Computational Physics Student Summer Workshop

Investigation of fluid-mediated particle-particle interactions using high-resolution simulations

Min Wang (T-3), Duan Zhang (T-3)

Particle-laden flow refers to a type of fluid flow that contains a significant number of solid particles. These flows are prevalent in various natural and industrial processes, among which high-speed particle-laden flows associated to ejecta and explosion are of particular interest to LANL. Current particle-laden flow theories primarily consider fluid-particle interactions through force models, such as drag force and added mass force. However, these force-only models are generally deemed insufficient as they lack critical meso- and macro-scale physics, resulting in ill-posed governing equations that fail to converge numerically. The long-term goal of this project is to develop a physically sound fluid-particle phase interaction model for particle-laden flows. Specifically, we will propose a particle-fluid-particle (PFP) stress model to complement existing force models, utilizing a novel particle statistical approach and microscopic high-resolution simulations. This pioneering stress model will more accurately capture fluid-mediated particle-particle interactions and the evolution of particle phase structure, addressing the missing physics in current particle-laden flow theories. Finally, the developed model will be integrated into LANL’s production codes to enhance support for LANL’s key programs.

Modeling the Rayleigh Taylor Instability Driven by Volumetric Energy DepositionKinetic plasma simulations using super-particles

Ally Timko (XCP-8), Brandon Wilson (XCP-8),  Jasper Thrussell (XCP-8)

The Rayleigh Taylor instability occurs at the unstable interface between a heavy fluid and light fluid. In 2023, the volumetric energy deposition (VED) driven Rayleigh Taylor instability was observed by students in the Los Alamos Dynamic Summer School (LADSS). This was the first time this instability was observed experimentally. In this project, students will develop and perform simulations of the VED Rayleigh Taylor instability. Predictions will be compared to experimental measurements using verification and validation practices. If a parallel LADSS project is chosen, students will have the chance to work with experimentalists acquiring new measurements and inform experimental design for future VED Rayleigh Taylor experiments.

Kinetic plasma simulations using super-particles

Guangye Chen (T-5), Chengkun Huang (T-5)

We propose to develop and investigate a super-particle approach to the Particle-in-Cell (PIC) method, based on the Scovel-Weinstein Poisson map, which represents phase space using a finite number of discrete "lumps" or super-particles. Unlike conventional PIC methods that use macro-particles with a delta function in velocity space, this approach captures higher-order moments, enabling accurate phase space representation with fewer particles. This results in a system that conserves Hamiltonian across truncation orders—an advantage over traditional fluid or PIC models. We plan to implement this super-particle method in the GPU-ready CabanaPIC and VPIC codes (both based on Kokkos) to simulate fundamental plasma phenomena, including Landau damping, two-stream instability, and Weibel instability. By modifying the collective field solver to incorporate higher-order moments, this approach provides a more scalable and accurate representation of particle dynamics. By minimizing computational overhead without compromising accuracy, this project aims to prototype a high-performance plasma simulation method that can significantly enhance kinetic plasma modeling capabilities across various applications.

Fluid-structure interaction simulations using modern versions of the material point method

Tim Waters (XCP-1), Duan Zhang (T-3), Jiajia Waters (T-3)

One of the forefront simulation capabilities being developed by LANL researchers are algorithms for modeling fluid-structure interactions. Such algorithms have numerous applications, ranging from predicting the width of an electron beam necessary to melt and vaporize a target material to improving the design of hypersonic re-entry vehicles. The fluid dynamics and solid mechanics group is currently developing a novel fluid-structure interaction code that combines the Discontinuous Galerkin (DG) with variants of the material point method (MPM). The algorithms are based on a hybrid Lagrangian-Eulerian approach where Lagrangian material points (capable of carrying history-dependent quantities) deform and move through an Eulerian background mesh. The summer project would focus on demonstrating the advantage of modern versions of MPM. In addition to working through tutorials on the theory behind DG and MPM in order to learn the code Moya/Cartablanca++, you will learn how to run and analyze simulations on LANL’s high performance computing (HPC) clusters, as well as how to extract physical insights from the numerical solutions.

