Autoignition Temperatures for Mixtures of Flammable Liquids

with Air at Elevated Pressures

by Elisabeth Brandes, Werner Hirsch* and Thomas Stolz
Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, Germany
Abstract
The autoignition temperature (AIT) of pure compounds has been measured at pressures between 2 bar and 15 bar in
a 0.5 l autoclave. The AITs are found to decrease substantially with increasing pressure, following a Semenoff
relation allowing extrapolation of the AIT to higher pressures. The ignition delay times follow an Arrhenius-like
relation and may become very long at higher pressures. For some compounds both the AITs and the delay times, the
1 bar values found with the standard apparatus do not fit well to the respective relation, pointing to a profound
influence of the experimental differences. Although very high fuel concentrations are required to find the minimum
AIT, they remain within the explosion range as the Upper Explosion Limit is shown to shift dramatically to higher
values with increasing temperature and pressure.

Introduction
Today, many industrial processes are operated at
elevated pressures. It is known that the AIT of a
fuel/air-mixture drops with increasing pressure, a fact
explained theoretically by Semenoff /1/ as early as
1928. Since then, a lot of work has been done to determine AITs at elevated pressure (see, for example, the
work of Gdde /3/ and references cited therein). Nevertheless, the safety characteristics of many technically
important substances are still poorly known under these
conditions. The present work deals with the determination of autoignition temperatures (AITs) of several
single compounds under high pressures.
Even less knowledge exists about the Upper Explosion Limit (UEL) at high pressures. A second series of
experiments was set up therefore to determine the UEL
at elevated pressure and a temperature near their AIT
for some of the pure compounds investigated.

admitted as a liquid and evaporates in the reaction

vessel. However, the reaction does not take place under
isobaric conditions like in IEC 60079-4 but under
isochore conditions.
The present experiments covered the range from
2 bar to 10 bar total pressure. In some cases, AITs were
also obtained for 1.5 bar or 15 bar. The fuel concentrations usually varied between 15% by vol. and around
40% by vol..

Experimental set-up
AIT
Under atmospheric conditions, the AIT of a liquid is
usually determined by dropping the liquid into a preheated Erlenmeyer flask, ignition being indicated by the
occurrence of a visible flame /2/. Measurements at
elevated pressures require a closed reaction vessel as
shown in Fig. 1. We use a 0.5 l autoclave (1). Ignition is
indicated either by a temperature rise (observed by two
thermocouples (6) within the vessel) or by the pressure
rise observed by the pressure transducer (10). For an
experiment, air is first introduced into the autoclave
from its supply (4), regulated to the desired pressure by
the pneumatic valves (7), (8), and (9), then the liquid is
fed into the vessel from its supply (2) by a HPLC pump
(3). The resulting pressure is taken as the starting
pressure. As with the standard apparatus, the fuel is

Corresponding author: werner.hirsch@ptb.de

Proceedings of the European Combustion Meeting 2005

Fig. 1: Diagram of the apparatus for measuring AITs

UEL
UELs at 10 bar and 180C or 200C were determined in a separate 9 l autoclave (11) (Fig. 2) heated
uniformly by a thermofluid jacket (15) equipped with a
conventional capacitive spark igniter (19). Ignition is
indicated either by a temperature rise > 50 K (observed
by thermocouples (16) within the vessel) or by the
pressure rise > 5% of the starting pressure observed by

trations of 25% and higher. Therefore for several compounds the influence of pressure on the UEL was
explored. The results are presented in Tab. 1.
Tab. 1: UEL of some single compounds at elevated temperature
and pressures of 10 bar

22

12

21
P

15

14
11
20
P

17

18
19

Fig. 2: Diagram of the apparatus for measuring UELs

Dependency of AIT on fuel/air ratio
To find the AIT at a given pressure, the fuel/air ratio
also has to be varied. In the case of simple organic
molecules the lowest values have always been found at
very fuel rich mixtures (high fuel/air ratios). As Fig. 3
shows, the present results are in accordance with these
previous findings.

