Non-linear regression approach to ROP predicted in Geothermal well aerated-drilling at Field X, South Sumatera, Indonesia
Arjuna Sriwijaya, M. Taufiq Fathaddin, Pri Agung

Magister Petroleum Engineering Department, Universitas Trisakti, Jakarta, Indonesia

Abstract

Rate of Penetration (ROP) is an operation parameter in Geothermal wells drilling, Field X, Indonesia, for all eleven wells, which needs to predict, for drilling operations forecast plan. Many correlations have been invented to reveal correlations between Predicted ROP and actual ROP, with one approach is simulating ROP predicted versus Weight on Bits (WOB), Rotation per Minute (RPM) of drilling bit. One of the expected equations, have ROP predicted as functional equations ROP predicted = A x WOBB x RPMC . In addition, independent variables such as True Vertical Depth (TVD), Compressed air flowrate (CA), Cutting return removal flowrate at Total Loss Circulation (CRT), Hole Diameter (HD), Cutting return removal flowrate at Partial Loss Circulation (CRP), and Foam Flowrate (FM), are all expected for contributing to ROP predicted, besides WOB and RPM as mentioned.
To achieve the goal for getting equations correlating between ROP predicted versus the independent variables as mentioned, an attempt had been conducted to find the correlations, using non-linear regression, from SPSS (Statistical Product and Service Solutions) Software. Based on the assumed correlation equation, it is found that ROP predicted = A x WOBB x RPMC, multiplied with additional factor for the right-hand side, consisting of two variables out of TVD, CA, CRT, HD, CRP, and FM, each two have exponent factors D and E, respectively.
From the output simulation program SPSS analysis non-linear regression, summary tabulation for deviation error ROP predicted vs ROP actual data is made, comparison is made between deviation error based on each running simulated equations applied to all eleven wells as mentioned, and with deviation error based on each of 11 wells simulated with each of 11 ROP predicted equations, beginning with ROP predicted = A x WOBB x RPMC
From the summary tabulation as mentioned, and taking the best average percentage deviation error from all the eleven wells and from the eleven ROP predicted equations, it is got the final equation,
ROP predicted = 9.546 x 10-8 x WOB0.694 x RPM0.341 x CRT3.003

Keywords: ROP, predicted, non-linear, regression, deviation

Topic: Material Engineering