Construction of synthetic capillary pressure curves from the joint use of NMR log data and conventional logs

Abstract

Capillary pressure (Pc) curves are important petrophysical parameters to characterize reservoir rock properties in hydrocarbon fields. Determination of Pc values conventionally relies on a variety of experimental processes. Although the experiments provide accurate outcomes, they may be extensive, time consuming and discontinues through the reservoir interval. The current study demonstrates the feasibility of synthesizing capillary pressure curves in carbonate reservoirs from conventional and Nuclear Magnetic Resonance (NMR) logs by using a two-step approach. The first step is to simulate T2 (longitude relaxation time) distribution values from conventional logs by using intelligent systems. For this purpose, eight Combinable Magnetic Resonance Bin Porosities (CBPs) are estimated from well logs with a reasonable accuracy (Correlation Coefficient (CC)>0.90 for almost all CBPs). In the second step, the Pc values are predicted from CBPs through an inversion process. The simulated Pc curves show a good agreement with laboratory derived Mercury Injection Capillary Pressure (MICP) curves at low mercury saturations (<35%) which are important in petrophysical studies of carbonate rocks. The predicted Pc values indicate acceptable Mean Squared Error (MSE) and R2 values (>0.70) at different mercury saturations.