Optimization of last-mile delivery routing under stochastic demand using hybrid metaheuristics for sustainable Industrial and Production Engineering applications
Table Of Contents
Chapter ONE
INTRODUCTION
- 1.Introduction
- 1.1The introduction
- 1.2Background of Study
- 1.3Problem Statement
- 1.4Objectives of the Study
- 1.5Limitations of the Study
- 1.6Scope of the Study
- 1.7Significance of the Study
- 1.8Structure of the Research
- 1.9Definition of Terms
Chapter TWO
LITERATURE REVIEW
- Content
- 2.1Conceptual Framework and Theoretical Foundations
- 2.2Last-Mile Delivery: Trends and Challenges
- 2.3Stochastic Demand and Uncertainty Modeling
- 2.4Routing Problems: From VRP to Stochastic VRP
- 2.5Metaheuristic Approaches in Logistics and Transportation
- 2.6Hybrid Metaheuristics: Design and Applications
- 2.7Sustainability and Green Logistics in Industrial and Production Engineering
- 2.8Real-Time Routing and Dynamic Environments
- 2.9Data-Driven Decision Making in Supply Chains
- 2.10Case Studies in Last-Mile Delivery and Optimization
Chapter THREE
RESEARCH METHODOLOGY
- 3.1Research Philosophy and Approach
- 3.2Research Design
- 3.3Problem Formulation and Modeling Assumptions
- 3.4Mathematical Model for Stochastic Last-Mile Routing
- 3.5Uncertainty Modeling of Demand and Travel Times
- 3.6Hybrid Metaheuristic Framework Design (e.g., GA-PSO-AC, etc.)
- 3.7Algorithm Development and Pseudocode
- 3.8Data Collection and Preprocessing
- 3.9Simulation Environment and Tools
- 3.10Performance Metrics and Evaluation Criteria
- 3.11Experimental Design and Benchmark Instances
- 3.12Validation and Robustness Analysis
Chapter FOUR
DATA PRESENTATION AND ANALYSIS
- Results, Analysis and Discussion
- 4.1Experimental Setup and Instance Generation
- 4.2Baseline Comparisons and Benchmarking
- 4.3Computational Complexity and Convergence
- 4.4Route Quality under Stochastic Demand
- 4.5Impact of Uncertainty on Delivery Time and Cost
- 4.6Sensitivity Analysis of Key Parameters
- 4.7Sustainability Metrics and Environmental Impact
- 4.8Real-Time Adaptation and Dynamic Re-Optimization
- 4.9Practical Implications for Industry
- 4.10Case Study Discussion and Insights
Chapter FIVE
SUMMARY, CONCLUSION AND RECOMMENDATIONS
- and Summary
- 5.1Summary of Findings
- 5.2Theoretical Contributions
- 5.3Practical Implications for Industrial and Production Engineering
- 5.4Limitations and Assumptions Revisited
- 5.5Recommendations for Future Research
- 5.6Final Conclusions
Project Abstract
This study tackles the complex problem of optimizing last-mile delivery routing under stochastic demand by developing and evaluating a hybrid metaheuristic framework tailored for sustainable Industrial and Production Engineering applications. The core objective is to minimize total operational costs while ensuring service level requirements, balancing reliability, speed, and environmental impact in urban and peri-urban distribution networks. We model demand uncertainty through probabilistic distributions derived from historical order data, temporal patterns, and external factors such as weather and traffic, enabling robust route planning that hedges against variability. The proposed hybrid algorithm integrates elements of metaheuristics including genetic algorithms, ant colony optimization, and variable neighborhood search, augmented with machine learning-driven demand forecasting and adaptive route re-optimization mechanisms. A key feature is the incorporation of dynamic last-mile constraints such as time windows, vehicle capacity, route duration, and multi-depot considerations, alongside sustainable objectives like fuel consumption, emissions, and vehicle utilization efficiency. The methodology comprises four main components. First, a data-driven demand forecasting module leverages probabilistic models and ensemble learning to estimate demand distributions at the customer level, providing scenario-based inputs for routing under uncertainty. Second, a diversified hybrid search engine combines exploration and exploitation through staged operators that adapt to real-time feedback, allowing rapid re-optimization as new orders arrive or demand realizations deviate from forecasts. Third, a multi-objective optimization core uses a weighted Pareto approach to simultaneously minimize total distance, time, and emissions, with policy-driven adjustments to prioritize sustainability or service levels as needed. Fourth, an implementation and computational experiment framework evaluates the approach on real-world-like datasets and synthetic scenarios reflecting varying demand volatility, fleet sizes, and network topologies. To validate performance, the study conducts extensive computational experiments comparing the proposed hybrid framework against state-of-the-art routing algorithms and baseline heuristics, under multiple stochastic demand scenarios. Performance metrics include total travel distance, travel time, fleet utilization, lateness, penalty costs for unmet service levels, and environmental indicators such as CO2 and fuel consumption. Sensitivity analyses examine the impact of demand uncertainty, fleet heterogeneity, and parameter settings on solution quality and computational efficiency. The results are analysed with statistical tests to confirm robustness and generalizability across different market conditions. The research contributes a novel, scalable methodology for last-mile routing under uncertainty that harmonizes operational efficiency with sustainability goals. It provides practical insights for logistics practitioners seeking resilient distribution strategies in the face of demand volatility, and it offers a transferable framework adaptable to emerging urban logistics challenges, including micro-fulfillment integration and electrified fleets. The study also outlines guidelines for implementing the framework in industry contexts, including data requirements, computational considerations, and governance of algorithmic decision-making to support responsible and transparent logistics operations.
Project Overview
What This Project Is About
A practical look at how delivery routes can be planned more efficiently when demand is uncertain. The project combines smart routing ideas with flexible decision-making tools to improve speed, cost, and sustainability in last-mile delivery.
The Problem It Addresses
Deliveries often face unpredictable demand, traffic, and time windows. Traditional routing assumes fixed demand which leads to inefficiencies, higher fuel use, and longer wait times. This project tackles those gaps by exploring methods that adapt to changing situations.
Objectives of the Project
- Understand how stochastic (random) demand affects delivery routing.
- Explore hybrid metaheuristic methods to find good routing decisions quickly.
- Assess environmental and cost impacts of improved routing.
- Provide a simple framework that could be used by small businesses.
What You Will Do Step by Step
1) Learn basic concepts of routing and randomness and how they are modeled. 2) Review simple routing methods used in practice. 3) Implement a basic routing solver and then add hybrid metaheuristics. 4) Create test scenarios with uncertain demand. 5) Compare results to traditional routing. 6) Analyze cost and environmental effects. 7) Summarize findings and propose practical guidelines.
Expected Outcome
Clear insights into how hybrid metaheuristics can improve last-mile routing under uncertainty, plus a practical workflow and guidelines that can help businesses reduce costs and emissions.