as a function of gas superficial velocity. Figure 6.1 (b) Film velocity as a function of gas superficial velocity. The physics to support what is happening in 6.1 (a) and 6.1(b) can be stated as follows. An increase in gas flow rate at constant liquid flow rate result in an increase in the interfacial velocity at the liquid-gas interface and hence an increase in interfacial shear stress. More momentum is transferred from the highly energetic gas core to the liquid film, resulting in an increasein the film velocity in the basearea of the film. The increase in the interfacial shear stress increases in the interfacial friction factor. Pressuredrop increases as interfacial friction factor increases. Next, wave amplitude is calculated from Equation (6.1) utilizing calculated interfacial the measuredfilm velocity. Figures 6.2 (a) and 6.2 (b) are obtained when shear stressand wave amplitude and normalized wave amplitude are plotted against gas superficial velocity.
3 0 VSLz 0.05m/s D VSL O.1Sm/s 16
- 2.5 E
14
to
J 12
00
b= a d
1.5 a 3
10 DDO
: o
Qooo
M6
6 4 2 Z
DDD OD O
OO OD D q3
A jr 0.5 0
0 10 20 30
D0D
13
40 50
0
0 10 20 30 40
so
(c)
(d)
Figure 6.2 (c) Effect of gas superficial velocity on wave amplitude. Figure 6.2 (d) Effect of gas superficial velocity on dimensionless wave amplitude.
242 MB ALAMU2010
Chapter 6
In annular flow, droplet size and velocity distribution is function of gas and liquid phase flow rates. Initially at low gasvelocity, most of the liquid flow in the form liquid film and film thickness and disturbancewave amplitude are relatively large Sawantet al., (2008b). Since the disturbance waves are the major source of droplets in annular flow (Ishii and Grolmes, 1975; Kataoka et al. 1983, Azzopardi et al. 2006), the droplet size distribution directly dependson the disturbancewave amplitude Sawant et al., (2009). This statement be supported from the time-averagedfilm thickness from present experiments as can can in Figure 6.3(a) and Figure 6.3 (c). The Figures show that disturbance wave seen increasein gas superficial velocity. amplitude reduceswith
Figure 6.3 (b) and Figure 6.3 (d) are time resolved drop concentration signals produced from the disturbance wave. At lower gas superficial velocity Vs0 = 13m/s the wave is highest Figure 6.3 (a). This correspondsto high amplitude drop concentration amplitude Figure 6.3 (b) and hence large drop size. Therefore, under low gas velocity condition, droplet size is relatively large. As the gas velocity increases,more liquid gets entrained into the gas core and liquid film flow rate gradually decreasesresulting in decreaseof disturbance wave amplitude Figure 6.3(c). Consequentlythe averagesize of droplets also decreases the increasein gas velocity Figure 6.3 (d). with
Chapter
Bure Amplitude
und Dynamics
in Annular
Two-Phase Flow
0.6
1VINN
0.5 E
a
E 0.4
d
1000 0
c u C
0.3 V
E 0.2
a
100
0.2 0
5
S.1
5.4
5.5
10 L 5
5.1
5.2
5.3
5.4
5.5
Time is]
(a)
0.2
0.16 E
(b)
6000
-VSG = 40m/s = 42m/s VSL = 0.05m/s VSL= 0.15m/s
-- 5000 r 4000 0
3000
-VSG
E
0.12
Y
0.08 E
V
2000
0.04
0 o 1000 0
S S. 1 5.2 Time [s] 51 5.4 5.5 5 5.1 5.2 Time [s] 5.3 5.4
5.5
(c)
Figure Figure Figure Figure 6.3 6.3 6.3 6.3 (a) (b) (c) (d)
(d)
Film thickness time-series at low gas superficial velocity. Drop concentration time-series at low gas superficial velocity. Film thickness time-series at high gas superficial velocity. Drop concentration time-series at high gas superficial velocity.
droplet size distribution directly depends on the disturbance wave amplitude Sawant et Since (2009) drop size (i. e.D; o) and wave amplitude profiles are expected to some common al., features. divide the Figure 6.3(e) into three different regimes, namely: Therefore, we 21 (0 m/s ? VsG<_ m/s), 1. Wispy annular regime Vs(; < 30 m/s) and, (21 m/s >_ 2. Transition regime
Chapter 6
3. Mist flow regime (30 m/s z VsG5 40 m/s). This classification is basedon drop size hydrodynamics. In the same manner Figure 6.3(f) divided into three regions for the purpose of comparing the two figures with a view to was if there a kind of hydrodynamic link between the two figures. Observation based clarifying on Mass Median Diameter (D50) behavior as gas rate increases shows that only wispy annular regime in both figures appearsto have occurred from same processstrongly linked together and dominated by interfacial shearwave. The possible explanation for the difference behavior noticed in comparing the two figures be that the assumptionmade by Ishii and Grolmes (1975) that interfacial shearstressis may directly proportional to velocity gradient in the liquid film may only be valid for horizontal flow therefore not suitable for wave amplitude calculation in gas-liquid two-phase and
flow. The conclusion can be supported further by Figure 6.1 (a) and 6.1 (b) vertical qualitatively.
200
O VSLa0.05m/i
_ E
180
160 140
13 e
- 2.S E
b2 4 ti
Reg. 2
;o VSLi lSm/s O.
E 120 CL Q 100 0 80 60
Wispy
eee 113
Region 3
1.5
CL I1 d I.
O 0 Wispy Q lcc
000 d
1313q
00
Region 3
3 0.5 0
13
05 10 3S 40 Gas superficial velocity, Vx [m/s) is 20 25 30 45 SO 0 40 10 20 30 Gas superficial velocity, V36[m/s) 50
.n
(e)
(fl
6.3: (e) Characteristic drop diameter as a function of gas superficial velocity. Figure Figure 6.3: (f) Dimensionless wave amplitude vs. gas superficial velocity.
245 MB ALAMU20I0
Chapter 6
(6.10)
Vp SSW = CW ILL TWF
(6.11)
20 * VSL=0.05m/s QVSL=0.15m/s u u N
112
16
QQQ 13
IL
0Q
0
o o 000
Oo3
34
0
0'
0 10 20 30 40
50
246 MB ALAMU2010
Chapter 6
Q VSL=0.15m/s
p O 13 p pO 0
3.0
13QOO 13 O O 00
2.0
LL 1.0
0.0
0 20 10 30 40 Gassuperficial velocity VsG[m/s)
50
Figure 6.4(b): Film velocity profile. Liquid film wall shearstressis calculatedas follows:
rwF = fF
PLVF
2
(6.12) Where:
fF = 0.046 Re F i;
pLVFd F Re F=; PL
dF = 48F(15F)1
dF= hydraulic diameter of the liquid film, [m]; SF= film thickness, [mm]; fF = Liquid phasefriction factor, [-]; d= pipe diameter, [m]; The new wave amplitude is plotted against superficial gas velocity as in Figure the liquid film does not affect liquid film wall shearstressaccording 6.5(a).Chaotic nature of 6.5(b) within gravity dominated regime where 0 m/s? V50 5 21 m/s. This is to Figure unlike interfacial shear stressis not an interfacial phenomenonand becausewall shearstress
Chapter 6
therefore does not exposeto the turbulence and chaotic nature of the film. In the transition VSG5 30 m/s, there is a strong link between the two Figure 6.5 regime described by 21 m/s>_ (a) and Figure 6.5 (b). In the mist flow regime where VsG > 30 m/s, good link is also observed. In Figure 6.5 (c) and Figure 6.5 (d) where drop size and wave amplitude are normalized, the link between the two becomes obvious. It can be concluded that drop size distribution is strongly dependenton wave amplitude. According to Figure 6.3 (b), wave amplitude for VSL= 0.05 m/s and VSL= 0.15 m/s both superficial velocity increases.Wispy annular regime is dominated by reduces as gas huge wave and dispersedphaseis characterizedby incomplete atomization.
200
O AVSL-O.OSm/si DSO Q 10VSL "Q. 15m/s: DSO
180
E: E I w a
ai 3 CL 1 E 1.5
160
E
140
d 10 oo so (313
05 10 35 40 Gassuperficialvelocity,V,, (m/s] 15 20 25 30 4S 50
120
100
M 0.5 0
0 10 20 30 Gassuperficial velocity ,V 40 lm/sl
50
no
(a)
(b)
Figure 6.5: (a) Mean drop diameter vs. gas superficial velocity. Figure 6.5: (b) Wave amplitude vs. gas superficial velocity.
248 MB ALAMU2010
Chapter 6
*VSLc
0.05m/s: MMD
OVSL. 0.15m/i:
MMD
2.5
0.025
OVSL. 0.05m/s QS OVSLsO. lSm/s
oz
N
O 13
.10.02
pes=
gcis
G d
N
rib
wispy annular
dPa a
mist flow
d 7
0.015
w1 CL. O.S
CL E 0.01 A
0 'a
0 2
m
8 4" 0 2
O, 13 .,,, O00
0.005
s
a
0 10 20 30 velocity, 40 Vx [m/s] s0
91
0 10 20 30 velocity , Gas superficial 40 V. 1m/sJ 50
Gas superficial
( c)
(d)
Figure 6.5: (c) Normalized drop size vs. gas superficial velocity. Figure 6.5: (d) Normalized wave amplitude vs. gas superficial velocity.
Around VSG = 21 m/s there is a transition from wispy annular to transition (wispy-mist film seems to have more stability. Disturbance wave and huge is annular) regime where believed to co-exist in this flow regime although disturbancewave appearsto dominate. In Wispy annular regime, indistinguishablevalues of D50for VSL= 0.05 m/s and VSL= 0.15 be as a result of chaotic nature of the liquid film. In transition regime, m/s observed may increasein gas superficial velocity for VsL = 0.05 m/s, wave decreases with while amplitude increaseswith increasein superficial gas velocity for VSL= 0.15 m/s. amplitude After the transition to mist flow at Vso = 30 m/s, the wave changesfrom disturbanceto pack disturbance wave. Observationsfrom high speed camera show that these waves are of ring by high amplitude, Mantilla (2008). characterized
249 MB ALAMU2010
Chapter 6
2.5
0 VSI 0.05m/s : D32 "
Q. "
180
1 160
O VSL=0.05m/s Q VSL=0.15m/s
e'
pe
Ez E b d 1.5
140 E
120 0 100 O
80 60
g o o E3RQ
d QQ dQQ., 0
41 E
d c 0'00 00
v 3 0.5
0
40
05 10 15 20 25 30 35 40 45 SO Vw Gas velocity, (m/s] superficial (a)
50
Figure 6.6: (a) Mean drop size vs. gas superficial Figure 6.6: (b) Wave amplitude vs. gas superficial
velocity. velocity.
is a strong indication that shear stresses should be included in the In conclusion, there drop size model. This may be reason why the model prediction is poor as most existing do not include shear stressestheir formulations. models
250 2010
Chapter
Wave Amplitude
and Dynamics
in Annular
Two-Phase
Flow
6.5 Validating
wave amplitude
model
Model calculation N compared with measurements of disturbance wave amplitude in as vertical annular flow. The measurements were taken by Sekoguchi & Mori (1997); Sekoguchi et al. (1994). Sekoguchi & Takeishi (1989); Sekoguchi et al (1985) . in a
25.8mm tube as reported by Pols (1998) using Multiple Ring Electrode Probes (Super Multiple Point -Electrode Probes) at several axial positions along the tube measuring and the wave velocity. The intrusive probes are generally more
accurate than non-intrusive flush mounted wall probes (Ceccio, 1991; Hewitt, 1978).
--
-+i-V'SL
5 E4
A
LO 00 3
0
1
R
"
0 10 20 30 40 50
0
Gas superficial velocity VSG[m/s) ,
Figure 6.7: Comparison of measured wave height and the model prediction.
height is determined by addition of average film thickness and wave amplitude Wave by the model. The results are compared with the data of Sekoguchi et al. as calculated
Chapter 6
presented in Figure 6.7.There is good agreement between the model prediction and
measurements as shown in Figure 6.7.
50
4 40 CL
w a 1.5
30
d n
20
opo 00 Q.
0 oepo
13
E1 A d
Q d
b0 Q QQdP
10
fl
U
10 20 30 velocity, 40 VSG [m/s] so
9 03 0
0 10 20 30 velocity 40 VSG[m/s] , 50
Gas superficial
Gas superficial
(a)
Figure 6.8: (a) Dimensionless pressure drop vs. gas superficial Figure 6.8: (b) Wave amplitude vs. gas superficial velocity.
(b)
velocity.
MBAL.
1M '_0l0
Chapter 6
6.7 Conclusions
A new model to predict wave amplitude in vertical up-flow has been proposed. The proposed is a modification of the existing correlation based on experimental findings. The model model has demonstrated that wave amplitude is strongly dependent on wall shear stress. The predicted wave amplitude using the proposed model agrees reasonably well with the
published data.
