Reducing computational effort of high head Francis turbines

The present work shows how a complete full-wheel model of a high head Francis turbine can be reduced directly by leaving out parts of the computational domain, but without losing accuracy in the RSI pressure that is of primary interest. Instead great savings in mesh size and consequently computational time is gained. Two main modifications have been performed: 1) Reducing the draft tube and 2) Excluding the distributor except for the guide vanes. This is performed both on separate models as well as combining them obtaining an ultimate mesh reduction of about 60%. From this process it is shown that the volute could be replaced by a very simple set of inlet boundary conditions without affecting the results as well as simplifying the meshing procedure. A close to linear speed-up with mesh size is obtained. For comparison, simulations performed on passage models have also been included and all results are validated against full-wheel transient simulations as well as experimental data. The work has been performed as part of the HiFrancis project at the Norwegian University of Science and Technology and the Norwegian Hydropower Centre.