UNESP/ UNIVERSITY OF TWENTE

Internship report

Research in flammability limits and deflagration

to detonation transition of ethanol

Hedwich Bekius
S0147117
3/12/2012
Preface
This report is written during my internship at UNESP in Guaratinguetá, Brazil. The internship was
arranged by IAESTE, the international association for the exchange of students for technical
experience. The work I did in Guaratinguetá was under supervision of Professor João A. Carvalho Jr.
He is doing research in different kind of areas that deals with combustion.

One of his projects treats the topic of flammability limits in which I worked during my internship.
Already research and experiments were done at UNESP and INPE, but new equipment to determine
flammability limits was built at the university in Itajubá during my stay. This new equipment is able to
handle higher temperatures then the one used before. Unfortunately I couldn’t do the experiments
during my stay, but I had the chance to take a look on how the equipment was mounted.

Professor João A. Carvalho Jr. also wants to start a research program in the area of transition from
deflagration to detonation. He asked me to do pre-research on the subject to help writing a proposal
for this new program. So this report actually contains two topics, the one of flammability limits and
the one of transition of deflagration to detonation.

I want to thank Professor João A. Carvalho Jr. that I could carry out my internship at UNESP for three
months. I learned a lot about new subjects concerning combustion and overall it was a great
experience in Guaratinguetá.
Abstract
The flammability limits can be divided in an upper and lower flammability limit. The lower
flammability limit is defined as the minimum concentration of fuel for which flame propagation is
possible. For the upper flammability limit this is the maximum concentration for which flame
propagation is possible. The knowledge of the flammability range of a certain fuel is useful, because
for safety reasons it is important to know whether a fuel/air mixture can explode or not under
different circumstances like change in temperature. This report has the emphasis on ethanol as a
fuel. The flammability limits for ethanol in air are 3.3%vol (lower) and 19.0%vol (upper) at
atmospheric pressure and a temperature of 298 K, according to experimental data. There are
different methods to determine the flammability limits for various temperatures. In the experiments
done at INPE, a 20 L vessel was used with ignition energy of 90 J, the spark gap was 6.4 mm. The
temperature in the vessel could vary between 20 ⁰C and 200 ⁰C. The lower flammability limits found
in literature match with the ones determined by the experiments. The upper flammability limits
determined in the experiments differ from the ones found in literature, the experimental values are
4.5% lower than the literature values.

The phenomenon of deflagration to detonation transition (DDT) in a tube is explained in short. First a
fuel/air mixture is ignited at one end of a tube. The flame front is propagating and pushes the flow in
the fuel mixture with a weak compression wave. The compression wave gets stronger due flame
acceleration, until it develops into a shock of considerable amplitude. This shock preheats the fuel/air
mixture and the temperature behind the shock increases. The reaction time decreases extremely in
the compressed local exothermic centers, called hot spots. This decrease in reaction may result in
DDT ahead of the flame front, unless the compressed gas is burnt by the flame before active
explosion is initiated.

At the moment the basics of DDT are understood, previous studies in DDT treated the detonation cell
size and the DDT run-up distance. For understanding the details of this phenomenon still a lot of
research needs to be done. This is done with help of numerical models and experiments. A numerical
model that uses not too complicated physical submodels to describe the fluid dynamics and
combustion processes has to be used to get the right balance between the speed and accuracy in the
DDT simulations. Two general types of experimental configurations to study DDT can be defined; one
that allows the turbulence to evolve itself, and another with an initial shock to immediately create
the turbulence. In the last type the flow will develop more quickly, because the turbulent flow is
created immediately and the simulation will converge faster. Different schemes can be used for the
simulation, of course an appropriate scheme that can capture shocks and large gradients in density
and pressure is needed. Resolution tests should be carried out to check the convergence.
Contents

1. Introduction ................................................................................................................................1
2. Flammability limits ......................................................................................................................2
2.1 Definition flammability limits ....................................................................................................2
2.2 Properties ethanol.....................................................................................................................4
2.3 Experiments flammability limits for ethanol ..............................................................................5
2.4 Results experiments ..................................................................................................................7
3. Deflagration to detonation transition ..........................................................................................8
3.1 Definition DDT ...........................................................................................................................8
3.2 Previous studies on DDT .......................................................................................................... 12
3.2.1 Detonation cell size .......................................................................................................... 12
3.2.2 Deflagration to detonation transition run-up distance ...................................................... 12
3.3 Experiments ............................................................................................................................ 13
3.4 Numerical modeling ................................................................................................................ 14
3.4.1 DDT modeling approaches ................................................................................................ 14
3.4.2 Initial conditions ............................................................................................................... 14
3.4.3 Numerical scheme and resolution..................................................................................... 15
4. Discussion ................................................................................................................................. 18
4.1 Flammability limits .................................................................................................................. 18
4.2 Deflagration to detonation transition ...................................................................................... 18
Bibliography...................................................................................................................................... 19
 1. Introduction
In this report the topics of the flammability limits of ethanol and deflagration to detonation
transition will be discussed.

