Investigating the variations in depositional facies by investigating the accuracy of the neural network model within the St. Louis limestone, Kearny County, Kansas

K-REx Repository

Show simple item record

dc.contributor.author Reece, Chance
dc.date.accessioned 2016-04-21T20:46:09Z
dc.date.available 2016-04-21T20:46:09Z
dc.date.issued 2016-05-01 en_US
dc.identifier.uri http://hdl.handle.net/2097/32588
dc.description.abstract The Mississippian-aged St. Louis Limestone has been a major producer of oil, and natural gas for years in Kearny County, Kansas. Since 1966 two major fields in the County, the Lakin, and Lakin South fields, have produced over 4,405,800 bbls of oil. The St. Louis can be subdivided into six different depositional facies, all with varying lithologies and porosities. Only one of these facies is productive, and the challenge of exploration in this area is the prediction of the productive facies distribution. A previous study by Martin (2015) used a neural network model using well log data, calibrated with established facies distributed within a cored well, to predict the presence of these facies in adjacent wells without core. It was assumed that the model’s prediction accuracy would be strongest near the cored wells, with increasing inaccuracy as you move further from the cored wells used for the neural network model. The aim of this study was to investigate the accuracy of the neural network model predictions. Additionally, is the greater accuracy closest to the cored wells used to calibrate the model, with a corresponding decrease in predictive accuracy as you move further away? Most importantly, how well did the model predict the primary producing unit (porous ooid grainstone) within the St. Louis Limestone? The results showed that the neural network was not completely reliable in predicting total facies distribution. This can be attributed to many different inefficiencies in the data, including different resolutions between cuttings data and well logs, limited well cuttings available, and missing cuttings from the wells that were observed. Relating the neural network predictions to actual well productivity validates the neural network’s ability to predict the producing facies. There are also instances of the productive facies being present when not predicted. This is likely a function of different facies thickness in these wells from the cored wells used to calibrate the model, rather than distance from the cored well. en_US
dc.language.iso en_US en_US
dc.publisher Kansas State University en
dc.subject St. Louis Limestone en_US
dc.title Investigating the variations in depositional facies by investigating the accuracy of the neural network model within the St. Louis limestone, Kearny County, Kansas en_US
dc.type Thesis en_US
dc.description.degree Master of Science en_US
dc.description.level Masters en_US
dc.description.department Department of Geology en_US
dc.description.advisor Matt Totten en_US
dc.date.published 2016 en_US
dc.date.graduationmonth May en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search K-REx


Advanced Search

Browse

My Account

Statistics








Center for the

Advancement of Digital

Scholarship

118 Hale Library

Manhattan KS 66506


(785) 532-7444

cads@k-state.edu