To investigate the effect of climate on the electricity use for heating and cooling, the correlation between the electricity consumption of the city of Milan in the five years 2013-2017 and the outdoor dry bulb temperature of the same location was analyzed through statistical tools, with monthly, daily and hourly time scales. The monthly electricity consumption of each product sector was correlated to the average outdoor dry bulb temperature and analyzed for each product sector, using a five parameters model. These analyses showed a satisfactory correlation in the residential (R^2=0,84) and commercial sectors (R^2=0,86) and a poor correlation in the industrial sector (R^2=0,43). To filter out the effect of variables different from weather, the daily analysis was limited to working days. The daily use of electricity and the daily average outdoor temperature were correlated using a parametric model, within a parameter estimation approach, to highlight the relevant physical phenomena and to identify the value of the building stock characteristic parameters. An exponential model was used and a modified five-parameter model (5MPM) was proposed, based on the second principle of thermodynamics and taking into account the effect of the past temperature by adopting an effective temperature approach. The comparison between the actual data of the working days of the entire five-year period and the electricity consumption obtained with the 5MPM showed a good agreement between the two distributions, as confirmed by the value of the coefficient of determination (R2 = 0.93), the value of the normalized root mean square error (NRMSE = 2.3%) and by the value of the mean absolute percentage error (MAPE = 1.3%). The modified five-parameter model was then applied to the analysis of hourly electricity consumption using five-harmonic Fourier series to identify the baseline consumption, which does not depend on outdoor temperature. By increasing the number of data set and their disaggregation, the approach based on regressing the historical data series of energy consumption could be successfully adopted for applications that are currently approached by measurement or by direct simulation. In this perspective, big data analytics in combination with the parameter estimation approach remains a promising tool to facilitate the interpretability of the energy use model.
Il presente studio propone l’analisi, mediante strumenti statistici, della correlazione tra il consumo di energia elettrica della città di Milano nel quinquennio 2013-2017 e la temperatura esterna della stessa località con scala temporale mensile, giornaliera ed oraria. L’obiettivo di tale analisi è stato la determinazione dell’effetto del clima sulla domanda di energia elettrica per il riscaldamento ed il raffrescamento degli ambienti, in base alla realtà edilizia presente. I consumi energetici mensili della città di Milano sono stati correlati alla temperatura ed analizzati utilizzando un modello a cinque parametri applicato a ogni settore merceologico. I risultati hanno evidenziato una correlazione soddisfacente nei settori residenziale e commerciale ed una correlazione scarsa in quello industriale. L’analisi giornaliera è stata limitata ai giorni lavorativi per filtrare l’effetto delle variabili diverse da quelle climatiche. La domanda di energia elettrica e la temperatura esterna sono state correlate utilizzando un modello parametrico al fine di evidenziare fenomeni fisici rilevanti e di identificare il valore di parametri caratteristici del patrimonio edilizio. Sono stati proposti un modello esponenziale ed un modello modificato a cinque parametri (5MPM) basato sul secondo principio della termodinamica, utilizzando la temperatura effettiva che tiene in considerazione l’effetto della temperatura passata. Il confronto tra i dati effettivi dei giorni lavorativi dell’intero periodo quinquennale ed i consumi elettrici ottenuti con il 5MPM ha evidenziato una buona corrispondenza tra le due distribuzioni, confermata dal valore del coefficiente di determinazione (R^2 = 0,93), dal valore dell’errore quadratico medio normalizzato (NRMSE = 2,3%) e dal valore dell’errore assoluto medio percentuale (MAPE = 1,3%). Il modello di regressione è stato poi applicato all’analisi del consumo elettrico orario, utilizzando lo sviluppo di Fourier a cinque armoniche in funzione dell’ora del giorno per identificare il consumo indipendente dalla temperatura. Aumentando il numero dei dati e la loro disaggregazione, si potrebbe adottare l’approccio della regressione di serie storiche di dati di consumo energetico per applicazioni che, attualmente, vengono affrontate con simulazione diretta o con misurazioni sul campo. In questa prospettiva, l’analisi dei “big data” e la stima di parametri possono diventare strumenti per l’interpretabilità dei modelli di consumo energetico.
Effetto del clima sulla domanda di energia elettrica per la climatizzazione: una proposta di modellazione inversa mediante approccio top-down applicato alla città di milano
2020
Abstract
To investigate the effect of climate on the electricity use for heating and cooling, the correlation between the electricity consumption of the city of Milan in the five years 2013-2017 and the outdoor dry bulb temperature of the same location was analyzed through statistical tools, with monthly, daily and hourly time scales. The monthly electricity consumption of each product sector was correlated to the average outdoor dry bulb temperature and analyzed for each product sector, using a five parameters model. These analyses showed a satisfactory correlation in the residential (R^2=0,84) and commercial sectors (R^2=0,86) and a poor correlation in the industrial sector (R^2=0,43). To filter out the effect of variables different from weather, the daily analysis was limited to working days. The daily use of electricity and the daily average outdoor temperature were correlated using a parametric model, within a parameter estimation approach, to highlight the relevant physical phenomena and to identify the value of the building stock characteristic parameters. An exponential model was used and a modified five-parameter model (5MPM) was proposed, based on the second principle of thermodynamics and taking into account the effect of the past temperature by adopting an effective temperature approach. The comparison between the actual data of the working days of the entire five-year period and the electricity consumption obtained with the 5MPM showed a good agreement between the two distributions, as confirmed by the value of the coefficient of determination (R2 = 0.93), the value of the normalized root mean square error (NRMSE = 2.3%) and by the value of the mean absolute percentage error (MAPE = 1.3%). The modified five-parameter model was then applied to the analysis of hourly electricity consumption using five-harmonic Fourier series to identify the baseline consumption, which does not depend on outdoor temperature. By increasing the number of data set and their disaggregation, the approach based on regressing the historical data series of energy consumption could be successfully adopted for applications that are currently approached by measurement or by direct simulation. In this perspective, big data analytics in combination with the parameter estimation approach remains a promising tool to facilitate the interpretability of the energy use model.File | Dimensione | Formato | |
---|---|---|---|
TesiCarlottaBonfiglio.pdf
accesso solo da BNCF e BNCR
Tipologia:
Altro materiale allegato
Dimensione
4.22 MB
Formato
Adobe PDF
|
4.22 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/135475
URN:NBN:IT:UNIPR-135475