lable at ScienceDirect

Journal of Loss Prevention in the Process Industries 24 (2011) 194e199

Contents lists avai

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier .com/locate/ j lp

Effect of ignition position on the run-up distance to DDTfor hydrogeneair explosions

Robert Blanchard*, Detlef Arndt, Rainer Grätz, Swen ScheiderBAM Federal Institute for Materials Research and Testing, Chemical Safety Engineering, Explosion Protection and Risk Assessment, Unter den Eichen 87, 12205 Berlin, Germany

a r t i c l e i n f o

Article history:Received 7 September 2009Received in revised form20 December 2010Accepted 20 December 2010

Keywords:Deflagration to detonation transition (DDT)Ignition positionHydrogenDetonation and retonation

* Corresponding author. Tel.: þ49 174 636 8526; faE-mail address: robblanchard@hotmail.com (R. Bla

0950-4230/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.jlp.2010.12.007

a b s t r a c t

The method described in this paper enabled reliable and accurate positioning of an overdriven detonationby calculation of shock wave velocities (detonation and retonation) for hydrogen explosions in a closed18m long horizontal DN150 pipe. This enabled an empirical correlation between the ignition position andthe run-up distance to DDT to be determined. It was shown that the initial ability of the flame to expandunobstructed and the piston-like effect of burnt gas expanding against the closed end of the tubecontributed to initial flame acceleration and hence were able to affect the run-up distance to overdrivendetonation. Flame speeds and rates of initial pressure rise were also used to explain how these twocompeting effects were able to produce a minimum in the run-up distance to DDT. The shortest run-updistance to DDT, relative to the ignition position, for this pipe and gas configuration was found when theignition position was placed 5.6 pipe diameters (or 0.9 m) from the closed pipe end. The shortest run-updistance to DDT relative to the end of the pipe was recorded when the ignition source was placed 4.4 pipediameters or 0.7 m from the pipe end.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Most flames in process equipment begin as deflagrations, wherecombustion is propagated by the molecular transportation of heatand mass. However, given the right conditions flame accelerationand the transition from a deflagration to a detonation can takeplace. In detonative burning, combustion is sustained by adiabaticshock heating and in comparison to deflagrative burning; flamespeeds, overpressures and rates of pressure rise are significantlyhigher and are generally more destructive.

The first published scientific studies on gas explosions in pipeswere carried out at the end of the 19th century,when themotivationwas the protection ofmines andmining equipment fromnatural gasexplosions. A flamewas initiated at the closed end of a long pipe andpropagated towards an open end, Mallard and Le Chatelier (1883)were the first to show inversions in the axial direction of theseflames. This was later confirmed by photographic evidence pre-sented by Ellis (1928). This study also showed that flame propaga-tion in a pipe consisted, initially, of several stages and that the flamestructure at these stages was effected by the overall length of thepipe. Since then several explanations have been put forward toexplain this phenomena (Bychkov, Akkerman, Fru, Petchenko, &

x: þ49 30 8104 1217.nchard).

All rights reserved.

Eriksson, 2007; Ciccarelli & Dorofeev, 2008), although it is stilla subject of some debate.

If unhindered, immediately after ignition in a tube the flamewill start to spherically propagate from the ignition source. As thesurface area of the flame front increases during this period,the flame speed accelerates and the rate of pressure rise increases.As flame propagation continues the flame frontmoves faster axiallythan radially, resulting in an elongated (finger) flame (Bychkovet al., 2007), as the surface area of this elongated flame increasesthe rate of combustion and the rate of pressure rise also increases.However, at some point the flame front will reach the walls of thevessel, subsequently cooling the flame front due to conductive heatloses. This causes a deceleration of the flame front velocity andthe rate of pressure rise, relative to the earlier stage. The flamewill continue to propagate axially and at some point after this time,due to a combination of various processes the flame takes on aninverted configuration, increasing its surface area. It can thencontinue to accelerate along the length of the pipe until furtherquenching takes place or resistance to downstream flow isencountered.

