- Title
- Optimisation of architectural building design parameters for students’ thermal comfort and energy savings in educational buildings
- Creator
- Alghamdi, Salah
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2022
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- The thesis aims to optimise the architectural building design parameters (ABDPs) values selection to achieve the highest level of students’ thermal comfort (TC) and energy savings in educational buildings in New South Wales (NSW), Australia. In Australia, around 25% of total energy consumption (EC) and 27% of greenhouse gas emissions are attributable to buildings. Approximately 50% of the energy consumed by heating, ventilation and air conditioning (HVAC) systems occurs on university campuses in Australia. A primary reason for the increase in EC is the spread of the HVAC systems in response to the growing demand for better indoor TC. TC is a critical component of indoor environments, especially in educational buildings where learning is the main objective. Maximising the indoor occupant’s TC level while minimising the building EC requires an optimisation method to identify appropriate solutions over time. When applied in architecture, optimising a building’s performance must typically address multiple objectives (e.g. less thermal discomfort hours (TDHs) generally entails higher EC). Consequently, the optimisation problem becomes increasingly complex, mainly when many ABDPs (such as window-to-wall ratio, cooling set-point temperature, and building orientation) and conflicting objectives are found. However, because of the complex thermal characteristics of ABDPs, there have been few comprehensive and in-depth studies on leveraging the effects of ABDPs toward energy saving and TC improvement educational buildings in NSW. This study proposed using the multi-objective optimisation (MOO) method to determine the optimum scenario of ABDPs achieving the highest level of students’ TC and energy savings in educational buildings. The main objectives of this thesis are, through building energy simulation (BES) and field investigations methods, (1) to identify the most significant ABDPs that affect the TC and EC of an educational building; (2) to develop and validate the accuracy of the BES model for evaluating the effects of ADBPs on both students’ TC and buildings’ EC; (3) to develop the back propagation artificial neural network (ANN) model to predict EC and TC for any given combination of the identified ABDPs in the conceptual design stage; and (4) to determine the optimum scenario of ABDPs among students’ TC and buildings’ EC by using the MOO method. In the BES model adopted in the study, a single-storey education building at The University of Newcastle, Australia, was selected as a reference building and the base case model for the simulation process. DesignBuilder (DB) was used to simulate different scenarios of ABDPs. A comprehensive energy simulation was applied to analyse the performance of the different ABDPs of educational spaces in hot summer and cold winter environments to achieve an optimal level of TC and EC. Monte Carlo analysis (MCA) was adopted to cover all probability scenarios of input ABDPs. For every input parameter, 2000 samples were applied based on the selected distribution by using the Latin hypercube sampling (LHS) technique to provide appropriate accuracy in uncertainty analyses (UA) and sensitivity analyses (SA). The use of UA and SA in this study aims to support the design process by showing the effect of different parameters on the design outcome in terms of uncertainty (normal distribution and range) and sensitivity (order of the most influential parameters). Other statistical methods, such as bivariate Pearson correlation, parametric analysis (PA), and cluster analysis were used to analyse and interpret the data. To further validate the BES and ANN model for improving energy savings and students’ TC, the field investigations method was applied by conducting subjective measurements, objective measurements, and building energy auditing. The field investigations include questionnaire surveys (subjective method) and physical measurements (objective method) of indoor TC variables in three classrooms at The University of Newcastle. In the subjective method, the adaptive thermal comfort (ATC) method was used to represent the actual mean vote (AMV) based on the ASHRAE Standard 55. The objective method was used to measure the indoor thermal environment, which was simultaneously measured as the participants responded to the questionnaires about their TC. During the surveyed classes, measurements were recorded for indoor air temperature (Ta), globe temperature (Tg), relative humidity (RH), and air velocity (Va). Totally, 152 individuals participated in the field investigation to complete the questionnaires during the physical measurement of the indoor thermal environment. The simulation results show that there is an opportunity for energy saving and to increase the level of TC in educational buildings in climate zone 5 in NSW, Australia. The effect of ABDPs in indoor