Oil Viscosity Predictions from Low Field NMR Measurements

Bryan, J.L., Kantzas, A. and Bellehumeur, C.

DOI: 10.2118/89070-PA
SPE 89070 (previously 77329);
SPEREE, 8(1), February 2005, Pages 44-52.


Canada contains vast reserves of heavy oil and bitumen. Viscosity determination is key to the successful recovery of this oil, and low-field nuclear magnetic resonance (NMR) shows great potential as a tool for estimating this property. An NMR viscosity correlation previously had been developed that is valid for order-of-magnitude estimates over a wide range of viscosities and temperatures. This correlation was built phenomenologically, using experiments relating NMR spectra to viscosity. The present work details a more thorough investigation into oil viscosity and NMR, thus providing a theoretical justification for the proposed correlation. A novel tuning procedure is also presented, whereby the correlation is fitted using the Arrhenius relationship to improve the NMR viscosity estimates for single oils at multiple temperatures. Tuning allows for NMR to be potentially used in observation wells to monitor thermal enhanced oil recovery (EOR) projects or online to monitor the viscosity of produced-fluid streams as they cool.


With approximately 400 million m3 of oil in place, the Canadian deposits of heavy oil and bitumen are some of the most vast oil resources in the world.1Heavy oil and bitumen are characterized by high densities and viscosities,which is a major obstacle to their recovery. The waning of conventional-oil reserves in Canada, coupled with increasing worldwide demand for oil, has forced the industry focus to shift rapidly to the exploitation of these heavy-oil and bitumen reserves.

The most important physical property of heavy oil that affects its recovery is its viscosity.1 This parameter dictates both the economics and the technical chance of success for any chosen recovery scheme. As a result, oil viscosity is often directly related to recoverable reserves estimates.2 Unfortunately,laboratory measurements of oil viscosity become progressively more difficult to obtain as viscosity increases.3 The oil that has been removed from the core also may have been physically altered during sampling and transport. Thus, the viscosity at reservoir conditions may be different from the value obtained later from the laboratory.2 In light of the shortcomings of conventional viscosity measurements, low-field NMR is considered as an alternative for estimating heavy-oil and bitumen viscosity.

The main appeal of NMR as a tool for assessing reservoir-fluid viscosities and phase volumes is that the measured signal comes only from hydrogen, which is present in both oil and water found in hydrocarbon reservoirs.4,5 Most of the low-field NMR applications in the petroleum industry have been in conventional oil, contained in sandstone reservoirs.6 To use low-field NMR technology in heavy-oil and bitumen formations like the ones present inAlberta, new methods of interpretation are required. The eventual goal for using NMR to estimate viscosity is to make these predictions in the field through logs. Currently, research toward this goal is conducted in the laboratory.

In previous work,7-9 an oil-viscosity correlation was presented that is capable of providing viscosity predictions for samples with viscosities less than 1 mPa×s to more than 3 000 000 mPa×s. This is a wider range than any other viscosity correlation presented in the literature.10-15 The correlation is only order-of-magnitude accurate but still could be valuable for applications on a logging tool, where the goal would be to determine viscosity variations with depth or areal location in a reservoir. The theoretical justification behind the NMR correlation is given in this work, along with a procedure for tuning the correlation to improve the viscosity predictions for individual oils as a function of temperature. Low-field NMR experiments are simple to perform and nondestructive. The same test also can be run by different technicians to yield the same results, which is a concern for conventional viscosity tests.3 In this manner, a properly calibrated NMR model for viscosity can be a very accurate and useful tool for predicting heavy-oil and bitumen viscosity at different temperatures.

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