If the flame continues to accelerate the transition from defla-gration to detonation can occur (DDT). Whether working inchemical safety (Ciccarelli & Dorofeev, 2008), or with technologysuch as pulse detonation engines (PDE) (Roy, Frolov, Borisov, &Netzer, 2004), it is important that this parameter can be calcu-lated and predictedwith a reasonable degree of accuracy in order to

R. Blanchard et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 194e199 195

affect the end goal, control over flame acceleration and the processof deflagration to detonation transition.

The point at which ignition takes place is an important factorduring the initial stages of combustion (Phylaktou & Andrews,1991). When the ignition source is further from a sealing flangeor a vessel wall the flame will have a longer period in which tospherically propagate, leading to initially higher overall flamespeeds and higher rates of pressure rise. Changes in these initialconditions will also effect how combustion continues further alongthe pipe and ultimately the transition into in detonative mode ofburning. It is this link between the changes in the ignition positionand the run-up distance to DDT which is considered in the currentresearch. Of paramount importance to the observation of this effectis the accurate determination of the run-up distance to DDT, thiswas achieved by the close examination of pressure waves at variouspoints in the closed tube.

1.1. Background

Some effects relating to where the ignition source is positionedare reasonably well understood (Ellis & Wheeler, 1928; Kirkby &Wheeler, 1928), however these studies tend to concentrate on theinitial effects such as the local flame form and the initial rate ofpressure rise (Kindracki, Kobiera, Rarata, & Wolanski, 2007). Otherstudies have also shown the ability of changes in the position of theignition source to affect the flame speed of a methaneeair explo-sion in a long closed vessel (Phylaktou & Andrews, 1991). However,little data is available on how this initial parameter affects otherlater combustion processes such as the deflagration to detonationtransition.

Determination of the run-up length to DDT requires that theposition at which the event occurs can be accurately calculated.Early methods for the determination of this value included char-acterisation of the flame speed and Schlieren streak photography.Later stroboscopic flash Schlieren photography further increasedthe knowledge relating to this phenomenon (Dumanois & Laffitte,1926; Lafitte, 1928; Egerton & Gates, 1927), however it was notuntil Urtiew and Oppenheim (1966) employed Schlieren photog-raphy with a short pulse width laser and a high frame rate thatprecise location of the DDT event could be achieved. This work alsogave an extremely powerful insight into the DDT process as thephotographs where able to show a local explosion within theflame-shock front, the “explosion in the explosion” (or overdrivendetonation), which was shown to be able to occur at both the flamefront and the shock front.

Other recent methods for determining the position of DDTinclude; the measurement of the propagation speed of pressure orcombustion waves relative to the CJ-detonation speed, analysis ofvisible light in line with the axial propagation of the flame andextrapolation of the velocities of the detonation and retonationwaves produced by the DDT event (Li, Lai, Chung, & Lu, 2005).

The later of these methods was used by Li et al. (2005) to locatethe position of the DDT in propane-air explosions, however pres-sure-time traces used to determine the velocity of the retonationwave showed some ambiguity. An attempt was made in the pre-sented research to use this method with a greater degree of accu-racy, employing a greater number of more sensitive pressuretransducers.

2. Experimental

2.1. Apparatus

A horizontal DN150 steel pipe was used for the following tests(d ¼ 159 mm). This pipe was made up of a number of segments

ranging from 2 to 5 m in length, bolted together with a gasket sealin-between the connections and blind flanges on both ends. Evac-uation prior to introduction of the test gas confirmed no significantleaks were present in the pipe. Earlier work employing constanttemperature anemometry techniques (Lohrer, Drame, Schalau, &Grätz, 2008) had also shown that these connections introducedno significant turbulence inducing element to flow along the pipe.The total length of the pipe was approximately 18 m (L), giving anL/d ratio of 112.

2.2. Sensors and data collection

The pipe segments contained several sensor penetrations in theaxial direction. Penetrations not used were filled with plugs whichsat flush with the inside of the tube. The position, and hence thespeed, of the flame front was determined using BPY 62 Silicon NPNphototransistors (OSRAM) placed at regular intervals along theentire length of the pipe.

