Search

SiO2 glass has been shock-melted at 100 GPa and then ramp compressed on the Z Accelerator at Sandia National Laboratories. In a shock-ramp experiment, a shockless ramp wave follows the shock impact, isentropically compressing the sample. This technique provides access to off-Hugoniot states, which are of importance to both materials and planetary sciences. Here we present results from shock-ramp experiments for two compositions of SiO2 glass, one with 1000 ppm by weight and one with <1ppm OH. We observe an increased particle velocity at the window-interface for SiO2 melt with 1000 ppm OH relative to the anhydrous sample, suggesting higher sound speed for the hydrous silicate melt. We describe how the lower shock impedance of molten silicate relative to the LiF window precludes the usual approach of iterative Lagrangian analysis to extract compressibility from shock-ramp velocity measurements. To understand the observation of higher velocity for the hydrated sample, and to guide development of a new shock-ramp data analysis approach, we present results from LAMMPS molecular dynamic simulations for SiO2 with 0-1wt.% H2O. These simulations show that the incorporation of water into SiO2 melt decreases the compressibility at constant pressure, consistent with the experimental observation of a stiffer response for the hydrated SiO2 melt. [ABSTRACT FROM AUTHOR]

Understanding the nature of phase transitions of tin under high pressure requires atomic scale modeling, where the simulation accuracy is largely determined by the interatomic potential (IAP) used. We have compared the current state of classical IAPs for tin and have found that only low pressure phases are well represented. However, machine learned interatomic potentials (ML-IAPs) have shown improved accuracy compared to traditional potentials. We report the development of two different training sets, domain expertise (DE) and active learning (AL), for fitting a tin ML-IAP. Each methodology has particular benefits, with DE better reproducing material properties like cold curves and AL more efficiently spanning descriptor space more efficiently. In future work, we plan to further compare the effect of both the training set and type of ML-IAP used on the accuracy of the ML-IAP on tin property calculations and extrapolative modeling performance. [ABSTRACT FROM AUTHOR]

X-ray diffraction data collected under dynamic compression conditions is extremely challenging to interpret due to non-ideal sources and geometries, in addition to an extreme paucity of data. In this work, we leverage a robust simulated X-ray diffraction tool based within the LAMMPS molecular dynamics code to generate a diverse dataset of accurate and realistic diffraction patterns in a computationally efficient manner. This data is used to train machine learning models that are capable of extracting the crystallographic orientation from diffraction simulation results. Specifically, we have developed a convolutional neural network (CNN) capable of identifying the orientations of the underlying crystal lattice from diffraction patterns. [ABSTRACT FROM AUTHOR]

A two-phase analysis is very restrictive in the case when the gaseous phase of a two-phase material is of critical importance for description of the shock response of porous reactive and energetic materials because condensed constituents are often heterogeneous. A three-phase constitutive model is considered, where a three-phase material is described as a whole with a unified set of conservation laws and behaviour of individual phases is described with a set of constitutive equations responsible for the inter-phase exchange of physical quantities. The present work introduces a description of the gaseous phase in the model, which enables one to simulate behaviour of porous materials with two condensed constituents. The present analysis employing the model for graphite-metallic porous mixtures shows a good description of Hugoniot experiments available in literature. [ABSTRACT FROM AUTHOR]

High rate stress upturn for ductile materials at rates between 103-104/s has been known since the 1980s, and previously we've shown how to model this behavior based on our overstress approach to dynamic viscoplasticity. It turns out that brittle materials also undergo high rate stress upturn, but at a lower rate of 1-10/s. most available data on high rate stress upturn of brittle materials is for concrete (from obvious reasons), and they are usually represented by DIF as function of strain rate, where DIF=Dynamic Increase Factor. Here we model high rate stress upturn for brittle materials using our overstress brittle response (OBR) model. Our OBR model includes: 1) damage onset curve for both compression and tension; 2) damage accumulation rate as function of damage onset overstress; 3) damage onset reduction as function of damage level down to the fully damaged state; 4) plastic flow of the fully damaged material as for a granular material; and 5) finite limit for damage accumulation rate. This last item stems from the finite rate of fracturing, and from the finite speed of crack growth, and it turns out that this is the feature that leads to the high rate stress upturn response. To demonstrate how the model works we compute several examples in cylindrical symmetry with different ratios of radial to axial velocities. [ABSTRACT FROM AUTHOR]

