Uncertainty quantification of building energy model that assume ideal temperature control
MetadataShow full item record
An increasing number of studies conduct uncertainty analyses to investigate discrepancies between predicted energy performance of buildings and their actual measured energy use. Based on prior uncertainty quantification studies, there is evidence that there remain unquantified uncertainties related to the HVAC system. Most current studies of HVAC system uncertainties focus on investigating the probabilistic nature of building thermal loads and assume this nature to be the key factor to impact the accuracy of performance predictions of the HVAC system. To verify this one has to acknowledge that instead of reacting ideally to the “thermal load”, HVAC systems sense space temperature and use it as the control state in HVAC control loops, thus deciding on the heating or cooling requirement of a space based on sensed space temperature. For a VAV terminal box which serves multiple spaces, temperature controllability only applies to the space where a thermostat is installed. The temperature in other spaces may consequently not be maintained and the delivered cooling/heating from the terminal box will typically not (fully) satisfy the removal or supply of the room cooling and heating load. Commonly used EnergyPlus simulations introduce an idealization on the space temperature controllability by matching building zone partition with the HVAC supply network topology and using thermal zone as the atomic control object. This thesis targets an uncertainty quantification approach to identify model form uncertainties in EnergyPlus stemming from such idealized space temperature controllability. A high-fidelity co-simulation model which integrates an EnergyPlus building energy model with a Modelica HVAC system model is developed as the high-fidelity reference model. The differences between outcomes of the EnergyPlus simulation and outcomes of the reference model are then established. Two new characteristic parameters, “spatial-HVAC mismatch” and “occupant load diversity”, are introduced in this thesis. The first defines the area of non-sensed spaces in relation to directly controlled areas where the space temperature is sensed. Occupant load diversity expresses the variabilities of occupancy related load profiles in each space. The uncertainty analysis of the impact of the idealized temperature control of the EnergyPlus representation of VAV system considering the stochastic usage pattern of occupants in two space functions with five alignment configurations in three boundary situations focusing on the risk of underestimating energy consumption and over estimating occupant comfort (unmet hours in particular). The thesis quantifies the differences between low and high fidelity predictions in the outcomes of space air temperature, cooling energy in different time interval (hourly, daily, and monthly), fan power, and unmet hours as a result of the idealizations used in routine EnergyPlus simulations. It then correlates them with the mismatch and load diversity factors introduced above. Based upon the uncertainty analysis, this study explores the characterizations of the results from the case studies and discuss the methodologies and steps for “post corrections” or MFU inclusion in the low fidelity model by using fan power as an example. The research outcomes generate significant knowledge to the understanding of the origins of building energy model deficiency generated by idealization assumptions about temperature control and how it contributes to the performance gap.