Regional life cycle impact assessment characterization models

 

Special session coordinators:
      Shanna Shaked, University of Michigan, Applied Physics
      Sebastien Humbert, Quantis

Regionalization is recognized as an important step towards improving the accuracy and precision of life cycle assessment (LCA) results, thereby increasing its discriminatory power for comparative assessments. Many types of damage, such as acidification or (eco-)toxicological impacts on humans and ecosystems, often occur as regional or local impacts, making it important to evaluate them as a function of where the emission takes place. By regionalizing such impacts, decision-makers can have greater confidence in the non-global impacts presented in the LCA.

This session aims at addressing the state of the art in the development of regional life cycle impact assessment characterization models, addressing a wide range of impact categories related to human health and ecosystems, resource use and depletion, and social issues.

To motivate the need for regionalization of impact assessment, the first half begins with a talk addressing regionalization of life cycle inventories and the subsequent differences in impacts due to region of emission. This will be followed by presentations of recent developments and applications of general spatial impact assessment methods for human health, including an application to emerging biofuels.

The second half will focus on regionalization in specific impact categories, including acidification and eutrophication. Regionalization in the expanding field of water use impact assessment will also be discussed. Finally, because social issues clearly vary among regions, the most recent developments in regionalized social LCA are presented.

Presenters:

Shanna Shaked, University of Michigan, Applied Physics

Geographical extrapolation of crop life cycle inventories and impacts
Thomas Nemecek, Agroscope Reckenholz-Taenikon Research Station
Anne Roches, Agroscope Reckenholz-Tänikon Research Station
Frank Hayer, Agroscope Reckenholz-Taenikon Research Station
Daniel U. Baumgartner, Agroscope Reckenholz-Taenikon Research Station
Gérard Gaillard, Agroscope Reckenholz-Tänikon Research Station
Sarah Sim, Unilever - Safety & Environmental Assurance Centre
Llorenç Milà i Canals, Unilever - Safety & Environmental Assurance Centre
presentation

With the increasing use of LCA for life cycle management in the food sector and the globalisation of the latter, there is a strong demand for specific data on agricultural products originating from various regions of the world. A detailed data collection for each situation is not feasible for each food ingredient from each producing country. There is a need to derive generic LCA data describing averages and variability over a wide range of situations. As a first step, we developed a method for geographical extrapolation of crop life cycle inventories and impacts.

A modular crop LCA model was built, by splitting the LCA into nine modules: a basic cropping module (encompassing the minimum operations and inputs to grow a crop), variable machinery usage, tillage, nitrogen, potassium and phosphorus fertilisation, plant protection, irrigation and product drying. Each of the modules is driven by a single parameter that can be relatively easily found for any studied situation (e.g. a farm; a country; a group of countries…). In case such parameters are not available for a given situation, we developed a model to estimate the parameters from FAOSTAT data for any country growing a given crop. Therefore it is possible to calculate crop impacts for any producing country as well as global / regional average impacts and their variability. The validation of the model with crop and country specific data from the ecoinvent database (ecoinvent Centre, 2007) gives a relatively good fit for the energy demand, global warming potential, ozone formation potential, but is less reliable for acidification potential, eutrophication potential (due to soil and climate dependence of the emissions) and ecotoxicity potentials (due to lack of detailed data on pesticide use). The results reveal the huge variability in impacts between different countries for the same crop. By using the underlying variability of agricultural management parameters, we found that, depending on the crop, N fertilisation and irrigation are the key parameters driving most of the impacts; furthermore pesticide application dominates the toxicity impacts. The method can be applied to estimate global median impacts and their variability, as well as for the assessments of crop production in a larger group of countries. The main results are the impacts per input unit for each module, the amount of the inputs used for the world production as well as for each country and the related median impacts, together with their statistical distribution.

 

GLOBOX: a spatially differentiated global fate, intake and effect model for LCA – results for nitrobenzene
Reinout Heijungs, Leiden University
presentation

GLOBOX is a model for the calculation of spatially differentiated LCA toxicity characterisation factors on a global scale. It can also be used for risk characterisation purposes. The GLOBOX model contains equations for the calculation of fate, intake and effect factors, and equations for the calculation of LCA characterisation factors for human toxicity and ecotoxicity. The model is differentiated on the level of 239 countries/territories and 50 seas. Each region has its own set of homogeneous compartments, and regions are interconnected by atmospheric and aquatic flows. Multimedia transport and degradation calculations are largely based on the EUSES 2.0 multimedia model, and are supplemented by specific equations to account for advective air and water transport between different co untries and/or seas. Metal-specific equations are added to account for speciation in fresh and marine surface water. Distribution parameters for multimedia transport equations are differentiated per country/territory or sea with respect to geographic features, hydrology, and climate. The model has been tested with nitrobenzene as a test chemical, for emissions to all countries/territories in the world. Spatially differentiated characterisation factors turn out to show wide ranges of variation between countries, especially for releases to inland water and soil compartments. Geographic position, distribution of lakes and rivers and variations in environmental temperature and rain rate are decisive parameters for a number of different characterisation factors. Population density and dietary intake play central roles in the variation of characterisation factors for human toxicity. The countries that show substantial deviations from average values of the characterisation factors represent a significant part of global GDP. It is concluded that spatial differentiation between countries is an important step forward with respect to the improvement of LCA toxicity characterisation factors.

