A non-intrusive approach to the proper generalised decomposition for flow problems in OpenFOAM

LaCàN Seminar Series in Computational Science and Engineering

Vasilis Tsiolakis

June, 22th, 2018 at 12:00 Add this item to your iCal calendar

Vasilis Tsiolakis received his Diploma in Mechanical Engineering in the Mechanical Engineering Department of the National Technical University of Athens (NTUA) and specialised in “Air and Ground Transport Vehicles”. He is now a PhD candidate working in Volkswagen Group Research, with primary academic institution: UPC and secondary: Swanswa University. His works aims at introducing reduced order modelling in the daily industrial environment to allow for the real-time aerodynamic design and optimisation of cars.





Computational fluid dynamics (CFD) is an integral part of the automobile industry. However, the aerodynamic optimisation process is known to introduce a major bottleneck in the design cycle, due to the necessity to test a large number of configurations. Using reduced order modelling (ROM) techniques, a methodology to introduce real-time aerodynamic design would lift this bottleneck by compacting, after an initial computational investment, all steps of the optimisation cycle into one. OpenFOAM has become an industry standard for CFD. As such, in this work, a methodology to implement the proper generalised decomposition (PGD) for parametric flow problems in OpenFOAM is proposed, by considering parameters of the problem (e.g. the physical properties of the fluid) as extra dimensions of the generalised solution. PGD has gained popularity due to its ability to compute a reduced basis with no a-priori knowledge of the problem. However, compared to other ROM techniques, its implementation usually requires modification of the core routines of an existing solver. This work, as part of a research aiming towards a daily industrial application, presents a non-intrusive approach for the implementation of the PGD for the numerical solution of flow problems using OpenFOAM. Particular emphasis will be given on the formulation proposed to enable the non-intrusive implementation. Numerical examples will be used to demonstrate the potential of the proposed methodology for solving flow problems where the viscosity is considered an extra coordinate of the generalised solution. This project is part of the Marie Sklodowska-Curie ITN-ETN AdMoRe funded by the European Union Horizon 2020 research and innovation program with grant number 675919.