Yedek parça envanter yönetimi

Since there is a high competition in the aviation industry, reducing costs comes into prominence and is becoming more critical each day. In the aviation industry, maintenance and inventory holding costs of spare parts give the opportunity to managers to decrease their operational costs. Therefore, demand forecasting with high accuracy is indispensable matter in spare parts inventory management. In the literature, traditional demand forecasting methods and measures are claimed to be insufficient due to the variability in demand size and the uncertainty in demand occurrence. While comparing traditional forecasting methods with non-traditional methods; classical performance measures are usually preferred and these measures often give misleading results when inventory cost minimization is selected as a primary objective for service parts. The main reason is the nature of demand that contains a large percentage of zero values with less non-zero demand. In this thesis, cost-based performances are measured employing different inventory policies and ordering approaches that are proposed to compare the traditional forecasting methods with the non-parametric and parametric forecasting methods generated for non-smooth demand. In order to compare these non-smooth demand forecasting methods, 535 different items are selected from the inventory of Turkish Airlines Technic MRO. A methodology is presented that is consisting of data classification, initial parameter estimation, parameter search with optimization and evaluation. Although in the literature it is claimed that traditional methods may fail in forecasting non-smooth demand, it has been observed that non-traditional methods are not performing better than the traditional alternatives when the inventory cost is taken into account as the performance measure. In this thesis, it is also claimed that applying different inventory policies and newly proposed ordering approaches with optimized parameters can reduce inventory costs more than the existing methods. Generated outputs of this study may provide a framework that will guide the inventory planners to make their decisions on which forecasting method, replenishment policy and ordering approach give the minimum inventory cost based on demand data type.