thermal environments and both students’ TDHs and EC have been discussed through the PA and SA. According to the study results, there is a very weak relationship between the students’ TDHs and the building cooling/heating load. In addition, the results indicate that cooling and heating set-point temperatures and roof construction have a significant impact on the sensitivity of the ABDPs for both building EC and the level of student TC. Increasing the cooling set point temperature from 22.0°C to 28.0°C and using a U-value of 0.2 W/m²K in roof construction can reduce the operative temperatures by 14.2% and 20.0%, respectively. These reductions could significantly lower the TDHs by 6.0 and 3.25 times, respectively. Also, the reduction in EC was 43.7% and 41.0% in cooling set-point temperature and roof construction, respectively. ANN is a forecasting method that allows the identification of complex nonlinear relationships between the response variable and its predictors. A dataset containing over 1845 scenarios was used with a 70/30 split for training and testing. The ANN model was also validated using the field investigation data. The ANN model has a maximum of 15 input variables and 12 hidden layers. The output layers are either 1 or 2 depending on the configurations set by the user. The backpropagation applied the Levenberg-Marquardt algorithm to train the ANN model. The regression percentage for the training and testing have good agreements with the R2 for output set to TDHs and EC higher than 0.97. The ANN model was found to provide acceptable simulation approximations with average relative errors below 2.0% for the total EC and below 1.0% for the average TDHs. Regarding the validation of EC results, the absolute values of the percentage differences (PD) between computer simulation results and actual data were equal to or less than 15%, which was considered accurate and validated. Thus, in this study, simulations were valid for all three educational spaces; case 1 (-13.5%), case 2 (-11.5%), and case 3 (-13.1%). The predicted mean vote (PMV) and predicted percentage dissatisfaction (PPD) values are similar to the actual votes reported in the questionnaires. To meet the occupant’s thermal needs, they actively modify the environmental and personal variables of TC. Clothes play a significant role in determining the difference between AMV and PMV. Based on the discrepancy rate between the simulation models (EnergyPlus and ANN) and the field investigation of two classrooms, the ANN model performed better for ABDPs to predict EC than EnergyPlus model in educational buildings in NSW. The MOO method was applied by combining a BES model with an optimisation algorithm, which was very useful in finding the optimum solutions. The optimisation method was investigated using the non-dominated sorting genetic algorithm II (NSGA-II). The Pareto-front analysis results show that the amount of change in the annual EC with varying ABDPs is less than TDHs. Optimal solutions range from an annual EC of 37217 kWh for TDHs of 1269 to TDHs of 130 for an annual EC of 85571 kWh. Pareto-front provides 32 optimal solutions. Then, building EC and TDHs were calculated for the new optimised building scenario. Overall, there was a substantial decrease in building EC and students’ TDHs of the building model after optimisation. This optimal solution provides TDHs saving of 16.3% and EC saving of 23.8%. The preferred temperature occurs at 23.8°C, which is 3.8°C lower than the neutral temperature (27.5°C). The preferred temperature of students in this study is within the ASHRAE comfort range (20.0–26.0°C). The findings show the applicability of the ASHRAE Standard 55 to students in classrooms in climate zone 5, Australia. The finding is also supported by other studies in different climate zones suggesting that the neutral thermal sensations do not correlate to occupants’ preferred temperature. The preferred operative temperature should be increased from the ASHRAE’s minimum acceptable range (20.0°C) to 23.7°C to fully benefit from energy savings. The outcomes of this study have the potential to strongly support the selection of appropriate ABDPs for classroom design by providing effective simulation and field investigation analysis. The results of this study are promising and may facilitate the use of the ANN model as a building assessment method to accurately predict the amount of EC and the level of students’ TC in any educational building in climate zone 5.
- Subject
- architectural building design parameters; thermal comfort; educational buildings; heating, ventilation and air conditioning systems; HVAC systems
- Identifier
- http://hdl.handle.net/1959.13/1505368
- Identifier
- uon:55665
- Rights
- Copyright 2022 Salah Alghamdi
- Language
- eng
- Full Text
- Hits: 101
- Visitors: 123
- Downloads: 31
Thumbnail | File | Description | Size | Format | |||
---|---|---|---|---|---|---|---|
View Details Download | ATTACHMENT01 | Thesis | 9 MB | Adobe Acrobat PDF | View Details Download | ||
View Details Download | ATTACHMENT02 | Abstract | 480 KB | Adobe Acrobat PDF | View Details Download |