Decision Support System for Optimizing Spare Parts Forecasting for Training Aircrafts

The sporadic nature of demand pattern of aircraft maintenance repair parts renders the forecasting of these spares an immensely complicated task. This research effort deals with techniques applicable to predicting spare parts demand for the training aircrafts. The experimental results of 3 forecasting methods, including those used by aviation companies, are examined and validated through statistical analysis. Actual historical data of 118 training aircrafts for hard-time and condition-monitoring components from the Aviation Flying Training School was used, in order to compare different forecasting methods when facing intermittent and lumpy demands. The results confirm continued superiority of weighted moving average and exponential smoothing method for intermittent demand, whereas most commonly used naive methods used Aviation Flying Training School was found questionable. A new approach has been devised for forecasting evaluation; formulation of a comparative analysis matrix to isolate most critical parts in terms of their prices and quantities followed by a predictive error-forecasting model which compares and evaluates forecasting methods based on their factor levels when faced with intermittent demand. This research is useful to identify parts/spares critical to the operation of a training aircraft in terms of both their prices and quantities and application of reliable and robust forecasting method to predict the future demand requirements, thereby optimizing the logistic supply chain and aircrafts operational performance over the life cycle. These findings may be applied to other aircrafts with similar demand patterns. In this research, it has been assumed that all spares/components are non-repairable and are replaced upon failure.