The e-grocery sector is experiencing constant growth, yet last-mile distribution remains a structural challenge. Delivering fresh groceries in densely populated urban areas is costly, emission-intensive, and complicated by product perishability. A less-studied dimension in this field is food waste: in the existing literature, uncollected orders are typically treated as a fixed penalty, decoupled from the actual network design. This thesis aims to bridge this research gap by developing a comprehensive multi-objective optimization framework for urban e-grocery distribution. This project proposes a ``Depot-to-Locker'' distribution model, relying on pre-existing warehouses and a network of refrigerated Pick-up Points (PUPs). To capture the complexities of a metropolitan area, the model integrates a heterogeneous fleet routing system and accounts for Limited Traffic Zone (LTZ) restrictions. The core innovation of this research is the integration of a behavioral waste function, which correlates the physical distance between the customer and the locker with the probability of order abandonment (the ``no-show'' phenomenon), directly translating into physiological food waste. The model was solved through a two-stage approach that uses Mixed-Integer Linear Programming (MILP) with Nearest Neighbor Heuristic algorithm, allowing the analysis to be done on a city scale. It is shown that the Pareto Frontier Analysis suggests that there is an inherent conflict among the three objectives, where any improvement in one objective is achieved by sacrificing another. The model enables city managers as well as e-retailers to calculate the cost of trade-off among the economic, environmental, and social objectives.
Il settore dell'e-grocery sta registrando una crescita costante, tuttavia la distribuzione dell'ultimo miglio rimane una sfida strutturale. La consegna di generi alimentari freschi in aree urbane densamente popolate è costosa, genera alte emissioni ed la deperibilità dei prodotti ne complica le dinamiche. Una dimensione meno studiata in questo campo è lo spreco alimentare: nella letteratura esistente, gli ordini non ritirati vengono solitamente trattati come una penalità fissa, slegata dall'effettivo design della rete logistica. Questa tesi mira a colmare tale lacuna di ricerca sviluppando un framework completo di ottimizzazione multi-obiettivo per la distribuzione urbana dell'e-grocery. Questo progetto propone un modello di distribuzione ``Depot-to-Locker'', basato su magazzini preesistenti e su una rete di punti di ritiro refrigerati (PUP - Pick-up Points). Per descrivere le complessità di un'area metropolitana, il modello integra un sistema di routing con flotta eterogenea e tiene conto delle restrizioni delle Zone a Traffico Limitato (ZTL). L'innovazione centrale di questa ricerca è l'integrazione di una funzione comportamentale dello spreco, che correla la distanza fisica tra il cliente e il locker con la probabilità di abbandono dell'ordine (il fenomeno del ``no-show''), traducendosi direttamente in spreco alimentare fisiologico. Il modello è stato risolto utilizzando un approccio a due fasi che combina la Programmazione Lineare Intera Mista (MILP) con un'euristica Nearest Neighbor, consentendo calcoli su scala cittadina. L'analisi della Frontiera di Pareto rivela che i tre obiettivi sono strutturalmente in conflitto: qualsiasi vantaggio ottenuto in una dimensione, emissioni, costi o spreco alimentare, va inevitabilmente a scapito di un altro obiettivo. Il framework consente agli urbanisti e ai rivenditori online di quantificare il costo esatto di ogni compromesso strategico tra obiettivi economici, ambientali e sociali.
Optimizing last-mile e-grocery distribution: a multi-objective approach integrating logistic costs, environmental impact, and behvioral food waste
Roasio, Emanuele Luigi
2025/2026
Abstract
The e-grocery sector is experiencing constant growth, yet last-mile distribution remains a structural challenge. Delivering fresh groceries in densely populated urban areas is costly, emission-intensive, and complicated by product perishability. A less-studied dimension in this field is food waste: in the existing literature, uncollected orders are typically treated as a fixed penalty, decoupled from the actual network design. This thesis aims to bridge this research gap by developing a comprehensive multi-objective optimization framework for urban e-grocery distribution. This project proposes a ``Depot-to-Locker'' distribution model, relying on pre-existing warehouses and a network of refrigerated Pick-up Points (PUPs). To capture the complexities of a metropolitan area, the model integrates a heterogeneous fleet routing system and accounts for Limited Traffic Zone (LTZ) restrictions. The core innovation of this research is the integration of a behavioral waste function, which correlates the physical distance between the customer and the locker with the probability of order abandonment (the ``no-show'' phenomenon), directly translating into physiological food waste. The model was solved through a two-stage approach that uses Mixed-Integer Linear Programming (MILP) with Nearest Neighbor Heuristic algorithm, allowing the analysis to be done on a city scale. It is shown that the Pareto Frontier Analysis suggests that there is an inherent conflict among the three objectives, where any improvement in one objective is achieved by sacrificing another. The model enables city managers as well as e-retailers to calculate the cost of trade-off among the economic, environmental, and social objectives.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_07_Roasio_Tesi.pdf
non accessibile
Descrizione: Testo Tesi
Dimensione
10.85 MB
Formato
Adobe PDF
|
10.85 MB | Adobe PDF | Visualizza/Apri |
|
2026_07_Roasio_Executive Summary.pdf
non accessibile
Descrizione: Executive Summary
Dimensione
1.67 MB
Formato
Adobe PDF
|
1.67 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/260203