Autoignition temperatures of flammable liquids in closed vessels
PAPP Christian*), BRANDES Elisabeth*), HIRSCH Werner*), MARX, Marcus **)
*) Physikalisch-Technische Bundesanstalt, Braunschweig
**) Otto-von-Guericke-Universität, Magdeburg
Oral presentation by Ch. Papp at IXth ISPHMIE, Cracow 2012
ABSTRACT
Investigations on the autoignition temperature carried out in closed vessels (isochoric conditions)
result in some cases in remarkably lower ignition temperatures at of 1 bar compared to the autoignition temperatures (AIT) determined by standardized determination methods (e.g. EN 14522 or
IEC 60079-20-1). These standardized methods use an open vessel (isobaric conditions) and,
therefore, ambient pressure conditions (around 1 bar) are implicit. The standardized methods use
a visible flame as ignition criterion. For our determinations in a closed vessel the ignition criterion
was the signal of a photodiode, a steep temperature increase and/or a steep pressure increase.
The investigations result in a number of substances having a ‘standard AIT’ between 300 °C and
450 °C – especially esters and ketones – showing a significant decrease of the ignition
temperature in the closed vessel compared to the open vessel of the standardized methods. It is
shown that such ignition temperatures in a closed vessel correlate with cool flame temperatures as
determined in a slightly modified standard apparatus. A screening procedure based on the
standardized determination methods is described as well to identify such substances.
1. INTRODUCTION
The standardized methods to determine the autoignition temperature (AIT) use flasks open to the
atmosphere (DIN 57194 (2003) [1], EN 14522 (2005) [2], IEC 60079-20-1 (2012) [3] a 200ml
Erlenmeyer flask, ASTM E659 (2005) [4] a 500ml round bottom flask), but industrial applications
are often realized under conditions which are more comparable to a closed vessel. Furthermore, a
lot of industrial processes are carried out under non-atmospheric conditions. Although the
standardized methods aim first of all on the classification of flammable vapours and gases with
respect to explosion protected equipment, AITs determined by standardized methods are also
often used as a safety characteristic to prevent accidents due to autoignition. To determine the
influence of industrial process conditions (in particular elevated pressures), autoignition
temperatures were measured in a closed vessel [5, 6]. In addition to the pressure influence on
autoignition temperatures, these investigations show for some substances a significant difference
between the autoignition temperature measured in the closed vessel at 1 bar and the AIT
determined by a standardized method in an open Erlenmeyer flask. For such substances, the
ignition temperatures in the closed vessel can be up to 200 K lower. The ignition temperatures
measured in a closed vessel at 1 bar whether they differ from the standard AITs or not fit well the
respective Semenov correlation [7] which describes the pressure dependence of the autoignition
temperature [5].
This fact is shown in figure 1 for some substances.
These results raised two questions:
1. What are the reasons for such differences?
2. Is it possible to have a trustable screening procedure to identify substances showing such a
behavior?
This paper deals mainly with the second question. The affected substances or substance classes
as known so far are reported below. The aim of the current investigation was to develop a
screening procedure based on the standardized methods which would need as few modifications
as possible
-10
1,4-Dioxane
i-Butanal
p-Xylene
Cyclohexanone
-11
2
2
ln(p/Tz ) / (bar/K )
-9
-12
AIT
AIT
-13
AIT
Figure 1: Semenov correlations of the
autoignition temperature.
AIT
-14
0,0008 0,0012 0,0016 0,0020 0,0024 0,0028 0,0032
-1
1/Tz / K
2. EXPERIMENTAL
2.1 Determinations in a closed vessel
Ignition temperatures in the closed vessel were determined in a 320 ml autoclave made of
stainless steel. The vessel was equipped with a pressure transducer and two 0,5 mm
thermocouples. One thermocouple was positioned in the middle of the vessel, the other near the lid
of the vessel. To facilitate the comparison with the standardized methods which use an optical
ignition criterion, a photodiode was added via a quartz glass rod on top of the vessel. While for the
temperature and pressure recordings a sampling rate of 10 values per second was used, the
photodiode output had been recorded with 100 samples per second to ensure that even short light
impulses are recorded. In our former investigations a steep temperature rise of at least 50 K or a
steep pressure rise of at least 10% was fixed as a criterion for an ignition in a closed vessel,
because this corresponds for most substances with a light emission detected by the photodiode.
The procedure to determine the autoignition temperatures is as follows: The vessel which is heated
to the desired test temperature is evacuated to about 10 mbar. The amount of air necessary for the
desired concentration of the fuel air mixture is then fed to the vessel via an electromagnetic valve.
