Analysis of Production History for Unconventional Gas Reservoirs With Statistical Methods

Abstract

Unconventional gas resources have dramatically changed the future energy landscape. Developing these resources involves substantial risk. Such risk can be mitigated if information gathered at initial stages of the development of a field is used efficiently and effectively to guide future development. A variety of tools-such as decline-curve analysis (DCA), type-curve analysis, simulator history matching, and artificial intelligence (AI)-is used to that effect. These tools accomplish partially overlapping but different tasks. Additional tools that could not only facilitate the analysis of historical data but also guide future development would be of value. In this work, we propose an efficient methodology that can use historical production data from existing wells to answer questions such as the following: Which wells will behave similarly? Which wells will behave differently from each other or from standard expectations? Which factors will contribute to these differences? How can data from existing wells be used to anticipate the performance of new wells? The proposed methodology relies on standard principal-component analysis (PCA) and principal-component regression (PCR). The application of PCA to historical production data from twelve wells in the Holly Branch field quickly identified wells with distinct behavior. The subsequent investigation of pressure and the completion data for these wells revealed reasons for such distinct behavior. Finally, a simple linear model was built with PCR, with good ability to predict production from new wells, as assessed through cross-validation. The value of the efficiency offered by the proposed methodology would be much higher for larger data sets, for which manual analysis of production data is more cumbersome.

Practical implementation of knowledge-based approaches for steam-assisted gravity drainage production analysis

“…This is achieved by orthogonally transforming the dataset into a new set of uncorrelated variables or principal components, which are computed from eigenvalue decomposition of the covariance matrix (Smith, 2002). The PCA technique has been successfully applied in areas including history matching (Sarma, Durlofsky, Aziz, & Chen, 2007;Yadav, 2006), reservoir property estimation (Dadashpour, Rwechungura, & Kleppe, 2011;Lee, Kharghoria, & Datta-Gupta, 2002;Scheevel & Payrazyan, 2001), and production data analysis (Bhattacharya & Nikolaou, 2013).…”

Functional Approach to Data Mining, Forecasting, and Uncertainty Quantification in Unconventional Reservoirs

“…Similar to the procedure outlined in Bhattacharya et al (2013) where multivariate PCA was used, functional principal component analysis also allows for production profile reconstruction, and dimensionality reduction. To reconstruct one production profile one has to sum products of principal components and associated scores and add such sum to the mean function computed on entire ensemble (fPCA, Ramsay et al 2005):…”

“…The regression based forecasting framework outlined in Bhattacharya et al (2013), is another concept that can be extended into a functional context. Assuming that the mean function, and functional principal components are reasonably estimated from existing wells (training set), all that is needed to forecast production at some new well location are the functional principal component scores.…”

“…This is often challenging since production time series come as noisy, discrete observations of production rates over time. Conventional approaches to this problem rely on parameterizing the system with decline curves and work in the parameter space of the assumed decline model, or even simpler work with the raw data (Bhattacharya et al 2013). Here we take alternative non-parametric approach where we use the data to find the most appropriate smooth and continuous representation of declining production time series.…”

Statistical, Data-Driven Approach to Forecasting Production from Liquid-Rich Shale Reservoirs

“…The ability to use PCA to extract common trends and patterns from sets of data has made it applicable to production forecasting as well. Bhattacharya and Nikolaou [12] used PCA to analyze production history from unconventional gas reservoirs but did not forecast future production. Makinde and Lee [13] used the Principal Components Methodology (PCM) to forecast production from shale volatile oil reservoirs and compared the results to compositionally simulated data and production estimates from different decline curve analysis (DCA) models.…”