Invited: Holistic Physics-Based and AI-Enhanced Digital Twin for Large-Scale HIP Manufacturing

Holistic Physics-Based and AI-Enhanced Digital Twin for Large-Scale HIP Manufacturing

Simtec Soft

Abstract

Hot Isostatic Pressing (HIP) is a critical manufacturing technology for producing high-performance components in aerospace, energy, and advanced engineering sectors. However, scaling HIP to large components introduces significant challenges, including non-uniform thermal fields, complex gas flow behavior, and limited capability for direct measurement of key process variables. The HIP process involves strongly coupled multi-physics phenomena¡ªheat transfer, fluid flow, pressure evolution, and material deformation¡ªwhich are difficult to control using traditional empirical approaches.

This presentation introduces a holistic digital twin (DT) framework for industrial HIP, built on first principles and high-fidelity, fully coupled physics-based modeling to enable predictive, scalable, and efficient manufacturing of large-scale components. High-fidelity 3D Computational Fluid Dynamics (CFD) simulations are employed to model the complete HIP system¡ªincluding furnace, heaters, gas, and workload¡ªacross the full processing cycle. The approach captures transient temperature fields, gas flow dynamics, pressure evolution, and coupled thermo-mechanical behavior, showing strong agreement with experimental observations.

Unlike conventional methods, the DT computes full-field physical states, addressing the fundamental limitation that key variables, particularly temperature, are not readily measurable in large-scale HIP systems. This enables deeper insight into thermal gradients, hot spots, and process non-uniformities that influence densification, deformation, and final material properties.

While AI-driven surrogate modeling and optimization are still under development, their future integration is expected to enhance the DT toward faster prediction and real-time process optimization. Case studies of full-scale industrial HIP furnace simulations demonstrate the capability to reproduce real process behavior, identify critical non-uniformities, and support optimization of heater strategies, load configurations, and process cycles. This work highlights a shift from empirical, trial-and-error practices toward physics-based predictive manufacturing, enabling more reliable, efficient, and scalable HIP production, and establishing a foundation for next-generation digital transformation in HIP manufacturing.