Measurement of a material's longitudinal sound velocity (CL) at pressure can yield new information on its equation of state and strength, including sensitivity to phase changes [1-4]. Results are presented on a series of gas gun driven plate impact experiments on pure lead (Pb) samples at stresses between 5 and 17 GPa. Thin tungsten heavy alloy (WHA) flyer plates were used to generate a shock and release profile, with arrival times of both the shock and initial rarefaction waves measured using frequency shifted photon Doppler velocimetry (PDV) at multiple sample thicknesses via a stepped target setup. A comparison of these experimental results to current multi-phase material models are presented, with a reduced gradient in the sound velocity with pressure observed compared to that predicted. [ABSTRACT FROM AUTHOR]

SDOT: an explicit meshless detonation front tracking package is developed at Lawrence Livermore National Laboratory. The new particle-based algorithm solves a set of time-dependent ODEs in a front-normal coordinate rather than the level-set PDE solved in a global coordinate by existing mesh-based detonation front trackers. The new method keeps only front and boundary information at a time-step so has a smaller data space and is more efficient than existing mesh-based methods. The detonation-shock-dynamics (DSD) boundary-angle is explicitly applied by the choice of a boundary-normal coordinate. The front quality is maintained with an equal-distance smoothing method. With the flexibility of a meshless methodology, the new front tracker is capable of dealing with dynamical resolution change at run time, detonation acceleration effect and dead-zone formation in a HE region with an arbitrary boundary geometry. [ABSTRACT FROM AUTHOR]

The Preston-Tonks-Wallace (PTW) viscoplasticity model is valid in the wide range of strain, strain rate, and temperature, and therefore has been used in various applications. In particular, it has incorporated the flow properties in the strong shock limit, where the strain rates span the range of 10−3−1012 sec−1 [1]. In this paper, we examine how the Preston-Tonks-Wallace (PTW) flow stress was constructed: the closed form expression of the flow stress, which is known as the PTW model was obtained by integrating the differential form of the hardening law under the assumption that the strain rate is constant. We then consider there are cases, like explosively driven deformation and high-velocity impacts, where this is not true and how to use this model. As a case study, we choose a gaussian function as a strain rate history and compare two different ways to use the PTW model. First, we integrate the differential form of the PTW model numerically, coupled with this non-constant strain rate function. That way, we let the integral, and therefore the resulting flow stress path-dependent. Second, we use the closed form of the PTW, which was already integrated for the fixed strain rate and plug in the strain rate value at each discretized time. As will be seen, the discrepancy between the two methods is clear when the strain rate is decreasing. We draw the conclusion that based on the physical and mathematical arguments, one should solve the system of ODEs consisting of the differential form of the PTW model and the particular strain rate history of interest rather than using the closed form expression when the strain rate variation is large. [ABSTRACT FROM AUTHOR]

Shock to detonation transition (SDT) experiments are essential in calibrating and validating reactive burn models for explosives. This work investigates the large collection of SDT test data for the explosive PBX 9502 at ambient temperature that was presented by Gustavsen, Sheffield, and Alcon [Journal of Applied Physics, 99, 114907 (2006)]. We first analyze the experimental data and compare two different methods of determining the shock transition time/distance (namely, the bilinear method and the single-curve method). This reveals some of the uncertainty in estimating shock transition points, which contributes to scatter in Pop-plot data. Next, we compare the WSD, AWSD, and SURFplus reactive burn models for a collection of approximately 20 experimental shots using the FLAG hydrocode. Error estimates are used to quantify how well each reactive burn model (and their respective parameter calibrations) performs at predicting the SDT process for a range of loading conditions. Additionally, the importance of mesh resolution and numerical dissipation in SDT simulations will be assessed. [ABSTRACT FROM AUTHOR]

