BUILDING ENERGY DEMAND OPTIMIZATION CONSIDERING OPERATIONAL PHASE UNCERTAINTIES

Abstract

As the Indian residential sector's energy demand rises by 8% each year, careful steps are needed to control its rising energy footprint and mitigate the damaging effects of global warming and climate change. Building Simulation and Design Optimization tools can help architects and engineers select optimal layouts, materials, and systems to satisfy energy and thermal comfort criteria as per location and intended building usage. However, residential buildings are exposed to operational phase uncertainties linked to physical, weather and occupant-related parameters during their lifetime. Overlooking their influence on the building design optimization (BDO) process can result in sub-optimal building performance in the longer run. This research has presented a novel evidence-based decision support system that quantifies and integrates operational phase uncertainties into the BDO process. This research is focussed on residential buildings in Ahmedabad, India, bearing a semi-arid hot Bsh climate as per the Köppen-Geiger climate classification. The research begins with field investigations involving household energy demand (HED) surveys and real-time building energy use measurements to assess energy use diversity and the impact of operational parameters on building energy demand. HED surveys were spread across 481 households in different residential areas across Ahmedabad to gather information about building location, size, air conditioner and domestic appliances ownership, and usage. Further, sub-hourly cooling energy demand and ambient weather conditions were monitored across twelve residential buildings. The daily and peak hourly cooling demands and daily AC operation varied from (1.86 -12.4) kWh and (0.67, 2.72) kWh and (2-14) hours, respectively. AC cooling setpoint temperature and daily and monthly operation decisively impacted monthly energy demands followed by domestic appliance operations. In general, households with fewer air-conditioner (AC) installations displayed lower energy usage variations. One-size-fits-all energy demand reduction strategies can prove inefficient due to diversities in people's energy use preferences. Building modellers must gather empirical evidence to incorporate operational phase uncertainties into building energy estimations. Next, a probabilistic LHS-PAWN uncertainty and sensitivity approach is adopted to assess the relative influence of physical, weather and occupant uncertainties on building cooling energy demand predictions. After investigating several housing layouts from India's Central Public Works and Central Building Research Institute repositories, a representative building simulation model is developed. Occupant uncertainties dominated weather and physical parameters as AC operation and window opening setpoint temperatures, daily hours and months of AC operation, and weather were the five main parameters impacting annual and peak cooling demand predictions. It is recommended to adopt multiple (past, present and future) weather datasets and occupant behavioural models based on empirical evidence for developing realistic building performance estimates. Further, uncertainty aspects must be incorporated during the building design optimization for superior performance reliability. Next, a novel algorithmic framework coupling Matlab with Energy Plus was developed to apply existing and upcoming meta-heuristics algorithms for BDO. This optimization framework was then converted into an easy-to-use Graphical Use Interface software, i.e. NETZED: BOPT, for applying single and bi-objective versions of Particle Swarm, Grey Wolf and Genetic Algorithms for building design optimization. NETZED: BOPT can also identify influential design parameters by applying PAWN sensitivity analysis to the optimization results. Using the above framework, a multi-objective Grey Wolf Algorithm was applied to identify optimal physical (wall and window type, orientation) and operational parameters (air conditioner's operational profiles and cooling setpoints) to minimize residential annual and peak cooling energy demands. Further, a novel robust BDO approach was applied to assess the influence of operational phase uncertainties on the multi-objective BDO process. Nine real-world operational scenarios were developed using typical cold and warm weather datasets and low, moderate, and high energy usage patterns. For each scenario, the Non-Dominated Sorting Genetic Algorithm (NSGA-II) was applied using NETZED: BOPT to optimize energy efficiency and thermal comfort. The final robust optimal design was confirmed by selecting the most repeated parameter settings, representing the lowest sub-optimality risk. Further, PAWN sensitivity analysis is applied to identify influential building design features. Thermal transmittance (U) behaved as a deciding factor for envelope selection in the Bsh climate. Low U walls and glazing deliver the lowest annual and peak cooling demands and thermal discomfort. Concrete Bricks (U = 0.52 W/𝑚2𝐾) and Aerated concrete (AC) (U = 0.49 W/𝑚2𝐾) blocks, double-glazed windows (U = 3.094 W/𝑚2𝐾), small WWR (10%), deep overhangs (0.6 m) and low solar heat gain coefficient of 0.45 produce optimal results. Besides, AC operation with a high comfort cooling setpoint of 28 °C for six hours, i.e. 12:00 pm - 6: 00 am from March to June, produces the lowest cooling demands. PAWN indicates a decisive influence of window-to-wall ratios (KSWWR ~ (0.46−0.66)) and wall construction type (KSWall ~ (0.31−0.46)) on cooling demands. AC sizing has a more pivotal role in predicting thermal comfort (KSAC ~ 0.35) than cooling demands (KSAC ~ 0.10). Further, AC sizing involves higher sub-optimality risk than envelope-related parameters. Near-optimal solutions are also analyzed for greater design flexibility. By adopting this RBDO framework, designers can estimate the likely building performance range and select the most reliable parameter settings by considering diverse operational scenarios. Building envelope and system enhancements alone cannot reduce the built sector's energy footprint, and positively motivating households to adopt pro-energy-use behavioural attitudes is crucial to realizing objective energy savings. Thus, a Structural Equation Model (SEM) was developed to assess critical motivators and triggers of household energy use behaviour. Based on 330 households' responses, Socio-Environmental Influences (Γ= 0.328) has the most decisive influence on HEUB, followed by Government Regulations (Γ= 0.269), Education and Information (Γ= 0.246), Market Conditions (Γ= 0.217) and Social Pressure (Γ= 0.173). Based on SEM results, cost subsidies, competitive pricing and informative packaging for energy-efficient appliances, and regular energy use feedback can improve people’s energy use behaviour. Besides, incorporating sustainability-related aspects in the educational curriculum can enhance people's pro-environment behavioural attitudes.