Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 The physics, chemistry and dynamics of explosions Elaine S. Oran and Forman A. Williams Phil. Trans. R. Soc. A 2012 370, 534-543 doi: 10.1098/rsta.2011.0385 References This article cites 22 articles, 10 of which can be accessed free http://rsta.royalsocietypublishing.org/content/370/1960/534.full. html#ref-list-1 Email alerting service Receive free email alerts when new articles cite this article - sign up in the box at the top right-hand corner of the article or click here To subscribe to Phil. Trans. R. Soc. A go to: http://rsta.royalsocietypublishing.org/subscriptions Report Documentation Page Form Approved OMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. 1. REPORT DATE 3. DATES COVERED 09 JAN 2011 2. REPORT TYPE 00-00-2012 to 00-00-2012 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER The Physics, Chemistry And Dynamics Of Explosions 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION US Naval Research Laboratory,Washington,DC,20375 REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S) 11. SPONSOR/MONITOR’S REPORT NUMBER(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited 13. SUPPLEMENTARY NOTES Phil. Trans. R. Soc. A (2012) 370, pgs. 534-543 14. ABSTRACT 15. SUBJECT TERMS 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF ABSTRACT OF PAGES RESPONSIBLE PERSON a. REPORT b. ABSTRACT c. THIS PAGE Same as 12 unclassified unclassified unclassified Report (SAR) Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18 Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 Phil. Trans. R.Soc. A(2012)370,534–543 doi:10.1098/rsta.2011.0385 I NTRODUCTION The physics, chemistry and dynamics of explosions BY ELAINE S. ORAN1,* AND FORMAN A. WILLIAMS2 1Laboratories for Computational Physics and Fluid Dynamics, US Naval Research Laboratory, Washington DC, USA 2Department of Mechanical and Aerospace Engineering, University of California, San Diego, La Jolla, CA, USA The motivation for devoting a Theme Issue to explosions is discussed. As subsequent articles in the issue are written with the assumption that the reader has had a certain amountofpreviousexposuretothesubject,someofthehistoryandnecessarybackground information are presented here. The topics on explosions that will be encountered in the remaining articles are previewed. Finally, several important future outstanding research problems, beyond those addressed in the following articles, are discussed, with the objective of complementing the coverage of explosions in this issue. Keywords:explosion;detonation;deflagration;deflagration-to-detonationtransition;turbulent reactingflows;astrophysicalexplosions 1. Introduction Explosions are ubiquitous in the Universe. Beginning with the Big Bang, the history of the Universe has been determined largely by explosions, which disrupt orderly progression and initiate evolution in new directions. While intergalactic and galactic explosions are of many different types, explosions on the Earth involve a narrower range of phenomena that are somewhat more amenable to human understanding. Effects of explosions on humankind, both beneficial and detrimental, have been investigated from the beginning of science, resulting today in a degree of knowledge of their properties—their physics, chemistry and dynamics. During the past century, this knowledge has been used both to design explosions for specific purposes and to teach us how to avoid those that are harmful. Any knowledge, however, can be used or misused, for actual or misconceived benefit. Possible future advances include mitigation of explosion hazards, improved and new combustion engines and use of inertial confinement for the development of nuclear fusion as an energy source. *Authorforcorrespondence([email protected]). Onecontributionof12toaThemeIssue‘Thephysics,chemistryanddynamicsofexplosions’. 534 Thisjournalis©2012TheRoyalSociety Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 Introduction 535 The term explosion itself is nebulous. It generally refers to any type of scenario in which energy is injected into a system faster than it can be smoothly equilibrated through the system; that is, energy is deposited faster than a dynamical time scale. For chemical or nuclear explosions, this time scale (l/c ) s is based on the characteristic size (l) of the system and the acoustic velocity (c ). For magnetohydrodynamic explosions, c is replaced by the square root of s s the sum of c2 and the magnetoacoustic time scale c2 , where c is proportional s ma ma to the speed of Alfven waves, the square of which, in turn, is proportional to the magnetic pressure. The result of this rapid injection of energy is a local pressure increase. If the system is unconfined, or if the confinement is weak and can be broken, then strong pressure waves (shock waves) develop and spread outwards, travelling considerable distances before they are dissipated. As this happens, over-pressurized material begins to expand, and heated material cools. This very general description covers scenarios that range from the Big Bang, to thermonuclear explosions in stars, to magnetohydrodynamic explosions on the sun and to most of the chemical explosions on the Earth. The intent of this issue on explosions is to expose the reader to some of the many different aspects of explosion phenomena known today. The articles here address a variety of theoretical, computational and experimental studies of explosions, thereby demonstrating the range of methods employed in investigations. In addition to presentations of background knowledge, underlying theoretical formulations and their mathematical exposition, each article contains previously unpublished research that opens up new questions and issues for future study. An initial summary, given next, of a few basic concepts, which in subsequent articles are assumed to be known by the reader, may facilitate understanding of the material that is to follow. 