A Quantile Regression Approach to Modeling and Predicting Geothermal Well Drilling Costs
Loading...
Date
2025-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Kenyatta University
Abstract
Several factors influence cost of drilling a geothermal well. The most common ones consist drilled depth, type of drilling method used, drilling time, non-productive time among others. Accurate cost estimation is critical for a project’s planning and financial viability. In current practice, most drilling cost models estimate cost solely as a function of drilled depth. However, these models often overlook other critical factors such as drilling time and non-productive time that significantly influence drilling costs. Consequently, the models relied on do not explain the full range of variation in cost. Ordinary Least Squares (OLS) regression has been a widely used method for modeling drilling cost as a function of explanatory variables. However, the estimators derived from OLS are highly sensitive to outliers, which can significantly distort predictions and reduce the model’s robustness in the presence of non-normal error distributions. The objective of this study was to develop a robust model for estimating geothermal well drilling costs by incorporating key predictors that were previously overlooked using a quantile regression approach. The study accounted for the varying impact of predictors across different points of the cost distribution. This method offered a more comprehensive understanding of cost drivers and provided robust estimates that are less sensitive to outliers compared to traditional mean-based regression techniques like Ordinary Least Squares (OLS). Data from the Menengai geothermal project in Nakuru county was used in the study. The data comprised drilling data of 52 wells drilled between 2011 and 2019. The findings reveal significant correlations between drilling cost and both drilling time and non-productive time. Quantile regression analysis demonstrated that the impact of these covariates varies across the 0.25, 0.5, and 0.75 quantiles, with non-productive time exerting a more substantial influence on higher-cost wells. Compared to traditional Ordinary Least Squares (OLS) regression, quantile regression provides a more detailed understanding of the cost drivers. The model's coefficients for drilling time and non-productive time at different quantiles indicate that Drilling cost sensitivity varies, underscoring the importance of using quantile regression for more accurate and tailored cost estimations in geothermal drilling. The proposed model outperforms the traditional Ordinary Least Squares (OLS) approach, offering improved predictive power and more nuanced insights into cost determinants.
Description
A Research Project Submitted in Partial Fulfilment of the Requirements for the Award of the Degree of Master of Science in Statistics in the School of Pure and Applied Sciences of Kenyatta University, June 2025