The aim of this research endeavour is the validation of a virtual digital twin in terms of thermal comfort for forecasting future scenarios, particularly in situations where real-time data acquisition devices are not available. After thorough consideration, a residential apartment building located in the city of Monza was identified as an appropriate candidate for this study. Subsequently, permission to access the building's information and the recorded data set was granted by the BIM4EEB project. The building consists of nine floors, housing sixty-five residential apartments, and occupies a total footprint just shy of 800m2. The building serves as a representative example of typical apartment buildings found in Italy, owing to its size and construction age. Under the framework of the BIM4EEB project, a total of 13 apartments within the building were selected for sensor placement, aimed at providing a representative sample of the entire structure. It was however noticed that of the 13 apartments, not all the apartments had an available data set. This was primarily due to the formalities of the data capture not forming part of this study and as such there was no control over the completion of the data sets. To further refine the data sets, only the apartments possessing a complete data capture period were considered. As such the analysis was limited to apartments D, G and H. The apartments have an associated data capture period of 287 days for the bedroom and 332 days for the living room. It was evident that the format did not align with the output format generated by the virtual digital twin modelling software IESVE. Consequently, a manipulation of the data was required to ensure adherence to this specific format, facilitating the later stages of comparative analysis. Furthermore, the dataset analysis revealed multiple occurrences of incomplete hourly coverage within each month. Due to the sporadic nature of these gaps, predicting or interpolating missing temperature values was deemed unfeasible. Instead, the number of these missing temperature readings was systematically calculated and documented. Additionally, instances were observed when the sensor became non-operational, resulting in the repetitive recording of the last temperature measurement until the sensor resumed normal function. To address this, a criterion was established: any continuous recording of a constant temperature for four-hours or more was indicative of sensor malfunction. These non-conforming data points were retained within the dataset and tabulated to serve in the later analysis. The apartment building was modelled using the IESVE software, in accordance with the provided floor plans from the Revit model. Material characteristics were assigned based on available building information. In cases where specific details about the building's envelope were lacking, conventional construction stratigraphies consistent with buildings of similar construction age in Italy were applied. For the simulation of internal heat gains, profiles were derived in adherence to national standards and subsequently integrated into the building template manager. This approach was necessitated by the limited information regarding the building's precise use, aside from its residential classification. Similar methodologies were employed for devising ventilation and heating profiles, given the absence of comprehensive data. Furthermore, a corrective measure was implemented to account for thermal bridge issues present on each facade of the building. This entailed the calculation of an adjusted U-value, facilitated by the utilization of the Therm software. Subsequent to the simulation of the virtual digital twin, comprehensive comparisons were drawn for all six rooms, encompassing both the number of CDH (cooling degree hours) and CDD (cooling degree days) experienced in the physical rooms and their corresponding virtual digital twins. Notably, the model demonstrated discernible deviations from the actual room conditions, primarily in the context of CDD. The maximum variance observed in this regard remained below 125 CDD. Similarly, when comparing the living rooms, analogous patterns of deviation between the digital twin and actual room conditions were observed, with the maximum variance being just over 300 CDD. Following a deeper analysis, scrutinising each studied space on a monthly basis, three key findings were derived based on the comparison between the deviation in the number of CDH and CDD between the physical rooms and their corresponding virtual digital twins: • The correct functioning of the sensor, over the full extent of the study period is detrimental in the validation of the virtual digital twin. • Not only do the quantity of missing data points recorded by the sensor affect the number of CDH but, more precisely, the continuity and temperature at which these unrecorded data points manifest also exert a significant influence. • Due to the difficulties in acquiring specific findings pertaining to the ventilation profiles, which may continuously change, it is suggested to apply a scaling factor in the calculation of the number of CDH for the virtual digital twin. This will account for the disparity between the results which occur from the variation in the applied ventilation profile. From this study it was found that on average the differences between the number of CDH in the bedrooms with that of their virtual digital twin amounts to 37%. Whereas the differences between the number of CDH in the living rooms with that of their virtual digital twin amounts to 59%, remembering that the sensors in the living room were deemed as not being fully functional. In conclusion it becomes evident that adjustments are warranted in the final calculation of the CDH associated with the model. It is suggested to either apply a scaling factor to the CDH counts or implement more precise ventilation profiles to facilitate a genuine comparison between the digital twin and the physical building. Such refinements are essential to validate the utility of the virtual digital twin in assessing the potential impacts of alterations or modifications to the building accurately.