2. Fundamentals Thebasisofourpresentunderstandingbeganwiththeexperimentalobservations of Berthelot & Vieille [1] and Mallard & LeChatelier [2], and the theoretical explanations of Chapman [3] and Jouguet [4], more than a century ago. Notable advances were made by Zel’dovich [5], von Neumann [6] and Döring [7], among many others, who identified elements of the structure (the ZND structure) of detonations, which are the stronger of the two classical types of explosions. The names of these nine scientists have been thoroughly associated with the fundamental concepts that they developed. These concepts identify two distinct steady-flow regimes, to which time-dependent explosions may evolve, or between which time-dependent explosive transitions may occur. It is instructive to consider the pressure–volume (p–v) diagram of a fluid as shown in figure 1. The pressure p and specific volume v of the initial mixture 0 0 areindicatedasapointinthisdiagram.Forsteady,planar,one-dimensionalflow, principles of conservation of mass, momentum and energy can be used to find the locus of possible final states of the fluid after the heat release has occurred in an explosion. That locus, called the Hugoniot curve, as illustrated in the figure, consists of two disconnected parts, an upper branch and a lower branch. The two branches must be separated for exothermic processes because application of steady mass and momentum conservation alone requires that a straight line Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 536 E. S. Oran and F. A. Williams p upper branch a Rayleigh line for a detonation Hugoniot curve a Rayleigh line for deflagration p 0 lower branch v v 0 Figure 1.Traditional pressure–volume (p–v) diagram showing the Hugoniot curve with possible statesofdeflagrationsanddetonations. connecting the initial and final states in this diagram, called the Rayleigh line (known in gas dynamics as a trajectory for constant-area compressible flow with heat addition), cannot have a positive slope. When the final state lies along the lowerbranchoftheHugoniotcurve,theprocessiscalleda deflagration,andwhen it lies along the upper branch, it is a detonation. Thus, there are two very different and distinct types of one-dimensional, steady reaction waves propagating through homogeneous reactive material, consistent with conservation conditions. The figure illustrates that detonations exhibit large pressure increases and small volume decreases, while deflagrations have large volume increases and small pressure decreases. Propagation velocities of such fronts are proportional to the square root of the negative of the slope of the Rayleigh line, two representative examples of which are indicated on the figure. Propagation velocities of detonations are thus seen to exceed those of deflagrations; detonations are found to be supersonic and deflagrations subsonic. The figure also shows that there is a minimum velocity for detonations and a maximum velocity for deflagrations, limiting conditions that correspond to tangency of the Rayleigh line and the Hugoniot curve, and the associated final conditions are called the (upper and lower) Chapman–Jouguet (CJ) points of the curve. Further analysis demonstrates that final fluid velocities are sonic at the CJ points. Except for CJ conditions, when there is any intersection at all between a given Rayleigh line and the Hugoniot curve, there are two, as can be seen in the figure. Therefore, there are two different types of each process. From the slopes, it is seen that both types of deflagrations travel slower than CJ, and both types of detonations faster. The first intersection is called weak and the Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 Introduction 537 second strong. Final fluid velocities are supersonic for strong deflagrations and for weak detonations, while they are subsonic in the other two cases. Although strong deflagrations have been postulated for certain galactic winds, all known deflagrations on the Earth are weak, approaching CJ only for highly turbulent, energetic flames. On the other hand, most detonations are CJ, although both weak and strong detonations occur. Strong detonations often arise for periods of time in transitions from deflagrations to detonations (DDTs). Laboratory studies of chemical kinetics in shock tubes, when exothermic overall, in principle also are strong detonations for the entire shock-tube test time. On the other hand, condensation ‘shocks’ in wind tunnels are weak detonations. Explosions, however, are dynamic processes that often are neither the classical