First the topic of flammability limits is treated, it is an important tool used for safety reasons. The
flammability limits define whether a certain fuel/air mixture will ignite with help of an ignition
source. In the flammability research at UNESP it’s especially about the safety in aeronautical
applications, so for example in the fuel tank of an airplane. An upper and lower flammability limit is
determined for ethanol for various temperatures.

Next the topic of deflagration to detonation transition (DDT) is treated. The details of this
phenomenon are still not fully understood, so UNESP wants to do more research on this subject. For
safety reasons it’s important to understand DDT, it is of course very undesirable that detonation
occurs in for example a fuel tube. The prediction of flame acceleration and DDT in a tube is very
important for estimating potential explosion hazards. So the goal is to gain more insight in this
phenomenon and predict the critical conditions in deflagration to detonation.

1
 2. Flammability limits
In this chapter the flammability limits are discussed. First an introduction about flammability limits in
general is given. Next the differences in flammability limits of ethanol compared to other fuels and
the most important properties of ethanol are treated. Then the equipment used for the experiments
and the experiments itself with the results are discussed.

2.1 Definition flammability limits

Flammability limits are the borders of an air-fuel mixture region where flame propagation occurs and
does not occur. For safety reasons it is important to know whether a gaseous or vaporized liquid fuel
mixture with air will explode. A lower and an upper flammability limit is defined, respectively the
minimum concentration of fuel for which flame propagation is possible and the maximum
concentration of fuel for which flame propagation is possible.

Figure 1 Flammability limits ethanol as a function of temperature1

For a flammable liquid also a flashpoint can be defined. This is the lowest temperature at which there
will be enough flammable vapor to ignite when an ignition source is applied. The concentration of
vapor in air necessary to sustain combustion is different for each flammable liquid. For example the
flash points for gasoline and ethanol are respectively < 233 K and 285.8 K. It must not be confused
with the auto ignition temperature; that doesn’t require an ignition source.

The auto ignition temperature is the lowest temperature of a mixture at which it will spontaneously
ignite in a normal atmosphere without an external source of ignition. For example the auto ignition
temperatures for gasoline and ethanol are respectively 519-553 K and 638 K. To avoid pre-ignition
the fuel must have a low flash point and a high auto ignition temperature.

1
C. Coronado, J.A. Carvalho Jr., J.C. Andrade, E.V. Cortez, F.S. Carvalho, J.C. Santos, A.Z. Mendiburu, Flammability limits: a
Review with emphasis on ethanol for aeronautical applications and description of the experimental procedure. (2012)

2
In the project of UNESP the emphasis is on the flammability limits of ethanol for aeronautical
applications. There is a risk in the ullage in aircraft fuel tanks, because this can contain a mixture of
fuel, air and vapors, which can be ignited by an ignition source when the mixture is within the
flammability range. After the flight TWA 800 accident in 1996, there is paid even more attention to
the safety regarding flammability limits. A flammable mixture was formed in the ullage region of the
airplane due to increased fuel vaporization due to heating the tank by the air conditioning system. At
some point the flammable mixture was ignited, probably by a short circuit, which caused the
accident. The plane exploded and crashed into the Atlantic Ocean near New York and all 230
passengers died. This shows how important it is to have knowledge under what conditions flame
propagation will occur.

In this report experiments to measure the flammability limits are shown, but there are also a few
theoretical methods to determine flammability limits. It is shown that for lower flammability limits
the theoretical values fit the experimental data, but for the upper flammability the differences are
bigger between theoretical and experimental values. The theoretical values can be calculated with
the help of the stoichiometric concentration, combustion enthalpy, vapor pressure or the required
amount of moles for flame propagation. These methods can be found in articles of Tareq2, Piqueras3
and Petrobras4.

For fuel mixtures the flammability limits can be determined by Le Chatelier’s law. The lower
flammability limit by

, where C1, C2, … , Ci are the proportions of each fuel gas in the fuel mixture in %vol.

The upper flammability can be determined in the same way by

Kondo5 states that Le Chatelier’s law can predict the lower flammability limit accurately for many
different combustible mixtures, but for most cases concerning the upper flammability limit the
experimental values are lower than the values predicted by Le Chatelier’s law. According to
Glassman6 Le Chatelier’s law is only valid if the components are chemically similar, the law isn’t
accurate for mixtures that contain and unsaturated hydrocarbons.

