抄録
With the increasing adoption of renewable energy, thermal power plants are expected to shift from baseload power sources to load-adjusting power sources. This change in operating environment may lead to temperature and strain variations during load fluctuations, including start-stop cycles, which can cause creep-fatigue damage in addition to creep damage due to internal pressure in large-diameter pipes of thermal power plants. As the damage mechanism may change with the operating conditions, it is crucial to develop a digital twin system that monitors temperature and strain during operation and estimates the remaining life of the components. In this study, we developed a system integrating data assimilation techniques with finite element method (FEM) to construct a digital twin for estimating the creep-fatigue life of high-temperature piping structures. We conducted numerical twin experiments using the developed system to estimate the state of a welded pipe structure subjected to internal pressure and bending loads, and verified the validity of the system. Additionally, we performed state estimation with measurement errors to demonstrate the robustness of the system against disturbances. When assuming errors in the measurement locations, the estimation errors increased compared to the case using simulated experimental data without errors. However, the system was able to estimate boundary conditions following the trend of the phenomenon. This trend was similar for both load estimation and creep-fatigue life estimation. When assuming errors in the measurement range, a similar trend was observed as in the case of measurement location errors, but with larger estimation errors. When random noise of up to 10% of the maximum strain was added to the pseudo-measurement data, the creep-fatigue life estimation was not significantly affected. However, when biased noise was added, the impact was relatively significant.