deflagrations nor the classical detonations described by steady-state flows in homogeneous backgrounds with perfectly unobtrusive geometries. As an example, consider the preceding conclusion that detonations are supersonic and deflagrations subsonic; this will be seen in a later article to not always be true for porous media. Although pressure changes associated with deflagrations are smaller than those associated with detonations, the expansion and heating that occur in deflagrations cause pressure build-up, especially in confined spaces. This can be quite damaging and thereby causes the processes to be classified as explosions, even though detonations do not develop. In many instances, these pressure waves and their associated heating produce DDT. The most damaging explosions, however, are generally detonations because they have the highest pressures. Pressures even higher than those in CJ detonations can occur in transient processes, such as shock–shock interactions and strong detonations, which precede the completion of DDT. Depending on the specific configuration and both the reactivity and the energetics of the combustible mixtures, periods may occur in which there is supersonic front propagation slower than CJ, and such slow (sometimes called ‘weak’) detonations are types of explosions that may nottransformfullytoCJandthatcannotbeclassicalsteadilypropagatingfronts. Nevertheless, they may be very damaging and therefore need to be understood. There thus exist many non-classical types of explosions requiring further study. 3. Topics in this Theme Issue The first article in this issue presents an examination of one of the largest recent accidental chemical explosions. The explosion at Buncefield, UK, in 2005 is unique in that it is unusual to have such a thorough investigation of a large- scale, accidental, statistically improbable event. Questions raised by the forensics committee included the following. What caused the fuel spillage? How did it ignite? Do we know enough to understand the dynamics of the turbulent flame development and propagation? What was the role of the obstacles in the path of the explosion front? And finally, did a detonation develop at any stage? This first article, by Bradley et al. [8], summarizes the work of the technical investigation committee, presents the current state of the analysis of this event, puts it in the context of other historic large-scale accidental explosions and then describes where future research is required. This needed research, such as the research reportedinthesubsequentarticlesinthisissue,goeswellbeyondtheidealizations identified earlier. Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 538 E. S. Oran and F. A. Williams The ZND detonation structure envisions steady, planar flow with heat release initiated by the temperature rise across a strong planar shock wave (as in many shock-tube experiments). Evidence that this picture is much too simplified for most real-world explosions can be at least traced to the observations of spinning detonations by Campbell & Woodhead [9], and clear recognition of the prevalent existence of multi-dimensional structures was provided by Denisov & Troshin [10]. We now know that the ZND structure is generally unstable, evolving into cellular detonation structures that range from regular diamond patterns to very complex dynamic interactions of shock waves, reaction regions and shear layers behind leading shock fronts. In the second article, Kessler et al. [11] use multi-dimensional numerical simulations to study these phenomena. By extending the computations to detonation propagation normal to a gradient of stoichiometry, they identify upstream (behind the detonation front) reaction regions into which reactants diffuse from opposite sides. This region is bounded by rich and lean reaction regions, and the result is a triple- flame detonation wave. This type of gradient of fuel concentration could develop in the more complex spatial and temporal fuel distributions that occur in explosion-prone situations, such as in tunnels in coal mines or during industrial fuel leakages. Complementingthiscomputationalworkaretheanalyticalstudiesinthethird article; Clavin & Williams [12] present a unified formulation that covers a range of detonation topics, including initiation, propagation, quenching, pulsation and cellularstructure.Asymptoticanalysesarediscussedhereforthelimitofstrongly temperature-sensitive rates of heat release and for the Newtonian limit in which the difference between constant-pressure and constant-volume heat capacities vanishes (assumed by Isaac Newton and found to be surprisingly accurate for high-temperature real gases in hypersonics). Planar detonations are shown to be unstabletoshear-wavedisturbances,evenwhentheheat-releaserateisinsensitive to temperature, with a weakly nonlinear analysis indicating that this bifurcation leads to distinct diamond patterns. This article also clarifies conflicting literature concerning the stability of shock waves in general non-ideal fluids, deriving for the first time the boundary between decaying and non-radiating, non-decaying stable shock discontinuities. The next two articles are theoretical and experimental investigations, respectively, of explosions in porous media. The