Prof Chris Pearce & Dr Lukasz Kaczmarczyk

University of Glasgow, Computational Mechanics Group

Our research is focussed on the computational modelling of materials and structures, with particular focus on multi-scale mechanics and multi-physics problems, applied to problems ranging from safety critical structures to biomechanics, supported by EPSRC, EU, TSB and industry. The Finite Element Method (FEM) is an extremely powerful numerical technique for finding approximate solutions to a broad range of science and engineering processes that are governed by Partial Differential Equations (PDEs). It has revolutionised simulation and predictive modelling in science and engineering and has had a pervasive impact on industrial engineering analysis.

Despite the undoubted success of FEM, there is a continuous drive to push finite element technology beyond current capabilities, to solve increasingly complex real-world problems as efficiently as possible. Established commercial FE software can be relatively slow to adopt new technologies due to the dominance of out-of-date software architecture. Perhaps the greatest part of FE code development is expended in dealing with technical problems related to software implementation, rather than resolving the underlying physics. The biggest challenge is to create a computationally tractable problem, which can be solved efficiently while simultaneously delivering an accurate and robust solution by controlling the numerical error.

The presentation will first present the context of our research, briefly describing some examples from our projects, before looking at more details of our software development platform (MoFEM). This is a flexible and adaptable framework, while tackling the conflicting requirements of accuracy and computational efficiency. The catalyst for the creation of MoFEM was the need for a flexible and numerically accurate modelling environment for multi-physics problems, driven by the need of our industrial partners.