An artificial neural network approach to predict asphaltene deposition test result

Asphaltene deposition in reservoir, completion string or flow lines causes flow assurance problem including wettability reversal, permeability reduction, increased pressure drop, well and pipeline plugging and finally production rate reduction. Generally, asphaltene deposition in a sample of live oil, in different pressures and temperatures, is measured by High-tech expensive apparatus and used in asphaltene study in pipelines and reservoir. Present study describes an innovative method for easy and fast prediction of the asphaltene deposition test by use of artificial neural network (ANN). Different ANNs are designed and trained with different solution algorithms to find the best predictor for target samples. The output ANN shows significant accuracy for validation data and conclusively is reliable for prediction unknown values of target samples. Prediction of asphaltene deposition test results, gathered by ANN, is much time and cost saving than the conventional experimental studies.

► A new method is stated to predict the result of the experimental test using artificial neural network (ANN). ► Different ANN designs with some solution algorithms are performed. ► The appropriate ANN is selected based on minimum AARE for validation data. ► The 3 hidden-layer back propagation AAN concludes a high accurate prediction. ► Prediction of asphaltene deposition test results, gathered by ANN, is much time and cost saving than the conventional experimental studies.