JAX-FEM-ANISO Differentiable GPU-Accelerated Finite Element Framework for Inverse Identification of Finite-Strain Anisotropic Plasticity

Jul 16, 2026

Upload by Aurora

Calibrating anisotropic plasticity models is crucial for high-fidelity metal forming, welding, and additive manufacturing simulations, yet conventional workflows demand large experimental campaigns, costly forward FEM runs, and poorly scaling finite-difference sensitivities. The JAX-FEM-ANISO framework provides end-to-end differentiable, GPU-accelerated finite element forward simulation and inverse parameter identification for finite-strain anisotropic plasticity. By parallelizing key FEM bottlenecks on NVIDIA GPUs, it delivers up to 9.4× speed-up versus a CPU Abaqus baseline. Automatic differentiation yields consistent Jacobians and accurate gradients for PDE-constrained inverse analysis, avoiding step-size sensitivity and reducing cost. Using information-rich heterogeneous specimens with full-field displacement data, it identifies advanced constitutive parameters from a single test and recovers anisotropic yield and hardening parameters under increasing difficulty.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages46

Reading Time1 h 56 min

Introduction

Motivation: anisotropic plasticity calibration challenges

Experimental data and inverse identification approaches

Material Testing 2.0 paradigm and identification strategies

arXiv :2606 . 17390v2 [ cs .CE] 13 Jul 2026 JAX-FEM-ANISO: Differentiable GPU-Accelerated Finite Element Framework for Inverse Identification of Finite-Strain Anisotropic Plasticity Deepak Sharmaa, Itzel Salgadoa, Lu Huangb, Hui-Ping Wangb, and Jian Caoa,∗ aDepartment of Mechanical Engineering, Northwestern University, Evanston, 60208, IL, USA b General Motors, Warren, 48090, Michigan, USA ∗ Corresponding author e-mail: [jcao@northwestern.edu](jcao@northwestern.edu) Abstract Calibrating advanced anisotropic plasticity models is essential for high-fidelity simulations of metal forming, welding, and additive manufacturing. Conventional workflows, however, often require extensive multi-test experimental campaigns, computationally intensive forward simulations, and finite-difference sensitivity calculations that scale poorly with the number of material parameters. We present a fully differentiable, GPU-accelerated finite element framework JAX-FEM-ANISO, for forward simulation and inverse parameter identification of finite-strain anisotropic plasticity. Built on JAX-FEM, the framework exploits modern accelerator architectures by parallelizing the three major computational bottlenecks in nonlinear FEM: elemental weak-form and tangentstiffness evaluation, global sparse matrix assembly, and sparse linear solution. For a large-scale forward problem with 3 million degrees of freedom, JAX-FEM-ANISO on a single NVIDIA H100 GPU achieves up to 9.4× speed-up over a 24-core CPU Abaqus baseline. Automatic differentiation is applied through the constitutive update and solver workflow, providing consistent Jacobians for complex constitutive models without manual derivation and accurate gradients for PDE-constrained inverse analysis. Compared with finite differences, the JAX-AD gradients avoid step-size sensitivity and provide the required sensitivities at substantially lower computational cost. For inverse characterization, we combine information-rich, topology-optimized heterogeneous specimens with full-field displacement data to identify advanced constitutive model parameters from a single test, replacing what would otherwise require many conventional experiments. We demonstrate accurate recovery of anisotropic yield and hardening parameters in progressively challenging settings, including uniform and spatially varying material properties. The resulting AD-based formulation enables efficient optimization in high-dimensional parameter spaces where finite-difference approaches are computationally infeasible. These results establish differentiable, GPU-accelerated FEM as a practical high-throughput engine for simulation, characterization, and optimization workflows in advanced manufacturing. Keywords Inverse material parameters indentification · Finite-strain anisotropic plasticity · Differentiable simulations · JAX-FEM · PDE-constrained optimization · GPU acceleration 1 Introduction Accurate simulation of metal forming, welding, and additive manufacturing requires computational models that capture the geometric and material nonlinearities inherent in finite-strain plasticity [1, 2] . Anisotropic plasticity models, such as Hill–48 [3], Barlat’s Yld2000-2d [4], and Yld2004-3D [5], are widely used in automotive and aerospace applications to simulate complex sheet-metal forming processes. These constitutive models account for directional variations in yield stress and plastic flow that develop during manufacturing, enabling more accurate predictions of springback, formability limits, and final part geometry. Their implementation and calibration remain challenging because finite-strain anisotropic plasticity couples geometric and material nonlinearities [6] . Practical finite element implementations therefore require iterative Newton–Raphson schemes at both local material-point and global structural levels, together with consistent tangent operators and robust return-mapping algorithms for history-dependent internal variables

FAQs

What problem does JAX-FEM-ANISO address in anisotropic plasticity calibration?

It targets the high cost and poor scaling of conventional calibration workflows that rely on many experiments, expensive forward simulations, and finite-difference sensitivity calculations for finite-strain anisotropic plasticity.

How does the framework achieve differentiability and correct gradients?

It applies automatic differentiation through the constitutive update and solver workflow, producing consistent Jacobians for complex constitutive models used in PDE-constrained inverse analysis.

How is GPU acceleration used to improve performance compared with a CPU baseline?

JAX-FEM-ANISO parallelizes the main computational bottlenecks in nonlinear FEM—element weak-form/tangent evaluation, sparse matrix assembly, and sparse linear solving—achieving up to 9.4× speed-up on a single NVIDIA H100 GPU versus a 24-core CPU Abaqus baseline.

Jul 16, 2026

Upload by Aurora