Abstract

Evaluation of the induced seismicity risk that potential saltwater disposal wells (SWD) pose is a common practice in the oil & gas industry and at state agencies that regulate such activity. As the state agency that regulates this industry in Texas, the Railroad Commission (RRC) assesses the induced seismicity risk of new SWD applications prior to issuing (or denying) a permit. It does so by analyzing variables that are thought to predict the likelihood of increased seismicity should that well become operational. In this study, we construct a dataset of 4616 Texas SWD wells and employ machine learning to assess the predictive strength of the model as a whole and of the individual variables that it contains. Results show that simple averaging of variables in the RRC model and employing an optimal threshold achieves a recall of ∼77% (i.e., approx. seventy-seven out of every hundred SWD wells that are labeled as being associated with seismicity are predicted by the model to be high risk prior to their operation). Additionally, use of these variables in a Random Forest machine learning model produces a similar recall of ∼76%. Results also show that, although all RRC variables have some predictive power, the strongest RRC variables are those that assess the presence of nearby faults (which is where induced seismicity occurs). Those with lower relative predictive strength are associated with proximal historical seismicity and total permitted disposal volumes in the area. Learnings from this analysis can be used to construct future variables and models that are more predictive of induced seismicity associated with SWD activity.

Introduction

The Permian produces well over 10 million barrels of water per day as hydrocarbons are extracted from unconventional assets like the Delaware and Midland basins (Bennett, 2023). The most common means of managing this produced water is to dispose of it in formations either above or below the producing formations through SWD wells. This process elevates the pore pressure and stress in such formations, which in turn can induce faults that encounter such changes to fail and become seismogenic. This is exemplified by the fact that the Permian Basin has experienced roughly 700 magnitude 3 (M3) seismic events over the last 5 years, according to the USGS and TexNet (USGS, 2024; TexNet, 2024). It is therefore important to understand the potential induced seismicity risk to that SWD wells pose prior to permitting and drilling them.

This content is only available via PDF.
You can access this article if you purchase or spend a download.