Machine learning to assess successfulness of xylene soaks for asphaltene deposition management

In this Case Study, AFS discusses an analysis that was performed to determine the effectiveness of xylene soaks for asphaltene deposition removal in wells.


One of the few remediation options available when asphaltene deposition occurs in a wellbore is regularly performing xylene soaks of the wellbore to dissolve the asphaltenes causing restrictions. These kinds of operations are generally costly both from an execution perspective in terms of purchasing chemical and the equipment required to pump it, but also from a loss of production standpoint as the xylene soaks can last over a day in some cases. AFS was asked by a client to try and evaluate the success of regular xylene soaks for a large 7 well asset in the Gulf of Mexico with known asphaltene deposition issues.


Field data was supplied for the various wells and after initial review, periods of well performance were isolated before and after xylene soaks. Following that, a machine learning algorithm was used to assess the overall impact of the soaks by predicting future production assuming the xylene soak had not occurred and comparing it to the actual field data after the soak. All field data was utilized by the machine learning algorithms including oil and water flowrates as well as pressures at multiple points in the wellbore.


Figure 1: Wellbore pressure drop predictions with and without xylene soaks

In this figure, parity plots were created for each well to show the actual pressure drops from the reported field data as compared to predictions made using a machine learning algorithm assuming the xylene soak had not occurred. Values where actual pressures drops are lower than predicted, indicated by markers below the diagonal parity line, show xylene soaks that were effective at removing asphaltene deposition. Additionally, the effects of water cut and volume of xylene pumped can be seen by the color and size of the individual markers.


Results of the analysis showed that one of the largest factors effecting the success of the xylene soaks was the water production of the well, which has also been verified experimentally as higher water production reduces the risk of asphaltene deposition. Interestingly, very little effects were found correlating the amount of xylene pumped and the reduction in pressure drop following the soak. It is assumed, however, that a minimal volume of xylene is still required, after which diminishing returns are found. Overall, the analysis was able to conclude that out of the 7 wells regularly performing xylene soaks, only 3 wells were still actively benefiting from the operations, which allowed the client to confidently reduce the amount of xylene soaks being performed resulting in a significant cost savings.

For more information about AFS Technology services, please visit our dedicated service page or contact us to discuss your specific project challenges in more detail.