Discrete-continuous optimisation of an axial flow blood pump

This paper presents results of discrete-continuous optimisation of an axial flow blood pump. Evolution Strategies (ES) are used as a global optimisation method in order to localise the optimal solution in relatively short time. The whole optimisation process is fully automated. This also applies to geometry modelling. Numerical simulations of the flow inside the pump is performed by means of the Reynolds-Average Navier-Stokes approach. All equations are discretised by means of the finite volume method and the corresponding algebraic equation systems are solved by the open source software for CFD, namely OpenFOAM. Finally, optimisations results are presented and discussed. The objective function to be maximised is simply pressure increase. The higher pressure increase the lower angular velocities required. This makes it possible to minimise the effect of haemolysis because it is mainly caused by high shear stress which are related, among others, to angular velocities.