469324 Data-Driven Modeling of Gas Leakage from Shale Natural Gas Wells

Wednesday, November 16, 2016: 1:42 PM
Monterey II (Hotel Nikko San Francisco)
Shobhit Misra and Michael Nikolaou, Chemical & Biomolecular Engineering, University of Houston, Houston, TX

Availability of large amount of data recorded during natural gas well construction and completion stages can be put to use for studying underlying reasons for some of the critical problems which occur during the production phase. One such problem is natural gas leakage from the cemented annulus of the gas wells. The gas leakage may severely pollute the environment and underground water aquifers. Therefore, it is important to understand the reasons responsible for gas leaks and subsequently seek solutions to limit such leakage from wells drilled and completed in the future.

A number of factors may cause poor cementing, which may lead to development of cracks and micro-channels in the cement, providing a continuous path for gas to migrate from the production zone to the well head (Nelson, 2006). The extent of gas leakage from the cemented annulus is measured by a parameter called Sustained Casing Pressure (SCP) which is simply the pressure exerted by gas in the cemented annulus at the well-head. In a previous case study that we conducted on a set of gas wells data, we developed linear models to identify factors causing high SCP (Wehrens, 2011). Although the linear models provided a reasonable approximation of the underlying cause-and-effect relationships, the model accuracy was somewhat restricted by nonlinearities existing in the process (Hastie et al., 2009). In this paper we present a nonlinear modeling approach to study underlying causes of high gas leakage. A number of data-driven nonlinear modeling approaches, revolving around dimensionality reduction, as well as model specifics will be included in the final presentation.

References:

  1. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning: Data Mining, Inference, and Prediction (2nd ed.).
  2. Nelson, E. B. (2006). Well Cementing (2nd ed.).
  3. Wehrens, R. (2011). Chemometrics with R: Multivariate Data Analysis in the Natural Sciences and Life Sciences. New York: Springer.

Extended Abstract: File Not Uploaded
See more of this Session: Modeling and Computation in Energy and Environment
See more of this Group/Topical: Computing and Systems Technology Division