2
Tareq A. Albahri, Flammability characteristics of pure hydrocarbons, Chemical Engineering Science p. 3629-3641. (2003)
3
C.M. Piqueras, J. García-Serna, M.J. Cocero, Estimation of lower flammability limits in high-pressure systems, The Journal
of Supercritical Fluids p. 33-40. (2011)
4
Safety information sheet of the chemical product – Anhydrous ethyl alcohol fuel. Brazilian Oil Inc. (2009)
5
S. Kondo, K. Takizawa, A. Takahashi, K. Tokuashi, A. Sekiya, A study on flammability limits of fuel mixtures, J. Hazard Mater.
p. 440-448. (2008)
6
I. Glassman, Combustion, 2nd Ed., Academic Press, San Diego, California, (1996)

3
2.2 Properties ethanol
Ethanol ( ) has been made by the fermentation of sugars since a long time. At the
moment all beverage ethanol and more than half of industrial ethanol are still made by this process.
Sugars are the raw material and an enzyme from yeast, zymase, changes the simple sugars into
ethanol and carbon dioxide. The fermentation reaction is very complex, but can be represented by
this simple equation

2 +2

Also starches from potatoes, corn, wheat and other plants can be used in the ethanol production by
fermentation. The starches first have to be broken down into simple sugars. The ethanol produced by
fermentation ranges in concentration from a few percent up to around 14 percent, after that ethanol
destroys the zymase enzyme and fermentation stops. The ethanol is normally concentrated by
distillation of aqueous solutions. The ethanol distillation is limited to a purity of 95-96%, i.e. hydrous
ethanol. When pure ethanol is desired, anhydrous ethanol, a dehydration process can be applied.

Ethanol can be used as an automotive fuel by itself or can be mixed with gasoline. Since Brazil is the
largest producer of sugarcane, where ethanol can be produced off, ethanol is a very popular
alternative fuel.

Ethanol is a clear, colorless liquid with a characteristic smell. Its melting point is at 158.9 K and its
boiling point at 351.5 K. The molar mass of ethanol is 46.07 g/mol and at 293 K the density is 0.789
g/mL. The stoichiometric concentration is 6.58 %vol and the combustion enthalpy 306 kcal/mol. The
flammability limits for ethanol in air are 3.3%vol (lower) and 19.0%vol (upper) at atmospheric
pressure and a temperature of 298 K, according to experimental data.

4
2.3 Experiments flammability limits for ethanol
For measuring the flammability limits for gases no standard method is developed yet, despite the
attempts to do this. At the moment different methods are used to measure the flammability:

 Flashpoint test (open pilot flame)

 Spark ignition test (capacitive spark)
 Temperature limit method (electrically heated fuse-wire)
 Concentration limit method (electric arc or fuse-wire)

The new experiments for flammability limits take place at the university of Itajubá where the
equipment is located, a spark ignition test is used for the experiments there. The flammability limits
also depend on the ignition energy and its distribution in time and space, it is therefore important to
characterize the amount of energy in the spark that is actually deposited in the gas. The higher the
activation energy of the reaction, the higher the ignition energy has to be. Too high ignition energy
will sometimes only show a dissipation of the ignition energy and not the flame propagation. When
the temperature increases, the lower flammability limit will decrease and less combustion energy will
be required to ignite the flame. The upper flammability limit is shifted up when temperature
increases. For pressure increase above atmospheric pressure the interval width of the flammability
limits doesn’t always increase, for some mixtures it even decreases. So a mixture that has flame
propagation at atmospheric pressure does not have to have it at a higher pressure.

The minimum ignition energy, the minimum amount of energy to ignite the combustible vapor,
decreases as electrode spacing increases, it reaches its lowest value at certain spacing, and then
starts to rise again. In the experiments carried out in Itajubá, the spark gap is 6.4 mm and the ignition
energy 90 J.

Figure 2 Schematic overview equipment for flammability experiments

5
 The lower and upper flammability limits were determined for hydrated ethanol (96% ethanol, 4%
water) as functions of temperature for atmospheric pressure. For the experiments a 20 L spherical
vessel is used, see Figure 3. The experimental values approach the values that would be obtained in
an open space when the vessel used is large enough, according to Takahashi7. When a small volume
with high ignition energy is used for test, it will result in a wider interval of flammability and
therefore higher safety costs. For tests with small ignition energies or small vessel volumes it is
possible that there is no flammability observed for certain combustibles, so it is important to use
sufficient ignition energy and vessel volume.

The vessel is equipped with a high precision pressure

transducer and two thermocouples, one positioned
near the wall and the other positioned in the inner core
of the vessel.

The fuel is injected in the vessel in vapor phase at all

times guaranteed by an evaporator, even at low
temperatures. The temperature of the vessel can vary
between 20 ⁰C and 300 ⁰C. The liquid fuel is injected
when the pressure in the reaction vessel is 2 kPa, this is Figure 3 Photo of the transparant vessel

below the saturation vapor pressure for ethanol, so the liquid becomes vapor when injected into the
reaction vessel. This liquid injection is used for
the lower flammability limits, for the upper
flammability limits the evaporator is used. After
the fuel is injected, the synthetic air flows in the
reactor and the concentrations of all the
components are determined by the partial
pressures. An uniform mixture of vapor with air is
ignited in the closed vessel and propagation of
the flame after ignition will be observed within
the flammability region. The concentration of
ethanol is varied between each test until the
composition of the mixture that will achieve
flame propagation is determined.