Fig. 3: Dependency of the temperature rise T of a

reaction on the composition. The filled symbols indicate
the runs regarded as ignition.
These high fuel concentrations are far beyond the
UEL measured under atmospheric conditions. Incomplete mixing of the evaporating fuel with the air in the
autoclave may play some role, but can account only
partly for that result. The known shift of the UEL to
higher values at high temperatures is also not sufficient
to explain the possibility of an ignition at fuel concen-

temperature of
measurement in C

16

UEL in % by vol.
at 10 bar

1-Propanol

28.8

41.8

200

2-Propanol

14.5

39.3

200

Cyclohexane

10.5

39.6

200

n-Hexane

22.1

42.7

180

n-Heptane

26.4

40.5

180

Pentane

10.7

44.4

180

Acetone

Compound

13

UEL in % by vol.
(atmospheric)

the pressure transducer (20). To start an experiment, air

is introduced into the heated autoclave from its supply
(14) to the desired pressure (21) controlled by pneumatic valves (17) and (18), then the liquid is fed into the
vessel from its supply (12) by a HPLC pump (13) using
a nozzle which generates a very fine spray. The mixture
is homogenized by stirring for some minutes. The
resulting pressure is taken as the starting pressure.

16.2

22.5

180

Butanone

12.6*)

22.5

180

Methylpropionat

13.0*)

26.7

200

Ethylacetat

12.8*)

24.6

200

Ethanol

36.4

52.8

200

Methanol
*) at 100C

54.1

59.4

200

Tab. 1 shows that the increase of the UEL with

increasing pressure is in most cases even more dramatic
than that due to the temperature increase. This is in real
contrast to the behaviour of the Lower Explosion Limit
(LEL) which is known to be nearly independent of
pressure at least for pressures up to 5 bar. The relation
between the range of autoignition, the UELs at 1 bar
and at 10 bar and the maximum possible fuel
concentration (due to limited vapour pressure) is displayed in detail for the example of n-propanol in Fig. 4
and for hexane in Fig. 5. In both cases the lowest
temperature of ignition at 10 bar is reached at fuel
concentrations above the UEL at 1 bar at the same
temperature.
A consequence of the high fuel concentrations at
AIT is, however, that the pressure increase after ignition
is usually rather weak. Reasons are the incompleteness
of the oxidation and the high heat capacity of the fuel
that remains unreacted. Near the AIT the reaction also
often does not proceed through the whole mixture.
Therefore near AIT the pressure only rises by a factor of
2 or less.

50

Explosion range at 10 bar

40

fuel concentration in % by vol.

Range of autoignition
at 10 bar

max. concentration at 1 bar

max. concentration at 10bar
UEL at 10 bar
Upper explosion limit at 1 bar
LEL at 1 and 10 bar
autoignition temp. at 10 bar

30

20

10

Explosion range at 1 bar

0
0

50

100

150

200

diffusion rates or the time to heat up the cold injected

liquid. As most of these factors have an Arrhenius-like
dependency on temperature similar to the reaction rate,
it is nevertheless possible to obtain a straight line in an
Arrhenius plot. This is demonstrated for four compounds in Fig. 7. The data taken for pressures from
2 bar to 15 bar all fall on a single line if a first order
dependence on pressure is assumed. In contrast, delay
times observed at 1 bar with the standard apparatus do
not fit on the line but are consistently longer than
expected from an extrapolation from the high pressure
values. They are therefore excluded from Fig. 7.

250

300

16
pressure
temperature at the centre of autoclave
temperature at the top of autoclave

T / C

Fig. 4:.Relation between the UEL at 1 bar and 10 bar, maximum

possible concentration and the range of autoignition for n-propanol
Temperature in C

40

R a n g e o f a u toig n itio n
a t 1 0 ba r

14
260
13
240

12
11

220
10

E xp lo sio n ra ng e at 1 0 b a r

fu e l con ce ntra tio n in % by vo lu m e

L E L a t 1 b a r a n d 1 0 b ar
m a x. c o n ce n tra tio n a t 1 b a r
m a x.co n ce n tra tio n at 1 0 b a r
UEL at 1 bar
UEL at 10 bar
A IT a t 1 0 b a r