L'obiettivo di questa ricerca è la validazione di un gemello digitale virtuale in termini di comfort termico per la previsione di scenari futuri, in particolare in situazioni in cui non sono disponibili dispositivi di acquisizione dati in tempo reale. Dopo un'attenta considerazione, un edificio residenziale situato nella città di Monza è stato identificato come candidato appropriato per questo studio. Successivamente, è stata concessa l'autorizzazione per accedere alle informazioni dell'edificio e al set di dati registrati dal progetto BIM4EEB. L'edificio è composto da nove piani, ospita sessantacinque appartamenti residenziali e occupa una superficie totale di poco meno di 800m2. L'edificio serve come esempio rappresentativo degli edifici residenziali tipici presenti in Italia, grazie alle sue dimensioni e all'età di costruzione. Nel contesto del progetto BIM4EEB, sono stati selezionati un totale di 13 appartamenti all'interno dell'edificio per il posizionamento dei sensori, con l'obiettivo di fornire un campione rappresentativo dell'intera struttura. Tuttavia, è stato notato che su 13 appartamenti, non tutti avevano un set di dati disponibile. Ciò è stato principalmente dovuto al fatto che la cattura dei dati non faceva parte di questo studio, e quindi non c'era controllo sulla completezza dei set di dati. Per raffinare ulteriormente i set di dati, sono stati presi in considerazione solo gli appartamenti con un periodo di cattura dati completo. Di conseguenza, l'analisi è stata limitata agli appartamenti D, G e H. Gli appartamenti hanno un periodo di cattura dati associato di 287 giorni per la camera da letto e 332 giorni per il soggiorno. È emerso che il formato dei dati non coincideva con il formato di output generato dal software di modellazione gemello digitale virtuale IESVE. Di conseguenza, è stata necessaria una manipolazione dei dati per garantire la conformità a questo formato specifico, facilitando le fasi successive dell'analisi comparativa. Inoltre, l'analisi del set di dati ha rivelato molteplici casi di copertura oraria incompleta in ciascun mese. A causa della natura sporadica di queste lacune, è stato ritenuto impraticabile prevedere o interpolare i valori di temperatura mancanti. Invece, è stato calcolato e documentato sistematicamente il numero di queste letture di temperatura mancanti. Inoltre, sono state osservate situazioni in cui il sensore non funzionava, con la registrazione ripetitiva dell'ultima misurazione di temperatura fino a quando il sensore non riprendeva la normale funzione. Per affrontare questo problema, è stato stabilito un criterio: qualsiasi registrazione continua di una temperatura costante per quattro ore o più era indicativa di un malfunzionamento del sensore. Questi punti dati non conformi sono stati mantenuti nel set di dati e tabulati per essere utilizzati nell'analisi successiva. L'edificio dell'appartamento è stato modellato utilizzando il software IESVE, in conformità ai piani del pavimento forniti dal modello Revit. Le caratteristiche dei materiali sono state assegnate in base alle informazioni disponibili sull'edificio. Nei casi in cui mancavano dettagli specifici sulla struttura dell'involucro dell'edificio, sono state applicate stratigrafie di costruzione convenzionali coerenti con gli edifici di età simile in Italia. Per