CREST is an entropy-dependent reactive burn model for shock initiation and detonation of explosives in hydrocode calculations. This paper describes the development and initial testing of a CREST model for an explosive that comprises 95% by weight TATB and 5% Kel-F 800 and having a nominal density of 1.905 g/cm(3). Following other recent CREST models, the equations of state are fitted to available shock Hugoniot, sound speed and overdriven detonation wave data, while the reaction rate coefficients are calibrated using an automatic method to fit sustained-shock gas -gun data and the explosive's detonation size -effect curve. It was found that the optimal reaction rate from the automatic parameterization was unable to predict recent detonation corner turning data used for validation purposes. Subsequent analysis of parameter sets close to the "best" found reaction rates that gave significant improvement to the corner turning data with only slight reductions in the quality of fit to the standard calibration data; one of these was chosen for the reaction rate in the final model. The implications of these findings for the future calibration of reactive burn models for TATB -based explosives will be considered and discussed.

The high explosive Composition B (Comp B), which comprises approximately 60% RDX and 40% TNT by weight, is still widely used for a range of applications, and reactive burn models able to simulate its shock initiation and detonation response are required for modelling assessments. CREST is unique among such models in using an entropy-dependent reaction rate to convert the solid unreacted explosive to gaseous detonation products. This paper describes the calibration of a CREST reactive burn model for Comp B. The equations of state are fitted to available shock Hugoniot, sound speed and overdriven detonation wave data, and are demonstrated to be suitably robust and compatible. The reaction rate has been tuned to fit recent sustained-shock-gas-gun data and, owing to a lack of modern data, a detonation size-effect curve dating from the 1950's on a slightly different Comp B variant. This made it necessary to accept a compromise fit to the calibration data, and which is explored further in the paper. The model is then tested against other Comp B data from the literature to assess its ability to predict a wide range of behaviour. CREST now adds to those reactive burn models available for simulating this well-studied explosive.

The shock and detonation response of high explosives has been an active research topic for more than a century. In recent years, high quality data from experiments using embedded gauges and other diagnostic techniques have inspired the development of a range of new high-fidelity computer models for explosives. The experiments and models have led to new insights, both at the continuum scale applicable to most shock and detonation experiments, and at the mesoscale relevant to hotspots and burning within explosive microstructures. This article reviews the continuum and mesoscale models, and their application to explosive phenomena, gaining insights to aid future model development and improved understanding of the physics of shock initiation and detonation propagation. In particular, it is argued that " desensitization" and the effect of porosity on high explosives can both be explained by the combined effect of thermodynamics and hydrodynamics, rather than the traditional hotspot-based explanations linked to pressure-dependent reaction rates.

We examine shock, release and reshock into the tri-amino-tri-nitro-benzene (TATB) based explosive PBX 9502 (95% TATB, 5% Kel-F 800) from both an experimental and modeling point of view. The experiments are performed on the 2-stage light gas gun at Los Alamos National Laboratory and are composed of a multi-layered impactor impinging on PBX 9502 backed by a polymethylmethacrylate window. The objective is to initially shock the PBX 9502 in the 7 GPa range (too weak to start significant reaction), then allow a rarefaction fan to release the material to a lower pressure/temperature state. Following this release, a strong second shock will recompress the PBX. If the rarefaction fan releases the PBX to a very low pressure, the ensuing second shock can increase the entropy and temperature substantially more than in previous double-shock experiments without an intermediate release. Predictions from a variety of reactive burn models (AWSD, CREST, Augmented Ignition and Growth, SURF) demonstrate significantly different behaviors and thus the experiments are an excellent validation test of the models, and may suggest improvements for subsequent modeling efforts.

The CREST reactive-burn model uses entropy-dependent reaction rates that, until now, have been manually tuned to fit shock-initiation and detonation data in hydrocode simulations. This paper describes the initial development of an automatic method for calibrating CREST reaction-rate coefficients, using particle swarm optimisation. The automatic method is applied to EDC32, to help develop the first CREST model for this conventional high explosive.