Previously we presented a model (which we call PORT) with rate dependent pore closing/opening, for which we assumed that all pores close/open with the same dynamics [1]. Here we upgrade PORT to take into account different dynamics as function of pore size. We represent different pore sizes by their volume (v), and we have k discrete pore sizes. We therefore call this model VK. For the ith pore size we have ni, i=1,k pores per unit mass. We're not aware of information on pore size distributions of porous materials. We therefore assume that pore sizes are initially distributed with a log- normal distribution. Similar to PORT, we define quasistatic pore closure curves that depend on pore volume v, and we compute the rate of pore closure with a linear overstress equation relative to these curves. From the values of v and vdot (rate of change of v) we then compute (for each cell and for each time) the overall porosity φ and its rate of change φdot. Finally we compute Pdot and Tdot (P=pressure and T=temperature) in the same way as in PORT, using the equation of state of the porous material. To show how our VK model works we apply it to a simple 1D problem. A 20GPa sustained pressure pulse enters a porous aluminum target. We show histories of pressure, temperature and porosity at several locations into the target. We compare these curves with the ones obtained for k=1 (as in PORT). [ABSTRACT FROM AUTHOR]

Elastic constants can be defined by either energy–strain or stress–strain relationships. Although the formal theory is well established, their actual computation via the energy–strain relationship can lead to erroneous results at non-zero pressure. This work explicitly illustrates the problem by computing the elastic constants of fcc-Al via both routes using density functional theory. Following an overview of proposals to address the difficulties, a solution based on adding a correction to the energy derivative which represents the work done by the strain is put forward. In some respects the method is equivalent to other solutions that have been introduced but has the advantages of physical transparency and reduced mathematical complexity. [ABSTRACT FROM AUTHOR]

Within the energetics community, considerable effort is being put forth to find a robust scale-bridging link between unreacted material microstructures and the observed material responses, e.g. detonation and sub-detonative phenomena. Specifically, one area where this scale-bridging capability is needed is mesoscale modeling of explosives initiation (MMEI); here, material microstructures are imported directly or as statistical reconstructions into a hydrocode. While MMEI is attractive for simulating the shock initiation process with ever-increasing model fidelity, a large gap remains between the data being generated at the mesoscale and the calibration of burn model parameters. In this work, stochastic burn models are explored as a paradigm-shift to address possible scale-bridging schemes. These stochastic, particle-based methods are similar to those used for granular and droplet-laden flows, with Langevin-type equations. Further parallels are drawn to turbulent combustion modeling and preliminary developments using probability density function (pdf) theory by Baer, Gartling, and DesJardin. In order to implement these new scale-bridging schemes, one example of a stochastic burn model is explained in greater detail. Results from the stochastic burn model and MMEI simulations are given to illustrate the proposed approach. Ultimately, the execution of this work will be a community endeavor; to achieve such a capability, research efforts should focus on full-field data mining and pdf evolution, in addition to new numerical techniques for hydrocodes. [ABSTRACT FROM AUTHOR]

Gold is commonly used as pressure standard thus an accurate equation of state (EoS) for this material is needed. A multiphase EoS for gold has been developed using the process described in Cox and Christie [1]. In this methodology, parameters for statistical-mechanics-based condensed matter physics models are obtained using a stochastic analysis technique called particle swarm optimization. Equation of state codes from both AWE and LANL were both implemented using this process. This report highlights the types of analysis that can be performed with this method and shows a discrepancy between high pressure isentropic compression data and the resultant gold EoS. [ABSTRACT FROM AUTHOR]