253 MB ALAMU2010
Chapter 7
Chapter 7
Split of Gas-Liquid Two-Phase Flow at Vertical Dividing Junction
7.1 Introduction
Simultaneous flow of gas and liquid occurs over in a wide range of industrial equipment from large diameter pipes found in the oil and gas production industry to the micromight be employed to cool electronic components. In many channels which it is necessaryto divide the flow into two or more channel motivated by arrangements or processrequirements.Partial separation across T -junction is one of area restrictions the mechanisms of separationwhich has been widely adopted in manifolds comprising in oil and gas production to reduce processing load on the main gathering system separator. A junction can be defined as three connectedpipes, Azzi et al. (2010). If one is an inlet the other two are outlets, it is termed a dividing junction. The system with two inlets and is termed a combining junction. Though, for dividing junctions, the outlet and one outlet be at any angle to the inlet, two geometries are commonly found. In that pipes can junction one outlet is in-line with the inlet; the other outlet is termed side arm to these.In impacting junctions both outlets are perpendicularto the inlet. perpendicular dividing junctions are considered. The orientation of the In the present work side arm is also important. Junctionscan be classified as horizontal or vertical dependingon pipes the inlet pipe. The most important dimensions of a junction are the the orientation of diameters of the pipes, Di, D2, D3, where 1 refers to the inlet pipe, 2 to the straight
254 MB ALAMU2010
Chapter 7
through continuation and 3 to the side arm. Azzopardi (1999) has noted that side arm junctions usually have DI = D2. The division of gas-liquid flows at a dividing junction depends on the resistance (pressure drops) in the two outlet pipes as well as the physical phenomenawhich affect the phase split. In practical applications, these resistances are caused by equipment downstream of the junction. In experimental studies, it is usual to place valves in the downstream lines. Altering the valve position changesthe downstreamresistance.In this way it is possible to go from zero take off through the side arm (infinite resistancein the to 100% take off (infinite resistancein the straight-through pipe downstreamof side arm the junction).
Phase separation across T -junction is a secondary mechanism of separation process dependsmainly on phasemomentum, the existing flow pattern before whose efficiency the junction and the force of gravity Dionissios (2007). If satisfactory separation be achieved; the device could be an excellent candidate for sub-sea efficiency could minimum control system is required. The application of this application where in avoiding slugging at multiphase transportation system is very separation principle advantageous. Works related to T -junction and its application has been reviewed in detail, Azzopardi et (1999). Recently, based on the Equation of mass, momentum and energy balance, al. Dionissios (2007) studied the dynamic separation of gas and liquid inside aT -junction horizontal run and a vertical branch. He proposed a mathematical model which with a
255 MB ALAMU2010
Chapter 7
predicts phase distribution and pressuredrop acrossT -junction of different orientations and pipe diameterstaking into accountthe flow pattern before the junction. The motivation for this work arises from the paucity of published data on separationof
flows at vertical T -junction. For instance, the recent model proposed gas-high viscosity
by Dionissios (2007) does not include flow after the junction as a parameter.The usual is that flow pattern before and after is always the same. However, results assumption from present study show that flow pattern often changes after leaving the junction. Therefore, this work provides new data from measurementcarried out by increasing liquid phase viscosity in a systematic manner. The problem is approachedby taking fraction waves at each segmentof the T -junction. measurementsof void
256 MB ALAMU20/0
Chapter 7
8
Probe8 Probe7 Probe4
Pressure
gauge
3 Probe
ProbeS
Side arm
Probe6
Probe2
Wet gas meter Wet gas meter
cl
C
Cycloneseparator Cycloneseparator
Gas rotameters
Liquid
gorge
pm
Liquid measurement
Liquid rotameters
test section.
Liquid glycerol solution carefully mixed to produce solution with varying dynamic liquid viscosity of 3.6,6,8,10,12,18,27 and 36 mPa s. The solution was then mixed
temperature to form two-phase flow. Air was drawn from the with air at ambient and supplied to the mixing unit where it combines with water / compressedair main
257 MB ALAMU20I0
Chapter 7
drawn from a storagetank by meansof a centrifugal pump. The mixing glycerol solution unit consisted of an annular section surrounding a porous wall. Water/glycerol solution the main pipe from the periphery to form a film on the wall whilst the air passes enters along the middle. Inflow of air and water/glycerol solution is controlled using separate banks of calibrated rotameters.Downstreamof the mixer, the two phasesflow vertically a 1.3 m development length of stainless steel pipe of 5 mm diameter upward along before entering the T -junction. The test section pressure was measured 0.04 m upstream of the T -junction. The Tjunction is smoothly machined inside a rectangular block of acrylic resin. The transparent block facilitates visualization while the flat external surface minimizes the distortion due to refraction. Beyond the T -junction, the two-phase flow streamin the run travels vertically upwards for 1.42 m, before flowing horizontally and vertically arm downwards to a separatorconsisting of a vertical cylinder of 100 mm diameter. The side 1.06 m of straight horizontal pipe followed by a vertically downward arm consists of to an identical separator.The run and side arms are 5 mm in diameter and made of pipe stainlesssteel. The pressureand phasesplit at the T -junction is controlled using gate valves positioned inlet to each separator.After separation,the liquid is collected from the bottom of at the air passesout through a pipe at the top. The exit liquid could either each separatorand be recycled or taken off to be measured.During measurements, water / glycerol solution diverted into a measuringcylinder where the volumetric flow rate was measuredfor was long time to minimize uncertainty of measurement. a sufficiently
258 MB ALAMU2010
Chapter
Split of Gas-Liquid
Dividing
Junction
Figure
with conductance
Air was metered at the top of a pipe using a calibrated wet gas flow meter which allowed the flow rates to be measured over a period of time. The liquid level in the separator were kept constant to ensure accuracy. vessels Time varying, cross-section averaged void fractions were measured using eight identical probes placed along the test section. The probes consist of two stainless conductance steel ring mounted in acrylic electrodes resin housing. They were carefully
that the electrodes had the same diameter as the test section. A manufactured so is placed at 0.07 in after the mixer and three probes are situated in the conductance probe at 0.13 in from the junction. Two probes were placed close main, run and side arms 1.1 in downstream of the junction on the run arm and two more were together at 0.58 in away from the junction on the side arm. The probes around the Tpositioned junction are pictured in Figure 7.2. employed for the conductance probe is similar to that of Fossa The electronic circuitry (1998) and Fossa and Guglielmini (1998). The conductance technique has been chosen
Al B ALAAfU 2010
Chapter 7
to measure void fraction becausewater is an electrical conductor, while air is essentially resistive. Glycerol in its raw form is not conductive. Sodium Chloride solution was added to the conductivity of the solution to that of water. Andreussi et al. (1988) and Tsochatzidis et al. (1992) are some of the researchers have to used conductanceprobe successfully. In this technique, a cross-sectionalaveragedvoid fraction can be determined once the relationship between the electrical impedanceand the phasedistribution has beenestablished.
Oprobe4
+ y= -1.0489x2 0.0833x+ 0.9942 RI = 0.9958 0 0.2 0.4 0.6 0.8 Conductance,Ge* [-] Dimensionless
260 MB ALAMU2010
Chapter 7
Fossa Guglielmini(1998) & O60mm & A 70mmFossa Guglielmini(1998) Nottingham G38mm 0 Calibrationcurves Sidearm (Probes & 6) 3,5 -
0y
0.2
0.4
0.6
0.8
For the ring conductanceprobes used in the present experiments, the distance between width of each plate are 1.7 and 0.5 mm, respectively. This gives the electrode plates and to pipe diameter ratio (DeID) of 0.34 and electrode width to pipe electrode spacing diameter ratio (s/D) of 0.1. The probes give a voltage output which is proportional to the the two-phase mixture. This response is converted to dimensionless resistance of by referring to the value obtainable when the pipe is full of liquid. Different conductance for void fraction were derived for the calibration relationships probes orientated horizontally and vertically in terms of dimensionless conductance. For the vertical
261 MB ALAMU2010
Chapter 7
probes, the calibration which was undertaken by Omebere-lyari et al. (2005) involved instantaneous void fractions using plastic plugs with cylindrical rods artificially creating of known diametersand relating this to the dimensionlessconductance(Figure 7.3). In the case of the horizontal probes, Mak et al. (2006) have compared test results for in a 38 mm diameter pipe with the work by Fossaand Guglielmini (1998) for 60 probes 70 mm diameter tubes with similar aspectratios (De/D and s/D ratios) to the 5 mm and diameter probes. This shows little differences in the relationship betweenvoid fraction and dimensionless the best fit line is applied for the presentexperimentsas given in Figure conductanceand 7.4.
.. rte n. r.
Direction of flow
Figure 7.5: Cross sectional view of the ring type conductance probe.
262 MBALAMU2010
Chapter 7
Viscosities of the glycerol solutions were measured with Brookfield Dial Viscometer (Model LVT) while the sample conductivity was measured with cconductivity meter WTW model LF340.
7.3 Results
Liquid hold-up and phase split measuredat each take-off are presentedand discussed. For each test, the flow rates emerging from both the run and side arms out of the Tjunction were recorded.The fractions of gas and liquid flows taken off through the side Mass balanceswere carried out between the inlet and outlet flows. arm were calculated. For gas and liquid, all reported data agreewithin 5 %. The results are consideredin increasing levels of complexity. At the simplest level, the void fraction were examined. Firstly simple time series of cross sectionally averaged statistical measuredwere extracted.
7.3.1 Variation
physical
properties.
Time-resolved information can be examined at a number of levels. A great deal of information can be obtained by considering the time series of the cross-sectionally fraction. An example of this for the two liquids studied is shown in Figure averagedvoid 7.6. These are taken at gas superficial velocity of 3.2 m/s and a liquid superficial 0.12 m/s. In both casesthe results show the characteristic alternateregions of velocity of higher and lower void fractions which epitomise slug flow. It is clearly seen that the in the liquid slug part is higher for the water than for the glycerol data for void fraction
263 MB ALAMU2010
Chapter 7
flow leaving junction in the vertical run arm of the pipe. In the case of flow diverted to the side arm, the result shows characteristic alternate regions of high and low void fraction for water which is an indication of slug flow. An obvious uniform distribution fraction around a mean value is seen for glycerol which is strong indication of of void flow. For flow approachingthe junction, the time traces suggesta developing stratified flow regime for water whereasthe signature displays for glycerol shows regions of slug low void fraction with fairly regular peaks of higher void fraction probably characteristic
264 MBALAMU2010
Chapter
Split of Gas-Liquid
Tiro-Phase
Flow at vertical
Dividing
Junction
Flow leaving 1
Tqunction
-Air-water
-air-viscous
liquid
o. 8
0 A : 0.6
0.4 0.2 n
012
3456
Time[s]
Flow diverted to side arm
1 0.8
C
-Air-water
-air-viscous liquid
0.2 0 01 23456
Time [s]
7.6: Time averaged void fraction for air/water and air-glycerol Figure solution. Gas = 3.2 m/s, liquid superficial velocity = 0.12 m/s, Liquid viscosity = superficial velocity 36.0 mPa s, Gas Take Off = 0.68, Liquid Take Off = 0.68.
Chapter 7
Churn-Annular transition
0.6
0.4
0.2
o
05 10 15
o0
0
20 25 30
Q
35 40
0
Gas superficial velocity, Vsc [m/s]
Figure 7.7 (a): Detection of Churn-Annular flow transition boundary at a gas of 15 m/s based on liquid hold-up variation within the system as superficial velocity increases. gas superficial velocity
266 MBAL4MU2010
Chapter 7
O leaving juction
O approaching junction O diverted to side arm
0.8
1 annular 1,
0
05 15 10 20 25 30 35 [m/s] Gas superficial velocity, VSG 40
Figure 7.7 (b): Further analysis of the transitions within annular flow based on information from drop size in chapter four. established
This quantity is important to many engineering calculations particularly when pressure drop is involved as it is central to the gravitational component. Figure 7.7 shows how this parameter increasessystematically with increasing gas superficial velocity within different segmentsof thejunction.
Figure 7.7 (a) shows that liquid hold-up increaseswith gas superficial velocity towards a transitional gas superficial velocity of 15 m/s. This marks the beginning of annular flow the value of liquid hold-up. After the transition, the liquid hold-up decreases considering increasein gas superficial velocity. monotonically with In Figure 7.7(b), transitions within annular flow are further examined based on the knowledge and depth of information available from drop size measurementin Chapter
267 MB ALAMU2010
Chapter 7
Four. Thus, it has been establishedthat co-current annular flow occurs at gas superficial 21 m/s basedon subtle changesin the profiles of characteristic drop diameter velocity of (SMD) and entrained fraction. The evidences were further strengthened by the characteristic behaviours exhibited by disturbance wave properties corresponding the transition. A lot of studies agree on this according to Table 4.2. It is equally around in Table 4.2 that transition to co-current annular flow is independenton of pipe observed diameter, flow orientation and liquid superficial velocity. Based on these facts and figures, the transition seen in Figure 7.7 (a) can be thought of to be transition to wispy flow as indicated in Figure 7.7 (b). This transition marks the onset of annular liquid film. However, this atomization is not complete due to the atomization processof influence of huge waves. This regime continues until gas superficial velocity of 21 m/s transition to co-current annular flow occurs as previously explained in Chapter where Four. Becauseof the limitation of the pipe diameter drop size measurementwas carried this facility. out on Flow pattern prevailing in the main pipe is considered next. A strong theme in the two-phaseflows is to use different models for each flow pattern. Therefore modelling of it is important to identify which flow pattern is present. In order to identify flow pattern inlet condition various operating points are plotted on vertical flow pattern map at each Taitel et al. (1980) Figure 7.8. Very obviously, only two inlet conditions fall into slug of flow regime i. e. VSG= 3.2 m/s, VSL= 0.1 and VsG= 3.2 m/s, VSL= 0.2 m/s respectively. This agreesvey well with visual observation during the experiment and the information from time seriesin Figure 7.6.
10
Dispersedbubble flow
E
Bubbleflow
1
V C
Slugflow
E] i
C.
