The application of deep learning to predict fracture porosity of a fractured granite basement reservoir | |
| Author | Sandunil, Kohona Walawwe Kushan Oshadi |
| Call Number | AIT Thesis no.GE-16-11 |
| Subject(s) | Reservoirs. Predictive control. Petrology. |
| Note | A thesis submitted in partial fulfillment of the requirements for thedegree of Master ofEngineering inGeotechnical Earth Resources Engineering with area of specialization in Geosystem Exploration and Petroleum Geoengineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. GE-16-11 |
| Abstract | Porosity is one of three important petrophysicalparameters which playa vital role in estimating petroleum initially in place. Conventional methods used to estimate primary porosity of clastic reservoirs often do not work well to estimate fracture porosity. In this study,well log data from a well of an oil field in theCuu Long basin, Vietnam was used to predict fracture porosity using deep learning neural networks for the depth interval from 2515 m to 3015 m.Calculated fracture porosities using a conventional method proposed by Elkewidy and Tiab (1998) was used to train the model along with the well log data. Three analyses were successfully implemented to obtain a deep learning model. The first analysis with singlehiddenlayerartificial neural network model, second analysis with multiple hidden layerartificial neural network modeland the third analysis with deep learningneural network model.It was found that the conventionally-calculated fracture porosity ranges from 0 to0.084 while the deep learning predicted porosity was found in the range from 0 –0.082, showing a good match between them. The main finding of this study was the successful development of a deep learning model by a 3-step development process. The final deep learning model consists of 5-input set of G, LLD, DT, RHOB and NPHI, having5 hidden layers with 14 neurons per layer. The transfer function of RELU, typical for a deep learning analysis was successfully implemented instead of sigmoidal function |
| Year | 2017 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. GE-16-11 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Civil and Infrastucture Engineering (DCIE) |
| Academic Program/FoS | Geotechnical Engineering (GE) |
| Chairperson(s) | Pham Huy Giao; |
| Examination Committee(s) | Noppadol Phien-wej;Chao, Kuo-Chieh; |
| Scholarship Donor(s) | Asian Institute of Technology Fellowship; |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2017 |