Abstract
As part of the design of a new particle accelerator at CERN, a
research is conducted to study the challenges and opportunities of
multi-stage turbocompressor machines operating with light gases
and more specifically with a mixture of helium and neon. First, a 1D
stage performance prediction model is implemented and coupled
with a genetic algorithm in order to generate an impeller database.
Then, a stacking method is developed considering design philosophies
and technological limitations observed in the industry. This
model is coupled with a second loop of the same genetic algorithm,
which provides multi-stage architectures optimised for either compactness,
i.e. number of stages, or efficiency. For both objectives,
an ideal number of stages can be determined which increases significantly
as the operating gas becomes lighter. The impellers diversity
within the database also plays an important role on the overall machine
architecture. Finally, in alignment with potential technological
improvements, the motor maximum rotational speed is varied to
study the achievable reduction in the required number of stages.