The pressure at various points along the length of the pipe wasrecorded using piezoelectric pressure transducers (PCB M113A22,1,5 � 0.005 mV/kPa) with the signal being processed by a PCB 481Sensor Signal Conditioner.

All sensors were flush with the inside surface of the tube inorder to avoid any additional turbulence enhancement. Data fromall sensors was collected at a frequency of 0.18 MHz.

2.3. Gas mixtures

Gas mixtures were produced using the partial pressure methodand mixed by a paddle in a rocking pressurised (approximately15 bar) tube. The gas was mixed for between 5 and 10 min whichwas considered sufficient time to produce a uniform composition.The gas mixture was then introduced into the evacuated pipe, tothe desired pressure. All gas mixtures used during these experi-ments were at stoichiometric concentration and 1 bara, unlessotherwise stated. After introduction of the gas into the evacuatedpipe, ignition was affected as soon as safely possible in order toavoid any stratification of the gas mixture.

Gas mixtures were ignited at various points axially along thepipe using a melting wire (ignition energy 1e10 J). Ignition wasdetected using a phototransistor placed close to the ignition source,this phototransistor also triggered the data logging system. Formost experimental configurations a number of tests (minimumthree) were carried out depending on the reproducibility of theflame speeds, overpressures and the measured position of the DDT.

3. Results and discussion

3.1. Determination of run-up distance to DDT

An important parameter in any safety or engineering applica-tion relating to gas explosions in long vessels is the run-up distancerequired to produce a transition from deflagration to detonation.This is defined here as the distance between the ignition source andthe point at which DDT was observed. Pressure measurementstaken along the length of the pipe described in Section 2 were usedto obtain detailed information about the point of transition fromdeflagration to detonation. An example of the overpressuresmeasured is shown in Fig. 1, where the responses from nine pres-sure transducers are shown on the same time scale. The box in thetop-left of each plot details the point along the pipe (ignition at0.40 m in this case) at which the pressure was measured. Pressuremeasurements, for these experiments, were taken over a total of2.5 s, however for clarity, only a small time range is shown in Fig. 1,

Fig. 1. Over pressure at various points along a closed tube during a hydrogeneair explosion with the ignition source at 0.4 m from the tube end.

R. Blanchard et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 194e199196

during which the transition from deflagration to detonation isbelieved to have taken place.

Initial analysis of the shape and peak overpressure of the pres-surewaves showed that the approximate position at which the DDToccurred was between 6.57 m and 8.49 m. The pressure waverecorded at 4.21m showed no sign of a detonation-like shock wave,whereas true detonation shock waves were seen between 10.49and 16.56 m. Although it is difficult to see from the presented data(due to the pressure scale of the graph), the overpressure of thedetonation wave recorded at 8.49 m was significantly higher thanthose recorded between 10.49 and 16.56 m, this is further evidencethat the overdriven detonation associated with the DDT took placeclose to this point. In the pressure traces measured at 6.57 m and

8.49 m a lot of noise was seen after the initial explosion pressurewave had past, this is thought to be due to vibrations from theoverdriven detonation wave interfering with the cables connectingthe pressure transducers to the data logging system.

During the transition from deflagration to detonation a volumeof pre-compressed turbulent gas ahead of, or coupled with, theflame front detonates (Ciccarelli & Dorofeev, 2008), developingrelatively high velocities and pressures which later subside to thosetypical of a stable detonation. During the detonation of this pre-compressed volume a pressure wave is generated which thentravels downstream continuing the detonation process by adiabaticshock heating of the un-burnt fueleair mixture. This pressure waveis shown in the green circles in Fig. 1. When this pressure wave

Fig. 3. Flame speed and DDT position for a hydrogeneair explosion with the ignitionsource at 0.4 m from the tube end.

R. Blanchard et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 194e199 197

reached the end of the pipe it was reflected back into the reactionproducts, producing a pressure wave which travelled back downthe pipe, this pressurewave is shown in the blue circles of Fig.1. Theinitial denotation of the pre-compressed volume of gas alsoproduced a pressure wave which travelled upstream into thereaction products of the initial deflagration, a so called retonationwave (Urtiew & Oppenheim, 1966). This pressure wave, which isshown in the red circles in Fig. 1, was much weaker in comparisonto the propagating detonation wave.