In a hydrocode simulation the knowledge of pressure as a function of density and energy is needed to solve the hydrodynamic equations. This is obtained from the equation of state (EoS) of a material. For routine use the EoS should be accurate, time efficient, and robust. It is thus advantageous for complex models to tabulate the EoS beforehand, and during the hydrocode simulation to interpolate (or extrapolate) using this grid. A common EoS table format is the SESAME format developed at LANL. However, when this format was developed EoS models that accurately capture the discontinuous changes seen with phase transitions were not commonplace. This talk describes additions to the SESAME format that provide a route for tabulating a multiphase EoS. The final product is found to give an accurate and robust response as well as having an acceptable time penalty. [ABSTRACT FROM AUTHOR]

Three solid materials that exhibit steady structured shock-wave fourth-power dependence of shock pressure step versus strain rate are examined in further detail. They are, respectively the compound aluminum oxide, the metal uranium and the molecular crystal HMX. For Al2O3 and HMX, both polycrystalline and single crystal structured wave data are examined. For both materials the plot of polycrystal and single crystal data overlay in the pressure versus strain rate plot. Further, for polycrystalline Al2O3 and uranium the data span of steady-wave widths exceeds the crystal grain size by one to two orders of magnitude. Structured wave data for unalloyed alpha uranium and uranium six percent niobium are compared in the pressure versus strain rate presentation. Both metals exhibit fourth-power pressure versus strain rate behavior, however, U6Nb reveals markedly reduced shock viscosity relative to alpha uranium. In contrast, selected shock compaction Hugoniot data for granular materials are described that exhibit pressure versus strain rate power-law dependence with powers ranging between one and two. Underlying implications and possible sources of the experimental observations are discussed. A statistics base perspective is pursued that offers a sensible account for the underlying the shock failure transition including the universal power-law trends of the structure shock data. [ABSTRACT FROM AUTHOR]

We present advances in accurate, extremely rapid prediction of detonation performance for energetic molecules. These models may be integrated into larger efforts for high-throughput virtual screening, molecular optimization, or an experimentalist's selection of molecules before attempting a hazardous synthesis. Our machine learning workflow utilizes (a) a reference dataset generated from quantum mechanical calculations and the Cheetah thermochemical code, and (b) a directed message-passing neural network (D-MPNN) for nonlinear regression. The D-MPNN is a graph convolutional deep learning model best used with large datasets such as the one in this study. We create models to predict detonation velocity, detonation pressure, heat of formation, and density. Critically, prediction of the detonation properties requires absolutely no information other than the skeletal formula for a molecule. Molecules evaluated are CHNO-containing molecules from public datasets. Neural net architecture and training, including the Python workflow for parallel, automated dataset generation are discussed. The D- MPNN is also evaluated against LASSO and Kamlet-Jacobs models. [ABSTRACT FROM AUTHOR]

Quantum mechanical simulations of the structure of croconic and squaric acids under pressure up to 25 GPa predict the formation of a new high-pressure crystalline phases. Density functional theory (DFT) plane-wave calculations were performed to investigate the effects of isotropic compression on the structural transformations of croconic and squaric acids and to elucidate the details of the phase transitions The advantage of the method is that it allows to analyze a transition path and to monitor changes of bond lengths upon compression and decompression. The onset of pressure-induced polymerization of croconic acid was observed near 12 GPa, with the formation of strong hydrogen (O-H) bonds with a length of approximately 1.25 Å. The bond corresponds to an intermediate type of bonding between weak hydrogen (bond length ∼1.57 Å) and hydroxyl covalent (bond length ∼1.1 Å) bonds. Complete polymerization was noted at pressures near 25 GPa. Further compression up to 55 GPa did not significantly change the O-H bond length (∼1.2 Å), but was found to result in the O-H-O bond angle approaching ∼180 degrees. The simulations revealed a hysteresis upon decompression in the pressure-volume curve as well as the O-H bond lengths indicated possibility of stabilization of high pressure phases of the acids by pressure. Calculated Raman spectra of the acids at the high and ambient pressure are compared with published experimental vibrational spectra. The predicted structures are compared with that obtained using evolutionary method based on DFT plane-wave calculations. [ABSTRACT FROM AUTHOR]

