Hi Adrian,
Good questions. Firstly, it would be good to know whether they are comparing modules correctly which isn’t clear in your question. Are they including waste factor? Are they including lost in transport (if the eTool number does)? Are they accounting for transport impacts? These other factors that are possible causes for deviation.
Assuming that those factors have been ruled out. Yes, the ICE database is top down (average of many studies for a particular material) whereas EcoInvent is bottom up (detailed inventory data for one or many manufacturers to determine an environmental impact). eTool originally used ICE but then moved to EcoInvent based LCIs for the following reasons:
– Higher level of transparency in EcoInvent data, if we want to check the assumptions we can go very deep into the supply chain to understand what’s driving results
– Consistent methodology. Most LCA studies have some form of variation in the scope and/or methodology. Using EcoInvent eliminates this variation whereas it would be very difficult to do this with ICE. Whilst EN15804 has done much to align methodology and scope, there are still variations between EPD programme operators. The ECO EPD platform is working to address this but it’s still relatively early days. EN15804 2019 is still only just starting to get uptake as well (so in the next few years we’ll have scope / methodology differences between older EPDs (EN15804 2013) and newer ones.
– Alignment with standards. I actually personally like the ICE methodology but it doesn’t comply with LCA standards or best practice. One such aspiration of LCA is Energy Mass Balance which you forgo once you start averaging results. I also feel it would be hard for ICE to ensure no double counting.
– Inclusion of other environmental indicators.
– Larger range of LCI processes to select from and include in eToolLCD.
– More scrutiny on EcoInvent: Due to the commercial nature of EcoInvent, and the deep transparency of the LCI data there is more scrutiny on the datasets. My view on LCI data is it’s not a “static” thing. Impacts of materials will always be changing due to lower carbon energy etc. So even if you have your inventory collection perfect, there’s still a need for data to evolve. The transparency of EcoInvent means you can go deep on the assumptions and if warranted change those, or raise issues to EcoInvent so that the data can be improved for the next release.
– All life cycle modules are available for inclusion in studies.
I personally don’t think any of these issues are very significant in the larger scheme of things, particularly at the whole building level, and they are unlikely to significantly drive different design decisions. However we were certainly questioned and/or criticised by the LCA community on occasion in the past for using ICE for these reasons. We felt it was more defensible.
Regarding steel, it is important to note that the “UK” steel in eToolLCD is actually considered to be manufactured in the UK, so it’s not immediately comparable to EU steel. I also understand that EcoInvent steel is more of an industry average supply rather than “Blast Furnace” vs “Electric Arc Furnace”. Certainly Electric Arc Furnace is going to have a much lower impact than Blast Furnace produced steel, however there isn’t nearly enough recycled steel collected to meet current steel demand so if you specify 100% recycled steel in your project, the likely consequence of that is another project just relies a little bit more heavily on primary steel. One thing you can do in eToolLCD to better simulate the electric arc furnace (and I say “better” because it’s far from perfect) is to increase the recycled content up to 100%. However the benefits will only be seen in LC modules A1-A3.
In an upcoming LCI data release we’ll be improving the methodology around how we regionalise too. At the moment it’s pretty crude, we just swap the electricity sources within the supply chain to the local electricity source. We will be changing this to what we call “shallow regionalisation” where we only regionalise the genuinely local supply of products. This will improve accuracy of the data.
