A hybrid system design for the logistics of Unilever products to retailers with low sales potentials

The aim of this project is to suggest a profitable distribution system to deliver Unilever products to inaccessible retail points that have low sales potential but are large in number. First, factors that inhibit a profitable service to these retailers are listed. Based on the firm's requests and suggestions, a hybrid distribution system which aims to eliminate these factors - high transportation costs being the most relevant - is designed. The system which consists of forecast, phone-based delivery, and warehouse sales is first modeled as a mixed integer nonlinear program and later linearized with additional variables and constraints. Due to the size of the problem, a heuristic algorithm is developed after failing to solve the integer model using CPLEX. In order to test the effectiveness of the proposed algorithm, the results of 37 runs on smaller dimensional problems that have been solved to optimality are compared to results from the proposed heuristics. Having an average of 14.6% deviance from the optimum, the algorithm manages to solve 14 of these 37 instances to optimality. It is shown that the model also preserves its validity in a stochastic setting by a simulation study where deterministic parameters in the modeling stage are considered to be stochastic. The project also supported by applications such as database support and customer ID card design is in a testing stage. In summary, the project proposes a profitable and flexible distribution system to the firm and enables access to retail points with lower costs.