0.1
4)
11
0.01 0.01
100
Figure 7.8: Representation of test conditions on the flow pattern map of Taitel et al. (1980). Churn -Annular transition boundary agrees with churn-annular transition in Figure 7.7 (a) and Figure 7.7 (b). observed
The good agreement between the present data and the flow map may be as a result of inclusion of fluid properties in the formulation of the transition models employed by Taitel et al. (1980). The experiment by Taitel et al. (1980) was conducted with fluid of as used in the present study. A pipe of internal diameters of similar physical properties 25 mm (five times the size of pipe diameter used in this study), 51 cm long, with fluid at 25C and at 1.0 Bar (10 N/cm2) system pressure. Inlet conditions air/water as tested were also similar to the present study. finding is the transition from churn to annular flow which occurs The most outstanding gas velocity of 15 m/s as in Figure 7.8 which tie in with the around superficial in Figure 7.7 (a). observation
Chapter 7
Chapter 7
1 -0. , 0.6 L
Ilk ,`
LL O
CL 5
0
0 0.4 0.6 0.2 0.8 Fraction of gas taken oft 1-1 1
0 15
1 0.8
0.6 ti 0.4 ' ."
AIL
A
LA.
10
-Air-liquid
(36mPa s)
20.2 0
0
n oYO o
o
1
0
4
Void I,
PrIltion
IV
8 IL
13QnPa s)
4 2 n
VSG = 22m/s VSL = 0.2m/s: 36mpa s VSG = 16m/s V5L = 0.15m/s: 36mpa s A VSG = 15mfs VSL = 0. lm/s: 1&npa s A VSG - 12mf s VSL - 0.2m/s: 18mpa s O VSG = 23mfs VSL = 0. lrn/s: Smpa s f " VSG = 27m/s VSL = 0.5m/s: 3.6mpa s + VSG = 3m/s VSL = 0.2mJs: 3.6mpa s x VSG = 12m/s VSL = 0.5m/s: 3.6mpa s Q VSG = 15m/s VSL = 0.2m/s: 3.6nmpa s O VSG = 32m fs VSL = 0.3m/s: 3. bmpa s 0 VSG - 15m/s VSL - 0.1m/s : water
--
U.:, 0.4 _
II
rar Ge"
j O. 3 v X0.2 Vu
eo 0.0 O+
e of e
0
0 0.2 0.4 0.6 0.8 Faction of bas taken off 1"1 L
i o. a o. 6 ""i"
10
s LL 6
0.4 ;
ii0.2 O 0
OKCo
t+A ea .. + . +
oo
r1 12
CL 4
J.
0
0.2 Fraction
0.4
0.6
of gas taken
Figure 7.9: Comparison of air/water and air-viscous liquid data at the same flow liquid hold-up and PDF of void fraction. Gas superficial velocity = conditions using 3.2 m/s, liquid superficial velocity = 0.12 m/s. T-Junction Pressure = 1.4 bar.
Gas Take Off = 0.68; Liquid Take Off = 0.68.
"1 1 A/ BA 1'. 0 It wo
Chapter 7
272 MBALAMU2010
Chapter
Split of Gas-Liquid
Dividing
Junction
as flow pattern after the junction is usually not considered as an important design parameter.
For air/water considerably. system. flow before and after the junction does not seem to change emerge to be
similar after leaving the junction. However, flow to the side arm is stratified with little effect of liquid viscosity on hold-up in both instances.
VSL = 0.10 m/s VSL = 0.02 mis VSL = 0.02 m/s VSL = 0.10 m/s
100 10
W
--Air-watet -Atr41quid eta mia s)
U-
0.1
0.01 0.001
LL 5
The phase split plot does not indicate flow pattern. Identifying flow pattern based on be misleading. In order to identify flow pattern additional information is phase split can then be used in conjunction with phase split curve to determine required which can flow pattern in the system. Therefore, analysis is taken another level of prevailing
VC IWWWO 11B. IL 1.
Chapter 7
complexity. At this level, the idea of Gardner and Neller (1969) is adopted to identify flow pattern based dynamic information acquired by the probes at every take off. Gardner and Neller (1969) has demonstratedthe use of Froude number to identify flow pattern for the caseof bubble/ slug flow upstreamof a vertical bend. In the bend the gas flow either on the outside or the inside of the bend depending on the balance can between the centrifugal force tending to push the liquid phase to the outside and the forcing it to the bottom. The authors expressedthe competition between these gravity two forces by Froude number (ratio of centrifugal to gravity force or VSG/gD). It is that both phasesare flowing in radial equilibrium when the Froude number is claimed to unity. However, when the ratio is greater than unity the gas phase is displaced equal to the inside of the bend, while in the other case of a number less than unity the gas to the outside of the bend. moves Here, this analysis has been extended to dividing junction based on some similar features it shareswith bend. Gas take off velocity, liquid and gas densities were used in the computation of Froude number. Figure 7.10 presentsthe results. It has beenobserved that the flow pattern of conditions that lie on the constant line where FrG=1 is stratified flow basedon Probability Density Function (PDF) plot of void fraction (insert in Figure 7.10) belonging to those conditions. The data used are for air/water and air-18 mPa s They were taken at the horizontal side arm of the junction. glycerol respectively. Therefore, the claim of Gardnerand Neller is confirmed.
Chapter 7
(7.1)
M3(1-x3) 1111(1 x1)
(7.2) Where G' and L' are gas and liquid taken off respectively. th is the mass flow rate, kg/s. The quality, x, the ratio of gas mass flow rate to the total flow rate can be calculated from inlet condition using the following expression: Pg Vsc (PBVsc+ PLVSL)
x=
(7.3)
In terms of entrained fraction, fraction of gas taken off, Equation (7.1) can be re-written as: 27rL' G' _ -stn 2it K(1- E) 1 21rL' K(1E)
Chapter -
Junction
(7.4)
E= entrained liquid fraction. [-J. K is a factor which accounts for the effects of the ratio of side arm to main pipe diameters given as:
D ( 3)0.4 1.2 D1
(7.5)
The commonality
the entrained liquid fraction. I lowever, entrained measurement could not be carried out this facility because of the small pipe diameter. on Table 7.1 presents part of the output results generated by the computer programme. Govan et al (1988) model was used for entrained liquid fraction calculation while the Willetts (1987) was used for calculation of film thickness for the inlet equation of flow regime. conditions within annular Table 7.1: Fluid properties at different viscosities
(mPa 5]
[-1
0,00000 0,00000
Chapter
Split of Gas-Liquid
Dividing
Junction
70(
E B 601 c
m E E 50(
0.0075
61
40(
E
Asa
E
0 E 20( E
LL
101
0
05 10 15 ZU Liquid Viscosity 25 3U (mPa s) 35
Figure 7.11: Variation of film momentum, gas momentum and entrained fraction liquid viscosity. Gas superficial velocity = 15.2 m/s, liquid superficial velocity = with 0.13m/s, liquid viscosities = 1.0,12.0,18.0 and 36.0 mPa s.
4.50E-04 4.30E-04 4.10E-04 E3.90E-04 4,3.70E-04 3.50E-04 3.30E-04 E3.1 OE-04 `L2.90E-04 2.70E-04 2.50E-04 --0 5 10 15 20 25 30 Liquid Viscosity (mPas) 35 40
. P; '"
Chapter 7
Film and gas momentum including entrained fraction are plotted in Figure 7.11 while variation of film thickness is plotted with increasing viscosity in Figure 7.12. Critical examination of Figures 7.11 and 7.12 suggeststhat two mechanismsare operating; one at low liquid viscosity i. e. viscosity below 12 mPa s and another,after the point of inflection beyond 12 mPa s. Liquid film thickness, film momentum, gas momentum and entrained liquid fraction curves are characterizedby two gradients each correspondsto different Very obviously, as liquid momentum decreaseswith increase in mechanism or event. liquid viscosity, there is a corresponding increase in liquid film thickness as shown in Figure 7.11. low and high viscosity two-phase flows. The possible The observation further demarcates for this transition may probably be the change in the interfacial behaviour explanation by Mori and Nakano (2001) and Kondo et al. (1999) between low and high reported two-phase flow. McNeil et al. (2003) and Mori et al. (2001) have reported that viscosity increase in liquid viscosity increasesthe entrained fraction. However, using the method of Govan et al (1988) for calculation of entrained liquid fraction at the conditions of present liquid superficial velocities are 15.2 m/s and 0.13 m/s respectively. study where gas and Calculated entrained liquid fraction was given as zero and did not change with increase liquid viscosity. This is becauseall the liquid flowed in the liquid film. This may be becausethe critical Reynolds number for the onset of atomization was not exceededfor the conditions tested.
278 MB ALAMU2010
Chapter 7
Q 12 mPa s Liquid [VSG= lSm/s, VSL= 0.30 m/s] f36 mPa s Liquid [VSG: 16 m/s, VSL= 0.24 m/s] 1
y9-
0 C 41
0.8
a-
0.6
r Q
O'
0.4
0 C 0 u ru LL
0.2
0
0 0.2 0.4 0.6 0.8
Next, effect of liquid viscosity on phase split is considered. Figure 7.13 shows that the towards liquid dominated region as dynamic liquid phase viscosity phase split shift increases when inlet conditions are held constant. This can be attributed to the lower higher liquid viscosity. Low liquid momentum can promote film liquid momentum at the liquid film to be easily taken off. Film stop occurs when stop which could cause liquid film is brought to a halt because of its low momentum flux due to pressure increaseacrossthe junction after a critical gas take off has been exceeded. increase in the inlet gas flow rate was investigated by keeping the liquid The effect of in Figure 7.14. More liquid is diverted to the sidesuperficial velocity constantas shown increases. This trend agrees with the work of Azzopardi and arm as gas velocity
Chapter 7
Memory (1989), which showed that a change in the gas inlet velocity had an effect on the liquid take off. Azzopardi and Memory used air/water as test fluid.
O VSG =12 m/s VSL = 0.20 m/s f VSG = 27 m/s VSL = 0.2 m/s
1
0.8 0 c Cl)
0
0 0.2 0.4 0.6 0.8
At high gas superficial velocity of 27 m/s, liquid film is mostly transportedas drops in The drops could be re-depositedas film and can be taken off. Expansion of the gas core. increasescan push the liquid film towards the side arm as a the core as gas velocity created by re-circulation of liquid film or secondary flow result of pressure gradient thejunction. close to the entranceof
280 MB ALAMU2010
Chapter 7
More liquid was separatedin the case of air/water at liquid and gas velocities of 0.02 m/s and 14 m/s, Figure 7.15 (a). Azzopardi (1988) attributed this high liquid take-off to be as a result of low liquid film momentum resulting from film stopping.
VSG =16m/s
A VSG =12m/s VSL = 0.2m/s: l8mpa sO + VSG = 3m/s VSL = 0.2m/s: 3.6mpa sX QVSG = 15m/s VSL= 0.2m/s: 3.6mpa s *VSG= 14m/s VSL = 0.02m/s: water
VSG = 27m/s VSL = 0.5m/s: 3.6mpa s VSG =12m/s VSL = 0.5m/s: 3.6mpa s QVSG = 32m/s VSL= 0.3m/s: 3.6mpa s Z VSG = 3m/s VSL = O.lm/s: water
1
O a1
0.8
0
00
0.6
0.4
9O C 0 u LL
0.2
OD
0 0.2 0.4 0.6 0.8
0
Fraction of gas taken off [-]
Figure 7.15 (a): Effect of liquid viscosity on phase split.
In Figures 7.15 (h) and 7.15 (c). the liquids that are more viscous than water according Table 7.1. are expected to have lower momentum compared to water. Therefore, for air18 mPa s glycerol solution at VSG;= 15 m/s, Vsi = 0.1 m/s Figure 7.15 (b), and air/8 mPa s glycerol solution Vs(; = 23 m/s, Vs, = 0.1 m/s Figure 7.15 (c), more liquids are at
'41
Chapter 7
due mainly to film stopping effect. At these conditions liquid partially separated is expectedto high. Therefore, rate of drop deposition will be equally high. entrainment At these conditions, rate of atomization is very high as both conditions are in annular flow regime. It seemsthat the expansion of the gas core increasesthe mobility of the film around the wall and pushesit towards region of lower pressurein the side arm.
w 0.8 0
a ;; 0.6 'a M ;.
40
r.
O
fam.
f.
Figure 7.15 (b): Effect of liquid viscosity on phase split. Comparison s viscous liquid. air/water and air/18 mPa
Chapter 7
1
0.8 ""-
fVSG=15m/s VSL=O. im/s: l8mpas OVSG=23m/sVSL=0. lm/s: 8mpas *VSG =15m/s VSL = 0.1m/s: Water
; "'
0 C
0.6 0
f0O
a.
.. 0.4 0
0
0
"'O
c 0
0.2
LL
0
0 0.2 0.4 0.6 0.8
Next, an explanation is sought for increasein liquid take off as liquid physical property basedon the trend observedusing the plot of liquid hold-up against gas take off changes Figure 7.11. The figure shows liquid hold-up increaseswith increasing gas taken-off. Critical examination of the plot however reveals that liquid viscosity only have liquid hold-up when fraction of gas taken off exceedsa value of significant effect on 0.40. After fraction of gas taken off of 0.40, liquid hold-up is seen to increase with increasein liquid viscosity as gas taken off increases.Becausehold-up is high and liquid is low, the liquid is easily taken off. This explains why more liquid are momentum in Figure 7.15(b) and Figure 7.15 (c). separated
283 MB ALAMU2010
Chapter 7
r, I-I w
Y
= 4J
0.8
o. s
Z 0"
:: 0.4
_::.
