Optimization plays a central role in the design and development of chemical processes, enabling engineers to reduce costs, increase energy efficiency, and evaluate complex alternatives under uncertainty. In this context, superstructure-based methods offer a systematic and flexible way to identify optimal process configurations from a set of feasible technologies in a rapidly changing market. This thesis presents the development and implementation of a Branch and Bound algorithm for superstructure optimization in the process industry. The method is applied to a superstructure of a Rankine cycle modeled in Aspen HYSYS and controlled via Excel VBA. The algorithm systematically explores alternative process configurations by defining the flowsheet as a graph and partitioning it into interconnected "nodes" bounded by structural switches. Each node represents a subnetwork of units and streams. Its capital cost is evaluated using modular equations derived from Turton’s method. The decision tree is explored using a depth-first search (DFS) strategy. A costbased pruning mechanism is adopted to discard suboptimal branches. The algorithm successfully handles complex topologies including recycle streams, heat recovery, and shared units, without requiring manual intervention. Notably, this approach does not require any mathematical optimization model, making it intuitive and easily adaptable. The results confirm the robustness and computational efficiency of the method. Its graph-based logic allows for a clear and modular representation of alternative configurations, paving the way for future developments. In particular, such as the inclusion of operating costs, hybrid DFS/breadth-first search (BFS) strategies, and integration into more user-friendly interfaces., the approach could be extended to minimize environmental impacts via Life Cycle Assessment (LCA) by simply redefining the objective function.
L’ottimizzazione svolge un ruolo centrale nella progettazione e nello sviluppo dei processi chimici, permettendo agli ingegneri di ridurre i costi, aumentare l’efficienza energetica e valutare alternative complesse in condizioni di incertezza. In questo contesto, i metodi basati su superstrutture offrono un approccio sistematico e flessibile per identificare configurazioni ottimali di processo a partire da un insieme di tecnologie realizzabili in un mercato in rapida evoluzione. Questa tesi presenta lo sviluppo e l’implementazione di un algoritmo Branch and Bound per l’ottimizzazione di superstrutture nell’industria di processo. Il metodo è applicato a una superstruttura di un ciclo Rankine modellata in Aspen HYSYS e controllata tramite Excel VBA. L’algoritmo esplora sistematicamente configurazioni alternative del processo definendo il flowsheet come un grafo e suddividendolo in “nodi” interconnessi delimitati da switch strutturali. Ogni nodo rappresenta una sottorete di unità e correnti, il cui costo capitale è valutato utilizzando equazioni modulari derivate dal metodo di Turton. L’albero decisionale viene esplorato utilizzando una strategia di depth-first search (DFS). Un meccanismo di potatura basato sul costo viene adottato per scartare i rami subottimali. L’algoritmo gestisce con successo topologie complesse, incluse correnti di riciclo, recupero di calore e unità condivise, senza richiedere interventi manuali. In particolare, questo approccio non richiede alcun modello matematico di ottimizzazione, rendendolo intuitivo e facilmente adattabile. I risultati confermano la robustezza e l’efficienza computazionale del metodo. La sua logica basata su grafi consente una rappresentazione chiara e modulare delle configurazioni alternative, aprendo la strada a futuri sviluppi. In particolare, come l’inclusione dei costi operativi, strategie ibride DFS/BFS e l’integrazione in interfacce più user-friendly, l’approccio potrebbe essere esteso per minimizzare l’impatto ambientale tramite Life Cycle Assessment (LCA), semplicemente ridefinendo la funzione obiettivo.
Super-CORO : superstructure-based capex/opex robust optimizer : a process superstructure optimization framework
Lino Espinoza, Oscar Michael
2024/2025
Abstract
Optimization plays a central role in the design and development of chemical processes, enabling engineers to reduce costs, increase energy efficiency, and evaluate complex alternatives under uncertainty. In this context, superstructure-based methods offer a systematic and flexible way to identify optimal process configurations from a set of feasible technologies in a rapidly changing market. This thesis presents the development and implementation of a Branch and Bound algorithm for superstructure optimization in the process industry. The method is applied to a superstructure of a Rankine cycle modeled in Aspen HYSYS and controlled via Excel VBA. The algorithm systematically explores alternative process configurations by defining the flowsheet as a graph and partitioning it into interconnected "nodes" bounded by structural switches. Each node represents a subnetwork of units and streams. Its capital cost is evaluated using modular equations derived from Turton’s method. The decision tree is explored using a depth-first search (DFS) strategy. A costbased pruning mechanism is adopted to discard suboptimal branches. The algorithm successfully handles complex topologies including recycle streams, heat recovery, and shared units, without requiring manual intervention. Notably, this approach does not require any mathematical optimization model, making it intuitive and easily adaptable. The results confirm the robustness and computational efficiency of the method. Its graph-based logic allows for a clear and modular representation of alternative configurations, paving the way for future developments. In particular, such as the inclusion of operating costs, hybrid DFS/breadth-first search (BFS) strategies, and integration into more user-friendly interfaces., the approach could be extended to minimize environmental impacts via Life Cycle Assessment (LCA) by simply redefining the objective function.File | Dimensione | Formato | |
---|---|---|---|
2025_07_Lino Espinoza_Tesi.pdf
accessibile in internet per tutti
Descrizione: testo tesi
Dimensione
1.89 MB
Formato
Adobe PDF
|
1.89 MB | Adobe PDF | Visualizza/Apri |
2025_07_Lino Espinoza_Executive Summary.pdf
accessibile in internet per tutti
Descrizione: executive summary
Dimensione
365.07 kB
Formato
Adobe PDF
|
365.07 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/10589/239946