The times at which these various pressure waves were detectedby the pressure transducers were used to calculate their velocity inthe pipe, as shown in Fig. 2 where the reciprocal speed of thepressure waves are given by the gradients of the linear trend lines.Additional points in Fig. 2 were obtained from a second set ofpressure transducers, which were not displayed in Fig. 1 for reasonsof clarity.

The speed of the detonation wave travelling downstream afterthe DDT was approximately 1970 ms�1, showing good agreementwith the predicted CJ-donation velocity of 1968 ms�1. The speed ofthe shock wave reflected from the blind flangewhich travelled backinto the reaction products was 1330 ms�1, whilst the otherupstream shock wave generated by the initial DDT event hada velocity of 1270 ms�1. Extrapolation of the two lines associatedwith the initial DDT to the point at which they cross enabledaccurate positioning of the “explosion in the explosion” (or over-driven detonation) where the transition from deflagration todetonation took place. Using the velocities of the pressure wavesshown in Fig. 2 the transition point was found to be at a distance of8.09 m from the end of the pipe, or 7.69 m from the ignition source.This is in good agreement with the data shown in Fig. 1 as thepressure vs. time curve upstream of this point appears to be theshape commonly associated with deflagration, whereas the shapeof the curve immediately downstream of this point shows classicalcharacteristics associated with a detonation pressure wave.

The point at which the DDT took place is shown alongside theflame speed data for the same experiment in Fig. 3. This point wasvery close to where the maximum flame speeds were measured.This should be expected as the maximum flame speeds would haveoccurred during the overdriven detonation process during whicha pre-compressed volume of gas underwent detonation. However,while the flame speed can provide an indication of where theoverdriven detonation took place it should not be used in isolationas it relies on an average taken over a small length of the pipe and

Fig. 2. Velocity of pressure waves from a hydrogeneair explosion with the ignitionsource at 0.4 m from the tube end.

so the DDT could have taken place at any point between the twophotodiodes recording the time of the flame arrival. Using pressurerecords to calculate the velocity of the two waves originating fromthe DDT gave a much more accurate positioning of the transitionpoint.

Some simplifications and assumptions were used in the imple-mentation of this method, for example, when the overdrivendetonation occurs, both the flames speed and the rate at which theassociated shock waves travel will be faster than those of thefollowing stable detonation and retonationwaves. Therefore, it wasassumed that this deceleration from overdriven to stable shockwave occurred to the same extend for both the upstream anddownstream propagating waves.

The method presented in this research showed a significantimprovement on previous implementations of using the velocitiesof detonation and retonation waves to calculate the position of theDDT. The method presented here also has the advantage of cross-checking the DDT point with the flame speed data.

3.2. Effect of ignition position on the run-up distance to DDT

An important parameter in any explosion scenario is the posi-tion of ignition as it has been shown to influence initial flamepropagation, flame speeds and overpressures. However, no corre-lation has been made between these early effects and laterphenomena such as the run-up distance to DDT. Here the run-updistance to DDT is defined as the distance between the ignitionpoint and the point at which the DDT occurred.

The ignition point is also an important factor in current flamearrestor standards (EN ISO 16852:2010) which require only a “blindflange with ignition source”, no indication is given as to whetherthe ignition source should be flushwith the blind flange or whetherit can protrude into the flammable mixture. Additionally, apartfrom the materials science challenges of containing and controllinga sustained cycle of detonations, the pulse detonation engine(PDEs) community are also interested in the process of the defla-gration to detonation transition, including controlling and opti-mising the run-up distance to detonation.

Using the method described in Section 3.1 the run-up distanceto DDT for a number of ignition positions was calculated and isshown in Fig. 4. Initially, as the ignition point moved further awayfrom the blind flange there was a dramatic decrease in the run-updistance to DDT. With the ignition position flush with the blind

Fig. 4. Run-up distance to DDT. Fig. 5. Initial flame speeds for the ignition position flush with and at 0.74 and 5.10 mfrom the blind flange.

