This PhD thesis develops advanced methodologies for optimizing critical decision-making processes in real-world pharmaceutical logistics and healthcare services. Specifically, it focuses on the vehicle routing problem (VRP) and its complex variants, as well as on the optimization of emergency medical services infrastructure. The thesis presents innovative models and decision-support systems aimed at improving operational efficiency in these domains. The first part of the thesis addresses a variant of the VRP known as the multi-trip VRP with time windows. This complex, real-world problem arises in logistics settings, such as pharmaceutical distribution, where multiple depots, heterogeneous fleets, and time-constrained deliveries must be managed. A two-phase optimization approach is proposed: in the first phase, vehicle routes are generated based on customer demands and operational constraints, while in the second phase these routes are combined to minimize the number of vehicles required. For this optimization task, four solution methods are proposed and evaluated: two greedy heuristics, a metaheuristic algorithm, and a mixed-integer linear programming (MILP) model. Extensive computational experiments on realistic instances demonstrate the effectiveness of these methods, with the metaheuristic approach providing superior results in terms of solution quality and computational time. The second part extends the VRP analysis to a multi-period, multi-trip, multi-depot variant, considering logistic constraints over multiple planning periods. This phase develops a variable neighbourhood search (VNS) metaheuristic combined with a MILP model to handle large-scale, real-world problems. Computational results on practical instances confirm the VNS-based algorithm’s superior performance compared to traditional approaches, achieving significant reductions in the number of vehicles used across periods. The third part of the thesis shifts focus to healthcare, specifically the optimization of helicopter emergency medical services (HEMS) for trauma management in road traffic accidents. Two different bi-objective optimization models are developed to locate helipads and transfer points for helicopters, minimizing response time and maximizing geographic coverage under time-critical conditions. Using real-world data from the Parma province, the models offer valuable insights for HEMS infrastructure planning, providing actionable recommendations for public agencies to enhance emergency response efficiency. Overall, this thesis contributes to the fields of Operations Research and healthcare logistics by proposing state-of-the-art algorithms and decision-support systems. These solutions are validated through real-world applications, demonstrating their potential to improve both operational efficiency and service quality in logistics and emergency medical services.
Questa tesi di dottorato sviluppa metodologie avanzate per ottimizzare i processi decisionali critici nei settori della logistica farmaceutica e dei servizi sanitari. In particolare, si concentra sul problema dell’instradamento dei veicoli (VRP) e le sue complesse varianti, oltre all'ottimizzazione delle infrastrutture dei servizi medici di emergenza. La tesi presenta modelli innovativi e sistemi di supporto alle decisioni volti a migliorare l'efficienza operativa in questi ambiti. La prima parte della tesi affronta una variante del VRP nota come VRP con viaggi multipli e finestre temporali. Questo problema complesso e reale emerge in diversi contesti logistici, come la distribuzione farmaceutica, dove devono essere gestiti più depositi, flotte eterogenee di veicoli e consegne vincolate nel tempo. Viene proposto un approccio di ottimizzazione a due fasi: nella prima fase, i percorsi dei veicoli vengono generati in base alle richieste dei clienti e ai vincoli operativi, mentre nella seconda fase questi percorsi vengono combinati per minimizzare il numero di veicoli necessari. Per questo problema vengono proposti e valutati quattro metodi risolutivi: due euristiche greedy, un algoritmo metaeuristico e un modello di programmazione lineare intera mista (MILP). Ampi esperimenti computazionali su casi realistici dimostrano l'efficacia di questi metodi, con l'algoritmo metaeuristico che fornisce i migliori risultati in termini di qualità della soluzione e tempi computazionali. La seconda parte estende l'analisi del VRP a una variante multi-periodo, multi-viaggio e multi-deposito, considerando vincoli logistici su più periodi di pianificazione. In questa fase viene sviluppata una metaeuristica basata sulla variable neighbourhood search (VNS) combinata con un modello di MILP per gestire problemi su larga scala e casi reali. I risultati computazionali su casi pratici confermano le prestazioni superiori della VNS rispetto agli approcci tradizionali, ottenendo riduzioni significative nel numero di veicoli utilizzati su più periodi temporali. La terza parte della tesi si concentra sul settore sanitario, con particolare riferimento all'ottimizzazione dei servizi medici di emergenza in elicottero per la gestione dei traumi da incidenti stradali. Vengono sviluppati due diversi modelli di ottimizzazione bi-obiettivo per localizzare eliporti e punti di trasferimento per elicotteri, minimizzando i tempi di risposta e massimizzando la copertura geografica in condizioni di emergenza. Utilizzando dati reali provenienti dalla provincia di Parma, i modelli offrono preziose indicazioni per la pianificazione delle infrastrutture di emergenza, fornendo raccomandazioni operative per le agenzie pubbliche al fine di migliorare l'efficienza delle risposte alle emergenze. In sintesi, questa tesi contribuisce ai campi della Ricerca Operativa e della logistica sanitaria proponendo algoritmi e sistemi di supporto alle decisioni all'avanguardia. Queste soluzioni sono validate attraverso applicazioni reali, dimostrando il loro potenziale nel migliorare l'efficienza operativa e la qualità del servizio nei settori della logistica e dei servizi medici di emergenza.
Tecniche di Analisi Dati e Ottimizzazione per i Processi Logistici
CAVECCHIA, MIRKO
2025
Abstract
This PhD thesis develops advanced methodologies for optimizing critical decision-making processes in real-world pharmaceutical logistics and healthcare services. Specifically, it focuses on the vehicle routing problem (VRP) and its complex variants, as well as on the optimization of emergency medical services infrastructure. The thesis presents innovative models and decision-support systems aimed at improving operational efficiency in these domains. The first part of the thesis addresses a variant of the VRP known as the multi-trip VRP with time windows. This complex, real-world problem arises in logistics settings, such as pharmaceutical distribution, where multiple depots, heterogeneous fleets, and time-constrained deliveries must be managed. A two-phase optimization approach is proposed: in the first phase, vehicle routes are generated based on customer demands and operational constraints, while in the second phase these routes are combined to minimize the number of vehicles required. For this optimization task, four solution methods are proposed and evaluated: two greedy heuristics, a metaheuristic algorithm, and a mixed-integer linear programming (MILP) model. Extensive computational experiments on realistic instances demonstrate the effectiveness of these methods, with the metaheuristic approach providing superior results in terms of solution quality and computational time. The second part extends the VRP analysis to a multi-period, multi-trip, multi-depot variant, considering logistic constraints over multiple planning periods. This phase develops a variable neighbourhood search (VNS) metaheuristic combined with a MILP model to handle large-scale, real-world problems. Computational results on practical instances confirm the VNS-based algorithm’s superior performance compared to traditional approaches, achieving significant reductions in the number of vehicles used across periods. The third part of the thesis shifts focus to healthcare, specifically the optimization of helicopter emergency medical services (HEMS) for trauma management in road traffic accidents. Two different bi-objective optimization models are developed to locate helipads and transfer points for helicopters, minimizing response time and maximizing geographic coverage under time-critical conditions. Using real-world data from the Parma province, the models offer valuable insights for HEMS infrastructure planning, providing actionable recommendations for public agencies to enhance emergency response efficiency. Overall, this thesis contributes to the fields of Operations Research and healthcare logistics by proposing state-of-the-art algorithms and decision-support systems. These solutions are validated through real-world applications, demonstrating their potential to improve both operational efficiency and service quality in logistics and emergency medical services.File | Dimensione | Formato | |
---|---|---|---|
PhD Thesis - Mirko Cavecchia.pdf
embargo fino al 27/03/2026
Dimensione
19.31 MB
Formato
Adobe PDF
|
19.31 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/202075
URN:NBN:IT:UNIMORE-202075