To enable dynamic testing of diagnostic techniques designed to investigate phenomena that occur during the shock loading of materials by high explosive an experimental system which produces a 1-dimensional shock would be desirable. Typically to produce a 1D shocked area from a high explosive source a plane wave lens is used. Explosive plane wave lenses typically require precision machining of multiple explosives and depending on size can contain more than 1 kg of HE. To reduce the explosive mass and remove the need for precision machining of high explosive a hybrid multi-detonator/explosive track plane wave generator has been developed. To assess the simultaneity of the shock arrival and the pressure profile generated by the developed plane wave generator three experiments were conducted. Shock arrival times and the initial free surface velocities (after shock arrival) were recorded using up to 64 channels of PDV at the surface of a copper target placed at the output of the initiation train. The difference in shock arrival times were found to vary between experiments; the spread in arrival time was 241 ns in the first experiment and 125 ns in the second experiment. The pressures generated by the plane wave generator in 2 mm thick copper targets were found to vary between 38 GPa and 62 GPa. [ABSTRACT FROM AUTHOR]

Pulse Correlation Reflectometry (PCR) is a simple and robust new technique to measure the position of shock and detonation waves in a quasi-continuous manner by measuring the transit time of an electrical pulse in a coaxial electrical sensor. Experiments were performed to test the performance of different coaxial sensors with diameters from 0.36 mm up to 2.8 mm. The experiments were conducted using a trackplate with the PCR sensors in contact with the explosive. Piezo pins were fielded to validate the measurements made. The data shows the thin coaxial cables perform well, with the measurement of the velocity of detonation within 4 % of those obtained from piezo pins. A new approach to the analysis was developed which allows the frequency of pulses to be increased, thereby increasing the number of datapoints. This high frequency analysis has been validated with both synthetic data and experimental data which produces velocity of detonation results that fit within errors of data captured using lower frequencies. [ABSTRACT FROM AUTHOR]

Photonic Doppler Velocimetry (PDV) spectrograms are most often used to extract the single velocity of a single moving surface as a function of time. However, spectrograms regularly have further features than single surfaces with single velocities, including secondary surfaces, clouds and ejecta, and surface break-up. We present a method for analyzing PDV spectrograms using likelihood methods which can extract the spectrogram information more accurately and uniformly. We demonstrate on data that these methods give statistically-valid velocities and velocity uncertainties. We also show how these methods can be used to derive extractions of complicated surfaces, such as those with ejecta. Finally, we discuss how these methods can be used directly with models of the expected surface velocity to constrain model parameters, even for complicated observations. [ABSTRACT FROM AUTHOR]

When near-instantaneous shocks are recorded by a Doppler velocity interferometer (VISAR) they typically exceed the detector's ability to react, and "skipped fringes" result where its visibility briefly reduces. Traditionally, replacing skipped fringes required guesswork in analysis, which increased arrival time errors. Secondly, the use of long but velocity-sensitive interferometer delays with fast detectors which can resolve the delay has traditionally been avoided, because of the fear of confusing the arrival time signal. But shorter delays produce smaller fringe phase shift per velocity and thus can decrease velocity precision. We realize that while some loss of fringe information occurs at shock events, this is often just a partial loss, and the residual fringe information can still hold valuable information. We describe a forward model (FM) of the interferometer action and detector blurring that assists with VISAR fringe analysis at skipping events: (1) more precise shock arrival times, (2) even with long delays, and (3) improved ghost subtraction which improves accuracy over a broad time region. We demonstrate the utility of FM on National Ignition Facility (NIF) and Laboratory for Laser Energetics (LLE) Omega shots. [ABSTRACT FROM AUTHOR]