0 C
0
U cII
U.
i 0.2
S.. .
0.6
0
0
0.2
0.4
0.8
The present data has been used to test the predictive ability of published models, Figure 7.16. The model of Azzopardi (1988), Shoham (1987) and Sliwicki-Mikielwicz (1987) to predict the experiment phase split. Figure 7.11 shows the good are considered between the present data and the existing models. In the case of model of agreement Azzopardi (1988), it was found that a better performance was obtained if flooding was Mak et al. (2006) provided justification for omitting that phenomenon.It not considered. in their air/water experimentscarried out on the samefacility as usedin the was not seen Flooding was also not observed in the pipe above the junction in the present study. Azzopardi (1988) assumesthat the phenomenonof flooding causes present experiments. drops to deposit onto the wall film and hence fall back and be taken off. The entrained the liquid taken off is explained by the absenceof flooding. Better over-prediction of
284 MB ALAMU2010
Chapter 7
are obtained when the method of Azzopardi (1988) is modified to account predictions for the non-existenceof flooding. The models by Shoham et al. (1987), Sliwicki and Mikielewicz (1987) also give in Figure 7.16. Although the work of Shohamet al. (1987) excellent agreementas shown developed for a horizontal T -junction, it could be applied to vertical pipe was becausethe model assumes flow is a cylindrically symmetrical annular the arrangements flow. Where Shohamet al. (1987) under-predictedthe liquid take off, the probable reasonmay be the dominance of centripetal forces suggestedin their model which does not apply to this condition as liquid entrainment may not be significant. As the liquid superficial is decreased,the void fraction is increased and it leads to decreasethe film velocity As the film thickness is decreased,the film velocity can then be increased. thickness. This leads to an increasein the centripetal forces.
(7.6)
285 MB ALAMU2010
Chapter 7
White and Beardmore(1962) identified Morton number, M, as a property number which be used to characterise gas-liquid two-phase flow structures. They proposed the can following definition:
M= 9PI (PL - Pc )I Pp a3
(7.7)
Where: is the gravity acceleration,p is the viscosity, p is the density, and v the surfacetension. g Figure 7.16 shows dependenceof liquid hold-up on liquid viscosity. Liquid hold-up decreaseswith Archimedes number and inverse Morton numbers respectively as liquid increases.Both Archimedes number and inverse Morton numbers show similar viscosity of trend. This also explains dependence gas and liquid take-offs on liquid viscosity. This findings reported by Hong et al. (1978). Therefore, based on Figure 7.9, it agrees with be concluded that at low viscosity liquid hold-up is low and vice versa. Similarly, can basedon Figures 7.9 and 7.17 (a) , it can be implied that at high gas take off liquid holdis also high. Conclusively, division of two-phase flow at vertical dividing junction up be greatly influenced by liquid viscosity. can
286 MB ALAMU20I0
Chapter 7
0 VSG= 22m/s VSL= 0.2m/s: 36mpa s AVSG = lSm/s VSL= O.im/s: l8mpa s "VSG = 32m/s VSL= 0.3m/s: 3.6mpa s
A
VSG =16m/s VSL 0.15m/s:36mpas = 0 VSG 23m/sVSL O. = = im/s: 8mpas " VSG lm/s : water =1Sm/sVSL= O.
FlowleavingT-junction
Z.: 0.8
S
0.6
'O 0 0.4
13
C.
0.2
0O
0
1 10 100 1000 10000 100000 1000000
AP.
0.4
I
0
1 10 100 1000 10000 1000001000000
Chapter 7
Re. =
py"'d UL
(7.8)
VM = VSG+VSL where Liquid based Strouhal number was used to correlate structure frequency and plotted Reynolds number. Poor correlation is observedFigure 7.18. This may be against mixture because Strouhal number does not include physical property in its formulation. However, it is clear that more structures are created at higher liquid viscosity as shown in the Figure 7.18.
Chapter 7
lu
iwu
=IDO 3
0.1
10 3 c v 1
0 0
V1
0.01
n one -'--'
1 10 100 1000 10000 100000
0.1
I LU Mixture LUU Reynold number, 1000 Reu,, [- LUUUU 1Uuuuu
Mixture 10
Glow diverted O Vicous
Reynold
numcer,
Res,,, i-J
1000
Flow diverted
to sidearm
f water
wioo i
p viscousliquid O water
0
j L
A L 7 O 3
io 0 0 0 0
nm d1
0Q0 0 0%0
0.1
0o
o 00
0.01
3 o. i
0.001
1 10 Mixture loo Reynold 1000 number, 10000 Re.. 1-1 100000
10
100
1000
10000
100000
10
1000
Flow approaching T-juction
OO 0
O Vicous liquid 1 H a 3 c
io L 7 O
water
O
0.1
? 100 3
v
Owater
00
afi
10
OO
OOQ
d
0 00
I1
O0O6 0
0.01
O 0.1
1 10 100 1000 10000 100000
0.001
Mixture
10
Reynold
IDO
1000
number,
Rests J-J
10000 100000
Figure 7.18: Variation of structure frequency liquid phase Reynolds number. with
correlated
with Strouhal
number
to investigate this further, frequency of structure approaching the junction In an attempt is used as a yardstick to compare structure frequency at run arm and side arm
MB. -11.1.1I12010
Chapter 7
In order to study the distribution of the flow structure based on its respectively. frequency around the T -junction, structure frequencies at side arm and run arm are against structure frequency of approaching flow before the junction as seen in plotted Figure 7.19. The figure shows that structurefrequency before the junction is greaterthan the structure frequency exiting the junction after splitting as indicated in Figure 7.19. Overall, the structure frequency in the side arm is greaterthan structure frequency on the However, as liquid phase becomesmore viscous with lower surface tension run arm. structures are createdafter separationat the junction to the run arm. Immediately more the vicinity of the junction the flow tends to re-develop along the run arm. This after formation of more flow structures. In slug flow, coalescenceof smaller encourages Taylor bubbles, for instance, may take place to form bigger one. Therefore, more scannedby the sensor.The flow pattern observed in Probe 8, away from structures are the vicinity of the junction, according to Figure 7.9 shows slug flow while that at the is stratified flow, flow approachingthe junction being bubbly flow for the airside arm liquid. viscous Higher liquid hold-up noticed in Figure 7.9 when hold-up is plotted against fraction of taken off could be another testimony why structure frequency at Probe 8 is greater gas than the structure frequency at the side arm for liquid phase with same dynamic Another reasonmay the influence of low pressure, re-circulation zonejust at viscosity. the side arm causingblockage due to high hold-up. the entranceof
Chapter 7
1990
low
ioo
I 0
100
10
V
0 a
aofo
013
cr
13
cr
0.1
0.1
13
0.01
%A 0.01
oai
0.01
1000
:r
100 10O
&Ulm
1000
in
C
1
OD
p 13
0.1
'^ 0.01 0.01
0.01
0.01 1 0.1 Structure frequency, 10 100 100 Before the Junction tH1j
100
,,,
1o
S1
2
N
at
C.C1
0.01 Structure 0.1 frequency 1 . Before to the Junction 100 (Htj 1a
Sao
30 13a
DO OO. 3
0.1 O(0
O
ani
OA1 u1 Structure frequencc
10
1000
Figure 7.19: Variation of structure frequency correlated with Strouhal number with liquid phase Reynolds number.
Chapter 7
It seems there appearsto be a blockage of liquid around the entrance to the side arm caused by the recirculation of liquid. As a result, structuresare carried straight on pass the junction to the run arm where flow re-develops after travelling certain distance. It seems flow structure development is enhancedat higher viscosity. All these explain the around Probe 8 in Figure 7.19 where the structure frequency at that location observation is observed to be greaterthan the frequency of the structure diverted to the side arm.
1000
N
L W L
C. CJ
8t
10
L 1+
M1
w
N
IQ
0.1
00 0
0.1 1 10 100 1000 fl, Structure frequency approchingthe Junction [Hz]
Figure 7.20: structure frequency at side arm and main pipe as a function of frequency approaching the junction. structure
The difference in structure frequency after flow split has been plotted against the frequency of structure approaching the junction in Figure 7.20. The figure shows an inevitably linear relationship. Therefore, the linearity provides another opportunity to the data further. examine
292 Af BALAMU2010
Chapter 7
At the next level of analysis, the difference in structure frequency, that is the structure frequency of approachingflow towards the junction minus the structure frequency at the and run arm respectively, after flow division has occurred are plotted in Figure side arm 7.21.
1000
y=0.8162x+0.0787
R2 0.9958 =
N 100 I
1
.a
c lo
N
rl '1
oQ
'"
.-0.1
Q
10 100 Af1.3, Arm[Hz] Side 1000
Figure 7.21: Difference in structure frequency approaching the junction and frequency after separation at run arm plotted as function of difference in structure frequency approaching the junction and structure frequency after structure separation at the side arm.
The relationship also produces a linear behaviour with a straight line with a gradient of 0.82. The relationship can be expressedmathematically as follows: approximately fi-6_0.82
Af3
............................................................................
(7.9)
293 MBALAMU2010
Chapter 7
For instance, f3, slug frequency, based on the experimental data obtained for different pipe diametersin horizontal flow, according to Azzi et al. 2010, can be estimatedby:
0.0039 VSL OL
fSlug
=d
(7.10) Substituting for side arm pipe diameter, d=0.005m, in Equation (7.10) yields f3, slug frequency on the side arm:
TL f3 = 0.78 VSG
(7.11) Structure frequency in the run arm can be estimatedusing Equation (3.4) such that:
F; fZ = 0.009 VSc G
(7.12) Therefore, frequency of the approachingstructure, fl , can be estimated by substituting f2 and f3 respectively in Equation (7.4) as follows: P + 3.55VSL pG fi = 0.05 VSG
(7.13)
Equation (7.13) is advantageous because frequency of approaching slug can be if inlet conditions are specified. computed
294 MBALAMU2010
Chapter 7
7.4 Conclusions
From the results and discussions presented above the following conclusions can be drawn:
1. This work has demonstrated that flow pattern approaching the T -junction does change
leaving the junction along the main pipe as the viscosity of liquid phase increases. after It is therefore, important to include flow pattern after the junction as one of the for design. parameters 2. Change in liquid viscosity does not have a significant effect on flow pattern in the horizontal side-ann. 3. In some casestested, at superficial liquid velocity, VsL= O.lm/s and gas superficial Vs() = 15 m/s for air/water, VsG= 23 m/s for air-8 mPa s viscous liquid and velocities of Vs0 = 15 m/s for air-18 mPa s viscous liquid , an increase in liquid viscosity leads to liquid take off at the horizontal side arm. more 4. An increase in the gas superficial velocity at a fixed liquid superficial velocity diversion of the liquid into the sidearm. constant causesmore 5. Chum-annular transition occurs at a gas superficial velocity of 15 m/s. 6. Liquid viscosity does not affect significantly liquid hold-up until fraction of gas taken off exceeds0.40. 7. Thicker film or viscous liquid displays lower momentum and can be easily taken off.
Chapter 8
Chapter 8
Conclusions and Recommendations
8.1 Conclusions
Periodic structures in vertical gas-liquid two-phase flow have been identified as void in bubbly flow, liquid slugs and or Taylor bubbles in slug flow, huge waves in waves flow, disturbancewaves in annular flow and wisps in wispy-annular flow. In order chum to obtain detailed information on characteristics and behaviour of these periodic of experimental campaignswere carried out starting from simple pipe structures, a series flow arrangements through to complex pipe geometry employing pipes of different diameters (5mm and 19mm internal diameters). Experimental programmes were carefully controlled to minimise measurement to obtained good quality data. Results were extensively analysedleading to uncertainties the following accomplishments: in A. Dynamic drop size measurement vertical annular two-phase flow. This thesis has advanced knowledge in annular flow two-phase flow by making the following novel contributions: 1. In the history of drop size measurements, new, time-resolved drop size and data are reported for the first time. They have been obtained concentration film thickness and pressuredrop information by synchronizing the simultaneously with systems. All these parametersshow fluctuations with time. Some, various acquisition such as film thickness,are more obviously periodic than others.
296 MBAL4ifU2010
Chapter 8
Conclusionsand Recommendations
2. The time averagedvalues are in agreementwith prior data. 3. Drop frequency has been estimatedfrom fluctuations of drop concentration with time Power SpectrumDensity (PSD) of auto-covariancefunction. using 4. Interrelationship of drop concentration fluctuation and wave frequency has been Wave fluctuations are higher than drop concentration fluctuations. reported. 5. Transition boundarieswithin annular flow have been elucidated by careful analysis of the drop size and entrained fraction data. Interestingly, all instrumentation employed picked up thesetransitions: (i) Tra nsition to annular flow occurs when gas superficial velocity exceedsVsc = 21m/s with maximum entrained fraction, EF = 0.05. This value agrees with
data. published
(ii) Wisps were seen in the region identified as chum area by published transition This occurs below Vso = 21m/s. This is supported by video footage of flow models. during the experiments. Transition boundaries in the Hewitt & Roberts recorded (1969) flow map has been modified to reflect this observation. Characteristic wave features show that huge wave dominatesthe gas-liquid interface before transition to flow at VsG= 21 m/s. annular (iii) Transition to mist annular flow occurs at VS = 30 m/s. Normalized drop size that droplets shedding from disturbance wave are controlled by liquid shows The higher the liquid velocity, the higher the drop size. After superficial velocity.
