INVENTORY ROUTING PROBLEM MODEL IN MULTI-VENDOR SINGLE-BUYER CONSIDERING THE VENDOR CAPACITY, THE VEHICLE TYPES, THE PICK-UP TIME, AND THE PART GROUPING

• Main Author: Marpaung, Budi
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84832
• Institution: Institut Teknologi Bandung

To reduce inventory and pick-up costs, mutually beneficial cooperation between suppliers/vendors and buyers is the basis for developing an integrated inventory and pick-up model known as the Inventory Routing Problem (IRP). The IRP is an inventory problem that is integrated with the vehicle route problem (VRP), which needs to be solved simultaneously. This study aims to produce an IRP model for the multi-vendor single buyer (MVSB) system considering the vendor capacity, the types of vehicles, the vehicle trip’s duration, and the grouping of parts for pick-up, so that the total relevant costs consisting of set-up costs, ordering costs, storage costs, and shipping costs, are kept to a minimum. This research produced five models. Model 1 is an IRP model that considers limited vendor capacity with one type of pickup vehicle and a milk-run transportation system without considering the duration of the vehicle trip. Model 2 is a development of Model 1 by also considering the duration of the vehicle trip. Model 2 consists of 2 sub-varians: Model 2A (unlimited number of vehicles with no maximum limit on vehicle trip duration), and Model 2B ( limited number of vehicles with a maximum limit on vehicle trip duration). Model 3 is a development of Model 2 by also considering the type of vehicle. Model 3 consists of 2 sub-varians: Model 3A ( the number of each type of vehicle is unlimited and without a maximum limit on the duration of the vehicle trip), and Model 3B (the number of each type of vehicle is limited and with a limit on the maximum duration of the vehicle trip). Model 4 is a development of Model 2 by also considering part grouping for pick-up. Model 4 consists of 2 sub-varians: Model 4A (unlimited number of vehicles and no maximum limit on vehicle trip duration), and Model 4B, (limited number of vehicles and no maximum limit on vehicle trip duration). Model 5 as the final model of this dissertation is a development of Model 4 by also considering the type of vehicle. Model 5 consists of 2 sub-varians: Model 5A, (the number of each type of vehicle is unlimited and there is no maximum limit on the vi duration of vehicle travel), and Model 5B (the number of each type of vehicle is limited and there is a maximum limit on the duration of vehicle travel). This research problem cannot be solved with an analytical approach because a number of decision variables are integers, some of which are non-linearly functions. Next, the five models were formulated as Mixed Integer Non-Linear Programming (MINLP) and solved using an exact approach with the help of Lingo 18.0. The results obtained show that the computing time increases exponentially as the problem size increases, which is characteristic of NP-hard problems. Thus, just like the analytical approach, the exact approach cannot be relied on to solve this research problem, especially when the complexity of the problem increases. To overcome the limitations of analytical and exact approaches in finding solutions, in this research, a proposed algorithm based on Ant Colony Optimization (ACO) was developed. The ACO algorithm developed uses a decomposition approach, namely breaking down the problem into two parts, namely the inventory sub- problem and the VRP sub-problem. There are two decision variables that simultaneously exist in both sub-problems, namely pick-up cycle time and order allocation. The decision variable that only exists in the inventory sub-problem is the pickup frequency, and in the VRP sub-problem is the vehicle route. Test results for small-scale problems show that the ACO algorithm has the same solution as the global optimum solution obtained using the exact approach, with much shorter computing time. Solutions can also be obtained for medium- and large-scale problems with realistic computing times. Thus, the ACO algorithm can be used to find problem solutions for various problem scales. The research results confirm the findings of a number of previous studies, which show that there is a trade-off between inventory costs and pick-up costs. The research results also confirm the findings of a number of previous studies, which show that the total costs relevant to the milk-run transportation system are lower compared to the independent transportation system (direct pick-up). The findings of this research are in line with the facts in the manufacturing industry, especially in the automotive industry, which was revealed in a number of studies to use a milk-run transportation system to pick-up parts from various vendor locations using several types of vehicles. This research recommends the pick-up of parts using a milk-run transportation system. However, there is still a need to develop models that can accommodate more realistic problem situation, such as dynamic and probabilistic parameter values and production activities for vendors and buyers, which are generally multi-stage.