Two criteria are used, a visual criterion; the flame

that is produced by a spark in the air-ethanol
mixture has to be seen in the transparent vessel.
In this case for the visual criterion is chosen for
the American standard; a visual, flame
propagation up to 13 mm of the wall, horizontal
Figure 4 3D model of equipment used to determine flammability limits
or vertical, has to occur. Also a pressure criterion
is used; at the moment of ignition the pressure
and temperature are measured.

7
A. Takahashi, Y. Urano, K. Tokuhashi, S. Kondo, Effect of vessel size and shape on experimental flammability limits of
gases, J. Hazard Mater. p. 27-37. (2003)

6
2.4 Results experiments
Although the experiments in Itajubá couldn’t be performed, an explanation of how the flammability
limits are calculated from experimental values is given. Also photos that were recorded during the
previous experiments at INPE are shown, one without flame propagation, Figure 5, one showing
deflagration, Figure 6, and one showing detonation, Figure 7. The program LabView was used to
record the experiments. The experiments were repeated three times for the same temperature to
guarantee accuracy of the results.

The lower flammability limit can be calculated from experiments as

follows. When a sample produces the first flame with a certain
volume of ethanol, the last volume that did not produce a flame is
added up and the average of the volumes is taken to calculate the
LFL. The same for the upper flammability limit, but then with the
volume of the last flame that occurred and the first volume where no
flame propagation occurs anymore. The intensity will increase when a
higher amount of volume ethanol above the LFL is used. 1
Figure 5 No flame propagation

Figure 6 Deflagration1 Figure 7 Detonation1

Figure 8 Flammability experiments compared to literature data1

Figure 8 shows that the upper flammability limit differs with the values found in literature, these
values are more conservative than the experimental flammability limits. The lower flammability
limits do match with the values found in literature.

As was told before, there are different ways to determine the flammability limits. Here only one way
is explained to determine them. The initial conditions can vary a lot between different flammability
experiments. For example the size of the vessel is different for the EN, DIN and ASTM standard. Also
the type of igniter and the visual criterion can differ between the various standards.

7
 3. Deflagration to detonation transition
In this chapter the topic of DDT is treated. First the phenomenon DDT is explained, after that former
research in this area is treated. Then the proposal for experiments and numerical modeling for
UNESP for the future is discussed.

Research in DDT is done for a few air-fuel mixtures, but for an air-ethanol mixture little research is
done yet. So the goal of the future research program at UNESP is to gain more insight in DDT in a
tube for an air-ethanol mixture. Ethanol has a higher auto ignition temperature than for example
gasoline, and thus has different properties regarding DDT. Some fuels are more sensitive for DDT
than others.

3.1 Definition DDT

DDT can occur in several situations, in this report only DDT in a tube is treated. Normally,
deflagrations are initiated by for example a spark, while detonations are caused by shock waves via
localized explosion. A deflagration is a propagation of a combustion zone at a velocity less than the
speed of sound in the unreacted medium, subsonic velocity. A detonation is a shock wave which is
followed by a combustion wave, the shock and the combustion region are coupled and move
together. So detonation can be described as a supersonic combustion wave, see figure 9.

Figure 9 Pressure evolution in the detonation tube in the transition zone at different times

Flames generate shocks, especially in confined spaces like tubes, and increase the strength of the
shocks passing through a turbulent flame brush. In turn, shock interactions with flames increase the
production of turbulence. This enhanced turbulence promotes hot-spot formation due to a higher
energy-release rate. For isolated hot-spot ignition, static situation, the conditions for transition to

8
detonation are determined by the gradient in reactivity inside the hot spot. When the hot spot is too
small or the gradient too gentle or too steep, it will not lead to detonation, although a spontaneous
wave may still be formed. For hot spots in a dynamic situation, the temperature gradients inside the
hot spots can be disturbed by shock or rarefaction waves passing through them during the ignition
process. For some of these hot spots, the temperature distribution does not allow a detonation to
develop and the spontaneous wave produces a decoupled shock and a flame. It is possible that it
eventually will produce a detonation, because the decoupled shock can help to create reactivity
gradients in other nearby hot spots.