Numerical implementation of friction for granular simulations in three dimensions

Nitin Daphalapurkar (T-3), Mack Kenamond (XCP-1)

Existing numerical algorithms for normal contact in Lagrangian applications have proven versatile in 2D and 3D problems of granular simulations. The method for tangential contact needs further consideration for handling various shapes of granular particles and expansion to 3D. Programmed in FORTRAN, the ideal student would have interest and experience dealing with contact mechanics and proficiency in programming and numerical algorithms.

Fast numerical algorithms for simulating detonation waves

Tariq Aslam (T-1), Eduardo Lozano (T-1), John Rose (T-1)

The engineering design of high-explosive components requires fast and efficient computational methods. A new modeling framework based on hyperbolic PDEs is being developed to simulate the propagation of detonation waves in complex 3D geometries but the best choice of numerical method remains uncertain. Working closely with the modeling team, the selected candidates will (1) evaluate the performance of narrowband level set, Fast Marching (FMM), Fast Sweeping (FSM), and Fast Iterative (FIM) methods and (2) develop the necessary tools to parallelize the selected algorithms in LANL’s HPC machines using CPU or GPU hardware. The outcome of this project will be a fast wave propagation methodology capable of simulating detonation waves in complex engineering components.

Improving equation of state generation via anharmonicity and uncertainty quantification

Danny Rehn (XCP-5), Mark Mathis (XCP-5), Sabrina Li (XCP-5)

Equations of state (EOS) are important for understanding basic materials properties and for providing detailed materials property information in hydrodynamics simulations. One major challenge area for EOS is to incorporate anharmonic vibrational effects at high temperature. Our project will explore the use of free energy perturbation theory (FEPT) methods to accurately calculate anharmonic effects in materials using first-principles calculations based on density functional theory (DFT). We will also perform uncertainty quantification (UQ) using Markov Chain Monte Carlo (MCMC) and related methods to assess model uncertainty in fitting to the DFT data generated. This project will provide an opportunity for students to learn the basics of DFT, EOS, and UQ methods, with flexibility to focus on specific aspects that are of highest interest. We will focus on materials of interest such as aluminum and tantalum, with the goal of using the methods developed to inform on isobaric, shock, and diamond anvil cell experiments.

Plasma turbulence and particle acceleration in 3D magnetic reconnection

Xiaocan Li (T-2), Fan Guo (T-2), Hui Li (T-2)

Magnetic reconnection is a fundamental plasma process where magnetic energy is converted into plasma kinetic energy and accelerating high-energy particles. This process is thought to be responsible for producing energetic particles during Earth's geomagnetic storms, solar flares, and various astrophysical events. Recent research has shown that in 3D, magnetic reconnection can naturally generate plasma turbulence, which adds complexity and raises new questions about how particles are accelerated in these conditions. One of the main challenges in studying these processes in 3D is that they demand tremendous computational resources and data-intensive analysis. In this project, we will use LANL's advanced particle-in-cell code, VPIC, and its hybrid version, hybrid-VPIC, to study how turbulence develops and how particles gain energy during 3D magnetic reconnection. Students involved in this project will get opportunities to run large-scale simulations on flagship computers, perform advanced data analysis, and learn about the interactions between magnetic reconnection, turbulence, and particle acceleration.

Self-Consistent Relativistic Electron Scattering using the Sherlock Scattering Model for X-ray Diagnostics

Tyler Markham (XCP-1), Edward Norris (XCP-1)

The LANL Lagrangian Shock Hydro code currently is interested in expanding its scattering capabilities for the purposes of x-ray diagnostics. The current model is not self-consistent for slightly relativistic particles in electron beam-target interaction physics. The student would implement new self-consistent Braams-Karney potentials into the code to calculate this interaction correctly, which requires developing a new integration scheme coupled to the transport of charge particles.

Modelling of optical properties in Hot Dense Matter

Tan Hoang Bao Tran (XCP-5), Charles Starrett (XCP-5),  Jackson White (XCP-5)

Hot dense matter (HDM) is a plasma state that exists at the intersection of traditional plasma physics and condensed matter physics, presenting unique challenges due to its extreme density and temperature. In HDM, classical treatments that view charged particles as point-like objects fall short, and the disordered nature of the plasma complicates the application of conventional solid-state methods. Yet, accurate modeling of HDM’s radiative, thermodynamic, and transport properties is critical for applications in astrophysics, nuclear fusion, weapon physics, materials science, and developments in many-body quantum theory. This project focuses on modeling optical properties of HDM. The aim is to extend our existing code to incorporate various plasma effects on observable spectra of HDM.

