vehicle routing problem with hard time window to optimize the total cost a case study of a trucking company

HoChiMinh City, 5 March 2023

Vehicle Routing Problem By

Tran Duy Phuc

Submitted in partial fulfillment of the requirements for the Degree of Bachelor of Engineering in Logistics and Supply Chain

Approved by:

Dean of IEM school Dr Nguyen Van Hop

1.4 | Scope and Delimitation: Chapter 2: RELATED WORKS

2.1 Overview:

2.1.1: Company Overview 2.1.2: Case Overview: 2.2 Literature Review:

2.3 Key references/ Candidate Solution Methods Chapter 3: METHODOLOGY

3.1 | Approaches Comparison and Selection:

3.2 Proposed System Design or Proposed Solution Approach:

List of Tables

Table I: Method comparIsơn c cà cà cà cà cà cày Table 2: Data nofatIOn QC 00c ng TH H ng TH kg TH kg vn key

16 20

VRP Vehicle Routing Problem

VRPTC Vehicle Routing Problem with Time Constraints

VRPTW Vehicle Routing Problem with Time Window

MT-TDVRPTW Multi-trip time-dependent vehicle routing problem with time windows

MIP Mix integer programming

Chapter 1: INTRODUCTION

1.1 Background: Why study this topic?

The motivation for a paper about the Vehicle Routing Problem with Time Constraints (VRPTC) and its application in a real trucking business case is to address the specific challenges and complexities that arise in real-world logistics operations

In the trucking industry, efficient routing and scheduling of vehicles is essential to ensure timely delivery of goods, minimize transportation costs, and maintain high levels of customer satisfaction However, the presence of time constraints, such as delivery windows, a driver working hours, and traffic congestion, can make the VRP even more challenging

Therefore, the paper aims to develop new and improved algorithms and methods to solve the VRPTC and to apply them to a real-world trucking business case This research can help to identify optimal routes and schedules for delivery trucks, taking into account the specific time constraints and operational requirements of the trucking company

Furthermore, by applying the research to a real-world case, the paper can provide insights and recommendations that can be directly implemented in practice This can result in tangible benefits for the trucking company, such as improved on-time delivery performance, reduced transportation costs, and increased operational efficiency

Overall, the motivation for a paper about the VRPTC and its application in a real trucking business case is to address the specific challenges and complexities of logistics operations in practice and to provide practical solutions that can benefit real-world trucking companies.

1.2 Problem Statement — The Need for Study:

Vehicle routing with time constraints is a crucial problem in logistics and transportation management It involves determining the optimal route for a fleet of vehicles to visit a set of customers while adhering to time constraints

This problem arises in various real-world scenarios, such as parcel delivery, goods delivery, and public transportation In these scenarios, the customers have time windows within which they can be visited, and the vehicles have a limited capacity and a fixed operational time window

If an inland trucking company has an unattracted service level that directly resolved unsatisfied customers or provides quality service, customers may switch to a competitor who is able to offer a more reliable service For example, if a trucking company frequently fails to meet delivery deadlines, it may need to pay penalties or incur additional expenses for expedited shipping to make up for the delay Additionally, a low service level may result in higher driver turnover rates and increased recruitment and training costs This can lead to a loss of business and revenue for the company, as well as a loss of customer trust and loyalty

In Vietnam setting, to be more specific Ho Chi Minh city, with container government input a time constraint on the downtown part Containers can only access this path between 22:00 - 6:00 which has always been a challenge when come to vehicle routing Moreover, with import-export shipments where the free time for each delivery is limited, any service failure will directly resolve costs this problem becomes more crucial than ever

In this case study, if the company fails to arrange its internal trucking to deliver, it will have to outsource the truck to meet the deadline This is a factor that they want to reduce since outsourcing costs them and contains hidden risks The need for this study vehicle routing with time constraints stems from the fact that it can significantly impact the efficiency and cost-effectiveness of logistics and transportation operations.