In short DDT can be described as follows; first a flame front is propagating from a closed end and
pushes the flow in the fuel mixture with a weak compression wave. The compression wave gets
stronger due the flame acceleration, until it develops into a shock of considerable amplitude. This
shock preheats the fuel mixture and the temperature behind the shock increases. The reaction time
decreases extremely in the local exothermic centers, called hot spots. This decrease in reaction time
may result in explosion and DDT ahead of the flame front, see Figure 10, unless the compressed gas
is burnt by the flame before active explosion is initiated. 8

Figure 10 Shadow photographs of DDT in hydrogen-oxygen mixture, first frame at 3.3 ms, second at 3.4 ms. The preheat
zone, between shock and flame, has a width of 17 mm. 9

So in general two outcomes for flame acceleration can be expected. The first is DDT when the
reaction time behind the shock is sufficiently short. Ignition occurs and a second hot spot appears
before reflections can lower temperature. The second hot spot triggers the same events as the first
one, which triggers a third hot spot etc. This results in a sequence of ignitions and shock formations,
each of them appear a little more downstream and with shorter scales triggering the next shock and
hot spot in sequence. The shocks undergo a sequence of mergers and decay into a strong Chapman-
Jouguet wave, this is called the transition to detonation.10

Downstream a hot spot can trigger a weak detonation wave that slows down to CJ-speed and quickly
becomes a strong wave. This wave reaches the contact surface, chemistry stops and a reflection

8
N.N. Smirnov, V.F. Nikitin, Yu. G. Phylippov, Deflagration-to-detonation transition in gases in tubes with cavities, Journal of
Engineering Physics and Thermophysics Vol. 83 no. 6. (2010)
9
M.A. Liberman, M.F. Ivanov, A.D. Kiverin, M.S. Kuznetsov, A.A. Chukalovsky, T.V. Rakhimova, Deflagration-to-detonation
transition in highly reactive combustible mixtures, Acta Astronautica p. 688-701. (2010)
10
L. Bauwens, Zhe Liang, Shock formation ahead of hot spots, Proceedings of the Combustion Institute Vol. 29 p. 2795-
2802. (2003)

9
wave moves upstream. When the reflections move upstream faster than the chemical time behind
the weak shock, then pressure rapidly goes to equilibrium again and no noticeable decrease in
chemical time occurs anymore. So a transition criterion is that the reflection waves have to take
longer than ignition to occur.

DDT is triggered by auto-ignition, which is induced by pre-compression and preheating of the fresh
mixture adjacent to the advancing deflagration.11The auto-ignition temperature of ethanol is higher
than for example gasoline, so a gasoline-air mixture will be more sensitive for DDT than an ethanol-
air mixture.

The second outcome is when the explosion doesn’t occur. In this case the reaction time is longer
than the time available for the compressed gas package to travel between the shock and the flame.
The final state of flame acceleration is the Chapman-Jouguet velocity12.

The Chapman-Jouguet condition states that the detonation propagates at a velocity at which the
reacting gases just reach sonic velocity as the reaction ceases. Classical CJ-theory assumes infinitely
fast reaction, but finite rate chemistry was allowed for later in the ZND (Zeldovitch, von Neuman and
Doring) model.

Figure 11 Schematic view detonation ZND model, evolution of pressure and temperature

For a hot spot to evolve in detonation an interaction between a precursor shock wave and a
combustion wave must occur whereby the shock and the combustion wave feed into each other,
accelerating the shock at least to CJ-speed. This process must not be dampened by reflections from
downstream, otherwise transition is not going to take place. 13

In Figure 11 the lead shock propagates at CJ detonation velocity and the pressure and temperature in
the gas flow immediately behind this shock lead to auto-ignition of the gas in the reaction zone. The
pressure peak in Figure 11, about twice the CJ pressure, is called the von Neumann pressure, it

11
L. Kagin, G. Sivashinsky, On the transition from deflagration to detonation in narrow tubes. (2010)
12
V. Bychkov, D. Valiev, V. Akkerman, C.K. Law, Gas compression moderates flame acceleration in deflagration-to-
detonation transition. (2011)
13
L. Bauwens, Zhe Liang, Shock formation ahead of hot spots, Proceedings of the Combustion Institute Vol. 29 p. 2795-
2802. (2003)

10
develops at the head of the shock. The pressure increases at the flame front location, because when
a flammable mixture starts to burn in a tube, the expansion of combustion products accelerates the
reactants ahead of the flame front.

Zhu14 showed that when a deflagration transits into a detonation in a smooth tube, the deflagration
accelerates to a maximum velocity of about half the theoretical Chapman-Jouguet detonation
velocity, keeps at this state for a certain period, before an abrupt transition to detonation takes
place.

Macroscopic detonation properties like velocity and peak pressure can be predicted from one-
dimensional theory, but the ability to predict whether a detonation will develop under given initial
conditions is more limited. This is due the complex three-dimensional shock structure of the
detonation wave. Steady self-sustaining detonation can only evolve if this structure can be generated
and maintained. So it is important to know if detonation in a fuel mixture can be sustained in a pipe
of a given diameter to satisfy the required safe operating conditions. A prediction of the conditions
for DDT has to be done to avoid the extremely high pressures caused by DDT in the area where the
transition takes place and damages the material of the tube. To predict these critical conditions will
be the goal of the research program.

