The current industrial context is highly dynamic, with frequent variations in product portfolios and rapidly increasing customer demands. To thrive in this environment, manufacturing companies must continuously adapt their systems through frequent reconfigurations. In order to support performance evalution and continuous improvement, modeling of manufacturing systems has emerged as a common practice, facilitated by the development of digital solutions. As a result, many different models are present in modern manufacturing systems, but their integration remains challenging. This doctoral thesis addresses the issue of model integration by proposing a hybrid method that combines models of different sub-systems, independently developed with analytical or simulation approaches, into a unified system modeling architecture. The method provides steady-state performance evaluation to support long-term configuration management of evolving systems. The proposed method is based on a decomposition approach, implemented through the formalization of general interfaces that rely on the state-based representation of material inflow and outflow of sub-systems. By integrating diverse models into a cohesive framework, the method offers a comprehensive evaluation of system performance, considering the interdependencies between subsystems. Numerical evaluations demonstrate the accuracy of the proposed method compared to more commonly employed tools, such as discrete event simulation. Analysis of the solution algorithm reveals its robustness to the propagation of uncertainty deriving from the presence of simulation models in the algorithm itself and characterizes the variability of output performance indicators. The method is applied to the reconfiguration analysis of a manufacturing system for electrical distribution equipment. Specifically, it is adopted to evaluate the impact on system performance of major and minor reconfiguration actions, in presence of increasing customer demand, proving the utility of the approach in real industrial contexts.
L'attuale contesto industriale è estremamente dinamico, con frequenti variazioni nelle linee di prodotto e una rapida crescita della domanda. Per svilupparsi in questo ambiente, le aziende manifatturiere devono adattare continuamente i loro sistemi attraverso frequenti riconfigurazioni. Per supportare la valutazione prestazioni e il miglioramento continuo, la modellazione dei sistemi manifatturieri è emersa come una pratica comune, facilitata dallo sviluppo di soluzioni digitali. Di conseguenza, nei moderni sistemi manifatturieri sono presenti molti modelli diversi, ma la loro integrazione rimane una sfida. Questa tesi di dottorato affronta il problema dell'integrazione dei modelli proponendo un metodo ibrido che combina modelli di diversi sottosistemi, sviluppati indipendentemente con approcci analitici o di simulazione, in un'unica architettura di modellazione di sistema. Il metodo fornisce una valutazione delle prestazioni a regime per supportare la gestione di lungo termine della configurazione di sistemi in continua evoluzione. Il metodo proposto si basa su un approccio di decomposition, implementato attraverso la formalizzazione di interfacce generali che si basano su una rappresentazione a stati dell'afflusso e deflusso di materiali dai sottosistemi. Integrando modelli diversi in un quadro coerente, il metodo offre una valutazione completa delle prestazioni del sistema, considerando le interdipendenze tra i diversi sottosistemi. L'analisi numerica dimostra l'accuratezza del metodo nella valutazione delle prestazioni rispetto ad altri strumenti comunemente impiegati, come la simulazione ad event discreti. L'analisi dell'algoritmo rivela la sua robustezza alla propagazione di incertezza derivante dalla presenza di modelli di simulazione nell'algoritmo stesso e caratterizza la variabilità degli indicatori di prestazione in output. Il metodo è applicato all'analisi di riconfigurazione di un sistema manifatturiero per dispositivi di distribuzione elettrica. In particolare, viene adottato per valutare l'impatto di azioni di riconfigurazione di diversa entità sulle prestazioni del sistema, in presenza di una crescente domanda da parte dei clienti, dimostrando l'utilità dell'approccio in contesti industriali reali.
A hybrid method combining analytical and simulation models for performance evaluation of large manufacturing systems
Mastrangelo, Matteo
2023/2024
Abstract
The current industrial context is highly dynamic, with frequent variations in product portfolios and rapidly increasing customer demands. To thrive in this environment, manufacturing companies must continuously adapt their systems through frequent reconfigurations. In order to support performance evalution and continuous improvement, modeling of manufacturing systems has emerged as a common practice, facilitated by the development of digital solutions. As a result, many different models are present in modern manufacturing systems, but their integration remains challenging. This doctoral thesis addresses the issue of model integration by proposing a hybrid method that combines models of different sub-systems, independently developed with analytical or simulation approaches, into a unified system modeling architecture. The method provides steady-state performance evaluation to support long-term configuration management of evolving systems. The proposed method is based on a decomposition approach, implemented through the formalization of general interfaces that rely on the state-based representation of material inflow and outflow of sub-systems. By integrating diverse models into a cohesive framework, the method offers a comprehensive evaluation of system performance, considering the interdependencies between subsystems. Numerical evaluations demonstrate the accuracy of the proposed method compared to more commonly employed tools, such as discrete event simulation. Analysis of the solution algorithm reveals its robustness to the propagation of uncertainty deriving from the presence of simulation models in the algorithm itself and characterizes the variability of output performance indicators. The method is applied to the reconfiguration analysis of a manufacturing system for electrical distribution equipment. Specifically, it is adopted to evaluate the impact on system performance of major and minor reconfiguration actions, in presence of increasing customer demand, proving the utility of the approach in real industrial contexts.File | Dimensione | Formato | |
---|---|---|---|
PhD_Thesis_Mastrangelo.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Tesi
Dimensione
13.34 MB
Formato
Adobe PDF
|
13.34 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/10589/220193