The influence of manufacturing variations on product performance shows that
manufacturing uncertainties should be considered already in the development
phase. One way of doing this is by combining numerical optimization
and sensitivity analysis. This unification is referred to as Robust Design
Optimization (RDO). RDO methods can be applied to any type of product,
but the aim of this project is to develop and apply RDO methods for
turbomachinery applications that use inverse design methods. Existing
RDO methods are used to inspire new approaches that are tested on a
representative case. The RDO methods revolve around a CFD model created to
successfully facilitate RDO. Results show that while none of the tested methods
are successfully able to produce a more robust design than conventional
optimization in this case, designs performing similarly can generally have
different performance variations and hence sensitivities. In addition to this,
the hydraulic sensitivity of the blade has also been investigated through a
sensitivity analysis. This shows the trailing edge to be the most sensitive part
of the blade, and the tolerances should therefore be tighter at the trailing edge
in this design. Since the sensitivity analysis is performed on a single nominal
design, it cannot be extrapolated to apply to all impellers. Until more studies
are performed and potential general tendencies are identified, it is therefore
recommended to always investigate the hydraulic sensitivity of a design before
specifying tolerances.