la simulazione dei guadagni interni di calore, sono stati creati profili conformi agli standard nazionali e successivamente integrati nel gestore del modello dell'edificio. Questo approccio è stato reso necessario dalla limitata informazione sulla precisa destinazione d'uso dell'edificio, oltre alla sua classificazione residenziale. Metodologie simili sono state impiegate per la definizione dei profili di ventilazione e riscaldamento, data l'assenza di dati completi. Inoltre, è stata implementata una misura correttiva per tener conto dei problemi di ponti termici presenti su ciascuna facciata dell'edificio. Ciò ha comportato il calcolo di un valore U aggiustato, facilitato dall'uso del software Therm. Dopo la simulazione del gemello digitale virtuale, sono state effettuate comparazioni dettagliate per tutte e sei le stanze, includendo sia il numero di ore di raffreddamento (CDH) che il numero di giorni di raffreddamento (CDD) sperimentati nelle stanze fisiche e nei loro corrispondenti gemelli digitali virtuali. Va notato che il modello ha mostrato deviazioni evidenti dalle condizioni effettive delle stanze, principalmente in termini di CDD. La massima variazione osservata in questo contesto è rimasta al di sotto di 125 CDD. Allo stesso modo, confrontando i soggiorni, sono state osservate analoghe deviazioni tra il gemello digitale e le condizioni effettive delle stanze, con la massima variazione leggermente superiore a 300 CDD. Dopo un'analisi più approfondita, esaminando ogni spazio studiato mensilmente, sono emerse tre conclusioni principali basate sul confronto tra la deviazione nel numero di CDH e CDD tra le stanze fisiche e i loro corrispondenti gemelli digitali virtuali: • Il corretto funzionamento del sensore per l'intero periodo dello studio è fondamentale per la validazione del gemello digitale virtuale. • Non solo la quantità di punti dati mancanti registrati dal sensore influisce sul numero di CDH, ma più precisamente la continuità e la temperatura a cui questi punti dati non registrati si manifestano esercitano un'influenza significativa. • A causa delle difficoltà nel reperire risultati specifici relativi ai profili di ventilazione, che possono cambiare continuamente, si suggerisce di applicare un fattore di scala nel calcolo del numero di CDH per il gemello digitale virtuale. Ciò terrà conto della disparità tra i risultati che derivano dalla variazione del profilo di ventilazione applicato. Da questo studio è emerso che in media le differenze tra il numero di CDH nelle camere da letto e quelli dei loro gemelli digitali virtuali ammontano al 37%. Mentre le differenze tra il numero di CDH nei soggiorni e quelli dei loro gemelli digitali virtuali ammontano al 59%, tenendo presente che i sensori nel soggiorno sono stati considerati non completamente funzionanti. In conclusione, emerge che sono necessari adeguamenti nel calcolo finale dei CDH associati al modello. Si suggerisce di applicare un fattore di scala ai conteggi dei CDH o di implementare profili di ventilazione più precisi per consentire un confronto genuino tra il gemello digitale virtuale e l'edificio fisico. Tali perfezionamenti sono essenziali per convalidare l'utilità del gemello digitale virtuale nella valutazione degli impatti potenziali di modifiche o interventi sull'edificio in modo accurato.
