Development of an Artificial Neural Network for Predicting Fishbone Wellbore Performance in Tight Gas Sands

Open Access
- Author:
- Schumacker, Eric David
- Area of Honors:
- Petroleum and Natural Gas Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Turgay Ertekin, Thesis Supervisor
Turgay Ertekin, Thesis Honors Advisor
Yilin Wang, Faculty Reader - Keywords:
- natural gas
production
multilateral
wellbore
neural network
unconventional reservoir - Abstract:
- Fishbone wellbores are a type of multi-lateral wellbore structure that can be applied to unconventional natural gas reserves such as tight gas sands and shale gas reservoirs to increase natural gas production and make the development of natural gas plays plausible in otherwise uneconomic conditions. This study is aimed at developing an artificial neural network tool to forecast monthly production data for fishbone type wellbores in tight gas sand formation. Reservoir, fluid, and wellbore parameters, as well as their resulting monthly production data generated with reservoir simulation software, are used to train the ANN model. The resulting model is able to predict a combination of outputs for any parameter combination within the range of training. Three neural network structures will be designed. The first, described above, is a typical reservoir depletion study, with reservoir rock and fluid parameters known along with an existing wellbore geometry. In this case, monthly production data will be predicted by the neural network tool. The second structure will focus on wellbore geometry optimization, with reservoir rock and fluid parameters known, along with a desired production target. In this second case, possible wellbore geometry to yield aforementioned production targets will be predicted by the tool. In the final case, a formation test application, a known wellbore geometry along with early production data will be provided as known inputs to the tool. The tool will predict possible reservoir rock and fluid properties responsible for the given production data.