Spatial variability and optimal regional scale for intake fractions linked to a Canadian emission
Rima Manneh, CIRAIG
Manuele Margni, CIRAIG
Louise Deschênes, CIRAIG
presentation

The human health impact score of a pollutant in LCIA is obtained by the product of the emitted mass, its intake fraction (iF) and effect factor. The iF is calculated as the mass fraction of the pollutant emitted that is taken by the population (Bennett et al. 2002). When developing intake fractions for human health impact assessment, one has to deal with uncertainty. One source of uncertainty comes from neglecting to acknowledge spatial variability of the iFs. Questions arise concerning the level of regionalisation needed in LCIA for toxicity impacts.

The multimedia and multi-pathways model IMPACT 2002 was used (Pennington et al. 2005). iFs were developed for the Canadian context, for the following spatial resolutions: ecozones (15), provinces (13) and sub-watersheds (172). Spatial variability was assessed within and across each type of resolution, in order to determine the optimal regional scale for the evaluation of human toxicity impacts. The overall spatial variability (ratio of highest to lowest iF) was much higher for the sub-watersheds resolution and was up to 10 orders of magnitude for water emissions of acephate and benomyl. The highest iFs were obtained for emissions within regions of high agricultural and animal production. The analysis indicated that the resolutions based on the ecozones and provinces were unappropriate, since they did not bring additional discrimination capabilities among different emission locations.

Regionalization of life cycle impact assessment: geographic differentiation vs archetypes
Sebastien Humbert, Quantis
presentation

The regionalization of impacts in life cycle assessment is a topic of increasing interest. It can considerably increase confidence in results of ‘non-global’ impact categories, such as human health, ecotoxicity or water use. Regionalization reduces the uncertainty of generic characterization factors and enhances the accuracy of LCA. Several models have been developed to calculate characterization factors for specific regions in the world. However, because of the diversity and variations within each region, the number of situations required to calculate a specific characterization factor soon become computation and data intensive (e.g., regionalization based on country level already requires a calculation of about 200 CFs per pollutant and per media of emission). Rather than using specifi c global regions and sub-regions, another viable option is the approach of archetypes. This approach groups several regions with similar defining characteristics, such as population density (urban, rural, remote, etc.); ecoregions (temperate, tropical, boreal, arctic, etc.); water scarcity; etc. Such an approach can significantly reduce the number of regional characterization factors to be calculated, providing at the same time a simple and easy-to-implement solution for life cycle inventory modeling. To make this approach operational, different archetypes must be carefully defined and results must be compared with a geographically differentiated model to demonstrate to practitioners and decision makers that the results obtained with this approach are in line with a ‘true’ geographically regionalized approach.

Based on previous research focusing on regionalization, this work explores the discrepencies in results from the ‘true’ regionalized approach and the archetype approach for different impact categories (human toxicity, respiratory inorganics, ecotoxicity, and water use). For each of these categories, we propose a list of relevant archetypes and corresponding characterization factors, and we explain why the archetype approach proposes a practical solution toward addressing regionalization.

This approach uses a multi-criteria (i.e., different impact categories) analysis and will help life cycle assessment better address the issue of regionalization when considering extended supply chains.

A Life Cycle Impact Assessment Framework for Characterizing Human Health Benefits and Impacts from Emerging Biofuels
Agnes Lobscheid, Lawrence Berkeley National Laboratory
Thomas McKone, Lawrence Berkeley National Laboratory
presentation