After reaching the desired air pressure in the vessel, the valve closes and the liquid fuel is
introduced in portions via a second valve. The concentrations are calculated based on the ideal
gas law. For a maximum of 30 minutes it is observed whether an ignition occurs or not. Vessel
temperature and concentration of the fuel air mixtures were varied till just no ignition was detected.
The lowest temperature at which at least at one concentration an ignition occurred is taken as
ignition temperature in the closed vessel.
2.2 Modifications of the standardized determination methods
The major differences between the closed vessel and the open flask are shown in Table 1.
Studies which focus on the influence of different wall materials [8, 9] result in differences smaller
than the differences between the AIT and the ignition temperature in a closed vessel. This allows
the conclusion that the different wall materials are mostly not responsible for the ignition
temperature decrease and are therefore not checked with respect to a possible screening
procedure.
However, the differences in gas exchange as well as cool flame phenomena were considered as a
possible basis for a screening procedure.
Table 1: Major differences between the standardized AIT determination method and the
determination of the ignition temperature in a closed vessel
Standardized determination
Determination in a closed vessel
Isobaric conditions
Isochoric conditions
Gas exchange with the surrounding air No gas exchange
via the neck
Vessel made of glass
Vessel made of stainless steel
2.2.1 Gas exchange
Driven by natural convection, fuel is transported outside the open standard Erlenmeyer flask during
the ignition experiment, while cold air from the surroundings flows back into the flask. Also hot
spots which are the starting point for ignition may be influenced by this convection. As a result of
such convection processes, the fuel is reduced over time. To check the influence of such a gas
exchange on the ignition, different neck widths of the flask were tested. One flask was equipped
with a metal ring placed inside the neck opening at the top of the flask to reduce the width of the
neck to 1 cm. Also a flask with a neck width of 4.5 cm was tested. The standardized flask has a
neck width of 3 cm (Figure 2).
flask according
to standards
narrow-neck
flask
wide-neck
flask
Figure 2: Scheme of different types of flask opening
The ignition temperature determinations were carried out in accordance with EN 14522.
In addition, the gas exchange was simulated with ANSYS CFX Version 12 for water vapour as an
example (Figure 3).
Figure 3: Average mass fraction over time,
calculated in a CFD simulation
2.2.2 Cool flame phenomena
To cover the cool flame phenomena, a standardized apparatus was equipped with four additional
0,5 mm thermocouples. Two of them were located at the bottom, one was centered at half height
of the flask, and the fourth thermocouple was positioned at the lower end of the neck. For flame
detection, a single lens reflex (SLR) camera equipped with an additional infrared filter was placed
over the opening of the flask. The exposure of the camera is started by the control software when a
temperature higher than the wall temperature is detected by one of the additional thermocouples
inside the flask. The temperature measurement was performed with up to 30 samples per second.
The procedure for the ignition temperature determinations was in accordance with EN 14522.
2.2.3 Investigated substances
The investigated substances were mainly alkanes, alcohols, amines, esters and ketones in order to
take different substance classes into account.
3. RESULTS
3.1 Ignition temperatures in a closed vessel
Ignition temperature in a closed vessel at 1 bar / °C
Figure 4 compares the ignition temperatures in the closed vessel (ignition criterion: T > 50 K or
p > 10% or light emission) to the respective AIT. Some substances show a remarkable difference
(> 50 K) between the ignition temperature in the closed vessel and the standardized AIT. Those
are marked as in Figure 4. The AIT values are taken from the database CHEMSAFE [10]
460
0 K difference
50 K difference
100 K difference
200 K difference
420
380
340
300
260
220
Figure 4: Differences between the AIT and the
ignition temperature in the closed vessel
180
140
150
200
250
300
350
400
450
AIT / °C
Mainly substances which have an AIT between 300 °C and 450 °C are affected by such an ignition
temperature drop. Most of these substances are ketones and esters, whereas n-alkanes and
alcohols having an AIT in this temperature range showed no or just a small decrease in ignition
temperature (< 50K).
3.2 Tests with flasks of different neck widths
Some results of the tests with different neck widths are given in Table 2. In flasks with a wider
neck, the autoignition temperature is much higher than in the flask used in the standardized
apparatus. One of the main reasons is the higher temperature gradient inside the flask in this case.
Also the degree of the gas exchange with the surroundings is much higher compared to the
standardized flask (see Fig. 3). Reducing the diameter of the neck causes a decrease of the AIT
but not as significantly as the wider neck causes an increase.
As can be seen from Table 2, the effect of reducing the width of the neck on the ignition temperature is not distinct enough to develop a screening procedure for the substances which ignite at
remarkably lower temperatures in closed vessels.
