Performing hydraulic fractures on gas storage wells to improve their deliverability is a common practice in the eastern part of the United States. Most of the fields in this part of the country being used for storage are old. Reservoir characteristic data necessary for most reservoir studies and hydraulic fracture design and evaluation are scarce for these old fields. This paper introduces a new methodology by which parameters that influence the response of gas storage wells to hydraulic fracturing may be identified in the absence of sufficient reservoir data. Control and manipulation of these parameters, once identified correctly, could enhance the outcome of frac jobs in gas storage fields. The study was conducted on a gas storage field in the Clinton formation of Northeastern Ohio. It was found that well performance indicators prior to a hydraulic fracture play an important role in how good the well will respond to a new frac job. Several other important factors were also identified.
The identification of controlling parameters serves as a foundation for improved frac job design in the fields where adequate engineering data is not available.
Another application of this type of study could be the enhancement of selection criteria among the candidate wells for hydraulic fracturing. To achieve the objective of this study, an artificial neural network was designed, trained and applied. The paper will discuss the results of the incorporation of this new technology in hydraulic fracture design and evaluation.