The widespread of renewable energy sources and the continuous growth in electricity demand are leading electrical distribution networks towards difficult challenges. Among them, the development of reliable and flexible emergency restoration plans that can ensure service continuity and resilience, represent a difficult task for the Distribution System Operators (DSO). Within the Italian regulatory framework, standards such as CEI 0- 17 and ARERA Resolution 617/2023 explicitly require DSOs to periodically design and validate such plans. In practice, the definition of emergency strategies is still largely based on the expertise of engineers and technicians, whose deep knowledge of the network’s characteristics enables them to plan effective switching sequences. Although this approach has proven high efficiency, it is also true that it is time-consuming: for a large urban network such as Milan, drafting a different plan for each possible scenario can require several days of work. This thesis proposes an automated methodology to support the development of restoration strategies, testing the potential of Genetic Algorithms. The Milan medium-voltage network was reconstructed in Python through pandapower environment, and the methodology was first validated on simplified test cases and then applied to a portion of the real grid. The results show that the proposed approach is capable to automatically generate feasible switching plans for restoration processes that mirror real restoration strategies within a reasonable amount of time compared to traditional method. This highlights the potential of Genetic Algorithms as a decision-support tool for DSOs.
La grande complessità delle moderne reti di distribuzione elettrica, causata da una crescente penetrazione delle fonti rinnovabili, dall’aumento della richiesta di energia e da topologie sempre più articolate, impone ai gestori di rete di sviluppare piani di emergenza sempre più affidabili ed efficaci. In Italia, la normativa di riferimento (Norma CEI 0-17 e Delibera ARERA 617/2023) stabilisce l’obbligo per i DSO di predisporre, aggiornare e validare tali piani con una certa periodicità in modo da garantire la continuità e la qualità del servizio anche in situazioni critiche. Ad oggi, lo sviluppo dei piani di emergenza è affidato ad ingegneri e tecnici che, grazie alla loro esperienza, sono in grado di sviluppare piani di emergenza efficaci. Tuttavia, la crescita dimensionale e la complessità della rete di Milano rendono questo processo particolarmente oneroso in termini di tempo di lavoro e personale impiegato. L’obiettivo di questa tesi è quello di proporre una metodologia automatizzata che possa assistere il DSO nello sviluppo di piani di rialimentazione. Il metodo di ottimizzazione proposto si basa sull’applicazione di algoritmi genetici che riescono a garantire la qualità del risultato entro tempi computazionali ragionevoli. Dopo aver descritto il modello della rete di Milano su Python in ambiente pandapower, la metodologia è stata preventivamente testata su reti semplificate e successivamente applicata alla rete reale. I risultati ottenuti mostrano come l’approccio proposto sia in grado di ridurre sensibilmente i tempi di elaborazione e offrire agli operatori uno strumento di sostegno nelle fasi di elaborazione dei piani.
Genetic algorithms for distribution grid’s emergency plans
CARAFFA, ANDREA
2024/2025
Abstract
The widespread of renewable energy sources and the continuous growth in electricity demand are leading electrical distribution networks towards difficult challenges. Among them, the development of reliable and flexible emergency restoration plans that can ensure service continuity and resilience, represent a difficult task for the Distribution System Operators (DSO). Within the Italian regulatory framework, standards such as CEI 0- 17 and ARERA Resolution 617/2023 explicitly require DSOs to periodically design and validate such plans. In practice, the definition of emergency strategies is still largely based on the expertise of engineers and technicians, whose deep knowledge of the network’s characteristics enables them to plan effective switching sequences. Although this approach has proven high efficiency, it is also true that it is time-consuming: for a large urban network such as Milan, drafting a different plan for each possible scenario can require several days of work. This thesis proposes an automated methodology to support the development of restoration strategies, testing the potential of Genetic Algorithms. The Milan medium-voltage network was reconstructed in Python through pandapower environment, and the methodology was first validated on simplified test cases and then applied to a portion of the real grid. The results show that the proposed approach is capable to automatically generate feasible switching plans for restoration processes that mirror real restoration strategies within a reasonable amount of time compared to traditional method. This highlights the potential of Genetic Algorithms as a decision-support tool for DSOs.| File | Dimensione | Formato | |
|---|---|---|---|
|
TESI_MAGISTRALE.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
3.92 MB
Formato
Adobe PDF
|
3.92 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/10589/245397