Among current deep learning approaches for synthetic image generation, diffusion-based models stand out in terms of algorithmic stability and ability to retain high-fidelity image features with detailed resolution. In this work, we employ Dreambooth, a method for fine-tuning Stable Diffusion, on X-ray CT images of microstructure of the polymer-bonded form (PBX) of a commonly used high explosive, Pentaerythritol tetranitrate (PETN), which yields generative models for creating synthetic PBX images. The models developed here represent five classes (or 'lots') of microstructures and demonstrate successful generation of images of each class with high fidelity, as verified by computed classification accuracy of similar to 94% or higher. Data augmentation afforded by such image synthesis can be used to more reliably decipher underlying statistics, build processing-structure correlations, recognize off-normal structural anomalies, and identify age-related changes. Ideas related to converting image data into appropriate density mapping and performing mesoscale simulation or surrogate modeling of detonation are also discussed.

Physics-aware deep learning (PADL) has gained popularity for use in spatiotemporal dynamics simulations, such as those in computational modeling of energetic materials (EM). We show that the challenge PADL methods face while learning complex field evolution problems can be simplified and accelerated by decoupling it into two tasks: learning complex geometric features in evolving fields and modeling dynamics over these features in a lower-dimensional feature space. We build upon our previous work on physics-aware recurrent convolutional neural networks (PARC). PARC embeds knowledge of underlying physics into its neural network architecture for more robust and accurate prediction of evolving physical fields. PARC was shown to effectively learn complex nonlinear features such as the formation of hotspots and coupled shock fronts in various initiation scenarios of EMs, as a function of microstructures, serving effectively as a microstructure-aware burn model. Here, we further accelerate PARC and reduce its computational cost by projecting the original dynamics onto a lower-dimensional invariant manifold, or "latent space." The projected latent representation encodes the complex geometry of evolving fields (e.g., temperature and pressure) in a set of data-driven features. The reduced dimension of this latent space allows us to learn the dynamics during the initiation of EM with a lighter and more efficient model. We observe a significant decrease in training and inference time while maintaining results comparable to PARC at inference. This work takes steps towards enabling rapid prediction of EM thermomechanics at larger scales and characterization of EM structure-property-performance linkages at a full application scale.

Polymer-bonded explosives (PBXs) are composite materials whose structure-property relationships are poorly understood due to their complex microstructure. Microstructural adhesion between the energetic molecular crystals and the polymer binder in a PBX affects key properties including mechanical strength, safety, and aging characteristics, but is difficult to measure experimentally. To this end, we develop a molecular dynamics (MD) modeling framework to predict the cohesive properties of a prototypical PBX based on the insensitive high explosive TATB (1,3,5-triamino-2,4,6-trinitrobenzene) and a fluoropolymer binder. The MD framework rests on two key capabilities: (1) general protocols to construct MD simulations of crystal-polymer and crystal-crystal interfaces in low-symmetry materials, and (2) a numerically robust approach to compute the surface and interface energies. Formal analysis of the surface energy equation shows that its direct evaluation is poorly conditioned for systems involving polymers. In this respect, we adopt a regression-based approach for computing surface energies and extend it to interfaces. Application to a TATB-fluoropolymer PBX reveals several energetic drivers, including thermodynamic drivers for grain coarsening and a greater propensity for TATB crystallites to adhere to each other than to the polymer binder. These data indicate that the thermodynamic landscape governing formulation design and aging of PBXs is complex and involves additional variables besides crystal-polymer adhesion. While the present study focuses on explosives, the MD methodology is general and can be readily adapted to model other material systems involving low-symmetry crystals, including organic electronics, pharmaceuticals, and advanced polymer composites.

How can we build surrogate solvers that train on small domains but scale to larger ones without intrusive access to PDE operators? Inspired by the Data-Driven Finite Element Method (DD-FEM) framework for modular data-driven solvers, we propose the Latent Space Element Method (LSEM), an element-based latent surrogate assembly approach in which a learned subdomain ("element") model can be tiled and coupled to form a larger computational domain. Each element is a LaSDI latent ODE surrogate trained from snapshots on a local patch, and neighboring elements are coupled through learned directional interaction terms in latent space, avoiding Schwarz iterations and interface residual evaluations. A smooth window-based blending reconstructs a global field from overlapping element predictions, yielding a scalable assembled latent dynamical system. Experiments on the 1D Burgers and Korteweg-de Vries equations show that LSEM maintains predictive accuracy while scaling to spatial domains larger than those seen in training. LSEM offers an interpretable and extensible route toward foundation-model surrogate solvers built from reusable local models.