14
Y.J. Zhu, J. Chao, J.H.S. Lee, An experimental investigation of the propagation mechanism of the critical deflagration
waves that lead to the onset of detonation. (2007)

11
3.2 Previous studies on DDT
In previous studies of DDT, topics like detonation cell width and transition to detonation run-up
distance were treated. The detonation cell width can be used to characterize the detonability of the
mixture and is a fundamental property of the combustible mixture. The DDT run-up distance shows
the required length of the tube to achieve the critical flame speed for transition to detonation.

3.2.1 Detonation cell size

The detonation cell size can be used to describe the critical mixture condition for DDT in gaseous
mixtures, it represents a chemical length-scale for the detonation wave, see Figure 12. The mixture
detonation cell size has to be smaller than a certain characteristic length-scale of the tube for DDT to
occur. To make use of this property, data of the air-fuel mixture cell size is required. In the past
research is done for many air fuel mixtures and most of the detonation cell sizes can be found in this
data. In this way a correlation between the two length scales, cell size and length scale of the tube,
can be established.

Figure 12 Detonation wave structure

3.2.2 Deflagration to detonation transition run-up distance

Another relevant parameter is the required length of the tube to achieve the critical flame speed that
will trigger the transition from deflagration to detonation for a certain combustible mixture, the so-
called DDT run-up distance. The critical flame speed is close to the speed of sound in the combustion
products.

The run-up distance is influenced by the diameter of the tube. The smaller the diameter of the tube,
the shorter the run-up distance for DDT. According to Kuznetsov15 this can be explained because of
the thickness of the turbulent boundary layer, at flame position, linearly depends on the distance
between the flame and the shock. So to achieve the same influence of turbulence in the inner region
of greater tubes as in small tubes, a larger acceleration distance is needed.

The presence of obstacles in the flame path and wall roughness also influences the distance. The
increasingly turbulent fluctuations in the flow ahead of the flame front increases the rate of
combustion, which leads to flame acceleration and pressure build-up.

15
Kuznetsov, M., Matsukov, I., Alekseev, V., Breitung, W. and Dorofeev, S. (2005) Effect of the boundary layer on flame
acceleration and DDT.

12
3.3 Experiments
To gain more insight in what happens when transition from deflagration to detonation occurs,
experiments have to be done. The plan for the future experiments at UNESP for flame propagation in
tubes is as follows.

At the beginning, halfway and at the end of the tube photodiodes and pressure and temperature
sensors are installed. Tests are done with ignition sources on different locations, at the beginning,
halfway and at the end of the tube. This can be seen in Figure 7.

Measuring point Ignition source

Figure 13 Schematic overview of the test of flame propagation in tubes

The tube is closed at both ends and contains air. One end of the tube will be used to inject the fuel
and the desired mixture is identified by the partial pressures of air and fuel. The fuel enters the tube
in gaseous state. When the desired mixture is reached, the other end of the tube will be opened and
the mixture will be ignited.

The goal is to identify the phenomenon of the flame propagation in the tube, whether deflagration
or detonation takes place. The system will start to store the low speed data at the time of ignition.
The high speed data is collected from the pressure sensor separately. The high and low speed data

13
are used to determine the average speed of the flame. The pressure and temperature data is
collected by the readings of fast response thermocouples and pressure transducers.

3.4 Numerical modeling

In combination with the experiments, numerical simulations can be used to predict the conditions for
DDT. Numerical modeling makes it possible to predict the phenomenon of DDT in more detail. The
experiments can observe whether deflagration or detonation take place, but can’t give insight in how
the flame develops. The detonation wave has a complex three-dimensional structure, this means it is
very difficult to model such a wave. The CFD model with chemical reacting flows will have high
computational costs, so an efficient method has to be found to model the transition to detonation.

3.4.1 DDT modeling approaches

Three basis approaches can be described for DDT modeling, according to Smirnov 16. The first is direct
numerical simulation based on the Navier-Stokes equations for chemically reacting mixtures, but
instability is revealing when a turbulent flow is simulated. Due to the difficulty in modeling, it
becomes necessary to limit the chemistry simulation to an one single-stage reaction. According to
Oran17, it’s still not possible to simulate the activation of DDT by a secondary shock wave colliding
flame zone, an intermediate regime of detonation initiation, because directly simulating of DDT
caused by flame-induced turbulence is not possible.

The next approach uses simplified quasi one-dimensional models and semi-empirical models. The
empirical model is however unable to match experiments in detail, as it predicts continuous
acceleration, while in reality a velocity jump in time and space occurs up to an overdriven mode and
slows down to the CJ speed after. Yet even simplified models can be used for engineering application
when supported by a suitable choice of empirical parameters.

The third approach makes use of direct numerical simulation in combination with integral turbulence
models of the k-epsilon or k-omega type. This approach makes it possible to perform direct
simulation of DDT, starting from weak ignition, through further flame acceleration due to induced
turbulence and leading shock acceleration, up to detonation onset in a hot spot.

