A Methodology for Optimised Energy Retrofitting of Higher Educational Buildings

In Australia, a staggering 70% increase in electricity consumption over the past three decades has been documented, correlating to a 47% rise in total CO2 emissions. The building sector accounts for 40% of the global energy demand and 19% of Australia's energy consumption. Higher educational buildings are observed to account for up to 18% of energy use in the total building stock. Operational stages account for up to 90% of a building's total energy use across its lifecycle (Alam & Devjani, 2021a). Energy retrofitting of buildings is a viable solution to the ever-growing climatic problem. However, current retrofitting processes can have the following problems. Retrofits can be ad hoc and carried out with minimal structure and understanding. Moreover, targeting one retrofit may impact another, as many of the end goals have conflicting outcomes. For example, retrofits with high energy reduction capacity can result in higher capital costs which cannot be recovered during the life cycle. Stakeholders are interested in achieving multiple objectives at the same time. Further complications arise when multiple retrofits are completed in parallel as the combined effect can have adverse unexpected outcomes towards energy-saving goals. Finally, retrofit decision-making can be complex and challenging as many quantitative and qualitative criteria need to be considered when initiating a retrofit project. The research explores a method for optimised retrofitting to address the above issues in higher educational buildings. Parameters that contribute to the energy use of educational institutions were explored from a case study in Melbourne, Australia. The most common types of functional areas available in educational buildings were identified with the said data and contribution to energy use was calculated based on the floor areas of these three types of functional areas. The correlations were found via two methods: Multiple Regression and Adaptive Neuro-Fuzzy Inference System (ANFIS). Following this, a simulation block model was constructed for the identified three space types, and the base model was simulated in EnergyPlus and validated against data from the case study. A sensitivity analysis was conducted to find the most influential factors affecting the model outputs and, in extension, the three types of functions when considering retrofits independently. In order to identify the combinations of retrofit options that can have the highest positive impact on energy and cost-saving goals, a Multi-Objective Optimisation (MOO) was completed using jEPlus+EA. This process provides the opportunity to observe the impact a combination of retrofits can have on one or more objectives. In order to streamline the selection of the most practical retrofit option, Multi-Criteria Decision-Making (MCDM) was implemented so that the MOO can be incorporated with other defining factors that are important to the organisation's overarching goals. A method of incorporating the optimised retrofits into the building lifecycle asset management program and the ideal time for carrying out these retrofits so maximum impact can be achieved, both environmentally and economically was also explored. The outcomes of the regression models showed that energy prediction based on the floor area of the three types of spaces, Office, Teaching and Common Space, have better energy estimation capacity for buildings compared to estimating the energy based on gross floor area alone for the data set analysed. It was also noted that the ANFIS model used for the energy prediction has a better regression coefficient than the multiple regression model. The outcome from the MOO provided a Pareto front of solutions for the two objective scenarios analysed (heating vs cooling load reduction and total energy vs capital cost minimisation). The mean optimisation values were found to be in the order of office, teaching, and common (47,35, 28 kWh/m2) for cooling from highest to lowest and teaching, common, and office (1.4, 1.0, 0.9 kWh/m2) for heating. When comparing north and south zones, the results showed means of 28%, 26% and 19% difference in cooling and 75%, 14% and 40% difference in heating load for the office, teaching and common spaces, respectively. A ranked set of MOO solutions is provided. These solutions can be funnelled based on a variety of functional criteria that are useful for an educational organisation based on MCDM. Look-up tables for retrofit solutions were provided for the three functional spaces that were analysed. Once the ranked optimisation solution is identified, the expected energy requirement change can be calculated based on the original space that is under analysis. The annual reduction of energy can be converted to emissions based on the fuel types used for heating and cooling. The optimised option can be incorporated into the existing CapEx program for a smoother transition of energy retrofits. The CapEx program can be calculated based on the lifecycle software CAMS developed by RMIT University and the proposed solution for energy retrofits can be incorporated in the software platform. The results of the study focussed on educational facilities in Melbourne. The process can be replicated in other locations by adjusting the weather file used for the calculation.