297 MBALAMU2010
Chapter 8
VSG = 30 m/s, huge wave disappears, liquid film surface being covered by disturbancewaves. The transition is characterizedby entrained fraction, EF= 0.20. 6. Considering MMD (D50), thinner film produces bigger drops only at this boundary 21 m/s 2 VSG5 30 m/s condition: 7. Before VSG = 21 m/s, and after VSG= 30 m/s, thicker film producesbigger droplets. 8. In terms of SMD (D32),thicker film producesbigger drop up to VsG = 21 m/s. After VSG= 21m/s, thinner film producesbigger drop. 9. Entrained fraction depends on gas and liquid superficial velocities. Entrainment of liquid droplets into the gas core of annular flow is a diffuse flow phenomenonand not a displacement. This is evident by the nature of entrained fraction- gas piston-like superficial velocity plot. 10. Below VSG = 21 m/s, both entrained fraction and entrained mass flux are indistinguishable. After VsG = 21 m/s, both entrained fraction and entrained mass flux higher the higher the liquid superficial velocity. are 11. Below entrained fraction of 0.05 (EF = 0.05) and Vso < 21 m/s droplet flow is inertial-driven and gravity appearsto have significant effect on entrained fraction. The flow is chaotic and huge waves dominate the gas-liquid interface. Drop size or MMD is large under gravity-dominated flow. Effects of gravity diminishes after entrained fraction exceeds 0.05(i.e. EF > 0.05).After EF = 0.05 pressure drops uniformly with fraction until a minimum is reachedwhere EF = 0.20, VsG= 30 m/s. entrained
Chapter 8
12. Pressuredrop recovers after EF = 0.20, increaseswith increase entrained fraction as increases.Mist flow occurs as a result of complete atomization gas superficial velocity of liquid film. 13. In mist flow regime droplet size do not change significantly with increase gas velocity. After this transition normalized pressure drop increases with superficial fraction as gas superficial velocity increases. entrained 14. Examination of the time series in amplitude and frequency spacereveals interesting features. From the probability density function of mass median diameter, it is seenthat distinct regions of large and smaller mean drop sizes are evident are visible at there are lower gas superficial velocities but that there is not this feature at higher gas velocities. PDF of MMD indicates multi-modal distributions before Vs0 = 30 m/s. After VSG= 30 PDF changesto mono-modal distribution. m/s, 15. For gas producer the best operating practice would be to produce natural gas within transition region with the following boundary conditions: the VSG< 30 m/s 21 m/s >_ 0.05 EF: _ 50.20
This region is characterized by minimum entrained fraction and pressure drop which meansless shut-downsand more revenue.
299 MB ALAMU2010
Chapter 8
B. Wave dynamics in vertical annular two-phase flow From the results and discussionspresentedthe following conclusions can be drawn
0.15 m/s.
2. The dominance of huge or disturbance wave depends on liquid superficial velocity is maintained at a constantrate. when gas superficial velocity
3. Wavefrequency:
9 " " is a strong function of gas superficial velocity increaseswith increasein gas superficial velocity is higher the higher the liquid superficial velocity
4. Wave velocity: " " increaseswith increasein gas superficial velocity is higher the higher the liquid superficial velocity
5. Wave spacing: " " increaseswith increasein gas superficial velocity is higher the lower the liquid superficial velocity
Chapter 8
C. Effect of liquid viscosity on periodic structuresin two-phase ag s-liquid flow. Since most published data were acquired from experiments using air and water as test fluids, this thesis has contributed to the existing knowledge and distinguished itself by demonstrating effect of liquid viscosity on periodic flow structures in two-phase gasliquid flow in the following ways: increasein liquid viscosity. 13. Void fraction decreases with 14. Structure frequency increaseswith increasedliquid viscosity. 15. Liquid viscosity and surfacetension causesshift to transition boundary. 16. Higher viscosity liquid phaseinduces higher slippage between gas and liquid in two-phasegas-liquid flow. phase 17. Flow distribution coefficient, C., has a strong dependenceon liquid viscosity. 18. Structure velocity is strongly dependenton viscosity and surfacetension. 19. Flow structuretravels faster at the centre-line in medium of higher viscosity and
301 MB ALAMU2010
Chapter 8
D. Split of gas-liquid two-phaseflow at vertical T junction From the results and discussions presented above the following conclusions can be drawn: 1. This work has demonstratedthat flow pattern approachingthe T -junction doeschange leaving the junction along the main pipe as the viscosity of liquid phaseincreases. after It is therefore, important to include flow pattern after the junction as one of the
Chapter 8
Conclusionsand Recommendations
7. PDF plot of void fraction has successfully captured the effect of liquid viscosity on phase distribution. 8. Archimedes and Morton numbers have demonstratedability to track effect of liquid distribution. viscosity on phase
303 MBALAMU2010
Chapter 8
it is recommended that cross-correlation of two drop velocity value, concentration signals from Spraytec be carried out to yield true and representativedrop velocity which can be used in the conversion with a view to improving the accuracy of the entrained fraction obtained from drop concentration. S. More experimental activities should be focussed on investigation of liquid flow distribution parametersand the coefficient of drift velocity by viscosity on testing with liquid of higher viscosity than used in this study. 6. More works need to be carried out to extend finding s of this work by expanding in slug flow regime using liquid of higher viscosity to track experimental matrix
to flow pattern leaving the T -junction. changes
References
References
Adechy D., Issa F. I. 2004. Modelling of annular flow through pipes and T -junctions. , Computers and Fluids, 33(2): 289-313. Andreussi, P., Di Donfrancesco,A., Messia, M., 1988. An impedancemethod for the liquid hold-up in two-phaseflow. International Journal of Multiphase measurementof Flow 14,777-785. Andreussi, P., J. C. Asali and T. J. Hanratty, 1985. Initiation of Roll Waves in GasLiquid Flows. AIChE Journal 31(1): 119-126. Andritsos, N., 1986.Effect of Pipe Diameter and Liquid Viscosity on Horizontal Stratified Flow. Ph.D. Dissertation. University of Illinois at Urbana-Champaign. Al-Sarkhi, A. & Hanratty, 2002. T.J. Effect of pipe diameter on the drop size in a horizontal annular gas-liquid flow. Int. J. Multiphase Flow, Vol. 28,1617-1629 Alves, N. I., E. F. Caetano,K. Minami and 0. Shoham., 1991. Modeling Annular Flow Behavior for Gas Wells. SPE Prod. Eng., pp.435-440. Asali, J.C., Leman, G.W., Hanratty, T.J., 1985.Entrainment measurementand their use in design equations.PCH Physicochem.Hydrodyn. 6,207-221 Azzi, A., Al-Attiyah, A., Liu Qi, W. Cheema,Azzopardi, B. J., 2010.Gas-liquid twoflow division at a micro-T junction. Chem. Eng Sc., 65,3986 - 3993 phase Azzopardi B.J., HernandezPerezV., Kaji R., Da Silva M. J., Byers M., and Hample U., 2008. Wire mesh sensorstudies in a vertical pipe. HEAT 2008, Fifth International in Conferenceon Transport Phenomena Multiphase Systems,Bialystok, Poland. Azzopardi, B.J., 2006. Gas-Liquid Flows. Begell House Inc., New York. Azzopardi, B.J., 2004.Bubbles, drops and waves - differences or underlying 42nd EuropeanTwo-PhaseFlow Group Meeting, Genoa,23-25 June. commonality. Azzopardi, B.J., Colman, D.A., Nicholson, D., 2002. Plant application of aT -junction as separator.Chemical Engineering Researchand Design, Vol. 80,87-96. a partial phase Azzopardi, B.J., 1999.Phaseseparationat T -junction. Multiphase Scienceand Technology, Vol. 11,223-329. Azzopardi, B. J., 1997. Drops in Annular Two-PhaseFlow. International Journal Multiphase Flow 23, Suppl.: 1-53. of
References
Azzopardi, B.J., Teixeira, J.C.F., 1994 a. Detailed measurements vertical annular two of phaseflow - Part I: drop velocities and sizes.J. Fluids Eng., Vol. 116,792-795. Azzopardi, B.J., Teixeira, J.C.F., 1994 b. Detailed Measurementsof Vertical Annular Two-Phase Flow - Part II: Gas Core Turbulence. Journal of Fluid Engineering. Vol. 116, pp 796-800. Azzopardi, BJ., 1994c. The split of vertical annular flow at a large diameter T -junction. International Journal of Multiphase Flow 20,1071-1083. Azzopardi, B.J., Piearcey,A., Jepson,D.M., 1991.Drop Size Measurementsfor Annular two-phase flow in a 20mm diameter vertical tube. Experiments in Fluids Vol. 11, pp. 191 -197. Azzopardi, B.J., Memory, S.B, 1989.The split of two-phase flow at a horizontal Tflow. 4t' Int. Conf. on Multi-phase Flow, Nice, France, 19-21 annular and stratified June (Pub. BHRA). Azzopardi, B.J., 1988.Measurements and observationsof the split of annular flow at a T-Junction. International Journal of Multiphase Flow, Vol. 14,701-710. vertical Azzopardi, B.J., 1987. Dividing two-phase flow at a junction part I: annular flow at a T. Third International Conferenceon Multiphase Flow, The Hague, Netherlands, vertical 18-20 May. Azzopardi, 13.J., 1986.DisturbanceWave Frequencies,Velocities and Spacingin Vertical Annular Two-PhaseFlow. Nuclear Engineering and Design 92(2): 121-133. Azzopardi, B.J., Whalley, P.B., 1982.The effect of flow patterns on two-phaseflow in a T -junction. International Journal of Multiphase Flow 8,491-507. Azzopardi, B.J., Whalley P.B., 1980a. Artificial Waves in Annular two-phaseflow. Basic Mechanismsin Two-PhaseFlow and Heat Transfer proceeding, The America Soc.of Mech.Eng., Chicago,Nov. 16 -21, pp. 1-8. Azzopardi, B.J., Freeman,G., King, D.J., 1980 b. Drop sizes and deposition in annular two phaseflow. UKAEA Report AERE R9634. Baker, 0., 1954. Design of Pipelines for SimultaneousFlow of Oil and Gas. Oil and Gas Journal, 53,185. Bankoff, S.G., 1960. A variable density single fluid model for two phase flow with to steamwater flow. Trans. ASME, J. Heat Transfer 82,265-272. particular reference
References
Barbosa, J.R., Hewitt, G.F., Konig, G., Richardson, S.M., 2002.Liquid Entrainment, Droplet Concentrationand PressureGradient at the Onset of Annular Flow in a Vertical Pipe. Int. J. Multiphase Flow, Vol. 28, pp.943-961. Barnea, D., 1986. Transition from Annular Flow and from Dispersed Bubble FlowUnified Models for the Whole Range of Pipe Inclinations. I nternational Journal of Multiphase Flow, 12 (5), 733-744. Belfroid, S.P.C., Schiferli, W., Alberts, G.J.N., Veeken, C.A. M., Biezen, E., 2008. Prediction Onset and Dynamic Behavior of Liquid Loading Gas Wells. Paper SPE 115567, ATCE, Denver, Colorado, USA. Bendiksen, K. H., 1984. An Experimental Investigation of the Motion of Long Bubbles in Inclined Tubes. International Journal of Multiphase Flow, 4,467-483. Bennett, A. W., Hewitt, G.F., Kearsey, H.A., Keeys, R.K. F., Lacey, P.N. C., 1965. Flow of boiling at high pressure.Proc. Inst. Mech. Eng. 180 (Part visualization studies
3C), 1-11.
Bonizzi M., Issa, R. I. 2003. A Model for Simulating Gas Bubble Entrainment in TwoPhaseHorizontal Slug Flow. International Journal of Multiphase Flow, 29,1685 Brown, K. E., 1977. The Technology of Artificial Lift Methods, PennWell Publishing Co., Tulsa, OK. 1. Bruno, Y,, M. J. McCready, 1988.Origin of Roll Waves in Horizontal GasLiquid Flows. AIChE Journal 34(9): 1431-1440. Caetano,E.F., Shoham,0., Brill, J.P., 1992. Upward Vertical Two-PhaseFlow through Annulus. Part II. Modeling Bubble, Slug, and Annular Flow. Proc., Fourth BHRA an Intl. Conference on Multiphase Flow, Nice, France . 1989. ASME J.Energy Resources Technology, 114,13. Ceccio, S.L., 1991. Electrical ImpedanceTechniquesfor the Measurementof Multiphase Flows. Experimental Techniquesin Multiphase Flows. ASME. Ceylan, S., Kelbaliyev, G., Ceylan, S., 2003. Estimation of the Maximum Stable Drop Sizes, CoalescenceFrequenciesand the Drop Size Distributions in Isotropic Turbulent Dispersions. Colloids and SurfacesA: Physicochem.Eng. Aspects 212, pp. 285 - 295. Chopra,A. K. , 1982.Characterizationand Modeling of Annular Two-PhaseFlows. Ph.D. Dissertation. University of Houston. Coddington, P.,Macian, R., 2002. A study of the performance of void fraction in the context of drift-flux two-phase flow models. Nucl. Eng. Design correlations used 215,199-216.
307 AI B ALAAMU2010
References
Collins, R., de Moraecs, F.F., Davidson, J.F., Harrison, D., 1978. The motion of large bubbles rising through liquid flowing in a tube. J. Fluid Mech. 89,497-514.
Corcoran, T. E., Hitron, R., Humphrey, W., Chigier, N., 2000. Optical Measurement of Nebulizer Sprays: A Quantitative Comparison of Diffraction, Phase Doppler Interferometry, and Time of Flight techniques. Journal of Aerosol Sci., Vol. 31, pp. 3550.
