Managing uncertainty in supply for a whitefish processor
MetadataShow full item record
This research aims to improve the capacity planning for LNS in the processing facility of Båtsfjord. LNS has problems planning the capacity, due to uncertainty in the supply of raw fish to the processing plant. LNS cannot catch fresh fish itself due to regulations by the Norwegian government. Hence, the company depends on supply of external fishing vessels, where the catch varies every shipment. The only information LNS has is which vessels are coming to deliver fish, three days in advance. The research consists of three steps: 1. Finding a method to reduce the uncertainty in the supply of raw material; 2. Implementing the information of step 1 in the capacity planning of the processing plant in Båtsfjord; 3. Estimating the gains of the reduction in uncertainty of the supply of raw material. In the first step we compare three methods to reduce the supply uncertainty. With the requirements that LNS set, we decide that the method where we forecast the fish catch per vessel, using information that is available fits the needs of LNS the best. We use the information about the vessel length, the vessel age, the wind speed from the west, and the visibility to forecast the number of kilogram Cod a vessel delivers to LNS. All the four predictors are significant with a maximum error of 5%. For step 2 we create a mathematical model of the processing plant. We decide to use Stochastic Programming since we can consider the processing times deterministic. The age and quality of the raw material dictates which products LNS can produce. Hence, when not all the supply is used at day T, the possible value of the raw material deteriorates. Therefore, there is a recursion in the calculations, making it too calculate the gains with exact formulae. We use simulation to simulate each day, where we plan the capacity three days in advance on each day. To calculate the gains, we need a base scenario. Because LNS cannot provide us with detailed data of the planned capacity in the processing plants, we create our own scenario where LNS uses the average catch of a vessel to forecast the supply. With this method we show the gains when we reduce the supply uncertainty. Also, we compare the results with the use of our forecasting model with actual data of the planned ours in the cutting department of LNS. We create the KPI productivity , which consists of the kilogram processed fillet per cutting hour. We compare these two scenarios with the case where we only use the forecasting model, the case where we use the forecasting model in combination with overtime against extra costs, and the case with overtime and the purchase of an extra filleting machine. Although the case with an extra filleting machine yields the best profits, we advise LNS to implement the case with overtime. The investment in the extra filleting machine has a 43% probability of yielding less profit than only using overtime, making it a risky investment.