Identifying flow units by FA-assisted SSOM—An example from the Eocene basin-floor-fan turbidite reservoirs in the Daluhu Oilfield, Dongying Depression, Bohai Bay Basin, China

Abstract

Accurate identification of flow units is essential in oil and gas development. In this study, integrated core, log, and production data from the Eocene basin-floor-fan turbidite reservoir demonstrates a new method to identify flow unit using factor analysis (FA) and supervised-mode self-organizing-map neural network (SSOM). The reservoirs were classified into four types of flow units (I, II, III and IV). Five principal factors were extracted through factor analysis on thirteen evaluation parameters for reflecting the characteristics of basin-floor-fan turbidite reservoirs. Then using the five principal factors as the input, the flow unit prediction model was established based on SSOM. The prediction results of flow unit based on FA-SSOM are consistent with the results of core analysis and test oil conclusion, which have a good classification effect. Therefore, the prediction model based on FA and SSOM provides an effective way for fine reservoir interpretation. The established FA-SSOM model is further compared with Linear Discriminant Analysis (LDA) and Back Propagation (BP) neural network and has the best prediction. This study also sheds light on the remaining oil development by linking the identified flow unit type with daily production data.