Characterisation: new developments for toxicity
This is the first in a series of articles where PRé's own consultants explain a step in the LCA process and the way their own expertise fits into that step. First up is Laura Golsteijn, who is not just an expert on characterisation in LCIA, but also earned a PhD based on her original characterisation research.
Today, I am defending my PhD thesis on enhancing methods of toxic impact modelling in LCA at the Radboud University in Nijmegen. These methods are part of the characterisation step of the impact assessment phase of a life cycle assessment (LCA) study. First, I’ll briefly explain the phases and steps of an LCA, and then I’ll give you a short overview of the new developments my PhD research has contributed to.
Why Impact Assessment?
An LCA assesses and quantifies the environmental impact of a product or service over its entire life cycle. The main phases of an LCA are goal & scope setting, inventory analysis, life cycle impact assessment (LCIA), and interpretation. An inventory analysis provides information on all relevant energy and material inputs, and on the emission of toxic and non-toxic pollutants, but that alone does not provide enough information to guide decision-making. To be able to understand the consequences of these inputs and emissions, we need to translate them into environmental impacts. The impact assessment phase provides this translation.
Steps in Impact Assessment
The LCIA phase consists of four consecutive steps.
1. Classification. All substances are sorted into classes according to the effect they have on the environment. A cause-effect pathway shows the causal relationship between the environmental intervention (for instance, the emission of a certain chemical) and its potential effects. LCA professionals can choose impact indicators at different stages in this pathway, for example the midpoint or endpoint. Classification is relatively straightforward, which is why we haven’t given it its own article.
2. Characterisation. All substances are multiplied by a factor which reflects their relative contribution to the environmental impact, quantifying how much impact a product or service has in each impact category. Characterisation is the focus of this article.
3. Normalisation. The quantified impact is compared to a certain reference value, for example the average environmental impact of a European citizen in one year.
4. Weighting. Impact categories are assigned an importance value, and the resulting figures are used to generate a single score.
The Characterisation Step Calculates Impact
To quantify how much impact a product or service has in the different impact categories, we use characterisation factors (CFs). CFs express how much a single unit of mass of the intervention contributes to an impact category; how much 1 kg of chemical emission contributes to ecotoxicity, for instance. Let’s take the example of ecotoxicity and see how characterisation factors can be calculated.
For a proper assessment of the toxic impacts of chemical emissions, researchers need to follow a systematic modelling procedure through the fate, exposure and effects of the chemical.
- Fate. The environmental fate of a chemical describes the proportion of chemical that is transferred through the environment, and the length of time the chemical stays in the different environmental media.
- Exposure. Various species in an ecosystem can be exposed to chemicals through different uptake routes, such as inhalation of polluted air or ingestion of polluted water. The fate and exposure of chemicals are generally modelled with multimedia fate and exposure models.
- Effects. The effect of a chemical is determined by the sensitivity of a species to that chemical, among other factors, and is often derived from experimental toxicity data.
Environmental fate, exposure and effects are combined into one quantitative measure, the substance-specific characterisation factor.
New Developments in Characterisation Factors
During my PhD studies, my former co-workers and I developed new methods to predict missing data and improve inclusion of overlooked impact pathways, to enhance toxic impact modelling in LCA. Obviously, large amounts of data are required to calculate CFs, and often, there are gaps. The first goal of my PhD research was to discuss the practical usefulness of methods to predict missing data for LCA. The second goal was to include potentially relevant impact pathways that are commonly neglected at present.
Methods for Predicting Missing Data
In many cases, CFs for toxicity suffer from uncertainty because the data on the physicochemical and toxic properties of the chemical of interest is not complete. To enhance limited experimental datasets, we can extrapolate from measurement data, e.g. between chemicals, environmental media, or animal species.
- Extrapolation between chemicals. Based on their similarity in chemical structure, it may be possible to extrapolate from measured physiochemical or toxic properties of one chemical to another, using quantitative structure-activity relationships (QSARs).
- Extrapolation between environmental media. Using the so-called equilibrium partitioning method, we can extrapolate from toxicity values for one medium to another, for example from freshwater to soil. This method uses the chemicals’ sorption equilibrium.
- Extrapolation between species. Extrapolating a measured chemical property, such as toxicity, from one species to another can be done with interspecies correlation estimation regressions.
In my PhD thesis, I quantified and compared various sources of uncertainty in impact assessments, e.g. uncertainty due to small size of a species sample, predictive uncertainty related to regression models, and uncertainty in input data. I then discussed which estimation methods are most suitable to supplement experimental datasets, depending on which type of uncertainty contributed most to the overall uncertainty. Since data estimation methods are inherently uncertain, as well, it is important to select a method that complements the existing dataset.
My comparison showed that uncertain environmental degradation half-lives and a small sample size of species contribute most to the overall uncertainty in a toxic impact assessment based on estimation methods. Uncertainty from chemical-group specific QSARs generally contributes less to overall uncertainty. When it comes to assessing the effect of chemicals, using estimation methods can have two important advantages: an increase in sample size and an increase in diversity (representativeness) of the sample. I found that the best way to enhance limited toxicity datasets for a toxic impact assessment is to supplement the experimental data with interspecies correlation estimates.
Methods for Including Missing Impact Pathways
In currently available methods for ecotoxicity LCIA, some potentially relevant impact pathways are often overlooked. One of these pathways is the bioaccumulation of chemicals in aquatic food chains. During my PhD research, my co-workers and I developed a method to include the impact of chemical emissions on warm-blooded predators in freshwater food chains.
Another pathway that is often neglected is the human toxicity of indoor chemical emissions in occupational and consumer settings. We developed a method to include this type of toxicity in LCA, while accounting for variation in exposure scenarios. A case study on metal degreasing showed that, for an assessment of the life cycle impacts for human toxicity, toxicity from indoor chemical exposure is of major importance.
For questions about my PhD thesis, or related peer-reviewed scientific articles, please feel free to contact me.
The next article in this series will feature the normalisation step of LCIA.
I am eager to increase the environmental awareness of our society, and I believe that everyone can contribute to a more sustainable world, every day. At PRé we provide companies with both the knowledge and the tools to improve their products and services. I am excited to work for an organisation that is involved in developing sustainable initiatives.