Fire-tube boilers are widely used in several kinds of industries to produce saturated steam. Being their energy consumption during exercise significantly high, the improvement of their efficiency is of paramount importance. Recovering waste heat from boiler exhaust flue gas has proven to be an effective way to improve energy utilization efficiency. Some of the most frequent uses for the heat recovered from flue gases include the pre-heating of the air or air-fuel mixture used in the combustion by means of an Air Pre-Heater or increasing the temperature of the water entering the boiler drum by introducing an Economizer. This thesis focuses on a real industrial application of waste heat recovery for a steam boiler. A non-linear MPC approach is applied to a complex heat exchanger used to pre-heat the combustion air and the water entering the boiler drum. After that an artificial neural network is trained in order to capture and replicate the behavior of the boiler connected to the waste heat recovery system. Such network plays a key role given the importance of the temperature of the fumes leaving the steam generator and is used inside a neural network MPC to maximise the efficiency of the complete steam production line, taking into account the interactions between boiler and waste heat recovery system. The introduction of a neural network inside the model used for prediction leads to a complex hybrid configuration, partially based on a non-linear first principle model and partially data-driven, so that the resulting optimization problem is more complicated than available MPCs in literature. Finally a comparison between the performances of the original PI regulator and the Model Predictive Control is carried out, highlighting the improvements of the newly implemented control strategy and suggesting some guidelines on future developments.
Le caldaie a tubi di fumo sono ampiamente utilizzate in diversi tipi di industrie per produrre vapore saturo. Poiché il loro consumo di energia durante l’esercizio è considerevole, il miglioramento della loro efficienza è di fondamentale importanza. Il recupero del calore residuo dai gas di scarico delle caldaie si è dimostrato un metodo efficace per migliorare l’efficienze energetica. Alcuni degli usi più frequenti del calore recuperato includono il preriscaldamento dell’aria o della miscela aria-combustibile usata nella combustione per mezzo di un preriscaldatore d’aria o l’aumento della temperatura dell’acqua che entra nel corpo cilindrico della caldaia introducendo un Economizzatore. Questa tesi si concentra su una reale applicazione industriale di recupero del calore residuo per un generatore di vapore. Un approccio MPC non lineare viene applicato a un complesso scambiatore di calore utilizzato per preriscaldare l’aria di combustione e l’acqua entrante nel corpo cilindrico della caldaia. In seguito viene addestrata una rete neurale artificiale per catturare e replicare il comportamento della caldaia collegata al sistema di recupero del calore residuo. Tale rete ricopre un ruolo chiave data l’importanza della temperatura dei fumi in uscita dal generatore di vapore e viene utilizzata all’interno di un MPC a reti neurali per massimizzare l’efficienza dell’intera linea di produzione del vapore, tenendo conto delle interazioni tra la caldaia e il sistema di recupero del calore. L’introduzione di una rete neurale all’interno del modello usato per la previsione porta ad una complessa configurazione ibrida, in parte basata su un modello non lineare costruito sul primo principio della termodinamica e in parte sui dati, così che il problema di ottimizzazione risultante è più complicato rispetto agli approcci MPC disponibili in letteratura. Infine viene effettuato un confronto tra le prestazioni del regolatore PI originale e del Model Predictive Control, evidenziando i miglioramenti della nuova strategia di con- trollo implementata e suggerendo alcune linee guida sugli sviluppi futuri.
An MPC approach for a complex waste heat recovery unit of an industrial steam production system
Gaboardi, Alessandro
2020/2021
Abstract
Fire-tube boilers are widely used in several kinds of industries to produce saturated steam. Being their energy consumption during exercise significantly high, the improvement of their efficiency is of paramount importance. Recovering waste heat from boiler exhaust flue gas has proven to be an effective way to improve energy utilization efficiency. Some of the most frequent uses for the heat recovered from flue gases include the pre-heating of the air or air-fuel mixture used in the combustion by means of an Air Pre-Heater or increasing the temperature of the water entering the boiler drum by introducing an Economizer. This thesis focuses on a real industrial application of waste heat recovery for a steam boiler. A non-linear MPC approach is applied to a complex heat exchanger used to pre-heat the combustion air and the water entering the boiler drum. After that an artificial neural network is trained in order to capture and replicate the behavior of the boiler connected to the waste heat recovery system. Such network plays a key role given the importance of the temperature of the fumes leaving the steam generator and is used inside a neural network MPC to maximise the efficiency of the complete steam production line, taking into account the interactions between boiler and waste heat recovery system. The introduction of a neural network inside the model used for prediction leads to a complex hybrid configuration, partially based on a non-linear first principle model and partially data-driven, so that the resulting optimization problem is more complicated than available MPCs in literature. Finally a comparison between the performances of the original PI regulator and the Model Predictive Control is carried out, highlighting the improvements of the newly implemented control strategy and suggesting some guidelines on future developments.File | Dimensione | Formato | |
---|---|---|---|
An MPC approach for a complex waste heat recovery unit of an industrial steam production system.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Thesis document
Dimensione
7.73 MB
Formato
Adobe PDF
|
7.73 MB | Adobe PDF | Visualizza/Apri |
An MPC approach for a complex waste heat recovery unit of an industrial steam production system - Executive Summary.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Executive Summary
Dimensione
756.82 kB
Formato
Adobe PDF
|
756.82 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/10589/188404