Capturing sharp, evolving interfaces remains a central challenge in reduced-order modeling, especially when data is limited and the system exhibits localized nonlinearities or discontinuities. We propose LaSDI-IT (Latent Space Dynamics Identification for Interface Tracking), a data-driven framework that combines low-dimensional latent dynamics learning with explicit interface-aware encoding to enable accurate and efficient modeling of physical systems involving moving material boundaries. At the core of LaSDI-IT is a revised auto-encoder architecture that jointly reconstructs the physical field and an indicator function representing material regions or phases, allowing the model to track complex interface evolution without requiring detailed physical models or mesh adaptation. The latent dynamics are learned through linear regression in the encoded space and generalized across parameter regimes using Gaussian process interpolation with greedy sampling. We demonstrate LaSDI-IT on the problem of shock-induced pore collapse in high explosives, a process characterized by sharp temperature gradients and dynamically deforming pore geometries. The method achieves relative prediction errors below 9% across the parameter space, accurately recovers key quantities of interest such as pore area and hot spot formation, and matches the performance of dense training with only half the data. This latent dynamics prediction was 106 times faster than the conventional high-fidelity simulation, proving its utility for multi-query applications. These results highlight LaSDI-IT as a general, data-efficient framework for modeling discontinuity-rich systems in computational physics, with potential applications in multiphase flows, fracture mechanics, and phase change problems.

Many high explosive (HE) formulations are composite materials whose microstructure is understood to impact functional characteristics. Interfaces are known to mediate the formation of hot spots that control their safety and initiation. To study such processes at molecular scales, we developed all-atom force fields (FFs) for Octol, a prototypical HE formulation comprised of TNT (2,4,6-trinitrotoluene) and HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine). We extended a FF for TNT and recasted it in a form that can be readily combined with a well-established FF for HMX. The resulting FF was extensively validated against experimental results and density functional theory calculations. We applied the new combined TNT-HMX FF to predict and rank surface and interface energies, which indicate that there is an energetic driver for coarsening of microstructural grains in TNT-HMX composites. Finally, we assess the impact of several microstructural environments on the dynamic melting of TNT crystal under ultrafast thermal loading. We find that both free surfaces and planar material interfaces are effective nucleation points for TNT melting. However, MD simulations show that TNT crystal is prone to superheating by at least 50 K on sub-nanosecond timescales and that the degree of superheating is inversely correlated with surface and interface energy. The modeling framework presented here will enable future studies on hot spot formation processes in accident scenarios that are governed by strong coupling between microstructural interfaces, material mechanics, momentum and energy transport, phase transitions, and chemistry.

In this work, the design and execution of an experiment with the goal of demonstrating control over the evolution of a copper jet is described. Simulations show that when using simple multi-material buffers placed between a copper target with a conical defect and a cylinder of high-explosive, a variety of jetting behaviors occur based on material placement, including both jet velocity augmentation and mitigation. A parameter sweep was performed to determine optimal buffer designs in two configurations. Experiments using the optimal buffer designs verified the effectiveness of the buffer and validated the modeling.

In this work, we detail a novel application of inverse design and advanced manufacturing to rapidly develop and experimentally validate modifications to a shaped charge jet analogue. The shaped charge jet analogue comprises a conical copper liner, high explosive (HE), and silicone buffer. We apply a genetic algorithm to determine an optimal buffer design that can be placed between the liner and the HE that results in the largest possible change in jet velocity. The use of a genetic algorithm allows for discoveries of unintuitive, complex, yet optimal buffer designs. Experiments using the optimal design verified the effectiveness of the buffer and validated the modeling.

Understanding the mechanisms of shock-induced pore collapse is of great interest in various disciplines in sciences and engineering, including materials science, biological sciences, and geophysics. However, numerical modeling of the complex pore collapse processes can be costly. To this end, a strong need exists to develop surrogate models for generating economic predictions of pore collapse processes. In this work, we study the use of a data-driven reduced order model, namely dynamic mode decomposition, and a deep generative model, namely conditional generative adversarial networks, to resemble the numerical simulations of the pore collapse process at representative training shock pressures. Since the simulations are expensive, the training data are scarce, which makes training an accurate surrogate model challenging. To overcome the difficulties posed by the complex physics phenomena, we make several crucial treatments to the plain original form of the methods to increase the capability of approximating and predicting the dynamics. In particular, physics information is used as indicators or conditional inputs to guide the prediction. In realizing these methods, the training of each dynamic mode composition model takes only around 30 seconds on CPU. In contrast, training a generative adversarial network model takes 8 hours on GPU. Moreover, using dynamic mode decomposition, the final-time relative error is around 0.3% in the reproductive cases. We also demonstrate the predictive power of the methods at unseen testing shock pressures, where the error ranges from 1.3% to 5% in the interpolatory cases and 8% to 9% in extrapolatory cases.

