Small and Medium-sized manufacturing Enterprises (SMEs) face significant barriers to Digital Twin (DT) adoption, mainly due to high implementation costs and complex data integration requirements. Moreover, from the operational perspective, SMEs operating in Make-to-Order (MTO) and Assemble-to-Order (ATO) environments experience difficulties in providing reliable order-level lead-time estimates, as high product variety, order-specific routings, and the coupling of capacity and material data are often distributed across heterogeneous systems. This thesis designs, implements and validates a lightweight, interoperable three-layer DT framework that horizontally integrates production and warehouse data, to enable order-level lead-time prediction and to support what-if analysis for planning. Considering the aforementioned challenges, the proposed framework is designed to be cost-efficient and to support seamless integration across existing systems; it is structured into three main layers: (i) the Data Source Layer leverages existing enterprise systems without modifying them; (ii) the Integration Layer manages the harmonization of datasets, ensuring that information from different sources is consistent and usable through a unified structure; and (iii) the Digital Twin Layer consumes the unified dataset to run order-level discrete-event simulations (DES), capturing capacity and precedence constraints, resource availability, and material synchronization. The framework is implemented and validated in an industrial case within the steel fabrication machinery sector, using real Enterprise Resource Planning (ERP), Manufacturing Execution System (MES), and Warehouse Management System (WMS) data to feed the DT simulation. Results demonstrate that the proposed architecture not only delivers accurate and reliable lead-time estimates with limited integration effort, but also provides a coherent approach to transform heterogeneous enterprise records into simulation-ready inputs. This enables SMEs to effectively instantiate and use an offline DES twin model through standardized data exports, confirming the feasibility and transferability of a low-cost Digital Twin solution in MTO/ATO contexts.
Le Piccole e Medie Imprese manifatturiere (PMI) affrontano ostacoli significativi nell’adozione dei Digital Twin (DT), principalmente per gli elevati costi infrastrutturali e requisiti di integrazione dati. In particolare, le PMI che operano in configurazioni Make-to-Order (MTO) e Assemble-to-Order (ATO) faticano a fornire stime affidabili dei lead time a livello d’ordine, poiché l’elevata varietà di prodotto, le regole di produzione, e l’integrazione tra capacità produttiva e disponibilità di materiali restano frammentati nei diversi sistemi informativi. Questa tesi progetta, implementa e valida un framework economico e interoperabile a tre livelli che integra orizzontalmente i dati di produzione e magazzino per fornire previsioni di lead time con il DT, e per supportare la pianificazione della produzione. L’architettura proposta è economica e interoperabile, ed è strutturata in: (i) un Livello Sorgenti Dati che utilizza i dati dai sistemi aziendali esistenti senza modificarli; (ii) un Livello di Integrazione che standardizza e uniforma i dati, garantendo che le informazioni provenienti da sistemi eterogenei siano consistenti e utilizzabili all’interno di un modello dati unificato; (iii) un Livello Digital Twin che consuma il dataset unificato ed esegue una simulazione a eventi discreti, rilevando vincoli di capacità e precedenze, disponibilità di risorse e sincronizzazione dei materiali. Il framework è implementato e validato in un’ azienda italiana che produce macchine di carpenteria, utilizzando esportazioni di dati reali per alimentare il DT. I risultati dimostrano che l’architettura proposta non solo fornisce stime dei lead time accurate e affidabili con un ridotto costo di integrazione, ma offre anche un approccio coerente per trasformare dati aziendali eterogenei in input pronti per la simulazione. Ciò consente alle PMI di implementare efficacemente un Digital Twin a eventi discreti offline, tramite esportazioni di dati standardizzate, confermando la possibilità di adottare e replicare una soluzione di Digital Twin a basso costo nei contesti MTO/ATO.
Proposal and validation of a low-cost, interoperable digital twin framework for production and logistics integration in manufacturing SMEs: an application for lead time prediction in a make to order strategy
SAVOINI, LUDOVICA;Palladini, Matilde
2024/2025
Abstract
Small and Medium-sized manufacturing Enterprises (SMEs) face significant barriers to Digital Twin (DT) adoption, mainly due to high implementation costs and complex data integration requirements. Moreover, from the operational perspective, SMEs operating in Make-to-Order (MTO) and Assemble-to-Order (ATO) environments experience difficulties in providing reliable order-level lead-time estimates, as high product variety, order-specific routings, and the coupling of capacity and material data are often distributed across heterogeneous systems. This thesis designs, implements and validates a lightweight, interoperable three-layer DT framework that horizontally integrates production and warehouse data, to enable order-level lead-time prediction and to support what-if analysis for planning. Considering the aforementioned challenges, the proposed framework is designed to be cost-efficient and to support seamless integration across existing systems; it is structured into three main layers: (i) the Data Source Layer leverages existing enterprise systems without modifying them; (ii) the Integration Layer manages the harmonization of datasets, ensuring that information from different sources is consistent and usable through a unified structure; and (iii) the Digital Twin Layer consumes the unified dataset to run order-level discrete-event simulations (DES), capturing capacity and precedence constraints, resource availability, and material synchronization. The framework is implemented and validated in an industrial case within the steel fabrication machinery sector, using real Enterprise Resource Planning (ERP), Manufacturing Execution System (MES), and Warehouse Management System (WMS) data to feed the DT simulation. Results demonstrate that the proposed architecture not only delivers accurate and reliable lead-time estimates with limited integration effort, but also provides a coherent approach to transform heterogeneous enterprise records into simulation-ready inputs. This enables SMEs to effectively instantiate and use an offline DES twin model through standardized data exports, confirming the feasibility and transferability of a low-cost Digital Twin solution in MTO/ATO contexts.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Palladini_Savoini_Executive_Summary.pdf
non accessibile
Descrizione: Executive Summary
Dimensione
6.35 MB
Formato
Adobe PDF
|
6.35 MB | Adobe PDF | Visualizza/Apri |
|
2025_12_Palladini_Savoini_Tesi.pdf
non accessibile
Descrizione: Testo Tesi
Dimensione
21.4 MB
Formato
Adobe PDF
|
21.4 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/10589/247482