Material Model Calibration with Integrated Experiment Data

JeeYeon Plohr (XCP-5), Lauren VanDervort (CCS-6)

Material models of strength and damage are crucial components in a multi-physics simulation. Such models are required in a wide range of applications, such as structural engineering, astrophysics, and fusion reactor design. Calibration of these models is challenging since the experimental data being used have imprints of strength and damage but do not yield direct measurements of them. We use machine-learning aided methods (emulator) to utilize various types of experimental data, which enhances the predictive capability of the models and reduces the uncertainty. This will involve running multi-physics simulation codes to generate training data for the integrated experiments and using the Bayesian inference framework to provide the uncertainty estimation as well as point estimation for the parameters.

Efficient prediction of melting curves from atomistic simulation of liquids

Alfred Farris (XCP-5), Jerome Daligault (XCP-5)

The melting curve of materials, i.e., the locus of temperatures at which a solid material melts as a function of pressure, is of great significance in many areas ranging from production engineering to planetary sciences. Reliable and consistent measurements and theoretical predictions remain very challenging or impossible at the extreme conditions of pressure and temperature typically reached in certain applications. Likewise, under extreme conditions, materials often exhibit unexpected and complex modifications in their electronic and ionic structures that compound the difficulty of theoretically predicting the melting conditions. Our team is exploring a new, original, and promising simulation method to determine the melting curve of a material. Unlike all existing theoretical methods, our method does not require any input knowledge about the solid phase into which the material freezes. Requiring simulations of only the liquid phase is a significant advantage as the solid phase is not known a priori beyond experimentally achievable conditions. The student will contribute to the development and validation of the new method. They will perform molecular dynamics (MD) simulations (quantum MD or classical MD with machine-learning potentials) to determine the melting curve of several materials and compare to existing experimental results. They will learn how to use state-of-the-art simulation codes (VASP, LAMMPS), automize the simulation process, and extract the relevant data from the simulation output.

Algorithm Development for Arbitrary Lagrangian Eulerian Contact and Sliding

Nathan Vaughn-Kukura (XCP-1), Misha Shashkov (XCP-4)

Lagrangian contact algorithms allow materials to collide, slide, and separate throughout a simulation.  This enables significantly higher fidelity simulations in cases where materials in contact slide, rather than stick, and is widely used in Lagrangian and Arbitrary Lagrangian Eulerian (ALE) simulations at LANL. There are two limitations of Lagrangian contact algorithms that we will address with this project. First, Lagrangian contact algorithms are limited to the free surfaces of meshes, which does not accommodate cases where materials severely deform and require ALE techniques. Second, the Lagrangian contact algorithms operate on multiple disjoint meshes, which complicates and/or prohibits coupling the Lagrange contact hydrodynamics to the full range of multiphysics simulation capabilities supported by the code. In this project we will explore and develop ALE contact algorithms which can operate on sub-zonal material interfaces, rather than exclusively at mesh free surfaces, and on one simply connected mesh. We will implement and test the algorithms in one of LANL’s ALE codes. These algorithms will accommodate large deformation and alleviate the multi-mesh complexity, enabling higher fidelity simulations of a wider class of problems involving contact and sliding.

Using Simulation to Optimize Pulsed Beam DIagnostic Design and Analyze Data from Pulsed Power Experiments

Christopher Johnson (NEN-2), Kevin Meaney (P-4)

Radiographic pulsed x-ray sources such as the Dual-Axis Radiographic Hydrodynamic Test Facility (DARHT), Cygnus, and the upcoming Scorpius are fundamental tools for the Science-Based Stockpile Stewardship Program. In any beam-on-target experiment where beam-target interaction products are measured to infer properties of the target, enhancing the understanding of the beam characteristics (energy, intensity, and timing) improves the fidelity of the measurement. The goal of this project is to design a system of Cherenkov radiator detectors (of varying densities and signal generation thresholds) to be fielded in the DARHT beam to measure the time-dependent x-ray energy spectrum. This concept places design constraints on the system that require a unique light collection system composed of wavelength shifters, chromatic filters, optical fibers, irises and lenses to harvest light from the detector and transport the signal to a remotely housed light sensor. This project will provide many opportunities for students to contribute by using radiation and optical simulation calculations both in the detector design and characterization phase and upon analysis of experimental data obtained with the detectors when fielded. Students will learn to build applications with the Geant4 simulation toolkit to couple radiation and optical transport, optimize the detector and light transport system design, and simulate parts of the pulsed-power experiments. The ideal candidate would have some experience with physics, radiation detection, C++, and the root data analysis framework.

