Errors and Repeatability in VSARA Analysis of Heavy Oils

Abstract

Listen

Translate article

Abstract Increasing demands for world energy resources have accelerated the development of unconventional resources, especially of heavy oil reservoirs. Yet, the recovery of heavy oils remains challenging mainly due to variations in their viscosity. It is well known that chemical components of heavy oil control fluid properties, such as density, viscosity and shear modulus. Open column liquid chromatography is used to separate the Saturate, Aromatic, Resin, and Asphaltene (SARA) fractions. Although SARA fractions are a common method to report heavy oil compositions, they can have over 20 % errors. In this work we discuss potential error sources and establish a best-practice methodology to reduce the errors, which results in developing a modified SARA method (VSARA) to determine the composition of heavy oil. Experimental results show that the evaporative components of heavy oils, Volatile (V) fractions, are a major source of error in SARA fraction estimates. By tracking weight changes at every step of the SARA fractionation, errors are greatly reduced. Based on the comparison between SARA and VSARA results, VSARA fractions have a significantly lower error (within a 5% range) than SARA fractions alone. Multiple measurements for a single sample by different operators revealed that VSARA measurements are repeatable. Structural differences between the fractions have been verified using Fourier Transform Infra-Red (FTIR) spectroscopy, which shows the reliability of the proposed SARA method. We also compare our VSARA analyses with viscosity and show that viscosity of heavy oils correlates with resin and asphaltene fractions at concentrations above 25%; below 25%, it is uncorrelated. Since heavy oil composition can change with depth, viscosity can be expected to vary as well. Accurate information of changes in the VSARA fractions can be used to evaluate viscosity and viscosity heterogeneity in heavy oil reservoirs, select appropriate recovery methods, populate reservoir models with viscosity heterogeneity, and thus predict reservoir productivity more accurately.