Abstract
Asphaltene precipitation can sometimes pose operational problems in medium-light oils because of the low asphaltene solubility. The purpose of this work is to develop a methodology to predict the asphaltene envelope for all fluid systems in ADNOC’s reservoir portfolio based on existing laboratory experiments. Such a model would then be able to predict potential precipitation risks for current and future field development projects, especially the ones involving gas injection.
The starting point for development of the predictive model for asphaltene onset pressure (AOP) is the 100+ lab measurements carried out over several decades, of which 65% involve a number of injection gases such as CO2, lean and rich hydrocarbon gas, as well as sour gas. We then matched each data set with an equation of state (EOS) to generate the entire onset pressure envelope. Based on the envelope data points, we applied a data-driven method to reproduce the key trends, and used this trained model as a novel predictive tool for new production scenarios without experimental AOP data.
We first tested the PC-SAFT model for our phase behavior calculations but found that the method, as implemented in the software package, often experienced convergence problems. The PR-78 cubic EOS was found to be more reliable with the ability to match the experimental data despite limited predictive power. We find that availability of AOP data for reservoir fluids swollen with injection gas makes the thermodynamic model much more robust compared to tuning to a few AOP data on the original reservoir fluid alone. A single AOP point is generally not sufficient to fully constrain the EOS model unless model parameters from other studies are brought into use. SARA analysis is not mandatory for the EOS tuning itself and was found not to be required for training any of the data-driven methods. We limited the predicted data sets to temperatures below 350 °F, since all our reservoirs have temperatures below this threshold. From the calculated envelopes, we saw a clear impact of fluid composition on the shape of the AOP curve relative to the saturation pressure curve, as expected. We now have a tool, which can accurately predict the AOP curve for a combination of reservoir fluids and injection gases, as the long as the injection gas composition remains within the range tested experimentally.
