A Predictive Model of Wellbore Performance in Presence of Carbon Dioxide in Kizildere Geothermal Field

Abstract

Typically, geothermal wellbore model is used to predict the production performance of wells using a wellbore simulator based on flow tests. An iterative procedure is used to calibrate NCG content. In this study, a predictive modelling approach harnessing the power of machine learning is proposed. Several deep well data in Kizildere Geothermal Field have been used to calibrate the model. The results are compared to flowmeter data attached to a mini separator. It has been observed that flowmeter NCG results are consistent with predictive modelling results in most of the wells. Since NCG measurements with mini separator are challenging, it is possible to predict NCG values for the wells without actual measurements at the wellsite.