By optimizing vehicle routing, companies can reduce transportation costs, improve delivery times, and reduce outsourcing

Moreover, with the growing demand for same-day and next-day delivery services, there is a need for more efficient and optimized routing algorithms that can handle time constraints As such, studying vehicle routing with time constraints is crucial to help businesses improve their operations, leading to higher service levels and staying competitive in a rapidly evolving market

1.3 Objectives of Study:

e Improved Operational Efficiency: By optimizing the routing, the company can reduce travel time, fuel consumption, and vehicle wear and tear This can improve the overall operational efficiency of the company

e Reduced Costs: With optimized routing, the company can reduce transportation costs by minimizing the number of vehicles required and maximizing vehicle utilization This can lead to significant cost savings for the company

e Enhanced Service Level: By adhering to time constraints, the company can ensure timely deliveries and pickups, which can improve service levels e Better Resource Planning: The results of the study can help the company plan

its resources better For instance, the company can allocate vehicles and drivers based on the demand and expected delivery volumes

e Increased Revenue: By providing efficient and timely transportation services, the company can attract more customers, which can lead to increased revenue e Generating a mathematical model for routing with a consideration of various

constraints and time windows in real-life applications for import-export container trucking

e Implementing the mathematical model to provide optimal routing for the trucks.

1.4 Scope and Delimitation:

First, the location for this study only focuses on South Vietnam since the company we will investigate business mainly revolve around this area This means that we only examine the location of factory, warehouse, port, and ICD in the parameter of 300km around Ho Chi Minh city

Second, for the scope of this study, we assume that the vehicle will have a constrained velocity and no breakdown or damage will be investigated

Third, when considering only 20 and 40-inch containers with the same gross weight of 17 tons for 20-inch and 22 tons for 40-inch

Chapter 2: RELATED WORKS

2.1 Overview:

2.1.1: Company Overview

In this case, we are looking into DUY PHUC TRANSPORT TRADING COMPANY LIMITED This is a trucking business based in HCM city which was founded in 2014 The company's main palace of operation is located near Phu Huu Port (Terminal Cat Lai C) Quarter 4, Phu Huu Ward, District 9, City Ho Chi Minh, Vietnam

The firm specializes in container import-export trucking in South Vietnam, key segmentation of the company is garment Operation area from Binh Thuan to Ca Mau

10

11

Execute shipment Completed Is delay allow shipmnet able? Pick up new

shipments NO

The Multi-T VRP is an altered version of the classic VRP that enables vehicles to conduct multiple trips In their research, Wassan et al (2017) examined the Multi-T VRP with backhauls and created a two-tier variable neighborhood

13

search algorithm Nguyen et al (2013), Yan et al (2015), and Hernandez et al (2016) have investigated the combination of Multi-T VRP and VRPTW L Zhen et al (2020) demonstrated a case study on the multi-depot multi-trip VRP with time windows and release dates in their article published in Transportation Research Part E Additionally, Nguyen et al (2013) proposed a tabu search metaheuristic for the time-dependent Multi-zone Multi-T VRPTW, which enhanced route selection and vehicle allocation According to their experimental results, this approach produced superior outcomes in terms of the number of required vehicles and travel expenses

The constraint of vehicles’ arrival time at customers as well as time window constraints are taken into account by one well-establish form of VRP This alternative is known as the vehicle routing problem with time constraints (VRPTW) Many scholars have conducted advanced research in the VRPTW fields (Desrochers et al., 1992, Bent and Hentenryck, 2004, Alvarez and Munari, 2017)

Previous papers have investigated various vehicle routing problems such as time-dependent demands (Nguyen et al., 2013, 2017), or cost (Liu et al., 2018) While Ichoua et al (2003), Dabia et al (2013), and Sun et al (2013) consider time-dependent travel time, the multi-trip time-dependent vehicle routing problem with time windows (MTI-TDVRPTW) problem does not (2018b,a) MT-TDVRPTW problem is considered multi-trip because the vehicles may make multiple trips during the day, and it is time-dependent because the travel time between any two points depends on the time of day To the best of our knowledge, only Sun et al (2018c) have taken into account both time-dependent travel time and multiple trips at the same time Sun et al (2018c) developed a piecewise linear travel speed model that results in a quadratic travel time function, whereas the MI-TDVRPTW adheres to the widely used piecewise linear travel time function model (Ichoua et al., 2003).

