In the present thesis, an adaptive HVAC setpoint management strategy is proposed and implemented to enhance the rate of self-consumption from on-site photovoltaic (PV) generation in a warehouse (chosen as the case study and considered to be located in Bologna, Italy). The adopted methodology involves a simulation-based approach using EnergyPlus, while interfacing with Python to apply the interventions, along with implementation of a machine learning (ML)-based load prediction pipeline. In this context, the baseline (regular operation) energy behavior of the storage and office zones, along with the PV generation rate, is first simulated and analyzed. In the following step, an ML-based pipeline is developed using simulated data to predict the building's baseline consumption in the next hour. Additionally, the corresponding climatic data along with the hour-ahead predicted irradiance (that is then employed for forecasting the PV generation) is obtained from a publicly accessible database. Subsequently, a set of interventions is proposed and implemented through simulation to increase the HVAC consumption during periods of high PV generation and reduce it in the intervals with low PV generation. These interventions, which include heating setpoint relaxation and slight overheating (by one degree in both cases) in the indoor environments of the facility, are scheduled by a developed agent, taking into account the current and predicted values of the consumption and PV generation. The obtained results demonstrate that, by implementing these interventions, the building's self-consumption rate is increased by 9% in the chosen interval for interventions (and by 7.47% considering the whole heating season). Similarly, the building's self-sufficiency rate is shown to be improved, demonstrating that a greater portion of its energy needs are being fulfilled by the PV plant.
Nel presente lavoro di tesi, viene proposta e implementata una strategia di gestione adattiva dei setpoint HVAC per migliorare il tasso di autoconsumo da generazione fotovoltaica (PV) in un magazzino (scelto come caso di studio e considerato come situato a Bologna, Italia). La metodologia adottata prevede un approccio basato sulla simulazione utilizzando EnergyPlus, interfacciandosi con un Python per applicare gli interventi, insieme all'implementazione di una pipeline di previsione dei carichi basata sul machine learning (ML). In questo contesto, viene prima simulato e analizzato il comportamento energetico di baseline (funzionamento regolare) delle zone di stoccaggio e degli uffici, insieme al tasso di generazione fotovoltaica. Nella fase successiva, viene sviluppata una pipeline basata su ML utilizzando i dati simulati per prevedere il consumo di baseline dell'edificio nell'ora successiva. Inoltre, i dati climatici corrispondenti e l'irraggiamento previsto per l'ora precedente (che viene poi utilizzato per prevedere la generazione fotovoltaica) sono ottenuti da un database pubblicamente accessibile. Successivamente, viene proposta e implementata attraverso la simulazione una serie di interventi per aumentare il consumo di HVAC durante i periodi di alta generazione fotovoltaica e ridurlo negli intervalli con bassa generazione fotovoltaica. Questi interventi, che comprendono l' abbassamento del setpoint di riscaldamento e un leggero surriscaldamento (di un grado in entrambi i casi) degli ambienti interni della struttura, sono programmati da un agente sviluppato, tenendo conto dei valori attuali e previsti del consumo e della generazione fotovoltaica. I risultati ottenuti dimostrano che, implementando questi interventi, il tasso di autoconsumo dell'edificio aumenta dell'9% nell'intervallo scelto per gli interventi (e del 7.47% considerando l'intera stagione di riscaldamento). Allo stesso modo, il tasso di autosufficienza dell'edificio risulta migliorato, dimostrando che una parte maggiore del suo fabbisogno energetico viene soddisfatta dall'impianto fotovoltaico.
HVAC-driven load shifting strategies aimed at enhancing the PV self-consumption rate in a conditioned warehouse
Pineda Solis, Jose Luis
2023/2024
Abstract
In the present thesis, an adaptive HVAC setpoint management strategy is proposed and implemented to enhance the rate of self-consumption from on-site photovoltaic (PV) generation in a warehouse (chosen as the case study and considered to be located in Bologna, Italy). The adopted methodology involves a simulation-based approach using EnergyPlus, while interfacing with Python to apply the interventions, along with implementation of a machine learning (ML)-based load prediction pipeline. In this context, the baseline (regular operation) energy behavior of the storage and office zones, along with the PV generation rate, is first simulated and analyzed. In the following step, an ML-based pipeline is developed using simulated data to predict the building's baseline consumption in the next hour. Additionally, the corresponding climatic data along with the hour-ahead predicted irradiance (that is then employed for forecasting the PV generation) is obtained from a publicly accessible database. Subsequently, a set of interventions is proposed and implemented through simulation to increase the HVAC consumption during periods of high PV generation and reduce it in the intervals with low PV generation. These interventions, which include heating setpoint relaxation and slight overheating (by one degree in both cases) in the indoor environments of the facility, are scheduled by a developed agent, taking into account the current and predicted values of the consumption and PV generation. The obtained results demonstrate that, by implementing these interventions, the building's self-consumption rate is increased by 9% in the chosen interval for interventions (and by 7.47% considering the whole heating season). Similarly, the building's self-sufficiency rate is shown to be improved, demonstrating that a greater portion of its energy needs are being fulfilled by the PV plant.File | Dimensione | Formato | |
---|---|---|---|
Thesis_HVAC_Driven_Load.pdf
non accessibile
Descrizione: HVAC-Driven Load Shifting Strategies Aimed at Enhancing the PV Self-Consumption Rate in a Conditioned Warehouse
Dimensione
1.6 MB
Formato
Adobe PDF
|
1.6 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/10589/235266