Households consume 10.5% of all the energy consumed within Australia (Department of Industry, Science, Energy and Resources, 2020). Of this, 40% is used for heating and cooling (EnergyConsult, 2015). Reducing household heating and cooling energy consumption has multiple benefits: reduced energy bills; reduced greenhouse gas emissions; increased network reliability; and increased householder comfort and health.

One way to reduce heating and cooling energy consumption is to build houses that are more energy efficient. The Australian Building Code Board's National Construction Code (NCC) specifies a minimum energy efficiency standard that planned new dwellings must meet to get building approval.

Around 80% of new builds use the nationwide house energy rating scheme (NatHERS) to demonstrate compliance with the NCC energy efficiency provisions (CSIRO, 2023). NatHERS rates a home's thermal shell on a scale from 1 to 10 stars. The current minimum standard is 7 stars. NatHERS accredits four tools which conduct physics-based assessments of a building's shell. The tools model heating & cooling load and give a star rating. Based on the building's plans, information about the house — every wall, floor, roof, window, all the materials, orientations, shadings, and more — are entered into the tools. There are thousands of inputs. Assessors need to be trained to use these tools.

These tools are excellent for new build regulation. The physics-based model gives a good accuracy level.

However, there are situations where an energy assessment would be useful, but only limited input data is available, or the expertise to do a detailed assessment is lacking, or a quick rating is needed.

RapidRate, developed by CSIRO using machine learning techniques, is a statistical model that provides an estimate of heating & cooling load, and star rating. It requires fewer inputs, is quick to do, and does not require special training. The data used to train the RapidRate models come from the NatHERS physics-based assessments. This means that the RapidRate results are aligned with the NatHERS assessments. Comparison of prototype RapidRate predictions against actual NatHERS star ratings show a mean difference of 0.33 stars and a median difference of 0.2 stars.

RapidRate has 12 main inputs: dwelling type; floor area; external wall area by orientation; window area by orientation (including % double glazed); main wall, floor, and roof materials (including whether they are insulated); and Post code. RapidRate has 3 outputs: star rating, heating load, & cooling load.

An API enables connection of RapidRate to any user interface or application. CSIRO's current prototype user interface is a single page web-based form.

RapidRate has many potential applications. Housing data providers could enrich their data by using RapidRate to generate the energy efficiency attributes of dwellings. Financial institutions could use RapidRate to gain a better understanding of the energy efficiency qualities of their housing portfolios. Urban energy modellers could deploy RapidRate to quickly assess urban level energy consumption. Homeowners and renters could use RapidRate to assess their own homes and explore how they might be improved. RapidRate is good for using when the data available for an energy efficiency assessment is limited, and when a quick rating is needed, and specialist expertise is not available to do it.

As people become more aware of the energy efficiency properties of homes and the benefits they bring, demand for better performing homes will increase. This will put pressure on builders, developers, and landlords to provide more energy efficient homes. RapidRate is an ongoing project. It has the capacity to be a vital contributor in the push towards reduced household heating and cooling energy consumption and the associated benefits.

This project is a collaboration of a dedicated team led by Melissa James and team members (in alphabetical order of last names): Emily Dioguardi, Abhishek Godhani, Yong Bing Khoo, Felicity Splatt, Susannah Tweddle, and Chi-Hsiang Wang.