14

In this paper, we want to investigate further the application of VRPTW in a real case To be more specific in South Vietnam setting, the goal is to minimize the time travel of each route and still make sure that it fits with the constraints Ultimately, the method will increase service levels to satisfy the customers 2.3 Key references/ Candidate Solution Methods

In the case of DUY PHUC TRANSPORT TRADING COMPANY LIMITE there are many approaches to be considered

Le (2020) conducted a case study of a logistics distribution network in Vietnam and proposed a mathematical model for optimizing vehicle routes The model takes into account a variety of factors, including vehicle capacity, time windows for pick-up and delivery, and distance between different locations In this paper, the researcher used Mix integer programming (MIP) combined with algorithms flowcharts to minimize the total cost of a logistic trucking company Besides total cost, they also analyzed the internal vehicle usage, number of trips, distance per trip, and internal vehicle utilization rate Some of the Heuristics are:

Guided local search (GLS) is a metaheuristic algorithm that combines local search with a guidance function that directs the search to promising search space regions The guidance function can be designed using problem-specific knowledge or information gleaned from previous search trajectories

Simulated annealing (SA) is a stochastic optimization algorithm that simulates the metallurgical process of annealing to find the global minimum of a cost function Based on a cooling schedule that controls the probability of accepting worse solutions over time, SA accepts worse solutions with a certain probability.

Tabu search (TS) is yet another metaheuristic algorithm that iteratively seeks a solution by selectively prohibiting certain moves in the search space for a set number of iterations The forbidden moves are saved in a "tabu list" to prevent the search from returning to the same solutions over and over TS guides the search toward promising solutions by using long-term memory of the search history

Guided local search

(GLS) GLS combines local search with perturbation and guidance phases, allowing for more efficient exploration of the search space

GLS can be effective in finding reasonable quality solutions for vehicle routing problems, especially when combined with efficient local search procedures

GLS can use various guidance functions to guide the search towards promising regions of the search space, which can help avoid local optima

The effectiveness of GLS can depend heavily on the quality of the guidance function used GLS may be more sensitive to the specific problem instance and the quality of the guidance function used In contrast, SA and TS may be more robust and can perform well across a wider range of problem instances

Simulated annealing (SA)

SA is effective in exploring the search space, allowing for a greater chance of finding a better quality solution

SA can avoid getting SA can be slow to converge to a good solution, as it relies on a cooling schedule to control the probability of accepting worse solutions

15

stuck in local optima by randomly accepting worse solutions, which can lead to a more diverse set of solutions being explored

SA requires careful tuning of parameters, such as the cooling schedule, in order to achieve good results

Tabu search (TS)

TS can be effective in avoiding revisiting previous solutions, which can help in finding good quality solutions TS can use different types of tabu lists to restrict the search space, which can help avoid local optima

TS can be slow to converge to a good solution, as it can restrict the search space too much and limit exploration TS requires careful tuning of parameters, such as the tabu tenure, in order to achieve good results

Table 1: Method comparison

When solving vehicle routing problems with time windows using heuristics there are various approaches each has its own strength and weaknesses Le (2020) obtained that GLS is the best approach out of the 3 since it reduced the most cost and used the most internal vehicle Moreover, Moan's (2022) paper solution shows that GLS 1s the method which generates better solutions under a shorter run time (under an hour) In real cases, the run time often needs to be quick in order to take swift action toward routing so GLS is much preferred W Madushan (2022) also backed this up, as the study discovered that GLS outperformed in terms of solution quality and computation time when solving the selected industry application Therefore, my research will use the MIP along with GLS to minimize the total cost for our case

16