Time-Integrated Risks and Potential Consequences: The Fate and Exposure Modelling Community’s Role in Life Cycle Assessment (LCA)

Effective implementation of control measures has reduced many risks from single sources. Exceptions include heavy urban traffic and some indoor air emissions. At the same time, continual increases in production and consumption have resulted in an intricate network of emission sources. Nowadays, impacts are commonly the result of pollutant cocktails from multiple sources. The need exists for methodologies to also provide proactive decision support beyond the insights of the typical risk assessment practices; considering the time-integrated likelihood (risks) and potential consequences of impacts linked to marginal changes in chemicals emissions, rather than comparing concentrations to policy-driven thresholds at a particular point in time and space for chemicals in isolation.

Life Cycle Assessment (LCA) is one decision-support tool that has the potential to identify and quantify the time-integrated burdens of products and services; rather than for specific processes and sites. The identification of opportunities for preventing pollution and for reducing the consumption of resources using LCA ultimately leads to products with the potential, and advantages, of an improved life-cycle performance. Benefits range from commercial advantages (through eco-labelling, green purchasing programs, improved corporate responsibility, greater public awareness, ...); to supporting governmental obligations (reduction, rather than displacement, of pollution; improving best-available practice guidelines for industry sectors; …).

With a bias towards the fate and exposure community, we outline here some of the challenges and opportunities that are bringing different disciplines and stakeholders together to identify, implement, and advance best available practice in LCA to reach these objectives.

1. Life Cycle Assessment (LCA) Overview
LCA is similar to the common chemical engineering practice of conducting a mass and energy balance for a process flow diagram (PFD), only broader in scope. An LCA practitioner tabulates the emissions and the consumption of resources at every stage in a product’s life cycle, from its cradle to its grave (raw material extractions, energy acquisition, manufacturing, use and waste disposal). Figure 1 provides an example of such a life cycle. It is desirable that practitioners can then calculate indicators of the likelihood (risks) and the potential consequences of associated impacts in the context of climate change, stratospheric ozone depletion, tropospheric ozone (smog) creation, eutrophication, nutrification, acidification, toxicological impacts on human health and ecosystems, fossil fuel depletion, water consumption, land use, …

Figure 1 Simplified life-cycle flow chart for 1,4 Butanediol (BDO) derived from corn glucose (energy consumption and associated processes not shown) as evaluated for different impact categories in Pennington et al. (2001).

An LCA consists of the four phases in Figure 2 (ISO14040, ISO14041, ISO14042). We focus here on the Life Cycle Impact Assessment (LCIA) phase, but an understanding of the other three phases is first essential.

Figure 2 Phases and applications of an LCA (based on ISO14040)

The goal and scope definition of an LCA provides a description of the product system in terms of the system boundaries and a functional unit. The functional unit is the important basis on which alternative goods, or services, are comparable. The functional unit is not necessarily a quantity of material (although this is the case for butanediol in Figure 1). Practitioners may compare, for example, alternative types of packaging on the basis of 1 m³ of packed and delivered product. The amount of packaging material required can vary, depending on the option selected (paper vs. plastic, …).

Life Cycle Inventory (LCI) is a methodology for identifying and evaluating resource consumption and emissions at all the stages in a product’s life cycle. These emissions are likely to occur:
· at multiple sites,
· as different fractions of the total emissions at any one site (the fraction required to provide the specified functional unit; allocation amongst related and non-related co-products in a facility such as a refinery; …),
· at different times (use phase of a car compared to its disposal),
· and over different time periods (multiple generations in some cases).

Interpretation occurs at every stage in an LCA. In many LCAs, for example, a few emissions will dominate the category of human health impacts in the life cycle and the results will be readily interpretable. In others, normalisation (such as weighting indicators for different impact categories relative to the totals for a given geographical region) and multi-criteria decision making (panel-based weighting methods, economic valuation, …) provide useful aids to interpret results. See Finnveden et al. (Udo de Haes et al. 2001) for further discussion.

2. Life Cycle Impact Assessment (LCIA)
A large number of indicator methodologies are proposed in the literature for characterising impacts in different impact categories in LCA (Udo de Haes et al. 2001, Pennington et al. 2001, Bare et al. 2000, Bare et al. 1999). Not all these methodologies are suitable, or scientifically defendable, in the context of risks and potential consequences. Methods in LCA software are often outdated, providing conflicting and sometimes misleading results. These example issues present practitioners with serious problems when interpreting the decision relevance of LCA results (Owens 1999, Assies 1997) and when selecting a methodology. Current practice has given rise to potentially avoidable ISO 14042 statements such as “LCIA results do not predict impacts on category endpoints, exceeding of thresholds, safety margins or risks.” The important need for researchers, users, and stakeholders to understand the true relevance of LCA indicators and to identify, implement, and advance the best available practice is therefore clear.