3.4.2 Initial conditions

The interactions of shocks and flames create conditions under which DDT may occur. Numerical
simulations, using ignition as initial condition, take a very long time, due the slow development of the
compression wave. An initially laminar flame is created by igniting the mixture, accelerates and
becomes turbulent, and eventually forms a turbulent flame brush. When shock-flame interaction is
immediately used to create the turbulent flame it will be relatively fast to solve, because the flow
develops more quickly. In this case a flame will be ignited by sparks on one side of the tube and a
shock wave will be released at the opposite end of the tube. The experiments should also use this
shock-flame interaction to create the turbulent flame more quickly and to be able to compare with
the numerical simulation.

16
N.N. Smirnov, V.F. Nikitin, Yu. G. Phylippov, Deflagration-to-detonation transition in gases in tubes with cavities, Journal
of Engineering Physics and Thermophysics Vol. 83 no. 6. (2010)
17
Elaine S. Oran, Vadim N. Gamezo, Origins of the deflagration-to-detonation transition in gas-phase combustion, 2006

14
When a shock is used as initial condition, whether and where DDT occurs depends on the Mach
number of the incident shock. Scarinci18 observed that in reflected-shock experiments for the lowest
Mach numbers, the reflected shock was amplified by the turbulent flame, but no DDT occurred. For
intermediate Mach numbers DDT occurred between the flame brush and the reflected shock. DDT
occurred inside the flame brush for the highest Mach numbers. In these experiments details of DDT
could not be determined, because the transition was through a very rapid explosion.

Figure 14 Sketch with development of boundary layer ahead of accelerated

flame with time and flow velocity ahead of the flame, curve V(x)

3.4.3 Numerical scheme and resolution

For the numerical scheme a certain time integral has to be chosen, an option is for example the
Runge-Kutta scheme, which will be able to capture the shock and detonation waves effectively.
Resolution tests are necessary to test if a complicated numerical simulation is a good solution of the
equations used. The sizes of computational cells and time steps are varied to the point solutions no
longer change, the solution is then converged. Convergence is a critical test of the validity of the
computation, it is possible for different properties of a flow to become resolved at different
computational cell sizes. For example in DDT it is possible to not find the hot spot and still major
features of the solution are being resolved, or it’s possible to find different hot spots at different
resolutions. That’s why resolution tests are so important; it shows if a property of a flow is really a

18
Scarinci, T., and Thomas, G.O., Some Experiments on Shock-Flame Interaction, University of Wales, Aberystwyth, Wales,
U.K., 1990

15
property of the set of equations. Appropriate resolution test should be carried out to check the
convergence.

To resolve the uneven space scales involved in DDT and handle changes in the flow or reactivity, a
method for locally refining and coarsening the computational mesh is required. A.M. Khokhlov19
developed such a method; it uses an adaptive grid with the date structure based on the fully
threaded tree (FTT). Any cell can be refined or coarsened independently, this gives maximal flexibility
to the adapting algorithm. The mesh is refined around shocks, flame fronts and in regions of large
gradients of density, pressure, or composition. A.M. Khokhlov states in his article20 that adaptive
mesh refinement based on FFT accelerates the calculations significantly.

Figure 15 Example of numerical simulation of DDT showing density plots at different times

19
A.M. Khokhlov, Numerical simulation of deflagration-to-detonation transition: The role of shock-flame interactions in
turbulent flames, J. Comput.Phys. 143 (1998) 519-543
20
A.M. Khokhlov, Numerical simulation of deflagration-to-detonation transition: The role of shock-flame interactions in
turbulent flames, J. Comput.Phys. 143 (1998) 519-543

16
The times for the development of hot spots can be extremely short, in the order of microseconds,
see Figure 15. Adaptive mesh refinement allows to detect gradients of concentration and
temperature and then to increase the resolution early in the development of the gradients. This is
very hard to detect in experiments, where DDT appears as a sudden explosion inside the flame brush.

Different schemes can be used to simulate the transition to detonation, for example Yamada21 uses
the second-order Harten-Yee non-MUSCL modified –flux type TVD scheme to solve a strong shock
wave. A multi-threaded simulation is used to distribute the load of these high computational costs to
improve the performance.22 This is of course just an example of a scheme, but it is interesting to
investigate the advantages and disadvantages of various schemes.

21
E. Yamada, T. Shimada, A.K. Hayashi, Development of Java multi-threaded simulation for chemical reacting flow of
ethanol, Advances in Engineering Software p. 17-23. (2012)
22
E. Yamada, T. Shimada, A.K. Hayashi, Development of Java multi-threaded simulation for chemical reacting flow of
ethanol, Advances in Engineering Software p. 17-23. (2012)

17
 4. Discussion
Flammability limits and the transition to detonation are both used to define safe circumstances in the
applications a certain fuel is used for. Therefore it is important to have a good understanding about
these phenomena.

4.1 Flammability limits

It is desired to get accurate results of the flammability limit experiments to avoid unnecessarily high
prevention costs due to conservative flammability values, i.e. wider flammability range than
necessary. The advice is therefore to use a sufficient large vessel and ignition energy in the
experiments to gain reliable results.

