Hybrid modelling and digital twin development of a steam turbine control stage for online performance monitoring

Abstract

Increasingly tight integration of renewable power sources with thermal power sources leads to frequent operation at off-design modes of thermal power plants. It brings challenges including reduced efficiency, increased costs and safety issues in power plants. At a component level, load changes of thermal power plants are mainly carried out through actions of control stage systems of steam turbines. Accurate performance monitoring of control stage systems is key to enable faster and more flexible operation of thermal power plants thus provides opportunities to accommodate more renewable power with limited penalty to efficiency and safety. In this paper, a hybrid modelling method based on operation data and first-principle mechanism is proposed for performance monitoring of control stage systems. The hybrid modelling method features a) flow rate calculation of high-pressure control valves with limited measurement data, and b) flow and efficiency characteristics calculation over full working ranges of control stages. The proposed hybrid modelling method is validated via two case studies of a 330 MW subcritical steam turbine and a 1000 MW ultra-supercritical steam turbine, where digital twins of control stages are developed. Results show that average relative errors between simulated and measured values of exit pressure and exit temperature of control stages are within a range of one percent, providing great potential for plant-wide digital twin development and online performance monitoring.