Total organic carbon content (TOC) present in the potential source rocks significantly affects the response of several types of well logs. They are characterized by higher porosity, higher sonic transit time, lower density, higher gamma-ray, and higher resistivity than other rocks. This paper attempts to establish a quantitative correlation between standard well logs (sonic, density, neutron, gamma-ray and resistivity) and total organic carbon by means of intelligent systems with an example from the Upper Cretaceous reservoirs, in the eastern part of the North Western Desert of Egypt. The present work utilizes the ability of neural networks to discover patterns in the data important for the required decision, which may be imperceptible to human brain or standard statistical methods. Thus the idea is not to eliminate the interpretation from an experienced petrophysicist but to make the task simpler and faster for future work
In order to assess the validity of the ANN model developed, a comparison of TOC values determined by DlogR technique (Passey et al., 1990) was made with the ANN predicted TOC and the actual TOC values,. The error measurements, (correlation coefficient, root mean square error and mean absolute error), are 0.95, 0.17 and 0.005 for the predicted TOC using ANN technique and 0.16, 1.64 and0.76 for the estimated TOC using DlogR technique. In comparing results obtained from DlogR technique with those obtained from the multiple artificial neural system, indications are that the latter outperforms the first.