In-Situ Heavy Oil Viscosity Prediction at High Temperatures Using Low-Field NMR Relaxometry and Nonlinear Least Squares

ABSTRACT

Heavy oil and bitumen resources are the most abundant hydrocarbon energy source worldwide. However, due to their high viscosity, complex thermal enhanced oil recovery (EOR) methods are frequently applied to reduce it. Information about the distribution of oil viscosity within the reservoir can be used to help management of thermal EOR project. Nuclear magnetic resonance (NMR) downhole tools provide a non-destructive way to determine the oil viscosity without having to recover samples (core or produced oil) from the wellbore. Tests showed that the accuracy of the literature NMR viscosity models significantly decreases with temperature increase, making them ineffective for monitoring high temperature. Moreover, improved accuracy is required for high viscosity ranges of oil representative of original reservoir conditions.

An enhanced NMR viscosity model was developed and tested on a suite of 23 Canadian heavy oils recovered from different reservoirs. Subsequently, the model was tested on a single bitumen sample at temperature range 26–200 °C. Results were compared to nine well-known NMR viscosity models from the literature, and in both cases, the enhanced model scored the lowest root mean square error (RMSE) in this study. Furthermore, a simple model optimization procedure was presented, which employs nonlinear least squares (NLS) regression.

Experiments were carried out at the temperatures that correspond to those in the hot steam injection EOR treatments. The same methodology can be extended for use in cyclic solvent injection (CSI) where NMR model can detect the change of oil viscosity when the solvent vapor dissolves into the oil.

A full version of this paper is available on ScienceDirect Online.