Accurate Original Resistivity Inversion in a Heterogeneous Thick Flooded Reservoir Through Electrical Interpretation Unit Classification

Abstract

Low resistivity is a main feature of flooded reservoir in injection fields. The key to evaluate flooding severity is to reconstruct resistivity in its' original condition. In this process, the routine method is to find a relationship between resistivity and other logs to rebuild unflooded resistivity. However, it is very sensitive to formation parameters and prone to have larger error due to ignoring the difference of lithology, reservoir and electrical properties driven by geological background. This paper proposes a method to classify reservoir into different electrical interpretation units to address the problem that routine methods have. Firstly, reservoir was divided into different logging facies through multi-resolution graph-based clustering(MRGC). Secondly, a database of the relationship between logging facies and lithology in aimed reservoir was built based on the core, petrophysical logging data and so on. Eventually, electrical interpretation units were obtained based on the database and reservoir resistivity. Resistivity models of origin field were built for each electrical interpretation unit, which could improve the accuracy of original resistivity calculation dramatically. In the studied field, five logging facies were defined in a coring well using MRGC method. The relationship between each logging facie and its lithology were constructed as well. Each logging facie can be recognized though cross-plots from conventional logs. Meanwhile, considering reservoir fluid properties, through finding similar electrical properties in the reservoir with same logging facies and lithology, the studied well could be further divided into three interpretation units. With similar lithology and reservoir properties, reservoir heterogeneity from the same electrical interpretation unit decreases, plus bearing similar electrical property, the influence from saturation difference of the original field has weakened. Through the previous illustrated model, it has demonstrated a good effect in evaluating the original resistivity in thick layers of the M field from Bohai Bay. Comparing with the result from routine method, the resistivity from the proposed method has a better match with the resistivity from unflooded area, consequently, increase the evaluation accuracy dramatically. Eventually, the classification of logging facies and electrical interpretation units from key wells can be spread to flooded wells to calculate the unflooded resistivity, as a result, a more accurate flooding severity evaluation could be obtained. It could provide more reliable data for further residential oil development.