A new method for identifying reservoir fluid properties based on well logging data: A case study from PL block of Bohai Bay Basin, North China

Abstract

The characterization of reservoir fluid properties is a crucial component of oilfield operations, as it provides a vital data foundation for the development and optimization of oilfield work programs. However, the complexity of water-flooded, along with the mixed data from drilling and cable logging, and the inherently weak foundational research, make the evaluation of water-flooded formations difficult. Therefore, this article aims to address this challenge by proposing a new reservoir fluid identification method. In this article, an improved Markov variation field model is applied to map geophysical logging data and is integrated with a quantum hybrid neural network (HQNN) to address the nonlinear correlations between logging data. By integrating the non-standard Markov variation field with HQNN, this article constructs a novel reservoir fluid identification model. Experimental results demonstrate that the model achieves a recognition accuracy of 90.85% when trained on feature images mapped from logging data. Furthermore, the superiority of the HQNN was validated through eight sets of comparative experiments. Additionally, the model was further validated using logging data from blind wells within the block, demonstrating high predictive accuracy and proving its effectiveness for reservoir fluid identification in the PL block. The method proposed in this article not only addresses the challenge of evaluating water-flooded layers in the absence of key logging curves but also offers a novel approach to reservoir fluid identification using geophysical logging data. The non-standard Markov transition field model is employed to map logging data into feature images, offering a new perspective on the application of geophysical logging data in practical reservoir analysis.