15

30

20

10

Pressure in bar

50

280

200
9
180
0

500

1000
1500
Time (from start of injection) in s

8
2500

2000

Fig. 6:Autoignition of a benzene/hexane mixture at

197C: Igniton delay time > 35 min

E xp lo sio n ran g e a t 1 b a r
10000

0
0

50

100

150

benzene
butyl amine
cyclohexanone
propionic acid

200

T /C

Fig. 5: Relation between the UEL at 1 bar and 10 bar, maximum

possible concentration and the range of autoignition for n-hexane

E AZV

exp
RT
pn

where EAZV is an apparent activation energy. Therefore

it is expected that due to the lower ignition temperatures
remarkably longer ignition delay times can be observed
at higher pressures. An extreme example is displayed in
Fig. 6 where it takes more than 35 min (2100 s) for a
50%-benzene/50%-hexane mixture to ignite at a
pressure of 13.5 bar and 197C.
As can be also seen from Fig. 6, the ignition delay is
not only influenced by chemical factors (reaction rates),
but also by physical factors like vaporisation and

tZV *pZ in bar*sec

Ignition delay times

Due to limited reaction rates, some time will pass
between the admission of the fuel and the actual occur
rence of the explosion. For determination of the AIT it
is therefore necessary to wait for some time before the
outcome of a run can be regarded as "no ignition".
Under atmospheric conditions the standards require
a waiting time of 5 min which is usually sufficient to
avoid the possibility of overlooking an explosion due to
a very high ignition delay time. As ignition delay times
are closely related to reaction rates, they can, however,
be shown to follow an Arrhenius-like relation /4/:

1000

100

10
200

250

300
350
Temperature in C

400

450

500

550

Fig. 7: Representative Arrhenius plots for ignition delay times with a

first order dependency on pressure
Apparent activation energies can be calculated from
the slopes of the lines in Fig. 7. They may be used to
estimate the delay times for reactions at different pressures. As they are composed of several factors, they are,
however, not expected to agree well with activation
energies calculated or measured by different methods.
Autoignition temperatures
The primary objective of the present work is to
determine the autoignition temperatures at elevated
pressures. The results obtained so far for pure compounds are summarised in Tab. 2 and compared to the
values measured at atmospheric pressure with a standard DIN or ASTM apparatus. They include a number
of different groups of organic compounds such as
3

hydrocarbons, ketones, esters and amines. The

following conclusions can be drawn from this table:
1. The temperature of autoignition drops, as expected, substantially with increasing pressure.
2. The order of the compounds with respect to the
AIT at higher pressures is different from the one at
atmospheric pressure.

Tab. 2: Autoignition temperatures of several pure compounds at

elevated pressures compared to the standard values
Autoignition temperature in C at
Compound

Semenoff plots
According to Semenoff's theory of thermal explosion /1/, the relation between the pressure pZ of a
fuel/air-mixture and its autoignition temperature TZ is
described by the relation:
2
( +1)

E ASem
)
RTZ
where EASem is an apparent activation energy of the
reaction and n is the overall reaction order (usually
assumed to be 2).
pZ = k Tz n

exp(

pZ/T 2Z in bar/K

1E-4

1E-5

n-Heptane
Benzene
Ethanol
Propionic acid
Methyl propionate
Butyl acetate

1E-6
200

250

300
350
Temperature in C

400

450

500 550

Fig. 8: Semenoff plots for 1 - 10 bar for several pure substances

Fig. 8 gives the so-called Semenoff plots for a number of selected pure compounds. In general, the experimental values fall well on straight lines, from which
apparent activation energies in the range 100 kJ/mol to
350 kJ/mol can be calculated.
An exception is, however, some of the 1 bar values
measured with the standard apparatus which are much
higher than expected from extrapolation of the values
measured in the high pressure autoclave (open symbols
in Fig. 8). Apart from a possible switch in reaction
mechanism (low temperature/high temperature) differences in experimental conditions may cause this deviation:
1. The larger vessel volume (0.5 l compared to 0.2 l in
the standard apparatus) is known to decrease the
ignition temperatures.
2. Both the temperature and the pressure criterion for
ignition may be stricter than the visual criterion used
with the standard apparatus.
3. The closed vessel may make ignitions easier. In
some times a small pressure increase was observed
to precede ignition, which is not possible in an open
device.

