WP 5: Adjoint Optimisation of Unsteady Flows

Lead: QMUL, participants: VW, RR, NTUA, INR, ESI, ENGS

Objectives

• Development of techniques to minimise number and size of checkpoints
• Implementation of checkpointing techniques in the adjoint solvers
• Application to small-medium size test cases of industrial relevance.

Methodology

Higher-order time-stepping techniques will be implemented at NTUA and QMUL to minimise the number of timesteps that need storing, image compression techniques will be implemented to minimise their size. The multi-resolution and HPC developments of WP 4 will be used to reduce image size an accelerate the computation. ESR 6 at ESI will develop POD data compression and extend the current discrete formulation to support this.

The checkpointing techniques available with Tapenade (INR) will be implemented in the C++ AD tool dco by ESR 11 at RWTH and in OpenFOAM by ESR 4 at ENGS. The unsteady techniques will be applied to turbo-machinery cases at RR to obtain adjoint stability, and to simplified external aerodynamic DES flows at VW.

Lead: QMUL, participants: VW, RR, NTUA, INR, ESI, ENGS [months 12 - 48]

Participating Research Fellows

• Mehdi Ghavami Nejad (ESR 8 at NTUA): OpenFOAM-based continuous adjoint method for aerodynamic optimization of unsteady turbulent flows. (WPs 2, 3, 5)
• Christos Vezyris (ESR 9 at NTUA): Discrete and continuous adjoint methods: GPU-enabled and one-shot adjoints, for unsteady flow optimisation (WPs 2, 3, 4, 5)
• Georgios Ntanakas (ESR 10 at RR) Regularisation of turbo-machinery flows with unsteady flow separations (WPs1, 3, 4, 5)
• Nikolaos Magoulas (ESR 13 at VW): Application of continuous and discrete adjoint OpenFOAM to unsteady car aerodynamics, Adjoint-based optimisation workflows (WPs 1, 2, 3, 5)
• Sheikh Razibul Islam (ESR 14 at WUT): Adaptation for unsteady adjoint optimisation (WPs 3, 4, 5)