first, by Brailovsky et al. [13], summarizesimplicationsofasimplifiedmathematicalmodelbasedontheconcept of hydraulic resistance. Introduction of the hydraulic-resistance approximation reduces the order of the differential equations that describe the detonation and deflagration processes, thereby facilitating their solution and making it easier to understand their predictions. For these reasons, it is important to determine experimentally the extent to which the model can correspond to reality, and a new favourable comparison is identified (called Shchelkin’s effect there). The experimental observations and measurements reported by Ciccarelli [14], in the second of these articles, expose experimental similarities and differences between explosion propagation in porous media and in chambers filled with obstacles. Theseresultsarequiterevealing;theyshow,inparticular,structuresofexplosions in the low-velocity slow-detonation regime, where the propagation velocities of the fronts are between c and the CJ detonation velocity. It is made clear s here that acquisition of detailed, local data on explosions in porous media is Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 Introduction 539 a challenging experimental task. These studies are important, especially to the extent that porous-media results can be applied to industrial-accident situations with obstacles in the flow. The authors of the next article, Chiquete & Tumin [15], follow in the path of investigators who have performed mathematical stability calculations of planar detonations. Here, they specifically consider, for the first time, influences of slightly porous walls in circular tubes, and they compare their results with those for an impenetrable wall. The analysis and associated numerical results provide a clear example of effective methods for tackling problems of this kind. The next two articles, again the first somewhat more theoretical and the second somewhat more experimental, address the development of explosions in varioustypesofchambersandforvariouscombustiblemixtures.Bradley[16]first explainsawell-developed,coherentviewpointoftheexplosionprocessesinspark- ignition, internal-combustion engines. This study of explosions, which are at the smallest scale considered in this issue, indicates how and when engine knock may occur. Ideas are extended to explosion development in ducts, including a summaryofresultsforturbulentdeflagrationvelocities.Bycontrast,Thomas[17] summarizesresultsofextensiveexperimentalstudiesofDDT,bothingaseousand insolidexplosives.Inadditiontoaddressinganumberofdifferentprocesses,such asdetonationdevelopmentingradientsoffuelconcentrationandthetransmission anddiffractionofdetonationwavesatjunctions,theinteractionsofdifferentscales in explosion phenomena are described. A major concern of Thomas has been to produce experimental results on very hard problems that are amenable to further studyandanalysisbynumericalsimulations.Anexampleofthis,describedinthe article,isthe‘strangewave’phenomena,involvingthecouplingofashock,aflame and a boundary layer. The last two articles strike out in quite different directions. First, Starikovskiy et al. [18] describe the physics and chemistry of plasma-assisted ignition in combustion processes, and then they show how this can be used to promote DDT for use in engines that employ detonations for propulsive purposes, either continuously, as in rotating-detonation engines, or periodically, as in pulsed- detonationengines.Plasma-assistedcombustionisanextensivetopicin itself,one that requires consideration of ionized states of atoms and molecules. Although knowledge of the complex kinetics of such ionized states has been developed for specific applications, such as atmospheric and ionospheric chemical reactions or solar coronal thermonuclear reactions, they are not generally known for the usual hydrocarbon molecules involved in combustion problems. The authors of this penultimate article describe what is known, and we can thus infer how much we will need to learn to properly use plasmas to enhance combustion. Finally, Wheeler [19] addresses the many types of astrophysical explosions currently under discussion in astrophysics, bringing us up to date on some of the latest work and theory. These explosions include magnetohydrodynamic and thermonuclear energy sources, mentioned previously, and gravitational collapse is also considered. It is noteworthy how atomic, nuclear and plasma physics interact with turbulence and transport of particles, photons and neutrinos as criticalingredientsintheseproblems,whichinvolveextraordinarilywiderangesof relevant scales. Explosions of white-dwarf start lead to type Ia supernova (SNIa), whichhavebecomethe‘standardcandles’ofourUniverse.Wheelerdiscusseshow measuringSNIaspectraandredshiftsleadstothemostconsistentdetermination Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 540 E. S. Oran and F. A. Williams of the rate of expansion and size of the Universe. Understanding why SNIa are so uniforminvolvesunderstandingtheevolutionofturbulentflamesanddetonations in the thermonuclear environment in the star, which certainly is a challenging topic in explosions. 