Dwight Fulton

Structured approach to advanced candidate selection and treatment design of stimulation treatments

Because many oilfield workers see matrix treatments of wells as alow-tech operation, they often fail to pay attention to candidate selection and treatment design. This lack of attention may have led to the relatively low success rate of these treatments. A 1997 survey in a major oil company indicated that one out of every three to four jobs fails to produce more oil or gas after the treatment. This failure represents a loss to the company of over

Causal Analysis and Data Mining of Well Stimulation Data Using Classification and Regression Tree with Enhancements

10 million (U.S.), plus a missed extra production capacity of nearly 40,000 BOPD. The probable main cause for this poor performance is the lack of a structured approach to the following: ○ selecting the right candidate wells and the appropriate treatment ○ defining and implementing a structured treatment design procedure To improve the situation, a task force investigated the problem and mapped out a total process, which consists of the following steps: A candidate well is selected by comparing its actual performance against its theoretical potential. Then, the source of poor performance is identified, if applicable. Based on this information, the treatment type can be identified and designed, ultimately resulting in an operational stimulation program. The task force concluded that individual pieces of design software and some design rules existed for many elements, but they lacked an integrated overall approach. The team decided to create a software package by integrating fuzzy rules with appropriate mathematical models to guide field engineers through the individual design steps in a consistent, structured manner. This paper describes various elements ofthe integrated software package that was designed to meet these needs.

Multivariate Analysis of Job Pause Time Data Using Classification and Regression Tree and Kernel Clustering

In the well-treatment program certain variables, like Job Pause Time (JPT) and fracture screen-out, can affect its efficiency. JPT is the time during which pumping is paused in-between subsequent treatments and screen-out occurs when the fluid flow is restricted inside the fracture. We investigate whether it is possible to identify characteristic patterns in existing data that affect the extreme values of JPT as well as the most critical variables causing fracture screen-out. We apply Classification and Regression Tree (CART) analysis, validate the approach with well-stimulation case studies and enhance predictive capability by implementing normal score transform and data clustering.

Electroconductive proppant compositions and related methods

The well treatment program is an important part of the field development plan, and certain variables, such as job pause time (JPT), can affect its efficiency. JPT is the time during which pumping is paused between subsequent treatments of a job. The objectives of this work are to investigate whether, from existing data, it is possible to find patterns in significant variables that affect the extreme values of JPT in a particular region. The answers are sought by applying a classification and regression tree (CART) to both categorical and continuous variables in the database. The practical application of CART is presented using case studies first using classical CART analysis, then using CART analysis with enhancement tools such as the normal score transform (NST), and then dividing the large dataset into smaller groups using clustering. Significant variables are found that affect the response variables, and predictor variables are ranked in order of their importance. Such information can be used to control predictor variables that cause high JPT. The results are outlined in an intuitive way, including categorical, continuous, and missing values. Because CART is a data driven, deterministic model, we cannot calculate the confidence interval of the predicted response. Confidence in the results is purely based on the historical values, and the accuracy of the result produced by a tree model depends on the quality of the recorded data measured in terms of volume, reliability, and consistency. The prediction capability of CART is enhanced by the use of NST and clustering techniques. The approach presented in this paper analyzes a dataset with limited information and high uncertainty and should lead to developing a method for generating proxy models to find future success indices (e.g., for drilling efficiency or production from a fracture). This could standardize stimulation and generate decision ‘best practices’ to save costs in field development and the optimization process.