Costigan, G., Whalley, P.B., 1997. Slug flow regime identification from dynamic void in fraction measurements vertical air/water flows. International Journal of Multiphase Flow 23,263-282. Das, G., Das, P.K., Azzopardi, B.J., 2005. The split of stratified gas-liquid flow at a small diameter T -junction. International Journal of Multiphase Flow 31,514-528. Davies, R. M., Taylor, G. 1., 1950.The mechanicsof large bubbles rising through Extended liquids and through liquids in tubes.Proc Roy. Soc., 200A, 375-390. DeJong, P., 1999. An investigation of film structure and pressure drop in microgravity flow. M. Sc. Thesis, University of Saskatchewan, Canada. annular Delichatsios and `Probstein,. R. F., 1975. Coagulation in turbulent flow: theory and J. Colloid InterfLee sci. 51,394-405. experiment. Dionissios P.M., 2007. T -junction separationmodelling in gas-liquid two-phaseflow. Chemical Engineering and Processing,Vol. 46,150-158. Dressler, R. F. 1949.Mathematical solution of the problem of roll waves in inclined , Communicationson Pure and Applied Mathematics 2: 149-194. open channels. Dukler, A. E., 1960. Fluid Mechanics and Heat Transfer in Vertically Falling-Film Systems.Chem Eng. Prog. Symposium Series,56, No 30,1. Dumouchel, C., Yongyingsakthavorn, P., Cousin J., 2009. Light Multiple Scattering Correction of Laser-Diffraction Spray Drop-Size Distribution Measurements. Int. J. Multiphase Flow, Vol-35,277-287. Ferre D., 1979.Ecoulementsdiphasiquesa pochesen conduite horizontale Rev. Inst. Fr. Pet., 34,113-142. Fernandes,R. C., Semiat, R, and Dukler A. E., 1983: Hydrodynamic Model for GasLiquid Slug Flow in Vertical Tubes AIChE Journal, 29 (6), 981-989 Fore, L. B., Dukier, A. E., 1995. The distribution of drop size and velocity in gas -liquid Flow. Int. J.Multiphase Flow, 21,137-149. annular
308 MB ALAMU2010
References
Fossa, M., 1998. Design and performance of a conductanceprobe for measuring liquid fraction in two-phase gas-liquid flow. Flow Measurementand Instrumentation 9,103109. in Fossa,M., Guglielmini G. 1998. Dynamic void fraction measurements horizontal ducts with suddenareacontraction.' International Journal of Heat Transfer, Vol. 14, 3807-3815. Fukano, T., T. Morimoto, K. Sekoguchi A. Ousaka,1983. Air/water Annular Two-Phase Flow in a Horizontal Tube. 2nd Report, Circumferential Variations of Bulletin of the JSME 26(218): 1387-1395. Film Thickness Parameters. Gardner, G.C., Neller, P.H., 1969. PhaseDistribution in Flow of an Air/Water Mixture Rounds and Past Obstructions at the Wall of 76 mm Bore Tube. Proc. Inst. Mech. Engrs., 184,36. Geraci, G., 2005.Gas-Liquid Flows in Inclined Pipes and Venturis. PhD Dissertation, Chem. Eng. Dept. University of Nottingham. Gri ff ith, P., Wallis, G. B. 1 96 1. Two-phaseslug flow. J. Heat Transfer 83: 307-20. Govan, A. H., 1988. A note on statistical methods for comparing measured and UKAEA Report AERE-M 3621. calculated values. Govier, G. W., Aziz, K., 1972. The Flow of Complex Mixtures in Pipes. Van Nostrand Reinhold Co., New York, NY. Hagedom,A. R, Brown, K. E., 1964.The Effects of Viscosity in Two-PhaseVertical Flow, "JPT, Feb. Hale C. P., 1994.Modelling the slug flow regime. Technical Report, MPS/62, WASP/16, Departmentof Chemical Engineering, Imperial College of Science, Technology & Medicine, London, UK. Hall Taylor, N. S., Hewitt, G.F., Lacey, P.M. C., 1963. The Motion and Frequencyof Large DisturbanceWaves in Annular Two-PhaseFlow of Air/water Mixtures. Chem. Eng. Sc., Vol. 18. pp. 537 - 552. Hall Taylor, N. S., Nedderman,R.M, 1968.The coalescenceof disturbancewaves in two-phaseflow. Chem. Eng. Sci. Vol. 23,551-564. annular Hall Taylor, N. S., 1966. Interfacial wave phenomenain vertical annular two-phaseflow. Ph.D. Thesis. Trinity College, Cambridge.
309 MB ALAMU2010
References
Han, H., Z. Zhu, K. Gabriel, 2006. A Study on the Effect of Gas Flow Rate on the Wave Characteristicsin Two-PhaseGas-Liquid Annular Flow. Nuclear Engineering and Design 236(24): 2580-2588. Hanratty, T. J., A. Hershman, 1961. Initiation of Roll Waves. AIChE Journal: 488-497. Hanratty, T. J., Asali, J.C., 1983. Entrainment Measurementsand Their Use in Design Equations. Report forDesign Inst. For Multiphase Flow, U. of Illinois, Urbana, Illinois. Harmathy, T. Z., 1960.Velocity of Large Drops and Bubbles in Media of Infinite or Restricted Extent. AIChE J., 6, p. 281. Hassan, A. R, Kabir, C.S., 1988.A Study of Multiphase Flow Behaviour in Vertical Wells. SPEPE, 263, Trans., AIME, 285. Henstock, W. H., Hanratty, T. J., 1976.The interfacial drag and height of the wall layer in flows. AIChE J. 22 6, pp. 990-1000. annular Hewitt, G.F., 1961. Analysis of Annular Two-PhaseFlow: Application of Dukler Analysis to Vertical Upwards Flow in a Tube. AERE-R 3680. Hewitt, G.F., 1978.Measurements two-phaseflow parameters.Academic Press, of London. Hewitt, G.F., Hall-Taylor, N. S., 1970.Annular Two-PhaseFlow. PergamonPress, Oxford. Hewitt, G.F., Roberts, D.N., 1969. Study of Two-phase Flow Pattern by Simultaneous X-Ray and Flash Photography.UKAEA, Report AERE-M2159. Hills, J.H., 1997.The critical liquid flow ratesfor wave and droplet formation in flow. Exp. Heat Transfer Fluid Mech. Thermodyn. Ed. ETS 2, annular gas-liquid 1241-1247 Hinze, J. 0.1955. Fundamentals of the Hydrodynamic Mechanism of Splitting in AIChE J.1,289. Dispersion Processes. Hong, K. C., 1978.Two-phaseflow splitting at an impacting tee. J. Pet. Technology, 290296.
Ilobi, M. I., Ikoku, C. U., 1981,Minimum Gas Flow Rate for Continuous Liquid Removal in Gas Wells. PaperSPE 10170presentedat SPE Annual Technical Conferenceand Exhibition, San Antonio, TX, October 4-7.
310 MB ALAMU2010
References
Ishii, M., Grolmes, M. A., 1975. Inception criteria for droplet entrainment in two-phase concurrent film flow. "American Institute of Chemical EngineersJournal, Vol. 21,308318. Ishii, M. K. Mishima 1989. Droplet Entrainment Correlation in Annular Two, PhaseFlow 3. of Heat International Journal and Mass Transfer 2(10): 1835-1846. Issa RI., Bonizzi M., Barbeau S., 2006. Improved closure models for gas entrainment and interfacial shearfor slug flow modeling in horizontal pipes. Int. J. Multiphase Flow, 2006, Vol: 32, pp: 1287-1293. Jayanti, S., G. F. Hewitt, S. P. White 990. Time-dependentbehaviour of the ,1 liquid film in horizontal annular flow. International Journal of Multiphase Flow 16(6): 1097-1116. Johnson,G. W., NossenJ, BertelsenA. F. 2005. A Comparison between , Experimental and Continuous Theoretical Roll Waves in Horizontal and Slightly Inclined Pipes at High Pressure.BHR Group 2005 Multiphase Production. Kaji, R, 2008. Characteristicsof Two-Phase Structuresand Transitions in Vertical Upflow. PhD Thesis, Chem. Eng. Dept. University of Nottingham. Kataoka, I., Ishii, M., Mishima, K., 1983. Generation and size distribution of droplet in two-phase flow. Journal of Fluids Engineering Transactions of the ASME 105 annular (2), 230-238. Kondo,Y., Mori, K., Yagishita, T., Nakabo, A. 1999. Effects of liquid viscosity on wave behaviour in gas-liquid two-phaseflow. In: Proc. 5th ASME/JSME Joint Thermal. Kumar, It, M. Gottmann, K. R Sridhar., 2002. Film Thickness and Wave Velocity 634-642. Mesaurementsin a Vertical Duct. Transactionsof the ASME 124(September): Kuru, W. C., Sangalli M., Uphold D. D., McCready M. J., 1995. Linear Stability of Stratified Channel Flow. International Journal of Multiphase Flow 21(5): 733-753. Lin, P.-Y. 1985. Flow Regime Transitions in Horizontal Gas-Liquid Flow. Ph.D. , Dissertation. University of Illinois at Urbana-Champaign. Lee, J.K., Lee, S.Y., 2001. Dividing two-phaseannular flow within a small vertical a horizontal branch. Proceedingsof the Third International rectangularchannel with Conferenceon Compact Heat Exchangersand EnhancementTechnology for the Process Industries, Davos, Switzerland, July 1-6.
311 MB ALAMU2010
References
Lee, J.K., Lee, S.Y., 2004. Assessmentof models for dividing two-phase flow at small T -junctions. Fifth International Conferenceon Multiphase Flow, Yokohama, Japan,May 30-June 4. Lin, P. Y., and Hanratty, T. J., 1986. Prediction of the Initiation of Slugs with Linear Stability Theory. Int. J. Multiphase Flow, 12,79-98. Mak, C.Y., Omebere-Iyari,N. K., Azzopardi, B.J.,2006. The Split of vertical two-phase flow at a small diameterT -junction. Chem. Eng.Sci., 61,6261-6272. Mandhane, J. M., Gregory, G.A, Aziz, K., 1974. Flow Pattern Map for Gas-Liquid Flow in Horizontal Pipes.Int. J. Multiphase Flow, 1,537-553. Mandil, C, 2002 Environmental and Technical Issues Associated with NonConventional Oil. International Energy Agency (IEA) Conf. on Non-Conventional Oil, Calgary, Canada,Nov. 25 - 26. Mantilla I., 2008.Mechanistic Modeling of Liquid Entrainment in Gas in Horizontal Pipes. PhD Thesis, PetroleumEngr. Dept. University of Tulsa, Ok, USA. Martin, C. J., Azzopardi B.J., 1984.Wave in Vertical Annular Flow. PCH, Physico Chemical Hydrodynamics 6 (1-2): 257-265. Martin, C.J., 1983.Annular two-phaseflow. D.Phil. Thesis, University of Oxford. McGillivray, RM., Gabriel, K. S., 2002. Annular flow film characteristics in variable Ann. N. Y. Acad. Sci., Vol. 974,306-315. gravity. McNeil, D.A., Stuart, A. D.,2003. The effects of a highly viscous liquid phaseon two-phaseflow in a pipe. International Journal of Multiphase Flow, vertically upward Vol. 29,1523-1549. Miya, M., WoodmanseeD. E, Hanratty T. J, 1971.A Model for Roll Waves in Gas-Liquid Flow. Chemical Engineering Science26: 1915-1931. Moalem Maron, D., Dukler,A. E., 1984. Flooding and upward filmflow in vertical tubes II: Speculationson film flow mechanisms.International Journal of Multiphase Flow 10, 599. Mori, K. Nakano, K., 2001. Effects of liquid viscosity on inception of disturbance , behavior in gas-liquid two-phaseflow. Experimental Heat Transfer, Fluid waves Mechanics and Thermodynamics,Thessaloniki, 1829-1834. Nan Da Hlaing, Anuvat Sirivat, Kitipat Siemanond James 0. Wilkes, 2007. Vertical , two-phase flow regimes and pressure gradients :Effect of Viscosity. Experimental Thermal and Fluid Science,31,567 - 577.
312 A! B ALAMU2010
References
Nicklin, D.J., Wilkes, J.O. Davidson, J.F. 1962. Two-phase flow in vertical tubes. , Transaction of the Institution of Chemical Engineers,vol. 40, pp 61-68. Oliemans, R V. A., B. F. M. Pots, N. Trompe, 1986. Modeling of Annular Dispersed Two-Phase Flow in Vertical Pipes. International Journal of Multiphase Flow 12(5), pp. 711-732. Omebere-Iyari, N. K., Azzopardi, B.J. 2006. A study of flow patternsfor gas/liquid flow in small diametertubes."Chemical Engineering Researchand Design, Vol. 85, 180-192. Omebere-lyari, N. K., Azzopardi, B.J., Akinmade, A. A., 2005. Flow patternsfor gasliquid flow in small diametertubes.Ninth UK National Heat Transfer Conference, Manchester,UK, 5-6 September. Behaviour in Gas Pipelines is Predictable. Oil & Gas Oranje L. 1973.Condensate , Journal, 32: 39-44. Paleev, I. I. B. S. Filippovich, 1966. Phenomenaof Liquid Transfer in Two-Phase , Dispersed Annular Flow. International Journal of Heat and Mass Transfer, Vol. 9, pp. 1089-1093. Pan, L., T. J. Hanratty, 2002. Correlation of Entrainment for Annular Flow in Vertical Pipes. International Journal of Multiphase Flow, vol. 28, pp. 363 - 384. Paras,S. V., A. J. Karabelas, 1991.Droplet Entrainment and Deposition in Horizontal Annular Flow. International Journal of Multiphase Flow 17(4): 455-468. Paras,S. V., N. A. Machos, A. J. Karabelas 1994. Liquid Layer Characteristics in Stratified-Atomization Flow. International Journal of Multiphase Flow 20(5): 939-956. Pearce,D. L., 1979.Film waves in horizontal annular flow: spacetime correlator Central Electricity ResearchLaboratoriesReport RD/L/N/111/79. experiments. PetalasN., Aziz K., 1998. A mechanisticmodel for multiphase flow in pipes. 49th Annual Technical Meeting of Petroleum Society of the CanadianInstitute of Mining, Metallurgy & Petroleum.Paper 98-39, Calgary, Alberta June 8-10, Canada. Pickering, P.F., Hewitt, G.F., Watson, M. J., Hale, C.P., 2001. The Prediction of Flows in Production Risers- Truth or Myth?., FEESA Ltd., Camberley, Surrey, UK. Pinto, A. M. F. R., Campos,J. B. L. M., 1996. Coalescenceof Two Gas Slugs Rising , in a Vertical Column of Liquid. Chemical Engineering Science, 51 (1), 45-54.
313 MB ALAAIU2010
References
Pols, K M. 1998.Waves in separatedtwo-phaseflow. Ph.D Thesis. The University of , Nottingham, UK. Ros N. C.J.,1961. SimultaneousFlow of Gas and Liquid as Encounteredin Well , Tubing. JPT, Oct. Rosin, P Rammler, E., 1933. Laws Governing the Fitness of Powdered Coal. Journal ., of the Inst. Of Fuel, Vol. 7, pp.29-36. Sawant, P.H., Ishii, M., Mori, M., 2008a. Droplet entrainment correlation in vertical upward co-current annular two-phase flow. Nucl. Eng. Des. 238 (6), 1342-1352. Sawant,P.H., Ishii, M., Mori, M., 2008b. Propertiesof disturbancewaves in vertical flow. Nucl. Eng. Des. 238 (12), 3528-3541. annular Sawant, P.H., Ishii, M., Mori, M., 2009. Prediction of Amount of Entrained Droplets in Vertical Annular Two-phase Flow. Int. J. Heat Fluid Flow. Schmidt, J., Giesbrecht,H. Van der Geld, C.W.M., 2008. Phaseand velocity , distribution in vertically up-ward high-viscosity two-phase flow. Int. J. Multiphase Flow, 34,363 - 374. Sekoguchi, K., Mori, K., 1997.New developmentof experimental study on interfacial in gas-liquid two-phaseflow. In: Giot, M., Mayinger, F., Celata, G. (Eds.), structure -P. Experimental Heat Transfer, Fluid Mechanics and Thermodynamics. Edizione ETS 2, pp. 1177-1188. Sekoguchi, K., Mori, K., Tsujino, H., Ikeshita, M., and Kaji, M., 1994. Wave Venation in gas-liquid two-phaseflow. First report, time-spatial map of wave behavourand its characterisations. Sekoguchi, K., Ueno, T., Tanaka, 0., 1985.An investigation of the flow Characteristics in the disturbancewave region of annular flow: 2nd report, on correlation of principal flow parameters.Trans. JSME SeriesB 51,1798-1806. Sekoguchi, K., Takeishi, M., 1989. Interfacial Structures in Upward Huge Wave Flow Annular Flow Regimes.Int. J. Multiphase Flow, Vol. 15, pp.295-305. and ShohamO. 2006. Mechanistic Modelling of Gas-Liquid Two-PhaseFlow in Pipes. , SPE Book, U. of Tulsa, Oklahoma, USA. , Shoham,0., Brill, J.P., Taitel, Y., 1987.Two-phase flow splitting in a teejunctionChemical Engineering Science42,2667-2676. experimentaland modelling. Simmons,M. J.H. Hanratty, T.J., 2001. Droplet size measurements horizontal in , flows. Int. J. Multiphase Flow, Vol. 27,861-883. annular gas-liquid
314 MB ALAMU2010
References
Sleicher, C.A., 1962.Maximum stable drop size in turbulent flow. American Institute of Chemical EngineersJournal 8 (4), 471-477. Sliwicki, E., Mikielewicz, J., 1988. Analysis of an annular-mist flow model in a Tjunction. International Journal of Multiphase Flow 14,321-331. Soleimani, A., Hanratty, T. J. 2003. Critical Liquid Flows for the Transition from the Pseudo-Slugand Stratified Patternsto Slug Flow. Int. J. Multiphase Flow, 29,51-67. Stacey,T., Azzopardi, B.J., Conte, G., 2000. The split of annular two-phaseflow at a small diameter T -junction. International Journal of Multiphase Flow 26,845-856. Strand, B, Djuve, S.M. 2000. Pipeline Transport of Oil with Strongly Temperature DependentViscosity. BHR Group Multiphase Technology,pp. 17-33. Taitel, Y., Barnea, D., Dulder, A. E., 1980. Modeling Flow Pattern Transitions for Steady StateUpward Gas-Liquid Flow in Vertical Tubes. AIChE. J., 26, pp. 345-354. Taitel, Y., and Bamea, D. 990. Two-PhaseSlug Flow. Advances in Heat Transfer, 20, ,1 83-132. Taylor, G.I., 1961.Deposition of a viscous fluid on the wall of a tube. J. Fluid Mech., Vol 10, pp. 161-165 . Theron B., 1989. Ecoulementsdiphasiques instationnaires en conduite horizontale. PhD thesis, Institut National Polytechniquede Toulouse, France. Tobin, T., Muralidhar, R., Ramkarishna,D., 1990.Determination of Coalescence Frequenciesin Liquid - Liquid Dispersions: Effects of Drop Size Dependence.Chem. Eng. Sci. Vol. 45, No. 12,3491- 3504. Tsochatzidis,N. A., Karapantsios,T.D., Vostoglou,MY, Karabelas,A. J., 1992.A for measuringliquid fraction in pipes and packed beds.Int. J. conductanceprobe Multiphase Flow 18,653-667. Analysis and Prediction of Minimum Turner, R.G., Hubbard, M. G., Dukler, A. E., 1969. Flow Rate for Continous Removal of Liquid from Gas Wells. Journal of Petroleum Technology, pp. 1475- 1482. Uphold, D. D., 1997. Linear Stability of Multifluid Flows. Ph.D. Dissertation. Department of Chemical Engineering.University of Notre Dame. Hout, R.,Barnea, D., Shemer,L. , 2002. Experimental investigation of the velocity van field induced by a Taylor bubble rising in stagnant water. International Journal of Multiphase Flow vol. 28, pp 579-596.
315 MB ALAAIU2010
References
Westende,J.M. C., Kemp, H. K., Belt, R.J., Portela, L. M., Mudde, R.F., Oliemans, van't R.V. A., 2007. On the Role of Droplets in Cocurrent Annular and Chum-Annular Flow. Int. J. Multiphase Flow, Vol. 33, pp.595-615. Wallis, G. B., 1969.One-dimensionalTwo-phaseFlow. McGraw-Hill, New York. Wang L., Wu Y., Zheng Z., Guo J., Tang C., 2008. Oil-Water Two-PhaseFlow Inside T -junction. Journal of Hydrodynamics, 20(2), 147-153. Watson, M. 1989.Wavy Stratified Flow and the Transition to Slug Flow. Proceedingsof the 4th International Conferencein Multi-phase Flows, Nice, France. Weber, M. E., 1981. Drift in intermittent two-phaseflow in horizontal pipes, CanadianJ. Eng 59 (1981), pp. 398-399. chem. Weihong Meng, Xuanzheng T. Chen, Gene E. Kouba, Cem Sarica, James P. Brill, 2001.Experimental Study of Low-Liquid-Loading Gas-Liquid Flow in Near-Horizontal Pipes. SPE Production & Facilities (November), pp.240-249. Whalley, P. B., Hewitt, G. F., 1978.The Correlation of Liquid Entrainment Fraction Entrainment Rate in Annular Two-PhaseFlow. UKAEA Report, AERE-R9187, and Harwell. White, E.R. Beardmore,R.H., 1962.The velocity of rise of single cylindrical air , bubbles through liquids in vertical tubes.Chem Eng Sci, 17: 351. N. S., Azzopardi, B.J. , Thompson,C.P., 1983. Wave coalescence Wilkes, and in vertical annular two-phaseflow. Int. J. Multiphase Flow, Vol. 9,383-398 entrainment Willets, I. P., 1987. Non-aqueousannular two-phase flow. Ph. D. Thesis, University of Oxford, UK. Woldesemayat,M. A., Ghajar, A J., 2007. Comparison of void fraction correlations for different flow patterns in horizontal and upward inclined pipes. International Journal of Multiphase Flow, 33,347-370 Wolf, A., Jayanti, S., Hewitt, G. F., 2001. Flow development in vertical annular flow", Chemical Engineering Science,Vol. 56, pp. 3221-3235. Woodmansee, D. E. , 1968. Atomization from a Flowing Horizontal Water Film by a Parallel Air Flow. Ph.D. Dissertation.University of Illinois at Urbana-Champaign.
References
Woodmansee,D.E., Harrantty, T. J., 1969. Mechanisms for the removal of droplets from a liquid surfaceby a parallel air flow. Chem. Eng. Sci., Vol. 24, pp. 299-307. Wren, E., Baker, G., Azzopardi, B.J., Jones,R, 2005. Slug flow in small diameterpipes T -junctions. Experimental Thermal and Fluid Science,29,893-899. and Xiaodong Z., 2005. Mechanistic-based models for slug flow in vertical pipes. PhD Thesis.Texas Technology University. Yao, S.C., Sylvester, N. D., 1987. A Mechanistic Model for Two-Phase Annular Mist Flow in Vertical Pipes.AIChE J., 33,1008. Zabaras,G., Dukler, A. E., Moalem-Maron, D., 1986. Vertical Upward Co-current GasLiquid Annular Flow. AIChE J., Vol. 32, pp.829-843. Zaidi, S.H., Altunbas, A., Azzopardi, B. J., 1998.A Comparative Study of PhaseDoppler Laser Diffraction Techniques to Investigate Drop Sizes in Annular Two-Phase and Flow. Chem. Eng. J., Vol. 71, pp.135- 143.
Zhang, H. Q., Wang, Q., Sarica, C., Brill, J.P., 2003. A Unified Mechanistic Model for Slug Liquid Hold up and Transition between Slug and Dispersed Bubble Flows. Int. J. Multiphase Flow, Vol. 29, pp. 97-107.