Mesoscale calculations of hotpots created by a shock wave in a porous explosive show that the hotspots do not cool in times of order at least a microsecond. This suggests that simple models of porosity like the Snowplough model, which assume that a shocked porous explosive jumps to a point on the Hugoniot that is instantaneously in thermodynamic equilibrium, are not correct. A two-temperature model of shocked porous explosive has been developed in which a small fraction of the material, representing the hotspots, has a high temperature, but the bulk of the material is cooler than the temperature calculated by the Snowplough model. In terms of the mean state of the material, it is shown that the two-temperature model only minimally affects the pressure vs. volume and shock velocity vs. particle velocity plot of the Hugoniot, but that the mean state lies slightly off the equation of state surface. The results of the model are compared with two dimensional mesoscale calculations.

At the APS SCCM in 2009, Hill, Zimmermann and Nichols showed that assuming burn fronts propagate at constant speed from individual point hotspots distributed randomly in a volume, the reaction rate history could be determined. In this paper a simple analytic approximation is found for the time history of the pressure in the volume. Using acoustic theory, the time history of the pressure field for burning from a single spherical, isolated hotspot of finite radius is developed. Then at any point in the volume, the overall pressure history is determined from the sum of the pressure fields from all the individual hotspots. The results are shown to be in qualitative agreement with 1D mesoscale hydrocode calculations of the reaction and burning from a finite size spherical hotspot.

A series of gas gun-driven plate impact experiments were performed on vacuum melt-cast Composition B to obtain new Hugoniot states and shock sensitivity (run-distance-to-detonation) information. The Comp B (ρ0 = 1.713 g/cm3) consisted of 59.5% RDX, 39.5% TNT, and 1% wax, with ~ 6.5% HMX in the RDX. The measured Hugoniot states were found to be consistent with earlier reports, with the compressibility on the shock adiabat softer than that of a 63% RDX material reported by Marsh.[4] The shock sensitivity was found to be more sensitive (shorter run distance to detonation at a given shock input condition) than earlier reports for Comp B-3 and a lower density (1.68-1.69 g/cm3) Comp B formulation. The reactive flow during the shock-to-detonation transition was marked by heterogeneous, hot spot-driven growth both in and behind the leading shock front.

The Detonation Shock Dynamics (DSD) model allows the calculation of curvature-dependent detonation propagation. It is of particular use when applied to insensitive high explosives, such as EDC35, since they have a greater non-ideal behaviour. The DSD model is used in conjunction with experimental cylinder test data to obtain the JWL Equation of State (EOS) for EDC35. Adjustment of parameters in the JWL equation changes the expansion profile of the cylinder wall in hydrocode simulations. The parameters are iterated until the best match can be made between simulation and experiment. Previous DSD models used at AWE have no mechanism to adjust the chemical energy release to match the detonation conditions. Two JWL calibrations are performed using the DSD model, with and without Hetherington's energy release model (these proceedings). Also in use is a newly-calibrated detonation speed-curvature relation.

The CREST reactive-burn model has been remarkably successful in modelling shock initiation and detonation propagation behaviour in plastic-bonded explosives. This has been achieved by using a reaction rate that depends on shock strength (measured using a function of entropy of the unreacted explosive) and time since the shock passed, and not on evolving parameters of the flow like pressure or temperature which are the basis of most reactive-burn models. This paper investigates the advantages and disadvantages of entropy, temperature or pressure-dependent reaction rates for simulating a variety of one and two-dimensional shock initiation phenomena.

A critical hotspot is just large and hot enough that it can itself react, and go on to spread reaction into the cooler surrounding explosive, before it is cooled by heat conduction. This paper describes how previously-published hydrocode models were used to obtain critical hotspot criteria for HMX. The results compare well to those in the literature. In the simulations, which account for hydrodynamics, heat conduction and Arrhenius chemistry, reaction propagates outwards from hotspots via a flame driven by heat conduction. The flame propagation speed is compared to data for HMX from high-pressure diamond anvil cell experiments, leading to a new explanation for the negative pressure dependence sometimes observed in experiments.

