The increasing complexity of chemical operations demands flexible dynamic simulation tools. However, a gap persists between robust but closed commercial simulators and open-source equation-based libraries, which often lack the necessary abstraction for practical engineering. This thesis bridges this gap by introducing OpenProc, a modular Object-Oriented Modeling framework developed in Modelica. The core contribution is a semantic-aware connector architecture that decouples process topology from chemical definitions. Unlike traditional array-based approaches or unstructured expandable connectors, the proposed library utilizes smart interfaces TransportStream and ReactionStream combined with a receiver-side mapping algorithm. This ensures structural chemical consistency and allows unit operations to adapt automatically to different fluid mixtures. Thermodynamic calculations are delegated to an external backend (CoolProp) to guarantee robustness and maintainability. The framework is validated through the implementation of the Tennessee Eastman benchmark process. The model is validated against a literature reference to ensure physical consistency, while a monolithic hard-coded baseline is used to verify the numerical integrity of the modular architecture. Results demonstrate that the proposed framework reproduces the complex open-loop dynamics of the reference and achieves exact numerical superposition with the monolithic baseline, proving that modularity introduces no approximation errors. This work proves that the flexibility of Object-Oriented Modeling can be achieved without compromising simulation accuracy, providing a transparent foundation for control-oriented process simulation.
La crescente complessità dei processi chimici richiede strumenti di simulazione dinamica flessibili. Tuttavia, esiste un netto divario tra i simulatori commerciali, robusti ma chiusi, e le librerie open-source, spesso prive del livello di astrazione necessario per l'ingegneria di processo. Questa tesi affronta tale problema proponendo OpenProc, un framework modulare sviluppato in Modelica. Il contributo fondamentale è un'architettura a connettori semanticamente consapevoli che disaccoppia la topologia del processo dalla definizione chimica. A differenza degli approcci tradizionali o dei connettori espandibili non strutturati, la libreria utilizza interfacce intelligenti TransportStream e ReactionStream abbinate a un algoritmo di mappatura automatico. Ciò garantisce la coerenza chimica strutturale e permette alle unità operative di adattarsi a diverse miscele fluide. I calcoli termodinamici sono delegati a un backend esterno (CoolProp) per assicurare robustezza e manutenibilità. Il framework è validato attraverso l'implementazione del processo benchmark Tennessee Eastman. Il modello è validato rispetto a un riferimento di letteratura per garantirne la coerenza fisica, mentre una baseline monolitica è utilizzata per verificare l'integrità numerica dell'architettura modulare. I risultati dimostrano che il framework proposto riproduce le complesse dinamiche a ciclo aperto del riferimento e raggiunge un'esatta sovrapposizione numerica con la baseline monolitica, provando che la modularità non introduce errori di approssimazione. Questo lavoro dimostra che la flessibilità della modellazione a oggetti può essere ottenuta senza compromettere l'accuratezza della simulazione, fornendo una base trasparente per la simulazione di processo orientata al controllo.
Design and validation of an object-oriented Modelica library for dynamic chemical process simulation
ESPOSITO, FEDERICO
2024/2025
Abstract
The increasing complexity of chemical operations demands flexible dynamic simulation tools. However, a gap persists between robust but closed commercial simulators and open-source equation-based libraries, which often lack the necessary abstraction for practical engineering. This thesis bridges this gap by introducing OpenProc, a modular Object-Oriented Modeling framework developed in Modelica. The core contribution is a semantic-aware connector architecture that decouples process topology from chemical definitions. Unlike traditional array-based approaches or unstructured expandable connectors, the proposed library utilizes smart interfaces TransportStream and ReactionStream combined with a receiver-side mapping algorithm. This ensures structural chemical consistency and allows unit operations to adapt automatically to different fluid mixtures. Thermodynamic calculations are delegated to an external backend (CoolProp) to guarantee robustness and maintainability. The framework is validated through the implementation of the Tennessee Eastman benchmark process. The model is validated against a literature reference to ensure physical consistency, while a monolithic hard-coded baseline is used to verify the numerical integrity of the modular architecture. Results demonstrate that the proposed framework reproduces the complex open-loop dynamics of the reference and achieves exact numerical superposition with the monolithic baseline, proving that modularity introduces no approximation errors. This work proves that the flexibility of Object-Oriented Modeling can be achieved without compromising simulation accuracy, providing a transparent foundation for control-oriented process simulation.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_03_Esposito_Thesis.pdf
accessibile in internet per tutti
Descrizione: Thesis
Dimensione
1.24 MB
Formato
Adobe PDF
|
1.24 MB | Adobe PDF | Visualizza/Apri |
|
2026_03_Esposito_Executive_Summary.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
757.42 kB
Formato
Adobe PDF
|
757.42 kB | 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/10589/250865