Zhou, D., Yuan, H., 2009. New Model for Gas Well Loading Prediction. Paper SPE 120580,SPE Production and Operation Symposium, Oklahoma, USA. Zuber, N. Findlay, J. A. ,1965. Average Volumetric Concentration in Two-PhaseFlow , Systems. J. Heat Transfer, Trans. ASME, 87,453-468. Zukoski,E.E. 1966. Influence of Viscosity, SurfaceTension, and inclination angle on , long bubbles in closed tubes. J. Fluid Mech., 25(4), 821-837. motion of
317 AfBALAMU2010
Nomenclature
Nomenclature
AF Ac AP B C CF C1 Co Cv D, d, dp D50 D32 dF dmax d dcD Df dL dP dc Eo EM. EF f, fF fF FrL Fr, FrM fw fsc fE
g
Area of the liquid film Area of the liquid film Area of the pipe Flow distribution coefficient, used in Eqt. 2.25 Volumetric drop concentration Film friction factor constant,used in Eqt.2.48 Drift velocity coefficient Flow distribution coefficient Factor that accountsfor the effect of surface tension on the internal flow, used in Eqt. 6.1 Pipe diameter Mass Median Diameter SauterMean Diameter Liquid film hydraulic diameter Maximum stablediameter of dispersedbubble Critical bubble diameter Critical bubble diameter abovebubbles start to coalesceand deform Drop frequency Differential length, used in Eqt.2.32 Differential pressure,used in Eqt. 2.33 Gas core diameter, used in Eqt.2.36 Eotvos number Maximum entrainedfraction Entrained fraction Interfacial friction factor, used in Eqt.2.5 Film friction factor Liquid phasefriction factor, used in Eqt.6.12 Liquid Froude number Critical Froude number, used in Egt.2.14 Mixture Froude number Wave frequency, used in Egt.5.10 Gas core friction factor Entrained fraction
Acceleration due to gravity
m2 m2 m2 m m gm m m in m Hz m Pa m Hz m/s2
Interfacial friction factor PearceCoefficient constant,used in Eqt.5.5 Morton number Total liquid massflux Inverse velocity number Exponent, used in Eqt. 2.27 Liquid viscosity number Archimedes number Atmospheric pressure,used in Eqt.3.6
kg/m-s2 Bar
Nomenclature
Volumetric liquid flow rate in the film, used in Eqt.2.37 Volumetric liquid flow rate, used in Eqt.2.37 Volumetric gas flow rate, used in Eqt. 2.39 Auto-covariance function, used in Eqt. 5.1 Slug Reynolds number Reynolds number Core Reynolds number Superficial core Reynolds number Film Reynolds number Liquid Reynolds number, used in Eqt. 2.60 Minimum liquid Reynolds number, used in Eqt.2.61 Slip velocity ratio Annular liquid film perimeter, used in Eqt. 2.32 Strouhal number, used in Eqt. 3.1 Liquid basedStrouhal number, used in Eqt. 5.4 Interfacial annular parameter,used in Eqt. 2.33
Sampling time
superficial velocity Taylor bubble velocity, used in Eqt. 2.1,2.3,2.3 Structurevelocity Superficial gas velocity Superficial liquid velocity Drift velocity, used in Eqt.3.5 Superficial core velocity, used in Eqt.2.40 Superficial mixture velocity Translational velocity Core velocity Gasvelocity Drop velocity Film velocity, used in Eqt. 2.38 Liquid film velocity Critical gasvelocity at the onset of atomization Wave velocity Windowing function Weber number Critical Weber number, used in Eqt. 2.60 Lockhart-Martinelli parameter,used in Eqt. 3.2
m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s -
Nomenclature
Greek Letters
p e Eg a L C PL Pro PC Xw OP S1. SF SSW a 0 TWL T, it ASW
ac
Density Rate of energy dissipation per unit mass Cross sectional averagevoid fraction Viscosity Gasphaseviscosity Liquid phaseviscosity Gas core viscosity Pipe inclination from horizontal deltaldifference Liquid phasedensity Gas phasedensity Core density Wave spacing,usedin Eqt. 5.10 Pressuredrop Film thickness,usedin Eqt. 2.35 Film thickness Wave amplitude Surfacetension Dimensionlessconstant,used in Eqt. 2.52 Wall shearstress Interfacial shearstress 3.1415926 Wave amplitude
Core void fraction, used in Eqt. 2.42
Kg/m3 m2/s3 Kg/m-s or mPa s Kg/m-s Kg/m-s Kg/m-s degrees Kg/m3 Kg/m3 Kg/m3 m Pa/m m in m N/m Pascal Pascal in
-
aT
Subscripts
Ar F L G C M S SG SL Max CD CB f Archimedes Film Liquid Gas Core Mixture slug Superficial gas Superficial liquid Maximum Critical bubble diameter Critical bubble diameter Frequency
320 MB ALAMU2010
Appendix A
APPENDIX
Cross sectional view of advance technology of gas mass flow controller Bronkhorst, UK)
(source:
At B AI_AMU 20 10
Appendix B
APPENDIX
5 6 7 B 9 10 11 12 13 14 15 16 17 18 19 20
21
101.73 87.19 7467 63.96 54.78 4695 4022 34.47 2955 2533 2171 1860 1594 1367 1171 1004
8.61
5254 4503 3856 3303 2829 2425 2077 1780 1526 1308 1121 961 823 706 605 519
444
22 23 24 25 26 27 28 29 30 31
127.91 109.57 93.90 80.45 68.94 59.11 5066 4341 37.20 3189 27.33 23.43 20.08 17.21 14.76 12.65 10.84 929 7.97 6.83 5.85 5.02 4.30 3.69
6606 5659 4849 4155 3560 3053 2616 2242 1921 1647 1412 1210 1037 889 762 653 560 480 411 353 302 259 222 191
1024.69 85368 730.31 625.47 535.36 457.77 392.37 336.02 287.80 246.53 211.28 181.01 155.12 132.99 113.98 9768 83.70
71.75
52920 44088 37717 32302 27648 23641 20264 17354 14863 12732 10911 9348 8011 6868 5886 5044 4323
3705
18 19 20 21 22 23 24 25 26 27 28 29 30 31
61.50 52.71 45.18 38.73 33.20 28.46 24.40 20.91 17.93 15.37 13.17 11.29 9.68 8.30
3176 2722 2333 2000 1715 1470 1260 1080 926 794 680 583 500 429
Appendix C
APPENDIX C
Glycerol surface tension Temperature Coefficient 64.0 m N/ m at 20C m N/ mK -0.0598
A. Physical Properties of Air and Water at test conditions Equilibrium / steadystatetemperature 20C T-Junction pressure 1.4Bara Surfacetension (N/m) 0.073 Gas density (kg/m3) 1.40 Liquid density (kg/m) 998.0 Gas Viscosity (Pa s) 1.8E-05 (0.018 c P) 1.0E-03 (1.0 c P) Liquid viscosity (Pa s) B Physical Properties of Air and Glycerol-Water mixture at test conditions Equilibrium / steadystatetemperature 31C 1.4Bara T-Junction pressure 0.04582 Surfacetension (N/m) 1.40 Gas density (kg/m') 1261.0 Liquid density (kg/m3) 1.8E-05 (0.018 c P) Gas Viscosity (Pa s) 1.2E-02 (12 c P) Liquid viscosity (Pa s) C. Physical Properties of Air and Glycerol-Water mixture at test conditions Equilibrium / steadystatetemperature 40C 1.4Bara T-Junction pressure 0.04528 Surfacetension (N/m) 1.40 Gas density (kg/m') 1261.0 Liquid density (kg/m3) 1.8E-05 (0.018 c P) Gas Viscosity (Pa s) 1.8E-02 (18.0 c P) Liquid viscosity (Pa s)
D. Physical Properties of Air and Glycerol-Water mixture at test conditions Equilibrium / steadystatetemperature 35C I. 4Bara T-Junction pressure 0.04558 Surfacetension (N/m)
Gas density (kg/m3) 1.40
323 MB ALAMU2010
Appendix C
E Physical Properties of Air and Glycerol-Water mixture (3.6 mPa s) at test conditions Equilibrium/steady state temperature 71-junction (Sara) Surface tension 210C
pressure
1.4
(N/m)
0.071
Gas density
Liquid density
(kg/m3)
(kg/ms)
1.40
1097.0
Gas Viscosity
Liquid viscosity
(Pa s)
(Pa s)
Appendix D
APPENDIX
414249 1.1500 382017 1.1500 351549 0.1500 344018 0.1500 31.9401 0.1500 29.4945 1.1500 26.8131 1.1500 21.9189 0.1500 221756 0.1500 20.1558 0.1500 173860 0.1500 13.9613 1.1500 15.7700 0.0300 18,1200 0.0500 13,8200 0.0700 13.1300 0.0900 133100 1.1000 131800 0.1143 143300 0.1257 15.1800 0.1371 11.7800 0.1486 144400 0.1600 143300 0.1714 1416" 1.1829 1.1800 0.1943
0.9869 0.9860 0.9854 0.9843 0.9826 0.9801 0.9771 0.9735 0.9693 0.9745 0.9715 0.9555 0.9808 0.9782 0.9756 0.9736 0.9690 0.9660 0.9685 0.9654 0.9611 0.9562 0.9184 0.9417 0.9733 0.9645 0.9578 0.9528 0.9188 0.9443 0.9425 0.9408 0.938 0.9365 0.9342 0.9316 0.9303
ImmI
0.1094 0.1267 0.1470 0.1219 0.1364 0.2138 0.0916 0.1041 0.1166 0.1262 0.1484 0.1629 0.1508 0.1658 0.1866 0.2104 0.2483 0.2811 0.1277 0.1701 0.2026 0.2269 0.2464 0.2684 0.1772 0.2855 0.2992 0.3066 03179 03307 0.1371
470.23 494.84 423.06 382.22 305.47 247.52 203.13 175.26 145.27 120.99 152.45 120.32 2460,87 2281.90 2064.08 1744.02 1219.29 938.66 1168.74 820.51 601.73 407.96 292.75 246.57 5.69 34.08 172.56 149,6 77 1%. 151.88 132.23 152.07 159.54 187.96 191.85 208.18 956.06
1-1
0.4035 0.4092 03111 0.2797 0.2030 0,1542 0.1130 0.0862 0.0623 0.0464 0.0491 0.0323 0.6829 05811 0.4906 0.4023 0.2596 0.1846 01089 0.1199 0.0918 0.0548 0.0339 0.0229 0A030 0.0102 0.0341 0.0218 0.0266 0.0184 0.0151 0.0164 0.0159 0.0172 0.0160 0.0162 0.0693
1-1
171.04 0.029275 185.17 0.024657 6339 0.000569 150.11 0.003408 0,017256 0.014960 0.019677 0.015188 0,013223 0,015207
232.13 233.01 70.9 104.97 222.91 86.64 222.59 83.06 2365 99,7 235.31 97.69 103.65 107.16 106.38 121.02
225.11 0.015954 209.67 0,018796 199.55 0.019185 200.4 0.020818 241.72 0.095606
II
MB ALAMU 2010
325
Appendix E
APPENDIX E
Tabulated Film and Wave Data
VSG VSL Film Thickness STDEV
[m/s]
42.9011 41.3499 40.3161 36.5874 33.2238 31.1410 27.8176 24.5847 21.4446 19.1663 16.0963 13.4136 41.6249 38.2017 35.6549 34.6018 31.9401 29.4945 26.8132 21.9189 22.8756 20.1558 17.3860 13.9613 15.7700 15.0200 13.8200 13.1300 13.5100 13.8800 14.3300 14.7800 14.7800 14.6400 14.3300 14.2600 14.0800
[m/s]
0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.0300 0.0500 0.0700 0.0900 0.1000 0.1143 0.1257 0.1371 0.1486 0.1600 0.1714 0.1829 0.1943
[mm]
0.0627 0.0665 0.0694 0.0747 0.0827 0.0946 0.1097 0.1270 0.1470 0.1221 0.1364 0.2139 0.0913 0.1037 0.1168 0.1261 0.1487 0.1632 0.1509 0.1660 0.1867 0.2108 0.2484 0.2813 0.1275 0.1703 0.2026 0.2271 0.2464 0.2689 0.2775 0.2855 0.2995 0.3066 0.3186 0.3313 0.3372
[mm]
0.0226 0.0214 0.0226 0.0245 0.0342 0.0422 0.0483 0.0579 0.0579 0.0535 0.0997 0.0997 0.0289 0.0244 0.0289 0.0321 0.0374 0.0422 0.0380 0.0590 0.0501 0.0590 0.0768 0.0908
Appendix h
APPENDIX
Tabulated
VSG VSL VM
F
Pressure Drop Measured wave Velocity
Wave Data
[mm]
0.0627 0.0665 0.0694 0.0747
0.0827 0.0946
[m/s]
42.9011 41.3499 40.3161 36.5874
33.2238 31.1410
[m/s)
0.0500 0.0500 0.0500 0.0500
0.0500 0.0500
[m/s]
42.9511 41.3999 40.3661 36.6374
33.2738 31.1910
[-]
0.9869 0.9860 0.9854 0.9843
0.9826 0.9801
[-]
0.0131 0.0140 0.0146 0.0157
0.0174 0.0199
[Pa/ml
6718.4600 6518.4600 6381.5000 6256.9000
5858.8900 5481.6400
[m/s)
2.5250 2.4047 1.9423 2.1041
2.1956 1.7413
35.6549
34.6018
0.1500
0.1500
35.8049
34.7518
0.9756
0.9736
0.1168
0.1261
0.0244
0.0264
9760.7200
9218.0800
3.1563
3.1563
14.0800
0.1943
14.2743
0.9303
0.3372
0.0697
2268.7200
2.6579
Appendix G
APPENDIX
Tabulated
VSG [m/s]
42.9011 41.3499 40.3161 36.5874 33.2238 31.1410 27.8176 24.5847 21.4446 19.1663 16.0963 13.4136
G
Wave [Hz]
14.0000 16.0000 16.0000 14.0000 12.0000 10.0000 11.0000 9.5000 10.0000 11.0000 7.0000 9.0000
VSL [m/s] 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.0300 0.0500 0.0700 0.0900 0.1000 0.1143 0.1257 0.1371 0.1486 0.1600 0.1714 0.1829 0.1943
Measured
spacing
41.6249 38.2017 35.6549 34.6018 31.9401 29.4945 26.8132 21.9189 22.8756 20.1558 17.3860 13.9613 15.7700 15.0200 13.8200 13.1300 13.5100 13.8800 14.3300 14.7800 14.7800 14.6400 14.3300 14.2600 14.0800
29.0000 26.0000 26.0000 24.0000 22.0000 19.0000 21.0000 20.0000 21.0000 18.0000 13.0000 10.0000 13.0000 9.0000 10.0000 9.0000 8.0000 10.0000 13.0000 10.0000 10.0000 11.0000 14.0000 13.0000 12.0000
0.1024 0.1143 0.1214 0.1315 0.1435 0.1329 0.1202 0.1263 0.1002 0.1169 0.1619 0.1804 0.1143 0.2004 0.2196 0.2078 0.2425 0.2290 0.1850 0.2405 0.2525 0.2295 0.1899 0.2045 0.2215
2.5250 2.1041 2.1041 2.1041 1.8036 1.4853 1.8036 2.1957 1.8700 1.9400 2.2900 2.4048 2.4047 2.5250 2.5250 2.6579 2.6579 2.6579
Appendix H
APPENDIX
VSL [m/s]
0.0500
Measured
Measured
41.3499 40.3161
36.5874 33.2238
0.0500 0.0500
0.0500 0.0500
15.0000 4.5000
4.0000 1.5000
16.0000 16.0000
14.0000 12.0000
31.1410
27.8176
0.0500
0.0500
1.5000
1.1600
10.0000
11.0000
35.6549
34.6018
0.1500
0.1500
26.0000
24.0000
1.3300 0.1660
0.1660
8.0000
13.8800
14.3300 14.7800 14.7800 14.6400 14.3300 14.2600 14.0800
0.1143
0.1257 0.1371 0.1486 0.1600 0.1714 0.1829 0.1943
50.3300
0.1660 150.1600 2.3330 150.5000 50.1600 0.8333 0.1660
10.0000
13.0000 10.0000 10.0000 11.0000 14.0000 13.0000 12.0000