Cuttings transport modeling in underbalanced oil drilling operation using radial basis neural network

Underbalanced drilling is one of the drilling methods for better drilling according to its advantages. Cuttings transport effects on cost, time, and quality of oil/gas wells in drilling operation. Inefficient cleaning of wellbore may cause many drilling problems. Prediction and measuring of the cleaning efficiency in the wellbore annulus is a complex problem according to many effective factors. The field and experimental measurements of this parameter are time consuming and costly. This paper presents the radial basis function network (RBFN) method for prediction of cuttings concentration in underbalanced drilling condition to avoid the high cost experimental and field measurements. The average absolute percent relative error (AAPE) for train and test datasets in this study is 2.9e-13%, and 5.7% for the RBFN model. The comparison results of this study with literature review show the benefit of RBFN in prediction compared to back propagation neural network (BPNN) according to higher accuracy, faster training and simple network architecture. So, this network can be used in many mathematical problems for prediction and estimation instead of BPNN. Results of this study show that implementation of this developed model can be incorporated in drilling simulators for accurate estimation of cuttings concentration in wellbore instead of field and experimental measurements for hydraulic design in drilling operation.