How to deal with data gaps
In Life Cycle Assessment (LCA), life cycle inventory data (LCI) play a critical role. They contain the environmental profiles and impacts of specific products or technology. Map your product’s supply chain components with LCI data to receive a complete virtual picture of your product, including its environmental impacts – what we call an LCA. Naturally, the better and representative the LCI data, the more robust your LCA results are, and the more accurate the conclusions you can draw to benefit your business.
Representative data means product, technology and country
Representative LCI data correctly describes the product, technology and country of your supply chain component - what is the intermediate product, how was it produced, and where. But what can you do if you don’t have or can’t source THAT dataset? Today’s supply chains are complex, specialized and global. Little wonder if you don’t always find exactly what you need.
When modeling a new material or component it wouldn’t be unusual not to find LCI data to support your model. You need additional data to prove the new material is a benefit to the overall product LCA but you can’t find specific enough data.
How non-representative data affects your results
We ran a test scenario to evaluate the effect of non-representative LCI data. A chemical product was paired with different country information for five impact categories, and several chemical technologies were paired with the same country. The key indicator being how much the single results deviate from the average of all results, as a percentage. In the charts below, that means 0% is the average, and the arrows point to the highest and lowest deviation respectively. The blue bars indicate the range of values for 80% of the results, so the majority of the results lie in that range.
For both country and technology, the deviations are fairly significant – the taller the bar, the higher the range. For some impact categories, values are multiple times higher than the lowest value or even the average! Indeed, technology and region can significantly influence your results, underlining the importance of representative data (being country specific beyond electricity).
Strategy to deal with data gaps
thinkstep has extensive experience with LCA, having calculated 100’s of LCAs and several thousand datasets over the last 20 years. Our recommendation is this: For a start, use the dataset that comes closest to what you need. Then, check its contribution to the overall result. The tool for that is called sensitivity analysis. In GaBi, you can find it in GaBi analyst. A good way to keep track of your interim dataset is to pin point quality weaknesses using the data quality indicators in GaBi. That way, you can run your analysis and determine the dataset’s contribution to the end result. If the contribution is substantial, and data quality indicators low, you can consider replacing it with the specific dataset you need.
How to get the data you need
There are options to get the dataset you need. The first one is DIY: You can sit down and model it yourself. A less time consuming alternative would be to use thinkstep’s large data repository. We have more than 13,000 available datasets and we can also customize or create a new dataset cost effectively. This way, you can be sure that your data is consistent in how it’s modeled, upgraded annually and we are always available for questions or clarifications.
More than 13,000 datasets available
We offer data bundles but each dataset can also be purchased (and adapted) individually:
- GaBi Professional Database is the most used database that offers ‘the basics’ in over 3,000 datasets
- GaBi Extension Databases provide over 7,000 sector specific datasets, for example for end of life, plastics, electronics, construction materials etc.
- Data on Demand – single datasets that we can adapt or create specific to your requirements.
Results and charts taken from the whitepaper Addressing uncertainty in LCI data with particular emphasis on variability in upstream supply chains, thinkstep 2012.