Artificial neural network-based optimization of geothermal power plants

Abstract

In the world, due to limited energy resources, low efficiency of renewable energies, complex and costly energy conversion technology, and environmental pollution, human beings are trying to develop and improve innovative and efficient systems. Therefore, optimizing a complex system with optimization methods under real operating conditions makes it more efficient, economical, and environmentally beneficial. For simulation, monitoring, and failure prediction of a modern geothermal power plant (GPP), artificial neural network (ANN)-based models have proven to be very suitable, especially for existing systems without physical models. These models have good accuracy if they are trained with data-oriented, adaptive, rapid response, and appropriate data. This chapter introduces the ANN methodology to optimize the thermodynamic performance of a GPP. The mentioned methodology is performed on an existing binary GPP with low enthalpy. For this purpose, experimental data consisting of average hourly temperature, pressure, mass flow rate, and power data at certain reference points of the GPP are collected. A mathematical model including mass, energy, and exergy analyses is developed. As the output parameters of ANN, electricity production rate and exergy efficiency, which are the result of the model, are accepted. The optimization parameters have been created by selecting some parameters at the reference points of the GPP. Its power output and exergy efficiency are estimated by the ANN method using values from the GPP’s reference points. As a result, the best value of the selected optimization parameters is determined by the ANN method for estimating maximum power output and exergy efficiency. Thus, the effect and change of optimization parameters on the thermodynamic performance of the GPP can be observed.