Fig. 6. Initial pressure rise for the ignition position flush and at 0.74 and 5.10 m fromthe blind flange, measured close to each ignition position.

R. Blanchard et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 194e199198

flange the flame would have initially propagated hemispherically,giving it a relatively small surface area. In contrast, as the ignitionpositionmoved further away from the blind flange the flamewouldhave been able to propagate longer in a spherical mode, eliminatingheat losses to the walls and therefore increasing the burnt gasexpansion rate and flame speed. The extra flame surface area in thespherical mode would have also enhanced the burning rate andfurther increased the flame speed.

Aminimum in the run-up distance to DDTwas observed with anignition point at a distance of 0.9 m from the pipe end, giving anapproximate decrease of 40% in the run-up distance required forDDT compared to ignition at the blind flange. After this point therun-up distance to DDT increased as the distance between theignition point and the closed end of the tube increased, followed bythe run-up distance starting to level out at an ignition position of5 m.

It is thought that the expansion of the burnt gas against theclosed end of the pipe behind the propagating flame front wouldhave a piston-like effect which would have increased the flamespeed and increased the turbulent interactions between the flamefront and the gas mixture immediately before the flame front.Therefore, the further the ignition point moved away from theclosed end of the tube the less pronounced this effect became.These two competing effects gave an optimum position for ignition,which for this tube diameter and gas mixture was determined to be5.6 tube diameters, or 0.9 m, from the nearest closed end of thetube.

Initial flame speeds for three test configurations are shown inFig. 5, with the ignition source flush with the tube end and at 0.74and 5.10 m from the tube end. The competing factors in the initialacceleration of the flame can be seen by interpretation of the speedprofiles.

The flame front for the flush ignition configuration initiallypropagated relatively slowly as the hemispherical flame wouldhave extended away from the ignition point. As the flame reachedthe outside walls of the pipe there was a small decrease in the rateof flame acceleration. After this point the flame continued topropagate along the length of the pipe, accelerating until DDT wasachieved. With the ignition source at 5.1 m from the pipe endthe flame initially accelerated quickly, following this short initialphase the flame began losing heat to the pipe walls and as little orno piston effect existed to aid the flame acceleration the flamespeed decreased noticeably. The flame then began to accelerate as it

continued propagating along the pipe. The flame speed shown forthe 0.74 m configuration illustrates what happened close to theoptimum ignition position, the flame is able to initially acceleratequickly due to the spherical propagation of the flame front andhence minimise heat losses to the walls. Following this initialacceleration the flame was able to continue accelerating due to thepiston effect of the burnt gas expanding against the closed tubeend.

These effects can also be seen in the initial rates of pressure rise,recorded close to each ignition position, shown in Fig. 6. The 5.1 mconfiguration initially showed the fastest rate of pressure risebefore subsiding as the flame began to lose heat to the walls of thepipe. The configurationwith the ignition source flush with the wallshowed a much slower rate of pressure rise, however the periodover which the rate of pressure rise increased was considerablylonger. Close to optimum conditions were seen with the 0.74 mconfiguration, where a high and sustained increase in the rate ofpressure rise was observed.

As shown in Fig. 7, in order to affect DDT over the shortestdistance relative to the pipe end the ignition position should beplaced at 0.7 m or 4.4 pipe diameters from the pipe end.

Fig. 7. Distance between the closed pipe end and the DDT position.

R. Blanchard et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 194e199 199

4. Conclusions

The method described in this paper enabled reliable and accu-rate positioning of an overdriven detonation by calculation of shockwave velocities. This enabled an empirical correlation between theignition position and the run-up distance to DDT to be determined.It was shown that the initial ability of the flame to expand unob-structed and the piston-like effect of burnt gas expanding againstthe closed end of the tube contributed to initial flame accelerationand hence were able to affect the run-up distance to overdrivendetonation. Flame speeds and rates of initial pressure rise were alsoused to explain how these two competing effects were able toproduce a minimum in the run-up distance to DDT.

