A model for predicting plant maintenance costs

A model is presented that predicts the total cost of plant maintenance (i.e. direct cost of maintenance plus indirect cost of lost production) and is derived studying a random sample of tracked hydraulic excavators. Analysis is based on the machine history file data of 33 plant items, modelled using multiple regression (MR) analysis. Validation of the model was determined via the combination of an observed high R2 at 0.94 and various statistical tests which confirmed the prerequisites of a rigorous MR analysis. Machine weight, type of industry and company attitude towards predictive maintenance were found to be the best predictor variables of total plant maintenance cost. The paper also discusses reasons underlying the inclusion of predictor variables in the final model, and concludes with clear directions for future research in this field.