Focusing on emissions, the inventory data for a life cycle are in terms of the mass of chemical released to provide a given product or service (kg/functional unit). Emissions data are not in the form of flow rates (kg/year). It is therefore desirable to estimate the time-integrated likelihood (risk) and potential consequences of potential impacts associated with a unit mass of a chemical emitted into the environment; termed a characterisation factor.¹ Considering time-integrated impacts linked to a mass of a chemical emitted, rather than to a flow rate and resultant concentration, represents a clear and important departure from many current risk assessment practices.

¹While it may be ultimately desirable to calculate the likelihood (risks) and the consequences of the impacts linked to providing a given product service (so-called endpoint indicators), many decision makers prefer to multiply emitted quantities by midpoint indicators. Global Warming Potentials (GWPs), an example midpoint indicator based on an incomplete cause-effect chain description, reflect the relative infrared adsorption capacity of an emission in terms of CO2 equivalents. The likelihood and consequences of the impacts are not quantified or forecast – an advantage in many policy arenas. Udo de Haes et al. (2001) and Bare et al. (2000) summarize the merits, uncertainty tradeoffs, and disadvantages of midpoint versus endpoint indicator approaches for different impact categories.

Having multiplied the emissions data in a life cycle inventory by the impact characterisation factors, the resultant indicators are summed for each assessment endpoint (human health, aquatic ecosystem impacts, …) to provide overall indicators of the likelihood and potential consequences of the impacts associated with a product or service (on a per functional unit basis). Depending on the total demand for the functional unit, it is conceivable to estimate actual burdens attributable to meeting the product or service demands of a society for a given time period, such as one year. As these are risk-based measures they do not necessarily imply that actual impacts will occur. It is equally important to note that the risks and potential consequences are still summed over time and space in an LCA. The risks attributable to one particular site may not reflect the full extent of risks from that site (due to allocation amongst co-products in the inventory stage, consideration of the fraction attributable to the functional unit, …), and the risks may occur over multiple generations.

3. Fate and Exposure Modelling in LCIA
Now we focus on toxicological impacts; noting that analogous arguments are applicable for other impact categories such as global warming, eutrophication, water use, … (Udo de Haes et al. 2001)

Acknowledging that the relevance and suitability of many LCIA methods varies extensively and continual research advancements are essential, the likelihood and potential consequences of toxicological impacts associated with the mass of a chemical emitted into the environment can be calculated. For ecosystem impacts, this can be presented in the form of the change in the time-integrated exposure concentration associated with one kg of chemical emitted into the environment. Intake fraction² provides an exposure measure to help assess human health impacts.

Current techniques rely on time-integrated solutions to linear (or pseudo-linear) fate, exposure and effect models.³ Heijungs (1995) outlined the straightforward relationship between the steady state and time-integrated exposure solutions for such models. When, and over which time period, the emissions actually occur becomes less relevant – at least in the context of effects related to long-term (chronic) exposure.

Site-dependence: In all models in LCIA the location of a release can be very important (Udo de Haes et al. 2001). Specific-medium models are appropriate in LCA to characterise the marginal impacts of some pollutants such as particulate matter, NOx, SOx, and ozone. Localized dispersion modelling is relevant for short-lived chemicals and to account for the fraction of exposure that occurs in more densely populated areas (which may dominate some time-integrated exposure scenarios). If intermedia transport plays an important role, then we can calculate characterization factors using multimedia models.

²Fraction of an emission that results in human exposure – sometimes termed exposure efficiency; Exposure efficiency - “the fraction of material released from a source that is eventually inhaled or ingested” (Evans et al. 1999); Dose fraction - the fraction of a chemical mass emitted into the environment that eventually passes into a member of the population through inhalation, ingestion, or dermal exposure (Bennett et al. 2001).

³Pseudo-linear or marginal measures are adopted for air pollutants such as NOx, SOx, ozone, … Marginal measures linearly reflect the expected change in likelihood and consequences associated with a small change of an emission at a working point on a non-linear source-to-dose-to-response curve. For many other emissions, risks and consequences will be at low doses and exposure to complex mixtures of chemicals from multiple sources. Chemical-specific thresholds are not generally considered relevant in such a context (arguably a precautionary stance, but justified by many recent insights such as cumulative narcosis effects on ecosystems and given the current levels of contamination in the environment). Linear dose-to-response-likelihood curves are justifiable and some methods attempt to quantify associated consequences. See Crettaz et al. (2001a, b), Udo de Haes et al. (2001) and Potting et al. (1999) for further discussions.

