MODEL AND GENETIC ALGORITHM FOR DEPOT LOCATION AND VEHICLE ROUTING PROBLEM CONSIDERING MULTIPLE TRIPS, MULTIPLE TIME WINDOWS, AND SIMULTANEOUS DELIVERY-PICKUP

• Main Author: Jayadi, Okky
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/70480
• Institution: Institut Teknologi Bandung

The development of the location route problem model (LRP) is a combination of the vehicle route problem (VRP) with the selection of facilities. This study develops an MRL model by considering multiple trips, homogeneous fleet, multiple time windows, and simultaneous pickup and delivery. The development of this model is based on the system requirements regarding the selection of alternative facilities, and the selection of alternative time windows in customers. The objective function is to minimize the total cost of transportation which consists of total of fixed costs and total of variable costs. This research relaxes two assumptions in the classic LRP: (1) the customer can provide an alternative time to visit, and (2) the number of fleet available at the alternative depot is limited. Mathematical model development was carried out using LINGO 18.0 software, and algorithm development was carried out using Microsoft Visual Basic 6.0. The mathematical model and the proposed algorithm are subjected to unit verification and verification of the model run. Once verified, test the stability of the solution. The proposed algorithm uses a genetic algorithm that is adapted to the proposed model and has a percentage gap of 2.30% to the proposed mathematical model. The proposed algorithm is convergent, because the solution formed leads to a certain minimum value. The mixed time window type can result low objective functions, while the wide time window type can result in fast compute time. The proposed mathematical model and algorithm can be used for the following characteristics: (1) classical MRL, (2) mixed pickup-delivery, and (3) single time window.