The HERMES model (High Explosive Response to MEchanical Stimulus) [1] has been developed to study the behavior of energetic materials using computer simulations that are closely coupled with experiments. The material properties needed for the HERMES model to analyze a specific experiment vary, but the resistance of the unreacted solid to shear deformation as a function of stress state (including confining pressure) and deformation rate will generally be required. The resistance may include permanent deformation, widespread fragmentation, localized fracture, and porosity development, which all depend on the applied loads. Our non-shock ignition criterion follows the observation that shear localization with confining stress is the condition for ignition. Our shock initiation criterion is based on CREST [2]. We present examples of ignitions that self-extinguish quickly, ignitions that self-extinguish, but nevertheless produce measurable air blast, and ignitions that lead to delayed detonations. We assess the respective roles of the properties of energetic materials and the properties of the confinement on the violence of the response. [ABSTRACT FROM AUTHOR]

We performed tests with one of our propellants to determine its TNT equivalence. In these tests we detonate a charge of explosive on one end of a propellant cylindrical sample. We use two types of diagnostics: 1) to determine the propellant energy release rate, we wrap it with a thin aluminum shell and measure the radial velocity of the shell along the propellant sample with velocity interferometers (PDV); and 2) we use commercial blast gauges at a distance of about 10m to determine the blast effect of the propellant (after deducting the blast effect of the explosive),. We expected the two types of diagnostics to agree with one another, but this was not the case. From the velocity gauges we see that only part of the propellant near the explosive has reacted, and that the extent of this part depends on the size and configuration of the driving explosive. But from the blast diagnostics is large, and it seems that as if the whole propellant has reacted. We've seen some sporadic indications in the literature to such an enhanced blast effect from propellants [1-5]. We refer here to such an enhanced blast effect as a spreading effect. We claim that when a reacting propellant (or explosive) is spread out in space (without changing its total energy yield upon reacting), the blast effect at a given distance is enhanced substantially compared to the same energy yield without spreading. In our tests the relative amount of reacted propellant decreases substantially with distance from the initiating explosive, but because it does not react in a continuous fashion, the spreading effect makes its contribution to air blast quite high. To demonstrate the spreading effect we make 1D simulations of outgoing detonations of an HB (H=HMX, B=Binder) explosive in spherical symmetry. For different runs we dilute the explosive with the binder to various extents, from W=1 (pure explosive) to W=0.01 (1% explosive). For each such formulation we compute the detonation parameters using our in house chemical equilibrium code. For each formulation we adjust the explosive radius so that the total detonation energy stays the same. For all runs we monitor the reflected pressure (pr) at a radial distance of 12m. We get that as W decreases below 0.5, the total amount of HMX decreases down to 6% of the amount for pure HMX, but princreases up to 12% relative to that of pure HMX. We attribute this enhanced blast effect to the spreading of the explosive while keeping the amount of energy released unchanged. [ABSTRACT FROM AUTHOR]

Thermal explosion may happen when an explosive is in a temperature field, and when its boundary temperature is between the critical and ignition temperatures. Semenov and Frank-Kamenetzky solved the thermal explosion problem analytically in 1D and for a constant boundary temperature. Since the 1960s thermal explosion problems have been simulated numerically in 3D and for realistic boundary conditions. In thermal explosion tests it is easier to apply the boundary temperature gradually, and in this way they become cookoff tests. When thermal explosion happens in a small region of an explosive body, the events that follow, and their overall resultant violence, may be quite diverse, like: 1) slow decaying pressure wave; 2) fast non decaying wave; 3) strengthening pressure wave that builds up to shock initiation and detonation. The outcome of a thermal explosion depends on: 1) the sensitivity of the explosive; 2) the temperature field throughout the explosive body at the time of thermal explosion; 3) the geometry of the explosive body; and 4) the degree of confinement of the explosive body. To model what may happen after a thermal explosion event, we use our PDSR (= Pressure Dependent Shear Reaction) together with our TDRR (= Temperature Dependent Reaction Rate) reactive flow models. For each computational cell these two models work in sequence. Initially there is a shear reaction handled by PDSR. If as a result pressure and temperature there go beyond the threshold for reaction out of hot spots, TDRR takes over to compute shock initiation and detonation. We present computed examples of different outcomes of thermal explosion events. [ABSTRACT FROM AUTHOR]