Lithology identification using well logs: A method by integrating artificial neural networks and sedimentary patterns

Effective identification of lithology using well logs is one of the most important steps for reservoir characterization. A lot of methods have been developed to identify lithology automatically by analyzing the value or patterns of well logs. However, the response characteristics of log curves for some different lithologies are similar and indistinguishable. As a result, we may obtain results which do not follow the general geologic rules if we only use values or patterns of well logs as the criteria to identify lithology. In this paper, we propose an alternative automatic method to identify lithology by integrating both well logs and sedimentary patterns. First, we obtain the probability of lithology by applying the artificial neural networks (ANN) to the well logs. Then, we obtain the vertical combination patterns of different lithologies by applying a probabilistic statistical method to the cores. Finally, we produce an integrated lithology probability by combining the lithology probabilities calculated using ANN and lithology probabilities computed using sedimentary pattern. We validated our method by applying it to a real case study from China. The results indicate that the accuracy of lithology identification using the proposed method is much higher (91%) than that using neural networks method (83%).