Using Well Tests to Gain Operational Insights in 2016
Test rates provide insight well beyond the near term. One category of core insights provided by the DISCERN Energy platform consists of signals generated by initial well test data. These well tests offer the advantage of posing minimal delay with results often being available just 4-6 weeks after a well has been tested – frequently giving visibility to drilling results before quarterly results are announced. In contrast, production data is usually delayed at least 3-4 months, if not longer.
Also, production data on a well level may not be available (such as Texas where production is reported by lease), while well test data is specific to the well level.
However, investors often discount the value of initial well tests with the perception that 24 hours cannot predict longer-term well performance, or because of a fear that well tests can be manipulated. While the latter is undoubtedly true, our analysis shows that initial well tests often are, in fact, highly predictive of long-term well performance. Production data that is available as a new developing feature on the DISCERN Energy platform has let us examine the correlation between initial well test data and cumulative production.
In our study, while initial oil rates for all wells put on line in the Bakken in 2014 were only a weak predictor of 12-month cumulative oil production (r-squared = 0.27, Figure 1), we found that the predictive power was often very strong for individual companies or within focused regions. In Figure 2, we see that IP rates were a very strong predictor for Hess’ results throughout the Bakken with r-squared equaling 0.72. The predictive power was still solid though less pronounced for WLL and CLR across their respective portfolios (r-squared ~ 0.50, Table 1). For Continental, the predictive power varied by region with very strong correlation in McKenzie County (r-squared = 0.70, Figure 3) and a somewhat weaker correlation in Williams County (r-squared = 0.41, Figure 4).
In Billings County, the strong correlation reflected the presence of wells largely from Continental and Whiting, which both have strong correlations in that county (r-squared = 0.68 and 0.99 respectively, Table 1).
In contrast, correlations were poor for EOG across the board, and generally very poor for most operators in Mountrail county.
We think these differences from operator to operator, county to county, reflect the differing degrees of consistency of the well test conditions among operators and the variation of the geology within a county. An initial rate test is clearly too simple a metric to capture all these nuances, but at the same time what emerges is that the short 24-hour test is frequently an excellent predictor of longer-term performance.
Thus, although initial rate data may not always be useful in predicting long-term well performance, we view the data to be accurate on a frequent enough basis to warrant vigilant monitoring.
At DISCERN, we're fanatical about E&P data. To discover insights like this for yourself, learn more about DISCERN Energy. Just fill in your email address below to request a demo and begin building insight.