The research in flammability limits should also make use of standardized experiments. Till now a lot
of experimental research is done in flammability limits, but with various standard experimental tests.
This leads to different experimental values for flammability limits. When a standard is introduced all
values can be compared to each other and it will be of great value for everyone interested in
flammability safety issues.

4.2 Deflagration to detonation transition

The basics of the phenomenon deflagration to detonation transition are now understood, but still a
lot of research has to be done to explain DDT in detail. With the use of accurate numerical models in
combination with experiments more knowledge about DDT can be obtained.

In DDT simulations it is the difficulty to find the right balance between the speed and accuracy of the
simulation. The key is to use the most efficient numerical methods and not use too complicated
physical submodels to describe the fluid dynamics and combustion processes.

For more quickly simulations a shock-flame interaction can be used instead of igniting the mixture at
one end of the tube. The flow develops more quickly, because shock-flame interactions distort the
flame and makes the flow turbulent. So two general types of experimental configurations to study
DDT can be defined; one that allows the turbulence to evolve itself, and another with an initial shock
to create the turbulence.

It should be observed what happens when certain variables are changed in the numerical model. For
example the initial and boundary conditions need to be chosen carefully. Only small changes can
have a big influence on the results of the simulation. For example adiabatic boundary conditions, for
a thermally insulated tube, or isothermal boundary conditions, when the tube walls are maintained
at the ambient temperature, will give different results.

18
 Bibliography
[1] C. Coronado, J.A. Carvalho Jr., J.C. Andrade, E.V. Cortez, F.S. Carvalho, J.C. Santos, A.Z.
Mendiburu, Flammability limits: a Review with emphasis on ethanol for aeronautical applications
and description of the experimental procedure. (2012)
[2] Tareq A. Albahri, Flammability characteristics of pure hydrocarbons, Chemical Engineering
Science p. 3629-3641. (2003)
[3] C.M. Piqueras, J. García-Serna, M.J. Cocero, Estimation of lower flammability limits in high-
pressure systems, The Journal of Supercritical Fluids p. 33-40. (2011)
[4] Safety information sheet of the chemical product – Anhydrous ethyl alcohol fuel. Brazilian Oil Inc.
(2009)
[5] S. Kondo, K. Takizawa, A. Takahashi, K. Tokuashi, A. Sekiya, A study on flammability limits of fuel
mixtures, J. Hazard Mater. p. 440-448. (2008)
[6] I. Glassman, Combustion, 2nd Ed., Academic Press, San Diego, California, (1996)
[7] A. Takahashi, Y. Urano, K. Tokuhashi, S. Kondo, Effect of vessel size and shape on experimental
flammability limits of gases, J. Hazard Mater. p. 27-37. (2003)
[8] N.N. Smirnov, V.F. Nikitin, Yu. G. Phylippov, Deflagration-to-detonation transition in gases in
tubes with cavities, Journal of Engineering Physics and Thermophysics Vol. 83 no. 6. (2010)
[9] L. Bauwens, Zhe Liang, Shock formation ahead of hot spots, Proceedings of the Combustion
Institute Vol. 29 p. 2795-2802. (2003)
[10] L. Kagin, G. Sivashinsky, On the transition from deflagration to detonation in narrow tubes.
(2010)
[11] V. Bychkov, D. Valiev, V. Akkerman, C.K. Law, Gas compression moderates flame acceleration in
deflagration-to-detonation transition. (2011)
[12] Y.J. Zhu, J. Chao, J.H.S. Lee, An experimental investigation of the propagation mechanism of the
critical deflagration waves that lead to the onset of detonation. (2007)
[13] Kuznetsov, M., Matsukov, I., Alekseev, V., Breitung, W. and Dorofeev, S. Effect of the boundary
layer on flame acceleration and DDT. (2005)
[14] Elaine S. Oran, Vadim N. Gamezo, Origins of the deflagration-to-detonation transition in gas-
phase combustion, 2006
[15] Scarinci, T., and Thomas, G.O., Some Experiments on Shock-Flame Interaction, University of
Wales, Aberystwyth, Wales, U.K., 1990
[16] A.M. Khokhlov, Numerical simulation of deflagration-to-detonation transition: The role of shock-
flame interactions in turbulent flames, J. Comput.Phys. 143 (1998) 519-543
[17] E. Yamada, T. Shimada, A.K. Hayashi, Development of Java multi-threaded simulation for
chemical reacting flow of ethanol, Advances in Engineering Software p. 17-23. (2012)
[18] M.A. Liberman, M.F. Ivanov, A.D. Kiverin, M.S. Kuznetsov, A.A. Chukalovsky, T.V. Rakhimova,
Deflagration-to-detonation transition in highly reactive combustible mixtures, Acta Astronautica
p. 688-701. (2010)

19