Deep Learning techniques have been successfully applied to areas like image recognition and speech translation. It has shown a great potential on recognizing and generalization hidden patterns in data. Logs are critical to reservoir characterization. NMR logs contain valuable information on formation fluid, but it is more expensive than other logs. The purpose of our study is to generate NMR T2 based on formation mineral composition using a combined deep learning techniques with a limited number of NMR T2 data set. Our study shows that deep learning technique can predict NMR T2 with a desirable accuracy.
Presentation Date: Thursday, September 28, 2017
Start Time: 10:10 AM
Presentation Type: ORAL