This paper uses data from 450 gas wells drilled over a recent 4 ½ year period in an organic-rich mudstone reservoir from a Northeastern State. Results demonstrate the development of successful multivariate models explaining 60% to 70% of the variance in production. Reflecting both geology and completion attributes, the models reveal the relative importance of various key variables as well as how the importance changes as the production accumulates. Because the models control for both geology and completion variables, they also allow for statistical tests comparing:
Production from different formations where costs to reach the formations are quite different (thus generating significant savings going forward);
Linear versus non-linear effects of lateral length;
Competing fracking approaches offered by major suppliers; and, several other such tests.
The general approach, using this particular data set as an example, can be used to objectively and concretely address a large number of critical managerial decisions. The two key contributions of the paper are:
the demonstration that principal components factor analysis can be used to capture the information across several correlated independent variables and render that information in orthogonal predictor vectors;
thus allowing the successful modeling of production and facilitating meaningful tests of managerial decisions.
First, does production increase linearly with lateral length or is there a "flattening out effect" beyond certain lengths? Second, consider two potential target zones within a Formation where one has significantly better average rock quality than the other. Due to operational challenges, however, the higher quality rock zone costs, on average, an additional ﹩200,000 a well more to drill. It also increases the risk of drilling complications creating rig downtime or sidetracking scenarios. Are there significant differences in production that warrant pursuing the more costly (and higher quality) reservoir rock over the stratigraphically adjacent lesser quality but cheaper drilling target? Third, controlling for geology and completion design, is there a significant difference in production associated with one frack supplier versus another? And, does one frack supplier's method work better in certain geological formations versus others? Other questions: are there optimal stage lengths, clusters per foot or perforations per cluster? Is there an optimal amount of proppant to use? Can you predict well productivity from mapped attributes prior to drilling any wells in a field by learning from past experience?