Digitalisation of Ngatamariki Geothermal Power Plant

Abstract

With the commitment of the New Zealand Government to achieve carbon neutrality by 2050, the country is facing the challenge of replacing their coal and natural gas-derived electrical generation with renewable energy. One such method is increasing geothermal energy generation either by development of new plant, or optimization of existing plants. To assist in optimization, geothermal plants can be digitalised to assist the operator to predict working conditions of the plant. The ability to determine where loss in efficiency is occurring in the plant results in rectification becoming more seamless.
This paper discusses the methodology of taking large amounts of plant data of an existing plant and deriving a digital twin for Ngatamariki in the process simulator Symmetry. An initial set of real time data from an Organic Rankine Cycle plant were used to calibrate and calculate pressure drops and heat transfer coefficients across the plant components using regression that is built into Symmetry using different start and stop boundaries. These boundaries are important to allow regression on unit operation parameters to be more effective. An important aspect of this methodology is recognizing the importance of boundary locations. For example, inlets into pumps and turbines as typically boundary ends. This is due to the hard constraint of the material phase e.g liquid, or gas entering them. The results from using this approach are a quick and yet effective digital twin, with errors ranging from 0.5% to 4.5% absolute.