Development of an Advanced Monitoring Application for the Power and Efficiency of Air-cooled Geothermal Power Plants

Power generation from renewable energy resources becomes increasingly important, as CO2 emissions have to be significantly reduced for climate change mitigation. Several geothermal plants with an Organic Rankine Cycle have been built in the South Bavarian Molasse Basin in the last ten years. They produce electrical power, independent of solar radiation and wind, with a high annual full load hour percentage and comparatively low CO2 emissions. However, since the geothermal plants are operated with air condensers and scaling affects the performance of the electrical submersible pump and thus the lifted thermal water flow rate, the generated gross electrical power is strongly fluctuating and few operating points can be determined within a year with the same boundary conditions. Hence, the analysis and evaluation of the time-dependent course of the gross electrical power output proved difficult for operators of geothermal plants so far. Moreover, the gross electrical power output is a crucial key performance indicator for monitoring the entire process of Organic Rankine Cycle power plants in geothermal applications. For advanced monitoring of the gross electrical power of geothermal plants, an empirical simulation model based on linear regression is developed and will be presented in this paper. The ambient air temperature of the location of a geothermal power plant as well as the transferred heat to the power cycle by heat exchangers are the two variables of a two-dimensional polynomial function of the simulation model, whose regression coefficients are computed numerically. For this purpose, operating data from a four-year period of a geothermal power plant with an Organic Rankine Cycle in the south of Munich (Germany) were pre-processed and used. Polynomial functions of various degrees, different objective functions and varying input data sets for the numerical computation of polynomial coefficients with linear regression are examined. Regression models for four years of operation of the investigated geothermal ...