Our critique of the CSIRO Representative Meteorological Year (RMY) climate data sets highlighted several major shortcomings including a 60-minute offset error in several weather elements arising from a flawed transcription from the ACDB to epw format. Despite the faults, the free availability of this data sees it becoming the de facto standard for building energy modellers seeking to demonstrate compliance energy efficiency regulations, and the Australian Building Codes Board have even elected to reference it in the National Construction Code (NCC) 2022.
For this article we endeavoured to find a workaround to the timing offset errors by changing the building operating hours in the simulations.
Firstly, to understand how these offset errors were affecting building simulations, utilising Energy+ we simulated our archetypical 10-storey office building in the 8 Australian capital cities climates, analysing the cooling and heating requirements comparing CSIRO RMYs to Exemplary Selected Meteorological Year (SMY) data sets, produced utilising CSIRO RMYs years for each capital city. In all our simulations, the effect of the offset errors was obvious: utilising CSIRO RMYs, the monthly cooling requirements were lower and the monthly heating requirements were higher compared to Exemplary SMYs.
Our attempted workaround for circumventing the CSIRO timestamp errors involved re-running the simulations while changing the operational times of the archetypical buildings. That is, we shifted the simulations’ operational times by one hour to align with the CSIRO’s erroneous timestamps for the non-solar weather elements.1
The results from these simulations, together with the ones we previously carried out, allowed us to isolate the impact of this error and estimate the effectiveness of this workaround.
As expected, the tweaked operational times resulted in an increase in cooling requirements while heating demand decreased – closer to the results using the correctly-timestamped data. The results for Canberra are presented below, where Exemplary’s results are always taken from standard simulations.
These results show how this workaround is an effective strategy to correct the timestamp errors throughout CSIRO climate files. In fact, the percentage errors between CSIRO and Exemplary decrease when simulating CSIRO RMYs with tweaked operational times for both cooling and heating, where the latter is significantly more appreciable2.
Finally, to assure the quality of our analysis, considering that even after the workaround there are still some discrepancies in the results, we also ran simulations with a third weather files which was using CSIRO RMYs as a base, with GHI, DNI, and DIF substituted with the ones from Exemplary’s SMYs. As expected, in the results from these simulations utilising standard operational times, the effect of the offset errors of the other weather elements was still present, thus, although with lower % errors compared to CSIRO as published, the monthly cooling requirements were lower and the monthly heating requirements were higher compared to Exemplary SMYs. However, when simulating these weather files with tweaked operational times, the results were almost identical to the simulations with standard operational times utilising Exemplary SMYs, thus confirming that the discrepancy in the results is caused by the 30-minute time offset in the CSIRO solar data.
1 Note that a seperate error relating to the 30-minute time offset in the CSIRO solar data which still affects the modelling and is the main source of discprenacy in the results.
2 The cooling and heating percentage errors for off-peak seasons have not been reported, as the errors can be abnormally high due to the denominators’ low values.
This is a worry! I’m not sure of the licensing requirements etc. here but it would be terrific if you could publish patched datasets on Github or the like, rather than the odd CSIRO webstore platform. Community management of the data might blow some fuses w.r.t NCC but more eyes on it the better.