Artificial Neural Network Machinery Replacement Programme in Oil Industry

Equipment replacement is a major challenge in asset management, particularly in the oil industry where numerous machines deteriorate under harsh operating conditions. This study develops an artificial neural network (ANN) model to support machinery replacement decisions by predicting the effect of deterioration on salvage (resale) value and determining optimal replacement timing. Five years of cost and condition data for 100 machines were extracted from an eMaint computerized maintenance management system, including operating, maintenance, and repair information. The ANN model uses three key input variables—Increased Purchase Price (IPP), Decreased Operating Cost (DOC), and Decreased Maintenance Cost (DMC)—after extensive preprocessing, normalization, and train–test partitioning. Performance was evaluated using metrics such as Mean Square Error, correlation coefficient, F1-Score, recall, precision, and accuracy. The model achieved a test accuracy of 86.67%, indicating strong predictive capability for machine tool replacement decisions in oil industry operations.