Unconventional/complex reservoirs

Production Metrics To Predict Long-Term Performance of Unconventional Wells

In unconventional plays, comparing the effect of different completion designs or well-management strategies on well performance remains a challenge because of the relatively brief production history and lack of long-term field analogs of these plays.

jpt-2017-07-newfrontierhero.jpg
Getty Images.

In unconventional plays, comparing the effect of different completion designs or well-management strategies on well performance remains a challenge because of the relatively brief production history and lack of long-term field analogs of these plays. This study examines various production durations as potential candidates for reliable indicators of well quality. The results show that predictions of midterm performance begin to be reliable only near 180 days of cumulative well production. This study used actual daily production data to confirm that 30- and 90-day initial production are not correlated strongly to well actual performance in the 2-year range.

Introduction

It has been a common practice to quantify performance of unconventional wells through their 30-day initial production for purposes of information for the media and in investor presentations. The authors’ experience has been that the 30-day initial production often does not necessarily quantify a well’s potential. The goal of this study is to compare various production-duration periods to identify which can most accurately forecast longer-term production. This study tests different production durations by use of actual well-production history from the Utica play in Ohio, USA. The term “IP-nnn” in the paper refers to a specific number of days of nonzero production for the well; for example, “wellhead gas IP-30” refers to the total volume of the first 30 days of nonzero wellhead gas production for the well, whereas “condensate IP-90” refers to the total volume of the first 90 days of nonzero condensate production for the well.

Study Description and Assumptions

A data set of 676 unconventional producing wells for the Utica play was used to conduct this study. The authors began the study with several assumptions based on anecdotal observations from the early days of production in the play, with the goal of eventually using this data set to test the accuracy of their ­assumptions, which were as follows:

  • 30-day wellhead gas production is a poor predictor of long-term gas production.
  • 90-day wellhead gas production is a more-accurate predictor of long-term production than is 30‑day gas production.
  • 180-day gas production is not significantly better than 90-day gas production as a predictor of long-term gas.
  • 360-day gas production and 720‑day gas production are equivalent as gas-production-quality indicators.
  • Correlations of 30-, 90-, and 180‑day condensate production to long-term condensate production are significantly poorer than the equivalent for gas production.

Methodology

All 676 wells have at least 30 days of production history; 669 have 90 days of production history, 628 have 180 days of production history, 537 have 360 days of production history, and 247 have 720 days of production history. To test the previously described assumptions, the authors applied a basic ­regression-line technique to the production data, testing each production-interval predictive capacity against all longer intervals and plotting the results and generating a correlation coefficient (R2) for each comparison. The process was carried out for both wellhead gas production and condensate production.

Results

As described in a series of figures included in the complete paper, it was demonstrated that 30 days is not a strong predictor for longer-term wellhead gas production (360 and 720 days). One-hundred-and-eighty days is a significantly better predictor of longer-term production than the 30- or the 90-day value.

Fig. 1 shows an R2 value of 0.99, indicating that 360‑day gas production and 720‑day gas production are essentially equivalent as gas-production-quality indicators. Next, the ­authors repeated this process with wellhead condensate production in place of gas, starting with 30‑day wellhead condensate plotted against 90-, 180-, 360-, and 720-day condensate production. They observed 30 vs. 90 days, 30 vs. 180 days, 30 vs. 360 days, and 30 vs. 720 days, providing respective R2 values of 0.93, 0.88, 0.87, and 0.86, which is a substantially better correlation than that observed for wellhead gas at 30 days.

jpt-2017-07-184817fig1.jpg
Fig. 1—360-day production data plotted against 720-day gas production.

 

Further work revealed that, as before, the 90-day condensate production is a better predictor of long-term condensate production then the equivalent period is for wellhead gas.

Fig. 2 shows that 180 vs. 360 days provides an R2 value of 0.99, while 180 vs. 720 days provides a 0.98 correlation, indicating that 180-day condensate production and 720-day condensate production are essentially equivalent as condensate production-quality indicators.

jpt-2017-07-182638fig2.jpg
Fig. 2—180-day production data plotted against 360- and 720-day condensate production.

 

Furthermore, it was discovered that 360-day and 720‑day condensate production are essentially equivalent as condensate-production-quality indicators.

Summary

  1. 30-day wellhead gas production is a poor predictor of long-term gas production. Confirmed R2 coefficient is very low for 360- and 720-day production.
  2. 90-day wellhead gas production does not yield a significantly higher R2 coefficient.
  3. 180-day gas production is actually significantly better than 90-day gas production as a predictor of long-term gas.
  4. 360-day gas production predicts 720-day gas production effectively; a year of production gives a reliable forecast of the midterm behavior of the well.
  5. 30-, 90-, and 180-day condensate production correlate better to longer-term condensate production than does equivalent gas production for the same interval, an unexpected result.

Conclusion

The industry widely uses IP-30 as an indicator of well performance, for shale gas and tight oil wells. This study shows that on Utica, 180-day wellhead gas production would be recommended as a minimum interval for a reliable well-performance assessment. The authors also believed that, because of the physical phenomena affecting condensate production in the first 6 months of a well’s life on Utica (from crossing dewpoint pressure to variable unchoking practices), wellhead condensate production would be harder to predict than wellhead gas. This was revealed to be untrue: 180-day cumulative production is a reliable indicator of well performance for both gas and condensate wellhead production on Utica.


This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 184817, “Optimal Production Metric To Predict Unconventional Wells’ Long-Term Performance,” by Edward Ifejika, Bertrand De Cumont, and Nommie Kashani, Total, prepared for the 2017 SPE Hydraulic Fracturing Technology Conference and Exhibition, The Woodlands, Texas, USA, 24–26 January. The paper has not been peer reviewed.