TY - GEN
T1 - Alternate representations in numerical modeling of multistage hydraulically fractured horizontal wells in shale gas reservoirs
AU - Siripatrachai, N.
AU - Ertekin, T.
PY - 2012/1/1
Y1 - 2012/1/1
N2 - Natural gas from shale is becoming increasingly popular. To successfully produce gas from shale, drilling horizontal wells and implementing hydraulic fracturing are crucial. Conventional representation and design of hydraulic fracturing is by placing transverse fracture planes. Each transverse plane represents a single-stage hydraulic fracturing. However, microseismic field data show that high pressure frac fluid creates a stimulated reservoir volume instead of discrete transverse fracture planes. In this paper, an alternative representation for hydraulic fracturing called "crushed zone" representation is proposed. This crushed zone can be represented by an elliptical zone of high fracture permeability and smaller fracture spacing instead of high-permeability transverse fracture planes. Difficulties lie in the field optimization and characterization of hydraulic fracturing whether which model is more representative and what are their equivalent representations. Furthermore, utilizing a numerical reservoir simulator can be time-consuming as reservoir modeling and computation could take several hours for each scenario and the optimal cases could be overlooked. In this research, optimization of hydraulically fractured horizontal wells in shale gas reservoirs and establishing an equivalency between two different hydraulic fracture representations were studied. Artificial neural network (ANN) is a widely used technology in science and engineering because of its capability in establishing highly non-linear relationships. Due to its ability to provide simulation results in seconds, ANN can serve as a powerful tool in reservoir simulation. A commercial reservoir simulator is coupled with ANNs to create expert systems that can be used to overcome the aforementioned challenges. Reservoir simulation was utilized to generate production profile and establish equivalent hydraulic fracture representations. The simulation results were used to train the ANN. In this study, two ANNs are developed. The first ANN called "Gas Production Prediction ANN" can instantly predict production profile of a hydraulically fractured horizontal well completed in shale gas reservoirs with an error margin of less than ±10% from numerical reservoir simulation. The second ANN called "Equivalency ANN" is developed to establish equivalent hydraulic fracture representations between transverse hydraulic fracture representation and crushed zone representation. The equivalent representation from ANN when run with numerical reservoir simulator yields cumulative gas production within ±15% error compared to results obtained via conventional simulation.
AB - Natural gas from shale is becoming increasingly popular. To successfully produce gas from shale, drilling horizontal wells and implementing hydraulic fracturing are crucial. Conventional representation and design of hydraulic fracturing is by placing transverse fracture planes. Each transverse plane represents a single-stage hydraulic fracturing. However, microseismic field data show that high pressure frac fluid creates a stimulated reservoir volume instead of discrete transverse fracture planes. In this paper, an alternative representation for hydraulic fracturing called "crushed zone" representation is proposed. This crushed zone can be represented by an elliptical zone of high fracture permeability and smaller fracture spacing instead of high-permeability transverse fracture planes. Difficulties lie in the field optimization and characterization of hydraulic fracturing whether which model is more representative and what are their equivalent representations. Furthermore, utilizing a numerical reservoir simulator can be time-consuming as reservoir modeling and computation could take several hours for each scenario and the optimal cases could be overlooked. In this research, optimization of hydraulically fractured horizontal wells in shale gas reservoirs and establishing an equivalency between two different hydraulic fracture representations were studied. Artificial neural network (ANN) is a widely used technology in science and engineering because of its capability in establishing highly non-linear relationships. Due to its ability to provide simulation results in seconds, ANN can serve as a powerful tool in reservoir simulation. A commercial reservoir simulator is coupled with ANNs to create expert systems that can be used to overcome the aforementioned challenges. Reservoir simulation was utilized to generate production profile and establish equivalent hydraulic fracture representations. The simulation results were used to train the ANN. In this study, two ANNs are developed. The first ANN called "Gas Production Prediction ANN" can instantly predict production profile of a hydraulically fractured horizontal well completed in shale gas reservoirs with an error margin of less than ±10% from numerical reservoir simulation. The second ANN called "Equivalency ANN" is developed to establish equivalent hydraulic fracture representations between transverse hydraulic fracture representation and crushed zone representation. The equivalent representation from ANN when run with numerical reservoir simulator yields cumulative gas production within ±15% error compared to results obtained via conventional simulation.
UR - http://www.scopus.com/inward/record.url?scp=84861586346&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84861586346&partnerID=8YFLogxK
U2 - 10.2118/153813-ms
DO - 10.2118/153813-ms
M3 - Conference contribution
AN - SCOPUS:84861586346
SN - 9781618399298
T3 - Society of Petroleum Engineers Western Regional Meeting 2012
SP - 472
EP - 487
BT - Society of Petroleum Engineers Western Regional Meeting 2012
PB - Society of Petroleum Engineers
T2 - Society of Petroleum Engineers Western Regional Meeting 2012
Y2 - 21 March 2012 through 23 March 2012
ER -