Table 2: Influence of the neck width of the Erlenmeyer flask on the auto ignition temperature
Fuel
Measured ignition temperatures*
Flask
n-Propanol
Heptanone-3
Ethyl hexanoate
Closed vessel
385 °C
205 °C
253 °C
Standard flask
385 °C
408 °C
395 °C
Wide-neck flask
441 °C
444 °C
437 °C
Narrow-neck
380 °C
385 °C
365 °C
flask
*Temperatures are not rounded as it would be required by standards.
3.3 Tests with a flask equipped with additional thermocouples and a camera
The criterion for the ignition in the open flask is a visible flame. In case of ignition with a visible
flame, a steep and high temperature increase (> 200 K) is detected by all additional thermocouples. Lowering the test temperature below the AIT showed for some substances still a temperature increase of more than 50 K, which is mostly only detected by the thermocouples near the
bottom of the flask. Only a bale blue light was visible by eyes. However, along with such a
temperature increase a pale blue light could be recorded by the digital camera (time exposure
mode up to 60 seconds). The slow temperature rise of not more than 7 K/s, as recorded by the
thermocouples, indicates that this effect may be linked to a cool flame reaction inside the flask. In
general, temperature increases of less than 50 K have not been accompanied by a pale blue light.
It is important to note that in some cases such a temperature increase detected by the
thermocouples and accompanied by a pale blue light is found only where the two ignition regimes
(cool flames, ‘hot’ ignition) are separated by a temperature range with no obvious reactions. This
fact, known as negative temperature coefficient (NTC) [11], is shown for butyl butyrate in figure 5,
as an example.
in K in K
Temperaturanstieg
increase
Temperature
480
450
80
60
40
Figure 5: Negative temperature coefficient
of butyl butyrate
20
0
250
300
350
400
Kolbentemperatur
Flask
temperature inin°C°C
Remarkable differences between AIT and the cool flame temperatures are so far found mainly for
esters and ketones with an AIT in the range between 300 °C and 450 °C. 1,4-dioxane, i-pentane ihexane and di-isopropylether show a similar behaviour.
Comparing the ignition temperatures in the closed vessel with these cool flame temperatures in the
Erlenmeyer flask shows that both temperatures correlate well. There are only slight differences
(maximum to 20 K) between these temperatures. As could be seen from figures 6 and 7 as well,
esters and ketones with very short chains and such with long chains (e. g. ketones with a C-chain
length > 8) show no significant difference between the AITs and the ignition temperatures in the
closed vessel.
Other substances tested like n-alkanes, alcohols and amines show no or smaller differences in
ignition temperatures than those shown in figures 6 and 7 (see Table 3). In addition to that, in such
cases a cool flame reaction cannot be found inside the Erlenmeyer flask with the method
mentioned above.
500
AIT in Erlenmeyer flask
AIT in closed vessel
temperature in °C
450
cool flame temperature
400
350
300
250
200
te
te
te
te
te
te
te
te
te
te
te
te
te
yra eta era na yra yra ma era yra eta oa eta eta
but yl ac yl val propio yl but yl but yl for yl val yl but tyl ac hexan yl ac xyl ac
l
y
p
t
h
t h Bu
th Pro Eth pyl Eth rop Et
Bu thyl
Am He
P
Me
Me
E
Pro
Figure 6: AIT, ignition temperature in a closed vessel and cool flame temperature of the investigated esters
500
temperature in °C
450
AIT in Erlenmeyer flask
AIT in closed vessel
cool flame temperature
400
350
300
250
200
Bu
e-2
e-2
e-3
e-2
e-3
e-2
e-4
e-3
e-2
one
tan tanon tanon tanon xanon tanon tanon nanon tanon canon
c
n
n
p
p
p
O
He He
De
No
Pe
Pe
He
He
Figure 7: AIT, ignition temperature in a closed vessel and cool flame temperature of the investigated ketones
This means that the determination of the cool flame temperature with the aid of an additional
thermocouple fitted to the standardized vessel is suitable for identifying such substances that show
a significant difference between the AIT and the closed vessel ignition temperature (see Table 3 for
examples). A camera as an additional sensor for cool flames is not necessary, because in almost
any case in the cool flame regime the pale blue light was accompanied by a temperature increase
of more than 50 K but less than 150 K which is easily detectable via thermocouples.
According to literature [12] a cool flame can be the first part of a two-stage-ignition.
