Routing goods transportation in the supply chain under conditions of uncertainty using genetic algorithm

Dissertation for Master's Degree "MSC"

Field: Computer

Orientation: Software

Abstract

One of the issues that is very important in the supply chain today and extensive studies have been done in its field is the routing of freight vehicles to deliver goods to applicants. To solve this problem, the objective function should be optimized in such a way that criteria such as distance traveled, travel time, and the number of vehicles are minimized and the objective function is minimized and ultimately customer satisfaction is maximized. This problem is NP-hard and meta-heuristic methods are often used to solve it. In the real world, the existence of some factors makes the vehicle routing problem an uncertain problem. One type of uncertainty in this issue is the occurrence of variable demand from customers; That is, the amount of demand of some customers is uncertain, and only when the vehicle reaches the customer's place, his demand is determined. In this article, a method based on robust genetic algorithm is presented to solve the routing problem of cargo carrying vehicles with variable demand. In this method, an attempt is made to find resistant solutions for this problem that maintain their optimality in the face of changes. The evaluations performed and the comparison of the results have shown the effectiveness of the proposed method. Keywords: vehicle routing, uncertainty, variable demand, resistant genetic algorithm, stable solutions. Introduction In recent decades, meeting the needs of customers, who are the most important component of the supply chain, is one of the key goals of supply chain management. The vehicle routing problem (VRP) [1] is the most important and costly part in logistics. In the VRP problem, there are groups of vehicles that are required to move from the warehouse [2] to meet the demands of customers and return to the warehouse after serving all the customers. Limitations and assumptions have been added to this problem, which has led to the emergence of different types of it and has developed its application in different fields.

In real world practical problems, there are different needs that fulfilling more of them makes the problem more complicated; In other words, taking into account the existing limitations indicates a significant increase in complexity in the VRP problem. In fact, it is the existence of these limitations that makes VRP an indeterminate and variable problem; In other words, changing goals, the nature of the problem, or other factors over time, which are the uncertainties surrounding the problem, may cause a change in the optimality of these problems. If we include these uncertainties in the optimization process, the problem becomes a dynamic problem. Now, to solve such dynamic problems, there is a need for solutions that do not lose their optimality when dealing with the uncertainty of the problem; These answers are called "stable answers"[3]. Stable solutions are actually those solutions that can still perform well in the worst conditions of the problem. Instead of the title "stable", titles such as "stable" and "resistant" may also be used.

In this research, our goal is to present a method for solving VRP based on genetic algorithm (GA) [4] that can achieve robust solutions for this problem that can still maintain their optimality even in the conditions of uncertainty.

As we mentioned earlier, the addition of a series of restrictions causes the creation of different types of VRP becomes; In our study, the addition of uncertainty to the problem in the form of "uncertainty of customer demand" has led to the creation of VRP with random demand (VRPSD) [5]. VRPSD is also called Probabilistic Vehicle Routing Problem (PVRP)[6].

One of the most important factors of creating uncertainty in VRP is the uncertainty in customer demand, which in this research is also our focus on customer uncertainty. The occurrence of customer demand uncertainty is that the amount of demand is uncertain until the vehicle arrives at the customer's location; If the amount of demand is less than the load in the vehicle, the customer receives service and the vehicle goes to the next customer for service. Otherwise, a negative charge will be considered for the current customer, and the vehicle will be transferred to the next customer after going to the warehouse and reloading.Otherwise, a negative cost is considered for the current customer and the vehicle goes to the next customer after going to the warehouse and reloading. The objective function of the proposed method, in addition to a negative cost caused by the lack of service to customers, includes the cost of traveling between customers and the warehouse.

In order to evaluate the proposed method, we compared it with the methods of "genetic algorithm", "particle swarm optimization" and "combined particle swarm optimization". The obtained results show the effectiveness of the proposed method compared to other methods.

The structure of this thesis is as follows:

The first chapter describes the issue of vehicle routing in the supply chain. The second chapter presents uncertainty in problems and its management with the help of genetic algorithm, the third chapter has an overview of the problem of uncertain vehicle routing. The fourth and fifth chapters respectively state the proposed method and the evaluation of the proposed method, and finally, in the sixth chapter, the conclusions and future suggestions are mentioned. The first chapter

The issue of vehicle routing in the supply chain

Introduction

In today's industrial world where there is a tight competition between companies and manufacturers, producing a variety of high quality products according to the customer's demands and at the same time, the right price according to the customer's purchasing power, is the only way to stay and the competition between Companies are in competition among their components (Makui and Iftikhar, 2013). The customer's demand for high quality and fast service has increased pressures that did not exist before. As a result, companies can no longer handle all the work alone. Activities such as supply and demand planning, material procurement, production, quality control, product or service planning, product maintenance and inventory control, marketing, distribution of goods and services, delivery and customer service, which used to be done at the company level, have now been transferred to the supply chain level. The key issue in a supply chain is the coordinated management and control of all these activities so that customers can receive their service with the highest level of reliability, speed, quality and reasonable cost. Transportation activities occupy a relatively large part of the supply chain activities (Hospital Management Articles Bank, 2010; Nowrozi, 2018). Logistics and supply chain is a word that has been heard a lot in recent years in our country, especially these days under titles such as "supply chain management", "pure logistics" and "logistics and supply chain association" (Logistics Association of Iran, 2012). In this chapter, materials are presented about "supply chain" and also the role of "vehicle routing problem" in the supply chain.

Supply chain (SC) [1]

Supply chain includes all activities related to the flow and transformation of goods from the stage of raw material (extraction) to delivery to the final consumer, as well as information flows related to them; These activities include sourcing for the supply of raw materials, managing systems, assemblies, sales, etc. (Nowrozi, 2018; Makoui and Iftikhar, 2018).

In other words, a supply chain includes two or more organizations that are legally separate, but are connected through material, information, and financial flows.  These organizations can be companies that produce parts, components, and final products, and even include supply and distribution service providers (logistics) and the customer (final) as well (Hospital Management Articles Bank, 2019). A supply chain model is shown in Figure (1-1).

Supply chain management (SCM)[2]

The key issue in a supply chain is the coordinated management and control of all activities such as distribution, delivery and customer service. In SCM, the goal is to coordinate the flow of financial information and materials so that customers can receive their services with the highest level of reliability, speed, quality and reasonable cost (Hospital Management Articles Bank, 2013; Makoui and Iftikhar, 2014). The reason is that today, from the point of view of the final customer, only one organizational unit alone is not responsible for the competitiveness of its products or services. Therefore, the competition has moved from individual companies to the supply chain (Hospital Management Articles Bank, 2018).