The IODA project is curating a number of benchmarks for adjoint-based shape and topology optimisation methods, complementary to those of the AIAA ADODG. Four testcases are proposed:
- VKI U-Bend (VKI): a U-shaped hairpin bend cooling channel of a turbomachinery blade is to be optimised for minimal pressure loss. 3-D case with either shape or topology optimisation. Case proposed by T. Verstraete, VKI
- VKI LS89 cascade (VKI): 2-D shape optimisation of a turbomachinery cascade. Case proposed by T. Verstraete, VKI
- Berlin compressor (TU Berlin): 3-D shape optimisation of a turbomachinery compressor stator blade. Case proposed by M. Meyer, Rolls-Royce D.
- Drivaer vehicle (LSTM, TU München): 3-D shape optimisation of the side view mirror of the Drivaer vehicle (smooth underbody, closed engine bay, closed wheels, stationary floor, with mirror). Two objective functions are considered; Variant A: minimise drag on the side view mirror; Variant B: minimise the acoustic signature on the side window measured by the integral of the turbulent dissipation of a RANS model in a defined volume box. Case proposed by C. Othmer, Volkswagen AG.
Benchmark events are held at regular intervals at major EU conferences, the last event was the final conference of the About Flow project, Munich, 2017. Participants in the benchmarks are expected to present:
- A comparison of their flow simulation to the provided exp. data and a provided benchmark CFD simulation, initial values for objective functions and constraints.
- A description of their parametrisation approach and number of degrees of freedom used.
- A description which optimisation algorithm is used. If sensitivities are used, a documentation of these sensitivities along prescribed sections.
- Values for optimised objective and constraints, computational cost normalised by the cost of the flow simulation, profiles of the optimised shape along specific sections.
- A discussion of their results, in particular which flow alterations due to the optimisation are responsible for the improvement in objective function.
Further details can be obtained from Dr. Jens-Dominik Müller (firstname.lastname@example.org) .