In recent years, natural gas has assumed a pivotal role in the ongoing energy transition, becoming an increasingly significant component of the energy mix across Europe, and particularly in Italy. The transportation of natural gas relies on an extensive and complex infrastructure that demands advanced management strategies to reduce energy consumption and limit CO_2 emissions. Within the framework of a collaboration between SNAM and Politecnico di Milano, multiple projects have been initiated with the ultimate goal of optimizing the operation of the transmission network and developing a Digital Twin. This thesis contributes to that vision by proposing a framework based on the Modelica modeling and simulation environment, aimed at enhancing real-time monitoring through state estimation. Leveraging a physics-based dynamic model of the network and real-time measurements obtained from the SCADA system, the objective is to ensure consistency between the measured data and the system’s physical behavior, while compensating for errors and uncertainties in the data. To this end, an Extended Kalman Filter (EKF) is developed, incorporating an augmented state formulation to account for unknown model errors such as sensor bias. This approach enables the algorithm not only to track the system state accurately but also to infer hidden dynamics and anomalies that would otherwise remain undetected. The proposed methodology is validated on the whole Italian national gas network using synthetic datasets that have been deliberately corrupted to replicate realistic error sources. The results demonstrate the algorithm’s capability to reconstruct the true system behavior with high fidelity. This work lays the foundation for the future deployment of real-time state estimation on operational gas networks. In addition to improving network efficiency and reliability, such tools hold promise for facilitating predictive maintenance, enhancing safety, etc.
Negli ultimi anni, il gas naturale ha assunto un ruolo centrale nella transizione energetica in atto, diventando una componente sempre più rilevante del mix energetico in Europa, e in particolare in Italia. Il trasporto del gas naturale si basa su un’infrastruttura estesa e complessa, che richiede strategie di gestione avanzate per ridurre il consumo energetico e limitare le emissioni di CO_2. Nell’ambito della collaborazione tra SNAM e il Politecnico di Milano, sono stati avviati diversi progetti con l’obiettivo finale di ottimizzare la gestione della rete di trasmissione e sviluppare un Gemello Digitale. Questa tesi contribuisce a tale visione proponendo un framework basato sull’ambiente di modellazione e simulazione Modelica, finalizzato al miglioramento del monitoraggio in tempo reale tramite tecniche di stima dello stato. Sfruttando un modello dinamico fisico della rete e i dati di misura acquisiti dal sistema SCADA, l’obiettivo è garantire la coerenza tra i dati misurati e il comportamento fisico del sistema, compensando al contempo errori e incertezze presenti nei dati disponibili. A tal fine, è stato sviluppato un algoritmo basato sul Filtro di Kalman Esteso (EKF), che adotta una formulazione con stato aumentato per tenere conto di errori di modello non noti, come ad esempio i bias strumentali. Questo approccio consente all’algoritmo non solo di tracciare accuratamente lo stato del sistema, ma anche di inferire dinamiche nascoste e anomalie che altrimenti rimarrebbero non rilevate. La metodologia proposta è stata validata sull’intera rete nazionale italiana del gas naturale, utilizzando set di dati sintetici appositamente modificati per introdurre fonti di errore realistiche. I risultati dimostrano la capacità dell’algoritmo di ricostruire con elevata fedeltà il comportamento reale del sistema. Questo lavoro pone le basi per una futura implementazione della stima dello stato in tempo reale su reti di gas operative. Oltre a migliorare l’efficienza e l’affidabilità della rete, tali strumenti offrono un potenziale significativo per la manutenzione predittiva, il miglioramento della sicurezza e altre applicazioni innovative.
Advances in extended Kalman filtering applied to the Italian gas transportation network
PETROVIC, STEPA
2024/2025
Abstract
In recent years, natural gas has assumed a pivotal role in the ongoing energy transition, becoming an increasingly significant component of the energy mix across Europe, and particularly in Italy. The transportation of natural gas relies on an extensive and complex infrastructure that demands advanced management strategies to reduce energy consumption and limit CO_2 emissions. Within the framework of a collaboration between SNAM and Politecnico di Milano, multiple projects have been initiated with the ultimate goal of optimizing the operation of the transmission network and developing a Digital Twin. This thesis contributes to that vision by proposing a framework based on the Modelica modeling and simulation environment, aimed at enhancing real-time monitoring through state estimation. Leveraging a physics-based dynamic model of the network and real-time measurements obtained from the SCADA system, the objective is to ensure consistency between the measured data and the system’s physical behavior, while compensating for errors and uncertainties in the data. To this end, an Extended Kalman Filter (EKF) is developed, incorporating an augmented state formulation to account for unknown model errors such as sensor bias. This approach enables the algorithm not only to track the system state accurately but also to infer hidden dynamics and anomalies that would otherwise remain undetected. The proposed methodology is validated on the whole Italian national gas network using synthetic datasets that have been deliberately corrupted to replicate realistic error sources. The results demonstrate the algorithm’s capability to reconstruct the true system behavior with high fidelity. This work lays the foundation for the future deployment of real-time state estimation on operational gas networks. In addition to improving network efficiency and reliability, such tools hold promise for facilitating predictive maintenance, enhancing safety, etc.File | Dimensione | Formato | |
---|---|---|---|
2025_07_Petrovic.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Text of the thesis
Dimensione
8.63 MB
Formato
Adobe PDF
|
8.63 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/10589/239924