ISI Logo

Re-Stimulation Candidate Selection in Three U.S. Basins

Problem Statement

There is a significant potential for natural gas production enhancement via restimulation. In 1998 Gas Research Institute sponsored a research program to identify the most appropriate methodology for "Restimulation" candidate selection in tight sands. A team of five companies were involved in the project. The best methodology would have to prove its effectiveness in a controlled environment (Reservoir Simulation) as well as in the field in three basins in the US.

Methodology

Three methodologies tested for restimulation candidate selection were:

  • Statistical Approach (Moving Domain Analysis) [Offered by a major Service Company]
  • .
  • Traditional Reservoir Engineering (Type Curve Matching) [Offered by a major Consulting Firm].
  • Artificial Intelligence & Data Mining (AI&DM) Technique [Offered by Intelligent Solutions, Inc.].

The AI&DM-based technology offered by ISI included building data driven model using IDEA™ from the historical production and stimulation data. Once the model is developed and validated (using part of the dataset as blind data), a combination of Genetic optimization and Fuzzy Decision Support system will identify the best candidate wells.

The workflow used by ISI in restimulation candidate selection.

Results

The technologies offered by the three companies were applied to wells from three fields in the United States, as shown in the following map.

Restimulation candidate selection was tested in three fields in the United States.

After complete analysis each methodology offered a list of top 25 wells as restimulation candidates. A combination of several wells was selected by GRI for restimulation. The results are shown below. The well shown below was ranked #2 by ISI and ranked (low ranking) by other two techniques. Upon restimulation production increased significantly.

This well was a top ranked candidate well in ISI's list and was also ranked (much lower ranking) by other technologies.

The well shown below was ranked #15 by ISI and was NOT ranked (by other two techniques. Upon restimulation production increased by two folds.

This well was ranked #15 candidate well in ISI's list and was NOT ranked by other technologies.

The well shown below was NOT ranked by ISI but since it was ranked by other two techniques it was selected for restimulation. Upon restimulation of the no production increased was observed.

This well was NOT ranked by ISI's but was ranked by other two technologies.

The well shown below was NOT ranked by ISI but since it was ranked by other two techniques it was selected for restimulation. Upon restimulation of the no production increased was observed.

This well was NOT ranked by ISI's but was ranked by other two technologies.

The next step was testing the three techniques in a controlled environment. A reservoir simulator was used to model a field with hundreds of wells that were stimulated upon completion. Then many wells were restimulated and the results of restimulation were modeled. The data (same type of data that were accessible during actual exercises with field data) was provided to three companies. The three technologies were suppose to identify restimulation candidates.

A reservoir simulation model was used to test the accuracy of different methods for restimulation candidate selection.

In the following figure the well numbers are shown in circles. The wells in small green circles are the correct restimulation candidates. ISI's AI&DM-based technology was the most successful technology in identifying restimulation candidates.

ISI's technology proved to be the most successful technology for restimulation candidate selection.

References: