S16: Physics-grounded digital twins for computational mechanics: model updating, data assimilation and uncertainty
Co-chair:
Shady Adib (University of Hertfordshire; Imperial College London)
Keywords: Predictive Digital Twins, Computational Mechanics, Data Assimilation, Model Updating, Uncertainty Quantification, Physics-Informed Machine Learning.
ABSTRACT
Digital Twins are becoming increasingly important in computational mechanics as a means of linking numerical models with observations from physical systems. Their development is enabling computational models to move beyond offline simulation towards continuously updated representations capable of supporting state estimation, prediction and engineering decision-making. However, achieving reliable predictive Digital Twins remains challenging, particularly when models must operate under changing system conditions, incomplete or uncertain measurements, modelling errors and demanding computational requirements. Recent developments in data assimilation, computational model updating, reduced-order modelling, uncertainty quantification and data-driven methods provide new opportunities to address these challenges. At the same time, advances in physics-informed and physics-grounded machine learning are creating new ways of combining physical knowledge with monitoring and experimental data. These approaches offer considerable potential for improving the adaptability, computational efficiency and predictive capability of Digital Twins while preserving consistency with the underlying mechanics.
This mini-symposium will bring together researchers working on the computational methods required for the development of adaptive and predictive Digital Twins. Contributions are invited on topics including computational model updating, parameter and state estimation, data assimilation, system identification, reduced-order and surrogate modelling, Bayesian inference and uncertainty quantification, physics-informed machine learning, neural operators, hybrid physics–data-driven modelling, and verification and validation of Digital Twin predictions. Particular interest will be given to methods addressing incomplete, noisy or changing observations, sensor availability, observability and identifiability, model discrepancy, and the propagation of uncertainty through continuously updated computational models. Applications from across computational mechanics are encouraged, including structural and solid mechanics, structural dynamics, geomechanics, fluid mechanics, fluid–structure interaction, materials, manufacturing and other multiphysics systems. Both methodological developments and application-focused studies demonstrating the interaction between computational models and physical observations are welcome. The mini-symposium aims to provide a forum for discussing recent advances, outstanding challenges and emerging opportunities at the interface of computational mechanics, data-driven modelling and Digital Twins. It will encourage exchange between researchers developing computational methodologies and those applying them to monitored physical systems, supporting the development of more reliable, adaptive and predictive Digital Twin frameworks.
REFERENCES
[1] J.-S. Fan, M.-N. Xu, Y.-C. Li, Y.-F. Ding, Y.-F. Liu, B. Geng and X.-J. Liao, “Mechanics-based digital twin model for structural construction using 3D scanning and FE model updating”, Automation in Construction, Vol. 173, 106109, (2025).
[2] C. Zhang, Q. Chun, Y. Zhu, P. Li, Y. Lin and Q. Hua, “Digital twin technique for automatic damage detection of historical buildings with an adaptive model updating approach”, Mechanical Systems and Signal Processing, Vol. 234, 112831, (2025).
[3] L. Sun, Z. Hu, M.D. Todd and J. Zeng, “Generative artificial intelligence for Bayesian model updating in digital twins: a review and tutorial”, Structural and Multidisciplinary Optimization, Vol. 68, 249, (2025).
[4] D. Andrés Arcones, M. Weiser, P.-S. Koutsourelakis and J.F. Unger, “Model Bias Identification for Bayesian Calibration of Stochastic Digital Twins of Bridges”, Applied Stochastic Models in Business and Industry, Vol. 41, e2897, (2025).