Sensitivity analyses of simulations of fluid jetting from shocked metal surfaces

Bryan Kaiser (XCP-8), Jesse Canfield (XCP-4)

The students will run 2D simulations of shocked metals with grooves and bumps on the free surface of the metal. As the shock passes, the grooves and bumps eject mass at a much higher rate than the velocity of the free surface. The quantity of the ejecta is a nonlinear function of cavity geometry, shock loading, shock geometry, and other variables. The students will use already existing validated code to generate, run, and post-process large sets of Flag simulations over which the students will vary physical parameters (e.g., strength model parameters, shock loading) and/or numerical parameters (e.g., grid resolution, ALE scheme parameters) to generate sensitivity studies. The simulation sensitivity studies will be plotted against already existing experimental data to provide calibration information for physical parameters or to illuminate the sensitivity of the simulation output to numerical parameter choices. If successful, the sensitivities studies will be included in the section of a publication in which model parameter choices are defended. Students interested in turbulence modeling, baroclinic vorticity, Richtmyer-Meshkov Instability, secondary instabilities, equation of state modeling, and/or ejecta modeling are encouraged to apply.

Modeling optical properties of plasmas and warm dense matter

Joseph Kasper (XCP-5), Charles Starrett (XCP-5)

Understanding the properties of plasmas and materials at high densities and temperatures, such as those in stars or inertial confinement fusion remains an important field of research in atomic physics. Optical properties and equation of state are key physical parameters for physics models. In this project we will develop and apply approaches such as time-dependent density functional theory to model and evaluate these properties for warm dense matter. This will also provide an opportunity to implement or extend models in computer code and write a publication for a peer-reviewed journal.

Materials simulations under high impact

JeeYeon Plohr (XCP-5)

"Material response under extreme conditions is a challenging and interesting topic to explore. Examples include the structural engineering, fusion reactor environment, and astrophysical conditions.  Often the experimental data are not available in the pressure and temperature regimes of interest, and the predictive modeling plays an important role.

In this project, we will calibrate the material model parameters using the Bayesian framework and perform high impact, multi-physics simulations in the LANL's Lagrangian code, FLAG. Various loading scenarios as well as geometry will be studied and investigated.

Investigating the volumetric energy deposition Rayleigh Taylor instability with xRAGE

Brandon Wilson (XCP-8)    Filipe Pereira (XCP-8)    Ally Leffler (XCP-8)

The Rayleigh Taylor instability occurs when an unstable interface between a heavy fluid and a light fluid is subject to gravitational acceleration. In 2023, the volumetric energy deposition (VED) Rayleigh Taylor instability was observed by students in the Los Alamos Dynamic Summer School (LADSS). This was the first time this instability was observed experimentally. In this project, students will develop and perform simulations of the VED Rayleigh Taylor instability. Predictions will be compared to experimental measurements using verification, validation, and uncertainty quantification practices. If a parallel LADSS project is chosen, students will have the chance to work with experimentalists acquiring new measurements and inform experimental design for future VED Rayleigh Taylor experiments.

Triggering Kinetic Magnetic Reconnection

Ari Le (XCP-6), Adam Stanier (T-5), Fan Guo (T-2)

In space and astrophysical plasmas, energy may slowly be stored in the magnetic fields and plasma currents, and this energy is then explosively released when magnetic reconnection is triggered. Reconnection rearranges the magnetic field line topology and energizes the plasma in solar and astrophysical flares, geomagnetic storms in Earth’s magnetic tail, and disruptions in magnetic fusion devices. While reconnection releases the global stored energy of space plasmas, its trigger or onset typically depends on small kinetic-scale plasma physics dynamics. This project will use kinetic and hybrid (kinetic ion/fluid electron) versions of LANL’s high-performance particle-in-cell code VPIC to study the global-to-kinetic coupling of reconnection onset in multi-ion species plasmas.

