The background LCI source is AusLCI with ecoinvent shadow database. The numbers you’re referring to below for steel I believe are for default recycled content which may be lower for the EPDs. If you change the recycled content to match that of Infrabuild you may get a more aligned figure. There is some momentum in having the EPD producers contributing their data to the AusLCI. In the meantime though you can always use the EPDs directly in your eTool models. Note that Infrabuild appear at this point to be about the highest carbon steel producer in the world (hence why the eco-invent shadow database isn’t lining up too well despite being regionalised with Australian electricity supply). This may be due to inherent inefficiencies in the plant itself, poor quality metallurgical coal or low grade iron ore feedstock (that’s speculation though). Interestly the older Infrabuild EPD was not nearly as high impact so there’s been some change to the operations since they last published their EPD. The AusLCI appears to be more reflective of Australian steel supply (including imports) than production from Whyalla specifically.
I am of the same opinion that the generic “double glazed” window templates are unlikely to be reflective of a specific facade system. Particularly if the facade system has some structural requirements. If this is of concern for your design advice you can create a custom template that matches the facade system of the building itself (I think our team have done this on some projects historically in the UK).
One thing I’ve noticed over the years is that comparing embodied carbon data points will always raise questions. There is such large range in methodological approaches, varying quality of the data (and underlying LCA studies), geographical relevancy, temporal relevancy, technological relevancy, standards alignment, characterisation factors etc. Our approach at eTool is not to try to change the world but follow the strong global trends in alignment and choose data that is:
- Independently reviewed
- Transparent (any body can download AusLCI load it into an LCA tool and get into the weeds if they wish)
- Aligned with EN15804
- Supported by industry bodies such as Australian Life Cycle Assessment Society
- Actively maintained such that it is continually improved
The last point is extremely important in LCA. Continual improvement is necessary in all aspects. Methodology, standards alignment, data quality etc. In the case of national life cycle inventory, by the time you’ve completed an update for a data point that process needs to start again not long after as all the inputs are changing all the time (grid intensity being an obvious example). So none of this is “static”, it needs to keep evolving.
Another important consideration here is sensitivity. How sensitive is your model and the subsequent design directions to these differences if they were proven legitimate? Would it drive different strategies? The steel and windows example I don’t think would change things. It may bump certain strategies slightly up or down the priority list but I think the strategies would still stand. For example, using a lower carbon structural steel sourced from overseas, using timber in place of steel, using an alternative framing / mullion material for windows that is less intensive than aluminium, choosing a higher quality component with longer life span to reduce recurring impacts etc. These strategies I think are still going to stack up even if you didn’t refine the model.
eTool take data quality pretty seriously. We have a deep QA QC process when we receive a background data update from our supplier (Tim Grant from Life Cycle Strategies). This process has in itself led to improvements in the AusLCI database as it’s teased out issues in the data. We also prevent users from mixing data sources which is a common practice with other tool providers. As well as being contradictory to the ISO standards, I’ve heard first hand accounts of poor design decisions flowing from competing LCA software due to comparing the performance of products with impacts arising from different underlying methodologies and scopes (in one case the conclusion the software was giving was 100% portland cement concrete was lower impact than a fly ash blend which doesn’t stack up and must have been due to some data quality, scope or methodology misalignment). eTool doesn’t allow you to pick and choose data points from different LCI Sources (with the exception of EPDs) which greatly reduces this risk.
If you’d like to drill down into the assumptions that AusLCI uses let me know and we can set up a meeting to discuss in detail.