Virtual digital twin validation in terms of thermal comfort: a case study
de Oliveira, Dominic Tomaz
2022/2023
Abstract
The aim of this research endeavour is the validation of a virtual digital twin in terms of thermal comfort for forecasting future scenarios, particularly in situations where real-time data acquisition devices are not available. After thorough consideration, a residential apartment building located in the city of Monza was identified as an appropriate candidate for this study. Subsequently, permission to access the building's information and the recorded data set was granted by the BIM4EEB project. The building consists of nine floors, housing sixty-five residential apartments, and occupies a total footprint just shy of 800m2. The building serves as a representative example of typical apartment buildings found in Italy, owing to its size and construction age. Under the framework of the BIM4EEB project, a total of 13 apartments within the building were selected for sensor placement, aimed at providing a representative sample of the entire structure. It was however noticed that of the 13 apartments, not all the apartments had an available data set. This was primarily due to the formalities of the data capture not forming part of this study and as such there was no control over the completion of the data sets. To further refine the data sets, only the apartments possessing a complete data capture period were considered. As such the analysis was limited to apartments D, G and H. The apartments have an associated data capture period of 287 days for the bedroom and 332 days for the living room. It was evident that the format did not align with the output format generated by the virtual digital twin modelling software IESVE. Consequently, a manipulation of the data was required to ensure adherence to this specific format, facilitating the later stages of comparative analysis. Furthermore, the dataset analysis revealed multiple occurrences of incomplete hourly coverage within each month. Due to the sporadic nature of these gaps, predicting or interpolating missing temperature values was deemed unfeasible. Instead, the number of these missing temperature readings was systematically calculated and documented. Additionally, instances were observed when the sensor became non-operational, resulting in the repetitive recording of the last temperature measurement until the sensor resumed normal function. To address this, a criterion was established: any continuous recording of a constant temperature for four-hours or more was indicative of sensor malfunction. These non-conforming data points were retained within the dataset and tabulated to serve in the later analysis. The apartment building was modelled using the IESVE software, in accordance with the provided floor plans from the Revit model. Material characteristics were assigned based on available building information. In cases where specific details about the building's envelope were lacking, conventional construction stratigraphies consistent with buildings of similar construction age in Italy were applied. For the simulation of internal heat gains, profiles were derived in adherence to national standards and subsequently integrated into the building template manager. This approach was necessitated by the limited information regarding the building's precise use, aside from its residential classification. Similar methodologies were employed for devising ventilation and heating profiles, given the absence of comprehensive data. Furthermore, a corrective measure was implemented to account for thermal bridge issues present on each facade of the building. This entailed the calculation of an adjusted U-value, facilitated by the utilization of the Therm software. Subsequent to the simulation of the virtual digital twin, comprehensive comparisons were drawn for all six rooms, encompassing both the number of CDH (cooling degree hours) and CDD (cooling degree days) experienced in the physical rooms and their corresponding virtual digital twins. Notably, the model demonstrated discernible deviations from the actual room conditions, primarily in the context of CDD. The maximum variance observed in this regard remained below 125 CDD. Similarly, when comparing the living rooms, analogous patterns of deviation between the digital twin and actual room conditions were observed, with the maximum variance being just over 300 CDD. Following a deeper analysis, scrutinising each studied space on a monthly basis, three key findings were derived based on the comparison between the deviation in the number of CDH and CDD between the physical rooms and their corresponding virtual digital twins: • The correct functioning of the sensor, over the full extent of the study period is detrimental in the validation of the virtual digital twin. • Not only do the quantity of missing data points recorded by the sensor affect the number of CDH but, more precisely, the continuity and temperature at which these unrecorded data points manifest also exert a significant influence. • Due to the difficulties in acquiring specific findings pertaining to the ventilation profiles, which may continuously change, it is suggested to apply a scaling factor in the calculation of the number of CDH for the virtual digital twin. This will account for the disparity between the results which occur from the variation in the applied ventilation profile. From this study it was found that on average the differences between the number of CDH in the bedrooms with that of their virtual digital twin amounts to 37%. Whereas the differences between the number of CDH in the living rooms with that of their virtual digital twin amounts to 59%, remembering that the sensors in the living room were deemed as not being fully functional. In conclusion it becomes evident that adjustments are warranted in the final calculation of the CDH associated with the model. It is suggested to either apply a scaling factor to the CDH counts or implement more precise ventilation profiles to facilitate a genuine comparison between the digital twin and the physical building. Such refinements are essential to validate the utility of the virtual digital twin in assessing the potential impacts of alterations or modifications to the building accurately.| File | Dimensione | Formato | |
|---|---|---|---|
|
Dominic de Oliveira Thesis (PDF).pdf
accessibile in internet per tutti
Dimensione
8.09 MB
Formato
Adobe PDF
|
8.09 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/10589/209433