Modeling Terminal Decline Rate in Flow Regime Transition using Alternating Conditional Expectation Non-Linear Regression Methods
- Akash Sharma (Enverus) | Brandon Guttery (Enverus)
- Document ID
- Society of Petroleum Engineers
- SPE Liquids-Rich Basins Conference - North America, 7-8 November, Odessa, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2019. Society of Petroleum Engineers
- machine learning, EUR Prediction, decline curve analysis, multivariate statistics, data science
- 50 in the last 30 days
- 50 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 9.50|
|SPE Non-Member Price:||USD 28.00|
Accurate EUR estimation is a critical component of the oil and gas asset evaluation process and has become increasingly important in de-risking investments in shale plays. Hybridized decline curve analysis has emerged as an industry-wide best practice for this process. This method of interpretation is dependent on the accuracy of the estimated terminal decline rate or switch point. Accurately predicting the terminal decline rate for wells with insufficient production history, has proven difficult. This issue is further aggravated in emerging basins and regions of development, as well as complications associated with parent-child relationships in developed acreages. The present paper tackles these challenges by using a statistical approach to predict the switch point and onset of terminal decline rates in unconventional shale plays.
Typical production profiles in unconventional shales are marked by high initial declines and long transient flow regimes, followed by a transition to boundary dominated flow as the pressure transient reaches the boundaries of the effected reservoir. The decline rate at this switch point, as well as the duration to get to it can vary significantly and is dependent on a wide range of variables. The present paper tackles this multivariable problem by using an ACE (alternating conditional expectation) non-linear regression model to predict the switch point. Variables used to predict this change in flow regime include: Gamma Ray, Resistivity (Deep), RHOB, NPHI, formation thickness, WOR, GOR, completion design, proppant amount per foot, perforated interval, and production performance amongst others. In order to account for the impact on production, behavior from infill development and parent-child relationship discrepancies, date- dependent 3D well spacing was calculated and incorporated as a variable in the statistical model. This process was tested and applied to a large dataset of wells in the middle Bakken formation in the Williston Basin.
The transformations derived from the ACE models increased clarity on the most significant factors driving Terminal Decline. For this study, the resulting transformations from the geological attributes proved most intriguing and provided insight to the physical limits of each. A few important parameters include formation thickness, neutron porosity (NPHI), and reservoir thickness. The "sweet spot" for (NPHI) was determined to be ∼6-8% and lies flat with NPHI values greater than ∼8.5%. Physically, this would correlate with decreased hydrocarbon potential due to the decrease of pore volume below ∼8.5%. Interestingly, Terminal Decline increased as reservoir thickness increased which is counter intuitive to traditional rock properties. Finally, wellbores with a lateral length of ∼10,500' or greater experienced the largest decrease in terminal decline rate. One potential explanation is the variability of the geology the wellbore encounters. The ACE models derived during this study were used to determine the most significant factors impacting terminal decline and will be discussed further within the full study.
The present paper provides a novel approach to estimate the transition point to de-risk EUR estimates across shale plays. This analysis adds to the existing body of literature by using data analytics and machine learning to tackle this problem from a truly multivariable standpoint. The insights developed are widely applicable and may provide best practices for a varied range of challenges in EUR prediction.
|File Size||941 KB||Number of Pages||9|
Arps, J. J. (1945, December 1). Analysis of Decline Curves. Society of Petroleum Engineers. doi:10.2118/945228-G
Duong, A. N. (2010, January 1). An Unconventional Rate Decline Approach for Tight and Fracture- Dominated Gas Wells. Society of Petroleum Engineers. doi:10.2118/137748-MS
Joshi, K., & Lee, W. J. (2013, February 4). Comparison of Various Deterministic Forecasting Techniques in Shale Gas Reservoirs. Society of Petroleum Engineers. doi:10.2118/163870-MS
Kanfar, M., & Wattenbarger, R. (2012, January 1). Comparison of Empirical Decline Curve Methods for Shale Wells. Society of Petroleum Engineers. doi:10.2118/162648-MS
Long, D. R., & Davis, M. J. (1988, July 1). A New Approach to the Hyperbolic Curve (includes associated papers 18728 and 18942). Society of Petroleum Engineers. doi:10.2118/16237-PA
Sharma, A., & Burleson, J. (2019, October 1). Enhanced Type Curve Development using Date-Dependent Well Spacing Analysis. SPE Eastern Regional Meeting 2019. Society of Petroleum Engineers. Doi:10.2118/196604-MS
Sharma, A., & Lee, W. J. (2016, August 1). Improved Workflow for EUR Prediction in Unconventional Reservoirs. Unconventional Resources Technology Conference. doi:10.15530/URTEC-2016-2444280