Life Cycle Assessment data – its quality and uncertainties

The ambitious European Green Deal was presented at the end of 2019, and it is slowly being rolled out throughout the EU ever since. The Green Deal outlines a vision of the EU becoming the first climate-neutral continent by 2050, with a target to reduce greenhouse gas emissions by at least 55 percent by 2030.

With this target on the horizon, several EU legislative tools are starting to highlight the importance of Life Cycle Assessments (LCA). These include the EU Sustainable Finance Taxonomy (known as the EU Taxonomy) and the Energy Performance of Buildings Directive (EPBD) proposal.

The use of LCAs is also playing an important part in the upcoming national regulations within EU Member States, with LCAs for buildings becoming mandatory as well as Environmental Product Declarations (EPDs) for building products. Several Member States are looking to set limit values on the Global Warming Potential (GWP) in CO₂-eq per m². But what do we need to know about LCAs when using them as an instrument for policy making or as a way to measure the environmental impact of buildings?

Rule of thumb: 10 percent uncertainty in LCA results

LCA is a cradle-to-grave or cradle-to-cradle analysis technique to assess environmental impacts associated with all the stages of a product's life.^[1] Figure 1 shows the stages involved in the LCA for thermal insulation; raw material supply, manufacturing the product, (transport to site) and installation, use phase (in buildings), end-of-life stage (including demolition), waste processing, transport of waste and disposal, and lastly the benefits from reuse, recovery and recycling, which are often considered as being beyond the product life cycle.

In a similar structure, a building LCA covers the product stage, construction process stage, use stage and end-of-life stage, as well as the benefits and loads beyond the building life cycle. LCAs are vital instruments for measuring impacts on climate change as well as other environmental impacts, and through their use, we can develop more sustainable and energy efficient solutions at every stage. LCAs can also support decarbonisation in the building sector, especially when it comes to measuring the reduction in terms of embodied carbon which is in the spotlight of EU legislations.

However, it is necessary to recognise that there are several uncertainties associated with LCAs including; a) stochastic uncertainty (unpredictability), b) choice uncertainty and c) a lack of knowledge of the studied system^[2]. On average, LCAs provide results with an uncertainty of at least ±10 percent, and it is this inconsistency that has resulted in some criticism of the methodology. The ISO 14000 series of standards aim to provide requirements and guidelines to normalise the methodology, but the standards themselves are still ambiguous enough to allow different interpretations. In other words, an LCA of the same product completed by 10 different LCA practitioners has the potential to yield 10 somewhat different results.

Raw material data quality and choice

LCAs can depend heavily on the raw material databases available for the chosen LCA software, and this is the case when it comes to the manufacturing stage of LCAs for construction products. LCA practitioners usually choose the data-sets most representative to reality in terms of processes and geography, but sometimes they can be limited by what is available. Table 1 demonstrates the investigation of raw material data results from GaBi and Ecoinvent databases, the two prevailing databases often used in the construction industry for producing LCAs and EPD results.

For mineral wool production, several energy, binder, stone and packaging materials have been investigated and compared to steel production datasets – which are not typically used in mineral wool production – for reference. For stone materials such as stone and cement, the datasets from GaBi and Ecoinvent show very strong correlation, whereas for energy materials such as coke and coal or binder materials, the difference of the datasets can be as high as 20 to 60 percent^[3].

During the manufacturing stage of mineral wool production, differences within the dataset can lead to a difference in the GWP result of up to nine percent^[3], see Table 2. If the transportation of the materials to the site is included, the difference can be up to seven percent. In terms of the whole life cycle impact – when both use and end of life stages are included – the difference would be somewhat smaller. Nevertheless, the difference of datasets should be recognised at the raw material stage.

Data quality at product level

Material/product data: Up to 30 percent difference in results using generic data vs specific EPDs

According to the European Standard EN 15804 applicable to most construction products, EPDs are well established. With an aim to show a product’s environmental performance in a standard way, EPDs are widely used as input data when making the LCA of an entire building by applying EN 15978.

