The increasing deployment of renewable energy sources necessitates peak regulation services from thermal power plants, impacting their energy efficiency. Central to these plants, the steam turbine system significantly influences their operational efficiency. A digital twin model of this system was developed, integrating mechanism-driven and data-driven modeling methods. The neural network data-driven approach was specifically utilized for parameters such as feedwater pump speed and steam flow rate to the pump turbine. Other parameters were modeled with mechanism data hybrid driven modeling method. This model computes vital metrics such as low-pressure turbine exhaust steam enthalpy, work done and heat absorption per unit mass of steam, system efficiency, feedwater mass flow rate, and water-coal ratio—key for evaluating and enhancing the system's energy efficiency. An investigation into a reference case showed a decline in efficiency below design levels due to aging. By optimizing the live steam pressure and the cold-end system, relative improvements in energy efficiency of 0.35 % and 0.14 %, respectively, were achievable.
Read full abstract