eThesis - Kauppakorkeakoulun sähköiset gradut

Every modern company has to deal with some kind of forecasting. Whether it is capacity, demand, cash flow or some other kind of forecasting, there is always one common factor: uncertainty. This uncertainty has to be quantified in order to plan the optimal way of conducting future business.

The objective of this study is to research a case analysis of a capacity planning problem. The analyzed problem is a logistical capacity problem resembling newsvendor problem, where both overstocking and understocking carries an extra cost. A third party logistical company has to forecast the optimal amount of delivery trucks to hire for the next week based only on delivery data from previous deliveries. They have made a customer promise to deliver the packages on the promised date or they will be delivered for free during the next date. If they rent too few vehicles, they cannot make the deliveries on the promised date, but if they hire too many they will pay for half-empty cars for nothing. The aim of this study is to solve the optimal amount of ordered trucks for each week and to investigate if they newsvendor model is actually a suitable model for solving such problem.

Methodology of the study is to first review the scientific literature of the newsvendor model and it's applications. To object is to recognize how the model should be used to evaluate the problem at hand and which factors should be used to review the outcome. After the literature review is completed, the outcome of the constructed model is reviewed under the recognized criteria. The recognized important parameters are the delivery fleet size, net profit, delivery rate and expected utility. Each potentially optimal solution is reviewed under these terms and then compared to other options as a whole. After this, the findings are reflected and discussed.

The main finding of this study is that the solution provided by newsvendor model was well in line with recognized real life evaluation criteria, which would suggest that newsvendor model is a suitable model for solving such problems. The suggested optimal solution was found to be a very good solution in terms of each individual review criteria, but it was also the optimal solution when taking all review parameters in the account. These findings are reflected with limitations of the study before providing future research suggestions and conclusion.