Prediction of Calcium Carbonate Scaling in Pipes Using Artificial Neural Networks

Abstract

Scaling problems in oil production happen frequently. There are chemical and mechanical treatments to fight with this problem. Knowing if they will happen under particular circumstances of temperature and pressure is the best way to avoid or to control their apparition. This paper shows the most important characteristics that improve the calcium carbonate scaling accumulation in pipelines in order to understand how kinetic and thermodynamic characteristics affects the scaling process. Monitoring the scaling process using the Saturation Index was considered a method for scaling solutions, this method uses characteristics as solution pH, temperature, ion concentrations, among others. The influence that temperature and initial ion concentration have over the solution pH of the scaling system was studied using an experimental test in a batch reactor, this allows to follow the scaling process in real time through pH changes. The prediction of calcium carbonate scaling process represents another objective in this work. Numerical analysis based on artificial intelligence as Multi Layer Perceptron and Probabilistic Neural Network were used. A good learning process and a good prediction model for the ANN methods was shown when the experimental SI and predicted SI were compared. It was possible to confirm the accuracy of the ANN methods using external experimental tests.