Data Analytics for Production Optimization in Unconventional Reservoirs

Abstract

Considerable amounts of data are being generated during the development and operation of unconventional reservoirs. Statistical methods that can provide data-driven insights into production performance are gaining in popularity. Unfortunately, the application of advanced statistical algorithms remains somewhat of a mystery to petroleum engineers and geoscientists. The objective of this paper is to provide some clarity to this issue, focusing on: (a) how to build robust predictive models, and (b) how to develop decision rules that help identify factors separating good wells from poor performers. The data for this study come from wells completed in the Wolfcamp shale formation in the Permian Basin. Data categories used in the study included well location and assorted metrics capturing various aspects of well architecture, well completion, stimulation, and production.Predictive models for the production metric of interest are built using simple regression and other advanced methods such Random Forests, Support Vector Regression, Gradient Boosting Machine and Multidimensional Kriging. The data fitting process involves splitting the data into a training set and a test set, building a regression model on the training set and validating it with the test set. Repeated application of a "cross-validation" procedure yields valuable information regarding the robustness of each regression modeling approach. Furthermore, decision rules that can identify extreme behavior in production wells (i.e., top x% of the wells versus bottom x% as ranked by the production metric) are generated using the classification and regression tree algorithm. The resulting decision tree provides useful insights as to what variables (or combinations of variables) can drive production performance into such extreme categories.The main contributions of this paper are to provide guidelines on how to build robust predictive models, and to demonstrate the utility of decision trees for identifying factors responsible for good versus poor wells.

Data-Driven Analyses of Low Salinity Waterflooding in Carbonates

“…The DT method is a supervised ML algorithm which is based on a group of nodes called root, decision, and leaf nodes [66]. Each leaf represents a numeric value for an important independent variable [67]. The SVM model is also a supervised learning method that is used in handling regression and classification problems [67].…”

Machine learning on field data for hydraulic fracturing design optimization

“…Quite good results were shown by the authors of the article [14]; they also investigated various models. But as noted in the article from 476 wells, only 171 have no NaN values.…”

“…In [14] the authors developed a decision procedure to separate good wells from poor performers. For this purpose, the author investigated Wolfcamp well dataset.…”

Machine Learning-Based Production Prediction Model and Its Application in Duvernay Formation

“…where p is the number of predictor variables, is the noise terms, it is assumed that noise terms are uncorrelated and have independent and identical normal distributions with mean zero and constant variance [19,25,26]. The noise terms represent some variables that not include but may have contribution to Y.…”

“…The aim of SVR is to find a function f (x) that diverge from response y by a value no more than , and in the meantime the function f (x) as even as possible (Figure 2). More details about SVR are described in Al-Azani et al [28], Da Silva et al [29], El-Sebakhy et al [30], Li et al [31] and Schuetter et al [26]. In SVR, the choice of kernel function directly determines the performance of the model, the main kernel functions used in this study are shown as follows:…”