Comparison of fatigue loading in Francis turbine runners using extreme values interpolation

Abstract

Extreme values of strain signals are important for hydroelectric turbine runners fatigue assessment. However, due to measurements limitations such as short data length and limited subset of measured operation conditions, the extreme values are rarely fully captured. Our study aims to estimate the extreme values of runner strain at non-measured operating conditions by interpolating the extreme components from the measured ones. The method is based on the peaks over threshold technique and the kriging interpolation. A case study with two similar Francis turbines (same design and power plant) is presented. The comparison is made between the use of independent interpolation models and a combined model for the two turbines. This helps assess the assumption that similar turbines share similar fatigue loading. If that is the case, a common model for the whole fleet of similar turbines design in a given facility could be considered, and thus contribute to reduce the uncertainties related to unmeasured loadings without the need for in-situ measurements on every runner.