Meta-Model Based Feedback

The hybrid models discussed above inherit several drawbacks from their simulation component  

The supplier selection problem is adopted from Weber et al. (2000). It concerns procurement of one major raw material. Suppliers are selected for the length of a relatively long planning horizon. The number of suppliers to be selected is 

Zi represents the material price criterion, represents the delivery reliability criterion, and Z3 represents the quality criterion, x, is the quantity to be purchased from the /th supplier. N denotes the number of candidate suppliers, v, indicates whether the /th supplier is included in the supply network. and limits minimum and maximum total order quantities for the/th supplier, respectively, pj is the net purchase price per unit for /th supplier. A, is the percentage of late items for the /th supplier, fij is the percentage of items rejected for the /th supplier. 

The standalone optimization of this supplier selection problem is complicated by different dimensions of supplier selection criteria. The simulation model is used for evaluating the impact of delivery performance and quality of materials on manufacturing costs. It represents the actual manufacturing system at an appropriate level of abstraction. If all selection criteria are expressed in terms of the manufacturing costs incurred, then the multi-criteria problem is transformed into 

Alternatively, the optimization of a supply network could be performed using the simulation model directly or using a hybrid optimization-simulation model, which invokes the simulation model on each optimization trial, and is computationally challenging. 

---