CREST is an innovative reactive-burn model that has been developed at AWE for simulating shock initiation and detonation propagation behaviour in explosives. The model has a different basis from other reactive-burn models in that its reaction rate is independent of local flow variables behind the shock wave e.g. pressure and temperature. The foundation for CREST, based on a detailed analysis of data from particle-velocity gauge experiments, is that the reaction rate depends only on the local shock strength and the time since the shock passed. Since a measure of shock strength is the entropy of the non-reacted explosive, which remains constant behind a shock, CREST uses an entropy-dependent reaction rate. This paper will provide an overview of the CREST model and its predictive capability. In particular, it will be shown that the model can predict a wide range of experimental phenomena for both shock initiation (e.g. the effects of porosity and initial temperature on sustained-shock and thin-flyer initiation) and detonation propagation (e.g. the diameter effect curve and detonation failure cones) using a single set of coefficients.

CREST is a reactive-burn model that uses entropy-dependent reaction rates to model shock initiation and detonation behaviour in plastic bonded explosives. A CREST model for the TATB-based high explosive PBX 9502 was published previously at this conference. It is well known that changing the porosity of an explosive, like PBX 9502, can dramatically influence its sensitivity. The equation of state used in CREST incorporates the snow-plough model, allowing the porosity of the explosive to be selected at will, while keeping the reaction model constant. In this paper, it will be shown that CREST can predict the change in explosive sensitivity with porosity, as demonstrated by the experimentally determined Pop-plots for a similar explosive, LX-17. In contrast, it will be shown that pressure-dependent reactive-burn models are unable to predict this porosity effect without changing the reaction rate.

The Detonation Shock Dynamics model (DSD) is widely used for the propagation of detonation wave-fronts in hydrocode calculations of polymer bonded explosives. In DSD, a detonation velocity vs. curvature relationship D(K) is used to determine the speed at which the detonation front propagates through the explosive. The D(K) relationship for an explosive is usually obtained from rate-stick wave-curvature data. Recently, a series of very small rate-stick experiments, using samples of the HMX-based explosive EDC37 machined with a femto-second laser, have been performed at LLNL. In this paper, this data is combined with the results of previous 0.5 '' and 1 '' rate-stick experiments, to obtain a new D(K) relationship for EDC37. A wave-shape analysis code is developed which uses weighted least-squares fitting methods which improve on current fitting methods.

In recent years a comprehensive suite of gas-gun particle velocity gauge experiments have been conducted at Los Alamos National Laboratory. Detailed analysis of this data has led to advances in the understanding of the shock to detonation transition in polymer bonded explosives, and hydrocode simulations of the experiments are frequently used to calibrate reactive-burn models. In these simulations, the gauges are modelled using Lagrangian markers, with no physical representation of the gauge package. In contrast, the experimental gauge package consists of etched aluminium sandwiched between two sheets of FEP Teflon, using an urethane-based glue. The gauge package is approximately 60 mu m thick and is positioned between two wedge-shaped pieces of explosive at an angle of 30 degrees, to form a right circular cylinder. This paper investigates, with one-dimensional calculations, whether there is a need to include an accurate representation of the gauge package within future hydrocode simulations.

In many reactive flow models for explosives, the non-reacted and detonation products equations of state are treated independently, each just matching some appropriate experimental data. As a result, the two equations of state usually conflict in some parts of the pressure-volume-internal energy space. This paper suggests a general criterion for the relationship between the two equations of state and develops more specific criteria for the often used Mie-Gruneisen form of equations of state.

PBX9502 is an insensitive high explosive comprising 95% TATB and 5% Kel-F. Like some liquids, the shock velocity - particle velocity relation for non-reacted PBX9502 is initially curved, but tends to a linear relation for stronger shocks. Because of this, the method developed by Jeanloz of finding a Taylor expansion in finite strain form for the principal isentrope cannot be used to develop an EOS. Instead, the principal isentrope for PBX9502 has been found from the Hugoniot by integration, using an analytic form for the variation of Graneisen Gamma with specific volume. The isentrope has been extended by plausible extrapolations, both beyond the maximum Hugoniot compression and into the expansion regime. Finally an analytic fit has been made to the resulting principal isentrope to develop a Mie-Gruneisen form of EOS for use in reaction rate models.