Quantifying the Impact of Load-Following on Gas-fired Power Plants

Due to rapid penetration of renewables into the grid, natural gas combined cycle (NGCC) power plants are being forced to cycle their loads more frequently and rapidly than for which they were designed. However, the impact of load-following operation on plant efficiency and equipment health are currently poorly understood. The objective of this work is to quantify the impact of load-following on the gas-fired plants by developing high-fidelity multi-scale dynamic models. There are four main tasks in this project. First, dynamic model of an NGCC power plant has been developed. The main components of the NGCC plants are the gas turbine (GT), heat recovery steam generator (HRSG), and steam turbine (ST). The second task focuses on one of the undesired phenomena known as ‘spraying to saturation’ being faced by the NGCC plants during load-following, where the attemperator spray leads to saturation at the inlet of superheater and/or reheater causing damage and eventual failure of the superheater and/or reheater tubes due to two-phase flow. Different configurations of NGCC plants and operation strategies that can not only eliminate ‘spraying to saturation’ but can maximize the plant efficiency have been developed and evaluated. The third task focuses on modeling the unprecedented damages to the boiler components due to rapid load-following, which is leading to higher operation and maintenance (O&M) costs. Stress and wear models have been developed by accounting for creep and fatigue damages in key HRSG components. Multiple locations at the component junctions have been monitored and the most stressed part has been identified as the constraint in the dynamic optimization of the load-following operation. A multi-objective dynamic optimization algorithm has been developed for maximizing plant efficiency and minimizing deviation from desired ramp rates while satisfying operational constraints such as those due to stress and wear. The fourth task focuses on developing reduced order models. Since the modeling domain of interest includes multiple time scales and multiple spatial scales, it can be computationally intractable to use the iii detailed models for optimization/scheduling/control. Therefore, reduced order dynamic models have been developed for the NGCC system including the health models so that they can be computationally tractable for being used in dynamic optimization while providing desired accuracy