Sub-grid modeling of reactive burn in high explosive materials

Josh McConnell (XCP-2), CJ Solomon (XCP-2)

Resolving the reaction zone of a detonation front is important for predicting the behavior of detonating high explosive (HE). The practicality of running high explosives calculations with a reactive burn model is limited by the resolution requirements of the chosen burn model.  A potential work-around for the restrictive resolution requirements typical of reactive burn models is to presume the form of the joint probability density function (PDF) of a reaction model’s input variables (e.g. pressure, burn fraction, temperature) and computing the expected burn rate by convolving the rate law with the joint PDF conditioned on one or more scalar variances. 

The  goal of this summer project is to apply a presumed PDF approach to a reactive burn model, such as the Wescott-Stewart-Davis model, and determine the efficacy of this approach. The first part of this project will consist of running fully-resolved simulations of a detonating HE and using the resulting data to form an appropriate form of the presumed PDF for the burn model’s input variables. After a form of the PDF is determined, the presumed PDF approach will be applied to the target burn model and tested by running computations on a coarsened mesh and comparing to fully-resolved simulations.

Reduced order modeling for RANS verification

Daniel Israel (XCP-4), Arvind Mohan (CCS-2)

Conventionally, RANS models are calibrated using a self-similar solution.  However, much of the available data is not taken at the asymptotic self-similar state, nor is this the regime in which the models are primarily used.  We have developed a new method of creating a reduced-order surrogate model to use for model calibration and validation.  This model takes the form of a dynamical system.  Students will post-process data to obtain trajectories for this system, and use advanced optimization tools to calibrate and validate RANS models.

Kinetic Simulations of Magnetic Mirror Fusion Devices

Blake Wetherton (XCP-6), Scott Luedtke (XCP-6), Alex Seaton (XCP-6)

Magnetic mirrors are the simplest geometry magnetic fusion concept, and new devices are being built inspired by the advent of high temperature superconducting coils and some promising stability results in recent experiments. Students will run simulations of magnetic mirrors in the fully-kinetic VPIC code and/or its fluid electron/kinetic ion counterpart, Hybrid VPIC, with a focus on the physics of sloshing ions. Sloshing ions are energetic (tens of keV) fusion fuel ions injected as a beam, and they are believed to increase fusion efficiency and to help stabilize the mirror. Both the effects of the sloshing ions on mirror stability and the evolution of the sloshing ion distribution will be investigated.

Material Modeling for Dynamic Experiments

Kendra Van Buren (XCP-8), Sean Smith (XCP-8), Saryu Fensin (MPA-CINT)

This project will explore modeling choices used to simulate the deformation of metal samples undergoing planar shocks. These experiments, known as the "Lens" experiments to describe how the plane-wave shock is generated, were performed at LANL's Proton-Radiography (pRad) facility. The simulations will be performed with FLAG, a lagrangian hydrodynamics code developed at LANL. Of particular interest is to gain a better understanding of how metal samples with defects deform under shock-loading, and how our computational tools capture these effects. Some of the modeling choices that could be explored are: material strength parameters, material spall parameters, and defect sizes in the sample. Details and scope of the computational study may be adjusted based on student's research interests.

Application of Machine Learning in Nuclear Data Evaluation

Hirokazu Sasaki (T-2)

Nuclear Data libraries, which contain information about the interaction of particles with nuclei, are carefully curated from experimental data and theoretical predictions. This data includes details about nuclear reactions, such as their reaction probability (cross section), decay yields, spectra of the outgoing particles etc. and are used to understand/predict the behavior of particles in nuclear systems, such as nuclear reactors, astrophysical processes, radiography, gamma-based interrogation techniques etc. As such, any inaccuracies and imprecision in the nuclear data gets propagated to the uncertainties in the application of interest. Until recently, most general purpose nuclear data libraries have utilized Bayesian approach to tune the theory model input parameters to fit the experimental data. In this project, the student(s) will explore machine learning based approach to combine the theoretical models with experimental data to come up with evaluated nuclear data for specific reaction channels and physics observables. Upon the successful completion of this work, the work will be published in peer reviewed journals.