Identifying a hydraulic-turbine model from measured field data

An approach is proposed for identifying hydraulic-turbine model parameters from measured response data. The frequency response from the turbine-gate position to the machine power is measured at given steady-state operating conditions, and a structured model is fitted to the response. The fitting process is based on a nonlinear optimization method that provides numerical robustness. This approach allows one to fine tune and validate existing models or to identify a model without initial theoretical analysis. Theoretical development of the approach is discussed and results are presented for a 115-MVA unit at the Mt. Elbert pumped-storage powerplant operated by the Bureau of Reclamation