This study presents a thermo-hydro-mechanical reservoir model of the Chingshui geothermal field, coupled with a thermodynamic model of the binary cycle. A one-stage axial turbo-expander was designed and analyzed in three dimensions using ANSYS-CFX. The results were used to train the artificial neural network model, providing the mass flow rate of the working fluid, isentropic efficiency (eta is,exp), and shaft power of the turbo-expander under off-design conditions to the thermodynamic model of the binary cycle. When the mass flow rate of the geofluid (m(center dot)geo) exceeded 30 kg/s, eta is,exp gradually increased from its lowest value over the operation time and approached the design value. The performance of the power cycle degraded substantially owing to the decrease in the production temperature and considerable increase in the power consumption of the injection pump ( W(center dot)geo Owing to effect of W(center dot)geo p , the decrease ratios of the first- and second-law efficiencies at m(center dot)geo = 40 kg/s from the 1st to the 30th year were 5.7 % and 4.2 % for the cycle, and 23.2 % and 21.9 % for the total system, respectively. Finally, six scenarios of decreased m(center dot)geo were examined. The decreased m(center dot)geo had a relatively small impact on the payback period, and significantly affected the electricity production cost.