In other words, EPDs provide specific and necessary data of chosen building materials and products within a building in order to assess its overall environmental impact. Just as a nutrition label provides information about the vitamins, minerals and fats contained within a certain food type, an EPD provides information about the environmental impacts of certain construction materials. Staying with the food analogy, the amount of fat listed on a label only becomes relevant when considering a person’s entire diet. In the same way, the information provided in the EPD of a specific construction material should only be evaluated in the context of the entire building.

In reality, many compare the EPDs of individual products in the hope of finding the most environmental option – although it is not advised to do so according to EN 15804. As EPDs are also based on LCAs, the uncertainties and differences already mentioned are inherent to the EPDs of building materials and products. Decisions should therefore not be made based on a difference of EPD values between two products that is equal to or less than 10 percent.

The uncertainty of a building’s LCA can deteriorate depending on the availability of product specific EPDs in the building’s LCA software and its connected databases. Not all of the manufacturers of building materials have the resources and competences to generate and publish EPDs in the same way as the ROCKWOOL Group has since the early 2000s. This means that the designer and LCA practitioner can only look for a generic dataset to represent the chosen product – and as this is usually based on an average industry result, it is often outdated. Rambøll^[4] and the Department of the Built Environment of Aalborg University^[5] have each published a report on the LCA of buildings, and the conclusions of both are similar. The Rambøll report states that using product specific EPDs instead of generic data for best-in-class conventional building materials led to a reduction of CO₂-equivalents (unit for Global Warming Potential) of between 14-30 percent, depending on the building type. The Aalborg University report shows also a similar level of differences between generic data from Ökobaudat of three external wall types in comparison with using specific EPDs results, see Figure 2.

Accumulated uncertainties at building level

It is certainly a positive step that LCAs are becoming important tools to support the decarbonisation of the built environment and construction sector, both when it comes to in private companies and to the regulatory sphere in Europe. But it is crucial to consider the uncertainties potentially experienced at the raw material data level, the product level and lastly at the building level when estimating the environmental impacts and the whole life carbon of a building.

LCA is a great modelling tool, but it is not yet a precise science. Think of it this way; there are uncertainties associated with raw materials and EPDs as well as the lack of product specific EPD – and general uncertainties within LCA practice, so it makes sense that this general uncertainty also applies to building LCAs. For the End-of-Life scenarios, LCA practitioners need to be able to make their best expert judgement, as the design lifetime of a building is a minimum of 50 years and above. When it comes to looking forward, there is inevitably a degree of unpredictability about the future. End-of-Life scenarios can be one of the least harmonised parts of LCAs, but also one of the most crucial stage for whole life thinking. Some models are based on currently available End-of-Life practices, some consider more optimistically what the future looks like, and in other cases, the Product Category Rules impose either a country specific scenario or a worst case scenario. This hampers the level playing field of construction products used in a building, and can be exploited to show artificially low impacts at the End-of-Life.

Paving the way to set life cycle requirements

One major potential improvement could be the creation of a single European background database, as this would minimise differences at the raw materials level. At product and materials levels, EPDs should be harmonised with the same indicators, format and functional units, so these harmonised EPDs could be used for building LCA software with ease.

Secondly, the European Commission, industry and NGOs need to support the development of more clearly defined scenarios and more uniform Product Category Rules (PCR) for construction products. This is in order to ensure a maximum level playing field, improve understanding, and avoid incorrect comparisons.

The upcoming revision of the Construction Production Regulation can establish a single common background database and set mandatory requirement on EPDs according to harmonised standards. This way, it ensures that the data to be delivered by individual construction products, together with CO₂-reporting requirements at the building level by the EPBD, will pave the way to set correctly life cycle requirements (such as the Whole Life Carbon) for the EU built environment.