4. Other future research directions An important question not addressed in the following articles concerns explosion limits. It is commonly accepted that there are reactive mixtures which are too dilute,toofuelleanortoofuelrichtoexplodeunderanycircumstances.Moreover, the literature contains extensive tabulations of flammability limits for various combustibles, specifying, for example, upper and lower deflagration limits, such thatdeflagrationscannotpersistinmixturesricherthantheupperlimitorleaner than the lower limit. There are also explosion-limit diagrams, in which explosions are known to occur spontaneously in mixtures lying between these limits; these spontaneous explosion limits are narrower than deflagration limits and are of less interest in the present context. In addition to deflagration limits, there are detonation limits, although less data are available for detonations. Despite a common belief that limits of detonability are narrower than deflagration limits, fundamentally this need not be true. The chemical kinetics of detonations differs from deflagrations in that they require initiation steps in which chain carriers are formed, while the chain carriers are already present in deflagrations. To delineate mixtures in which the most severe types of explosions may occur, it is of interest to determine limits of detonation. It is well known that limits of detonation of combustible mixtures in chambers depend on the dimensions of the chamber. A detonation propagating past a stationary wall develops a boundary layer on the wall behind the leading shock, and the influence of this boundary layer on the flow can tend to weaken the detonation. For a detonation in a long circular tube of diameter d, the propagation velocity decreases with decreasing values of d, and at a critical diameter, the propagation ceases. The value of this limiting diameter depends on the composition of the mixture. Figure 2 is a graph of experimental results for the value of 1/d at this critical diameter for methane–air mixtures initially at roomtemperatureandnormalatmosphericpressure[20–22].Ithasbeenstandard procedure to extrapolate such plots to 1/d =0 to define detonability limits of combustible mixtures in the open. It may be seen from this figure that, prior to 2011, this extrapolation would have led to the conclusion that methane–air mixtures with less than 6 per cent or more than 14 per cent methane on a volume basis would lie outside the limits of detonation. However, the recent data seen in the wings clearly demonstrate that this conclusion would be false. Tabulated detonability limits for methane are in error. Further research to determine ultimate limits of detonation is warranted. As it becomes increasingly difficult to generate explosions in such near-limit mixtures, it is important to also study what is required for their initiation. Anothersignificantareaofinvestigationisthecontrolofexplosions.Depending on the particular situation, there is interest in generating explosions, preventing them,orquenchingthemoncetheydevelop.Therearemanydifferentideasabout explosioncontrol.Onearticleinthisissuediscussestheuseofplasmasforcontrol. Phil. Trans. R.Soc. A(2012) Downloaded from rsta.royalsocietypublishing.org on January 9, 2012 Introduction 541 25 20 1) – m 15 d ( 1/ 10 5 0 4 6 8 10 12 14 16 %CH 4 Figure2.Measuredleanandrichdetonationlimitsformethaneinairasafunctionofthepercent ofmethanebyvolume.Limitsareshownforfourvaluesof1/d,theinverseofthediameterofthe test device. Names and symbols refer to earlier studies (squares, Kogarko [20]; triangles, Matsui [21];circles,Gamezoetal.[22]). Reliefvalvesandblow-outcapsarecommonexplosion-controldevices.Therehave been studies of detonation arrestors, for example, employing narrow channels, porous materials, dust dispersions, water sprays, water-filled passages, etc. There have been methods for extinguishing explosions that attempt to disperse inert rock dust or placing buckets of water to be overturned during an explosion. But thetaskisconfusinganddifficult.Methodsthatappeartodampexplosionsunder someconditionsenhancethemunderotherconditions.Wewouldliketobeableto controltransients,tochangetheexplosionstate,sothatwecanignite,enhanceor quench at will, turn detonations into deflagrations or taylor the process of DDT. Much more understanding is necessary to achieve these objectives. Finally, more research is needed on the role of turbulence in explosions. A half-century ago, when the ZND detonation structure was widely accepted as universal, fine-scale shadowgraph observations surprisingly suggested that detonations were turbulent. Improvements in resolution soon showed that this turbulence, in reality, was the now-familiar regular cellular diamond pattern. Today, with greatly improved experimental techniques, especially for detonations having highly irregular, somewhat chaotic cell structures, reacting shear layers behind leading shocks are often described as turbulent. The true nature of this kind of turbulence, however, remains to be determined. As high-speed deflagrations interact with shock waves and boundary layers, they are subject to shock-related and shear-flow instabilities that may generate turbulence of character different from that which is commonly understood. More thought is thus warranted about turbulence in all types of explosions. Phil. Trans. R.Soc. A(2012)