Viscous flow serves as a significant heating mechanism during the formation of hot spots, but the shear viscosity which determines this response is poorly characterized for most high explosives. Recently, a model was proposed for the shear viscosity of liquid HMX (1,3,5,7- tetranitro-1,3,5,7-tetrazocane) that was fit to pressures reaching 5 GPa, but this work uncovered uncertainties in the viscosity at 0 GPa and remains untested at higher pressures. We use molecular dynamics (MD) simulations and the Green-Kubo formalism to predict the temperature and pressure-dependent shear viscosity of HMX over the pressure interval 0 GPa <= P <= 40 GPa. Reassessment of the viscosity at 0 GPa rules out several potential explanations for discrepancies between earlier reports; we tentatively attribute these differences to details of MD trajectory integration and analysis protocols. The shear viscosity of HMX exhibits an Arrhenius temperature dependence at each pressure considered, with exponential prefactor and activation energy terms that are also strong functions of pressure. An analytic form for the viscosity is developed based on an extension of the well-known Andrade equation that simultaneously captures the temperature and pressure dependencies in the MD data up to 40 GPa. Comparison against a recently developed model for the viscosity of liquid RDX (1,3,5-trinitro-1,3,5-triazinane) shows that both materials exhibit similar functional dependencies with the viscosity of HMX being higher by roughly an order of magnitude at a given temperature-pressure state.

Numerous crystal- and microstructural-level mechanisms are at play in the formation of hotspots, which are known to govern high explosive initiation behavior. Most of these mechanisms, including pore collapse, interfacial friction, and shear banding, involve both compressive and shear work done within the material and have thus far remained difficult to separate. We assess hotspots formed at shocked crystal-crystal interfaces using quasi-1D molecular dynamics simulations that isolate effects due to compression and shear. Two high explosive materials are considered (TATB and PETN) that exhibit distinctly different levels of molecular conformational flexibility and crystal packing anisotropy. Temperature and intra-molecular strain energy localization in the hotspot is assessed through parametric variation of the crystal orientation and two velocity components that respectively modulate compression and shear work. The resulting hotspots are found to be highly localized to a region within 5-20 nm of the crystal-crystal interface. Compressive work plays a considerably larger role in localizing temperature and intra-molecular strain energy for both materials and all crystal orientations considered. Shear induces a moderate increase in energy localization relative to unsheared cases only for relatively weak compressive shock pressures of approximately 10 GPa. These results help isolate and rank the relative importance of hotspot generation mechanisms and are anticipated to guide the treatment of crystal-crystal interfaces in coarse-grained models of polycrystalline high explosive materials.

The cover image is based on the Research Article Simulating the Effects of Grain Surface Morphology on Hot Spots in HMX with Surrogate Model Development by H. Keo Springer et al., https://doi.org/10.1002/prep.202200139

In this numerical study, we investigate the effects of layered high explosive (HE) charge design on Richtmyer–Meshkov instability (RMI) in metal plates with sinusoidal surface features. The detonation wave from the HE induces a shock in the metal target that subsequently interacts with the surface features; this results in vortex formation and ultimately RMI. We seek to modify RMI by altering the detonation wave characteristics. The modification is investigated in a twofold manner: first, by varying the initial design of the unconfined surface of the target and second, by varying the charge design and composition. Within a limited scope of this design space, a wide variety of behaviors related to RMI growth are observed. Mechanistic actions, including exaggerated front curvature, behind these modifications are proposed. Charge designs, which modify RMI the most for a select target design, are then presented.

We describe a two-phase model of combustion effects in aluminized high explosive (HE) charges. It is based on: (i) a Gas Dynamic Model of the expansion of the detonation product gases and their turbulent combustion with air; and (ii) a Heterogeneous Continuum Model of aluminum (Al) droplets and their combustion with the detonation product gases. Initial conditions are based on an analytical similarity solution for a cylindrical Chapman-Jouguet (CJ) detonation propagating at the CJ detonation velocity. The CJ jump conditions are computed at the thermodynamic equilibrium state by the Cheetah code, assuming the Al droplets are inert. We assume that the Al is 10% of the charge mass and occurs as droplets at the CJ state. Different initial droplet diameters, ranging from 10 to 100 microns, are studied. A hydrodynamic combustion model based on large Damkohler numbers is employed in this study, 3so1 thorn 0:276 Kd2 w.. It has a square-root dependence on the Reynolds number (Re) and inversed-squared dependence on the droplet diameter (dw). The burnout time (tB) of the Al droplets has a three-halves dependence on the droplet diameter, tB. od0wTHORN3=2. After burnout, the detonation products act as detonation products of the HE charge with active Al. They turbulently mix with air and form a combustion layer on the outer edge of the fireball. Details of the two-phase model, initial conditions and evolution of the flow field will be described.