Exemplary Energy executive director, Trevor Lee, exposes two years of misleading HVAC simulations from the CSIRO.
On 7 July, under the guise of the insignificant heading “CSIRO weather files for building energy modelling – update”, the CSIRO Data Shop notified users of an “update” to its “weather files” (sic)[i].
The email described in detail changes to the way in which timestamps are mapped from NatHERS[ii] climate files to the .epw format for use in non-residential building energy simulations, a change in how the irradiation data is timestamped and an update to the User Guide which reflects these changes.
Unhelpfully, the “update” advises that “some of these datasets” have been corrected (without saying which ones)[iii].
Even now, the CSIRO source describes them as ‘old’ and ‘new’ rather than the explicit and helpful ‘skewed’ and ‘corrected’. There is nothing new about the ‘new’ data.
The title of an “update” is grossly misleading, at best, and suggests an attempt at hiding the significance of the errors two years after the fact.
Yet the detail confirms what has been known by scientists, engineers, industry and professional associations and bureaucrats since April 2022: the original data contains gross errors which we estimate are responsible for 100,000 misleading simulations with serious implications for developers, building operators and the renewable energy and HVAC industries.
Background
Since August 2021, the CSIRO has distributed three important weather and climate datasets for use by building energy modellers:
- Typical Meteorological Year (TMY) – a collation of selected meteorological measurements, listing data at 8,760 hourly intervals to describe a ‘typical’ year of weather for a specific location. This data is published in the EnergyPlus (.epw) format and is available at https://acds.csiro.au/future-climate-typical-meteorological-year.
- Reference Meteorological Years (RMY) – conceptually representing the same information as TMY but presented in an amended version of the fixed record format of the Australian Climate Data Bank (ACDB). This data is the basis for climate information in the Nationwide House Energy Rating Scheme (NatHERS) software tools and is distributed with the CSIRO’s AccuRate software and its competitors (all using the same CHENATH computation “engine”).
- Predictive weather files – CSIRO’s predictive weather files are based on a typical meteorological year of historical weather data drawn from 1990 to 2015 and can be used to investigate the likely impact of climate change on building energy consumption. These are available in .epw and NatHERS-compatible formats at https://acds.csiro.au/future-climate-predictive-weather.
As the basis for NatHERS, the CSIRO’s RMY dataset is arguably the most commonly-applied pathway to demonstrate compliance with the residential energy efficiency requirements of the National Construction Code (NCC).
With its transcription to the .epw format, closely related data is commonly used for simulations of commercial buildings that are used to demonstrate compliance with section JV3 of the NCC, as well as simulations for other purposes.
Litany of errors
We undertook a detailed review of the CSIRO data sets shortly after they were published in 2021. Our work revealed three major shortcomings:
- Reliance on old weather data which fails to characterise recent trends in a warming climate;
- Blatant timing errors across weather elements in the .epw format; and
- A lack of coincident precipitation data despite the .epw format expressly inviting it.
Old data
The first issue is well-known: The CSIRO files are derived from historical data to December 2015. This was a conscious choice made in the interests of aligning the .epw data to data used in the NatHERS scheme.[iv]
In a warming climate, this decision means that the data fails to account for the hottest years, resulting in a dataset that is representative of a cooler climate than what future buildings should be designed to expect.
It is imperative for the NCC to accurately reflect real weather conditions. In a changing climate, this can only be achieved using regular updates that incorporate recent observations.
Written by Trevor Lee, ARAIA, Affil.AIRAH, Affil.CIBSE, M.IBPSA.
[i] Here, the CSIRO Data Shop makes another error: Weather is what you actually get, got or are forecast to get. Climate is what you expect to get, inferred from a long enough record of actual weather. The data available through the service is all Climate data, but CSIRO have given their products the erroneous title “weather data”.
[ii] Nationwide House Energy Rating Scheme
[iii] The notice even shows scientific semi-literacy by referring to Global Horizontal Radiation when the correct jargon is GHI Global Horizontal Irradiation.
[iv] Administrative staff in the NatHERS system believe this data cannot be updated without undertaking a major regulatory impact study; rather than just tweak the NatHERS starband thresholds (MJ/m²) to keep the stringency of the design and construction requirements unchanged.