1 bar

2 bar

5 bar

10 bar

n-Hexane

230

235

210

197

n-Heptane

220

201

197

190

n-Octane

215

Cyclohexane

246

245

225

215

Benzene

565

526

470

451

Toluene

535

457

261

Dioxan

375

212

197

189

Methanol

440

300

260

Ethanol

400

283

250

Propanol-1

385

300

265

240

Butanl-1

325

292

255

240

Pentanol-1

320

250

240

Hexanol-1

280

280

262

232

Acetone

525

350*

275

260

Butanone-2

475

290

235

210

Pentanone-2

445

260

210

Hexanone-2

420

196

187

Cyclohexanone

430

279

230

215

Propionic acid

470

358

299

266

i-Butyric aldehyde

165

143

122

Propionic aldehyde

190

108

98

93

Methyl propionate

465

400

284

253

Ethyl formiate

440

312*

280

225

Propyl propionate

445

315

251

Butyl propionate

425

320

240

i-Propyl acetate

425

296

241

245

Methyl acetate

505

470

415

338

Ethyl acetate

470

380

260

230

Propyl acetate

455

300

260

240

n-Butyl acetate

393

252

240

230

t-Butyl acetate

450

395

370

310

n-Pentyl acetate

350

226

i-Pentyl acetate

280

261*

240

224

Methyl butyrate

445

400

256

n-Butyl amine

310

280

258

= value at 2.5 bar

210

216

The Semenoff plots can be used to estimate the

autoigniton temperatures at even higher pressures.
Semenoff's equation seems to indicate that it is possible
to lower the AIT to arbitrarily low values by increasing
the pressure. But the higher the total pressure of the
fuel/air mixture, the higher is also the partial pressure of
the fuel required to reach the most critical fuel
concentration. At some temperature the required partial
pressure will become equal to the fuel vapour pressure.
At even higher pressures and lower temperatures it is
impossible to reach the critical fuel concentration, and
the depression of the AIT is expected to stop. A similar
mechanism has been employed to explain the reincrease of the standard AIT with chain length for very
long chain molecules /5/.
Conclusions
For a number of pure compounds, the pressure
dependence of autoignition temperatures and the upper
explosion limits at a temperature of 200C have been
determined. The results show that mostly the explosion
range widens dramatically with increasing pressure at
the UEL. Therefore, although the most sensitive fuel
concentrations in AIT determination lie at very rich
mixtures, they are always well within the explosion
range. No special mechanism for the ignition at AIT
needs to be applied.
The pressure dependence of the autoignition temperature has been shown to meet a Semenoff relation,
which can be used to predict the AITs at pressures.
Similarily, the ignition delay times follow an Arrheniuslike relation leading sometimes to very long delays near
AIT at high pressures.
When discussing the AIT and the UEL of a liquid,
its vapour pressure curve also must be considered. Due
to the rise of the UEL with pressure, it will often be
impossible to find an UEL at elevated pressure at a
specified temperature as the limited vapour pressure
prevents reaching the necessary fuel concentration. That
means it is not possible to exceed the UEL. Similarily
with respect to AIT a point will be reached when
increasing the pressure where the fuel vapour pressure
limits the mixture concentration. Rich mixtures that
would be necessary to find the AITs predicted by
Semenoffs relation cannot be obtained, and a further
decrease of AITs with increasing pressure will not be
observed. Therefore the AIT will stop to decrease
further. This offers the possibility of calculating a
lowest possible AIT for the substance under
consideration provided its vapour pressure curve is
known.
The values found at 1 bar with the open standard
apparatus often do not fit well to the Semenoff or
Arrhenius relations found in the autoclave experiments.
This shows that the differences in experimental set-ups
have substantial influence on the results. It is therefore
desirable to get AIT values for 1 bar under isochore
conditions, for example with the present autoclave.

Acknowledgements
We thank M. Gdde, W. Mller, G Riesner and J.
Scheffler for carrying out the experiments and operating
the autoclaves.
For financial support we thank the Hauptverband der
gewerblichen Berufsgenossenschaften
References
1. Semenoff, N.: Zur Theorie des Verbrennungsprozesses, Z.Phys. 48(1928), 571
2. IEC 60079-4: Electrical apparatus for explosive gas
atmospheres. Part 4: Method of test for ignition temperature
3. Gdde, M.: Zndtemperaturen organischer Verbindungen in Abhngigkeit von chemischer Struktur
und Druck, PTB-Bericht ThEx-8, Wirtschaftsverlag
NW Verlag fr Neue Wissenschaft, Bremerhaven
1998
4. Semenoff, N.N.: Some Problems in Chemical Kinetics and Reactivity, Princeton Univ. Press, v.2, 1959,
331 pp.
5. Gdde, M, Brandes, E. and Cammenga, H. K.:
Zndtemperaturen homologer Reihen Teil 2, PTBMitteilungen 108(1998), 437