Table 3: Cool flame temperature and AIT of several substances
Method /
Substance
Heptanone-3
Hexanone-2
Nonanone-2
Heptanone-2
Octanone-3
Ethyl hexanoate
Heptanone-4
1,4-Dioxane
Pentanone-3
n-Pentane
Ethyl valerate
Amyl acetate
n-Butyl butyrate
Di-n-butylamine
i-Octane
Methyl valerate
Propyl acetate
i-Pentane
Propyl butyrate
Butyl amine
Butyl acetate
Propyl propionate
Ethyl butyrate
1-Pentanol
3-Hexanol
Pentanone-2
Ethyl propionate
3-Pentanol
1,2-Propandiol
2-Butanol
Butanone
1,2-Hexandiol
Ethyl formate
Ethanol
Methyl butyrate
Ignition temperatures measured in °C
Standard
apparatus
410
420
240
305
234
395
420
375
455
260
450
350
395
260
410
420
455
420
435
310
395
445
445
320
300
445
455
360
387
390
475
362
420
400
455
Cool flame
detection
207
209
220
223
230
232
235
243
253
254
255
265
270
273
276
278
280
282
285
288
290
291
295
296
300
308
314
325
335
335
335
350
382
400
445
Closed
vessel
207
214
211
215
207
245
225
230
252
360
270
290
265
312
299
301
323
380
Under isochoric conditions a cool flame might be able to turn into a ‘hot’ ignition for such
substances. Such a ‘turn over’ is indicated by significantly steeper pressure and temperature rises
compared to the temperature rises which occurred with cool flames in the Erlenmeyer flask at
comparable temperatures. An example is shown in Figure 8 for 2-heptanone at a temperature of
256 °C, which is clearly below the AIT of 305 °C. As can be seen, the temperature and pressure
increase in the closed vessel is very sharp. The maximum temperature rise > 250 K/s can be
observed which is accompanied by a bright light emission detected by the photodiode (“Voltage”).
Temperature
350
2000
Pressure
Voltage
1500
250
1000
2
4
6
8
10
12
14
16
18
20
2
Voltage
300
Pressure in mbar
Temperature at the top of the vesssel in °C
4
0
Figure 8: Ignition of 3-heptanone at 1 bar
in a closed vessel below its AIT
of 305 °C
22
time in seconds
4. CONCLUSIONS
It is shown that the ignition temperature in the closed vessel correlates to the cool flame
temperature which can also be determined in a standard Erlenmeyer flask open to the atmosphere.
The isochoric conditions which exist in a closed vessel seem to cause a “hot” ignition even if the
cool flame temperature is much lower than the AIT. This is of high interest for industrial processes.
The current European standard EN 14522 does not explicitly specify the cool flame temperature as
a separate value to be determined, whereas American standard covers the determination of such
values. An implementation of the determination of the cool flame temperature into the European
standards seems to be useful. For this purpose the test apparatus should be equipped with a
thermocouple placed near the bottom of the flask. This would be the only modification of the
current standard apparatus necessary to identify substances with a remarkably lower ignition
temperature in closed vessels.
LITERATURE
1. DIN 51794 (2003): Prüfung von Mineralölkohlenwasserstoffen, Bestimmung der Zündtemperatur,
Beuth, Frankfurt am Main
2. EN 14522 (2005): Determination of the auto ignition temperature of gases and vapours.
3. IEC 60079-20-1 (2012): Material characteristics for gas and vapour classification - Test methods and
data
4. ASTM E659 (2005): Standard Test Method for Autoignition Temperature of Liquid Chemicals.
5. E. Brandes, W. Hirsch, W. Möller (2008): Autoignition temperatures of binary Mixtures at elevated
pressures, International Symposium on Hazard, Prevention and Mitigation of Industrial Explosions, pp.
94-101, St. Petersburg
6. E. Brandes, W. Hirsch (2007): Zündtemperaturen binärer Gemische bei erhöhten Ausgangsdrücken,
11. BAM/PTB-Kolloquium zu Fragen der chemischen und physikalischen Sicherheitstechnik, pp. 7-16
7. Semenov, N., (1928): Zur Theorie der Verbennungsprozesses, Zeitschrift für Physik, Volume 48, pp.
571-582.
8. Kaescher-Krischer, B. & Wagner, H., (1958): Die Zündungs von Brennstoff-Luft-Gemischen an heißen
Oberflächen. Brennstoff-Chemie, 39(3/4), pp. 33-64.
9. Frank, C. & Blackham, A., (1952): Spontaneous Ignition of Organic Compunds. Ind.Eng.Chem., 44(4),
pp. 862-867.
10. Physikalisch-Technische-Bundesanstalt, Bundesanstalt für Materialforschung- und prüfung, Gesellschaft
für Chemische Technik und Biotechnologie e.V.(2011): CHEMSAFE-Database, Frankfurt am Main
11. Fish, A., Read, I., Affleck, W. & Haskell, W (1969): The controlling role of cool flames in two-stage
ignition. Combustion and Flame, 13(1), pp. 39-49,
12. Barnard, J., Watts, A. (1969):. Cool-flame oxidation of ketones. Symposium (International) on
Combustion, 12(1), pp. 365-373.
Note:
The authors would like to thank BG RCI (Berufsgenossenschaft Rohstoffe und chemische
Industrie) for financial support.