While medium-specific models are usually considered to provide better spatial resolution, MacLeod et al. (2001) presented a multimedia model with spatial capabilities for the North American continent. Pennington (2001) presented a similar prototype for Japan and for Europe. The Japan and Europe models additionally facilitate calculation of both location-dependent and generic exposure measures for use in LCIA. The prototypes account for human population distribution, regional agricultural production yields and water supply sources, as well as watershed boundaries and wind patterns, to estimate the fraction of contaminants passing into the human population. The use of exact matrix-based solutions, for both the dynamic and steady-state multimedia mass balances, further improves the transparency and flexibility. The resultant generic and site-dependent exposure factors for LCA are accompanied by parameter uncertainty estimates; the confidence intervals on the generic indicators additionally accounting for the importance of not considering spatial variation and emission location in screening-level LCAs. Evaluation studies are now underway with several partners and potential end users.

Coupled Medium-Specific Models: Single-medium models are not considered appropriate in current practice for time-integrated exposure calculations for many so-called multimedia chemicals and for some emission scenarios (e.g. addressing human health impacts from benzene emissions to water). On the other hand, multimedia models are criticised for a lack of transparency and, in some cases, over-simplification of the transport/fate equations. Margni et al. (2001) proposed one solution to “have your cake and to eat it”: the integration of medium-specific models using robust coupling approaches in a modular framework. An exact multimedia solution is maintained using a “plug-in” matrix approach with inputs from separate medium models; as necessary and to best fit the assessment requirements. However, as feedback fractions (a measure of the extent of cyclic transfer of pollutants between media) are small for the majority of chemicals, the matrix step for integrating the medium-specific models is generally unwarranted. The single medium models can be directly coupled for at least ~95% of substances.

4. Best Available Practice in LCA
SETAC is now finalizing its document “Towards best-available practice in Life Cycle Impact Assessment” (Udo de Haes et al. 2001). The working groups that wrote this book addressed fate, exposure and effect modelling in the context of current LCA practice; considering climate change, ozone depletion, smog formation, acidification, eutrophication, toxicological effects to humans and ecosystems, … A clear need now exists to build on these insights by identifying, implementing and continually advancing the state-of-the-art and best available practice in LCA. This is a multidisciplinary challenge and must involve all stakeholders (industry, government, academia, NGOs). Here we outline two initiatives that are constructively channelling efforts towards achieving these goals:

OMNIITOX aims to enhance the capability of European industry to select more environmentally benign chemicals and processes; ultimately leading to products with the potential and all the advantages of an improved life-cycle performance. This program is building on: (1) industrial sector workshops (automotive, detergents, pulp & paper, cosmetics); (2) associated case study work packages; (3) a cross-comparison of site-dependent risk assessment and LCIA methods; and (4) the application, definition and identification of improvements for toxicological impact characterisation in LCIA. For further information please contact Sverker Molander (Coordinator, Chalmers, Sverkerm@vsect.chalmers.se ) or Pere Fullana (Technology Implementation Manager, Randa Group, pfullana@randagroup.es ).

The Life Cycle Initiative combines the worldwide resources of industry, government, academia, and NGOs under the umbrella of the United Nations Environment Programme (UNEP) and the Society of Environmental Toxicology and Chemistry (SETAC). This initiative will support further LCA progress by identifying and developing methods, tools, data and case studies to lead towards best available practice. For further information and to participate please visit the website (http://www.uneptie.org/sustain/lca/lca.htm ). To become directly involved as a funding partner please contact Anne Solgaard (UNEP DTIE, anne.solgaard@unep.fr ). For specific technical questions related to impact assessment in LCA, or this article, please contact the co-authors.

5. Concluding Remarks
Emissions into the environment and the consumption of resources associated with the provision of a product, or service, occur at many locations and over different time periods (multiple generations in some cases). These emissions and consumptions contribute to climate change, stratospheric ozone depletion, tropospheric ozone (smog) creation, eutrophication, nutrification, acidification, toxicological effects on human health and ecosystems, the depletion of resources, water consumption, land use, … A clear need exists to be proactive and to work beyond many current regulatory practices. In Life Cycle Assessment (LCA), practitioners from many domains come together to try to estimate the likelihood and potential consequences of time-integrated burdens linked to providing society with products and services. LCA supports the identification of opportunities for pollution prevention and for reducing resource consumption - ultimately leading to products with the potential, and all the advantages, of an increased life-cycle performance. But, there are, and will continue to be, opportunities for improvements.