Drilling data from an enhanced geothermal project and its pre-processing for ROP forecasting improvement

Drilling parameters are analyzed here to improve forecasting of the rate of penetration (ROP) in enhanced geothermal systems (EGSs). Data recorded during drilling a 4.2-km-deep well at a pilot EGS project in South Korea were analyzed. The greatly fluctuating ROP values were smoothed using a fast Fourier transform filter. Two drilling optimization methods (multiple regression and artificial neural networks) then evaluated the effect of smoothing: it improved ROP prediction in both cases, with over 90% correlation at relatively low degrees of filtering. A methodology to optimize the degree of smoothness for a given drilling data set is suggested.