MatAlytics Ltd | From Abaqus Simulation to Real-Time Structural Integrity Using Physics-Informed AI

We were honored to have Benedikt Engel from MatAlytics Ltd. present at the 2026 SIMULIA Americas Users Conference in Novi, Michigan, May 13-14, 2026.

Abstract: 

Across industries such as power generation, oil & gas, and heavy manufacturing, the integrity of high-value assets depends on accurately predicting stresses and material degradation under complex thermo-mechanical loading. Finite element analysis performed in Abaqus is the established gold standard for this task, enabling detailed evaluation of stress, strain, fatigue, and creep damage. However, despite its accuracy, conventional FE analysis is computationally expensive, time-consuming, and requires specialist expertise, limiting its use to offline studies and preventing real-time application during operation.

This presentation introduces CITRUS, a physics-informed AI framework that extends the value of Abaqus beyond offline simulation into real-time structural integrity assessment. High-fidelity training datasets are generated entirely using Abaqus, covering representative component geometries, material models, boundary conditions, and operating envelopes. These simulations produce full-field stress, temperature, and damage responses that form a rigorously validated physics foundation.

CITRUS leverages these Abaqus datasets to train physics-informed graph neural networks, ensuring predictions remain grounded in first-principles mechanics rather than purely data-driven correlations. Once trained, the model ingests live operational inputs, such as temperature and pressure histories, and instantaneously reconstructs stress states and fatigue and creep damage accumulation. Predicted results show agreement exceeding 98% compared to equivalent Abaqus analyses, while reducing evaluation times from hours to milliseconds.
 

 

Presenter: 

Benedikt Engel

CEO - MatAlytics Ltd

Dr-Ing. Benedikt Engel – Chief Executive Officer (CEO) and Co-Founder Co-founder and CEO of MatAlytics Ltd., with deep expertise in material behaviour, numerical modelling, and experimental validation. He holds a PhD in Engineering from the Technical University of Kaiserslautern, where he investigated the experimental and numerical behaviour of high-temperature materials in collaboration with Siemens Energy. Following this, he worked as a Postdoctoral Research Fellow at the University of Nottingham on a European research project with Rolls-Royce, developing models for crack growth behaviour in turbine discs. Benedikt specializes in high-temperature materials, including deformation behaviour, damage, and fatigue prediction for aerospace and power generation sectors, and has led major projects with Siemens, Rolls-Royce, and EDF. He has published his research in multiple international journals and delivered invited talks at leading conferences. As CEO, he combines technical and strategic vision to scale MatAlytics into a global leader of physics-informed AI solutions.