The shortest run-up distance to DDT for this pipe and gasconfiguration was found when the ignition position was placed 5.6pipe diameters (or 0.9 m) from the closed pipe end. The shortestrun-up distance to DDT relative to the end of the pipe was recordedwhen the ignition source was placed 4.4 pipe diameters (or 0.7 m)from the pipe end.

The conclusions given in this paper are valid only for the chosenpipe diameter; effects of scale also need to be investigated as wellas further ignition positions (axial and radial) and further gasmixtures with different expansion ratios in order to fully

understand the effects taking place and enable reliable predictionsto be made. It is also understood that, due to the length of the pipe,it is possible that initially produced sound waves could have beenreflected from the end of the pipe complicating the conditions ofthe DDT. It was not possible with the current facilities, however itwould be interesting to carry out these experiments with a longertube to be able to rule out any of these effects. Also of interestwould be to limit the piston effect of the expanding burnt gas byusing some manner of pressure relief close to the ignition point.

References

Bychkov, V., Akkerman, V., Fru, G., Petchenko, A., & Eriksson, L. E. (2007). Flameacceleration in the early stages of burning in tubes. Combustion and Flame, 150(4), 263e276.

Ciccarelli, G., & Dorofeev, S. (2008). Flame acceleration and transition to detonationin ducts. Progress in Energy and Combustion Science, 34(4), 499e550.

Dumanois, P., & Laffitte, P. (1926). The influence of pressure on the formation of anexplosive wave. Comptes Rendus Hebdomadaires Des Seances De L Academie DesSciences, 183, 284e285.

Egerton, A., & Gates, S. F. (1927). Further experiments on explosion of gaseousmixtures of acetylene, of hydrogen, and of pentane. Proceedings of Royal SocietyA, 114, 152.

Ellis, O. C. D. C. (1928). Flame movement in gaseous explosive mixtures, (Part 2).Fuel in Science and Practice, 7, 502.

Ellis, O. C.d. C., & Wheeler, R. V. (1928). Explosions in closed cylinders. (Part III). Themanner and movement of flames. Journal of the Chemical Society, 3215.

European Standard EN ISO 16852. (2010). Flame arresters e Performance require-ments, test methods and limits for use.

Kindracki, J., Kobiera, A., Rarata, G., & Wolanski, P. (2007). Influence of ignitionposition and obstacles on explosion development in methaneeair mixture inclosed vessels. Journal of Loss Prevention in the Process Industries, 20(4e6),551e561.

Kirkby, W. A., & Wheeler, R. V. (1928). Explosions in closed cylinder. Part I. Methaneair explosions in a long cylinder. Part II. The effect of the length of the cylinder.Journal of the Chemical Society, 3203.

Lafitte, P. (1928). Influence of temperature on the formation of explosive waves.Comptes Rendus de l’Académie Sciences Paris, 186, 951.

Li, J., Lai, W. H., Chung, K., & Lu, R. K. (2005). Uncertainty analysis of deflagration-to-detonation run-up distance. Shock Waves, 14(5e6), 413e420.

Lohrer, C., Drame, C., Schalau, B., & Grätz, R. (2008). Propane/air deflagrations andCTA measurements of turbulence inducing elements in closed pipes. Journal ofLoss Prevention in the Process Industries, 21(1), 1e10.

Mallard, M., & Le Chatelier, H. (1883). Recherches experimentales et theoriques surla combustion des mélanges gazeux explosifs. Annales de Mines, Paris series, 8,274e568.

Phylaktou, H., & Andrews, G. E. (1991). Gas explosions in long closed vessels.Combustion Science and Technology, 77(1e3), 27e39.

Roy, G. D., Frolov, S. M., Borisov, A. A., & Netzer, D. W. (2004). Pulse detonationpropulsion: challenges, current status, and future perspective. Progress inEnergy and Combustion Science, 30(6), 545e672.

Urtiew, P. A., & Oppenheim, A. K. (1966). Experimental observation of the transitionto detonation in an explosive gas. Proceedings of Royal Society A, 295, 1328.