Petroleum-based transportation fuels impose a significant disease burden. Life Cycle Impact Assessment (LCIA) provides a framework that addresses both benefits and impacts from biofuels for a number of key impact categories--including human health. LCIA addresses the links among emissions, transport, human exposure, and health damage. We have adapted LCIA to evaluate health impacts of emerging biofuels, such as cellulosic ethanol and butanol, relative to petroleum-based gasoline. This presentation describes the development of spatially resolved characterization factors (CFs) that yield important information on the human disease burden (expressed as disability adjusted life years, or DALYs) attributable to atmospherically emitted pollutants from the following life stages of petroleum-based and emerging bio-based transportation fuels—1) feedstock recovery/production; 2) transport to a fuel production facility; 3) fuel production/refining; 4) storage, transport and distribution of the fuel; and 5) fuel combustion (vehicle use). We use systems models that track the exchange of chemicals between air, water, soil, and plant compartments to provide fate and exposure factors. These factors combine to provide an intake fraction (iF). We combine the iF with chemical-specific effect factors to estimate spatially resolved CFs for human health impacts (DALYs per year) across all fuel life stages for: 1) the contiguous US; 2) a generic urban region; and 3) each US county. For gasoline, we find that the largest disease burden arises from primary and secondary particulate matter emissions during the vehicle-use stage. For gasoline VOC emissions, the majority of health damage occurs in large urban regions, and arises from fugitive emissions from service stations and local fuel distribution by truck. For cellulosic biofuels, we identify where emissions are likely to occur from the feedstock to vehicle use stages, and provide preliminary estimates of characterization factors. We estimate that a 10% reduction in gasoline use would avoid up to 20,000 DALYs per year in the US, primarily attributable to less primary and secondary particulate matter emitted from the vehicle use stage. We characterize the main sources of uncertainty in our impact estimates and prioritize information needs to better characterize the human health benefits and impacts of emerging biofuels.

Estimation of regional characterization factors for aquatic eutrophication
Alejandro Gallego, Department of Chemical Engineering, University of Santiago de Compostela
Luis Rodríguez, Institute for Environment and Sustainability
Almudena Hospido, Department of Chemical Engineering, University of Santiago de Compostela
Mª Teresa Moreira, Department of Chemical Engineering, University of Santiago de Compostela
Gumersindo Feijoo, Department of Chemical Engineering, University of Santiago de Compostela
presentation

LCA have been traditionally considered a site-independent tool, but nowadays there is a trend towards making LCA more site-dependent. Specifically for aquatic eutrophication, characterization factors have been reported for large geographical areas (mainly European and North American countries). Those factors are not detailed enough for countries which present large geographical, climatic and economical variability such as Spain. So, this work aims to calculate the characterization factors for aquatic eutrophication at a regional level, using Galicia (NW Spain) as a case study. Finally, the comparison of the factors here obtained with literature values have been used to analyse the influence of spatial differentiation along the causality chain.

Particular ecological and economic reasons justify the estimation of characterization factors in Galicia taking into account the specific characteristics of three different ecosystems: Atlantic Ocean, freshwaters and rias. Both the principal pathways of transport and the sources of nitrogen (N) and phosphorus (P) were considered to calculate the characterization factors. The analysis on uncertainty carried out identified the principal unsure values: the estimations of fractions of N-NHx and N-NOx deposited maritime waters, land and freshwaters and the amount of N and P deposited in the soil that reaches water ecosystems.

By comparing the results obtained with those available in the literature, it is clear that the application of transport factors in the calculation of characterization factors leads to a more realistic definition of aquatic eutrophication, especially when P inputs to the soil take place. When varying the spatial differentiation (continent, country or region), characterization factors do not vary significantly; however, this variation is likely to increase as long as the definition of the causality chain is improved as it has been reported for other impact categories. In this sense, the methodology here described can be adapted when those effect factors became available, being flexible and suitable for future applications in other regions.

Regionalisation of Ecosystem Sensitivity for Acidification: From the Local to the Global Scale
Pierre-Olivier Roy, CIRAIG
Manuele Margni, CIRAIG
Louise Deschênes, CIRAIG
presentation

With the globalization of markets, assuming that the entire life cycle emissions of a product system occur within a specific geographical area (eg. Europe or Canada) is no longer a suitable assumption for regional impact categories, since the impact of a same elementary flow varies from an emission location to another. Furthermore, current LCIA approaches do not allow to consistently assess and compare emissions occurring from different continents as the characterization factors (CF) are obtained from different characterization models. Thus, a model covering a worldwide scale, but also able to account for local conditions needs to be developed to provide consistent regionalised CFs worldwide. This paper aims to set the basis for the development of a global characterization model able to consistently evaluate acidifying emissions at both the local and global scale for the acidification impact category. Fate factors are evaluated with the GEOSCHEM model. Ecosystem sensitivity was evaluated using the PROFILE steady-state model. Simulation comparison of ecosystem indicators (pH and/or BC/Al ratio) with the PROFILE model using coarse worldwide input parameters and field observations evolution of pH in the soil and/or critical load exceedence over a restrained region (quarter size of Europe) of North America, proved that it was possible to represent local field observations adequately or at least in a better way than available worldwide soil interpolation data. The proposed approach could also overcome the current limitations of the North American LCIA methods which do not consider ecosystem sensitivity.

 

Capability and challenges of regionalized LCIA: the water case
Stephan Pfister, ETH Zurich, IFU
Annette Koehler, ETH Zurich
Stefanie Hellweg, ETH Zurich
presentation

Life cycle assessment is designed to facilitate comprehensive quantification of environmental impacts related to a product or service. However, spatial differences have been widely neglected in impact assessment so far. Especially in the case of water, global average impact factors make no sense as regional features of climate influence greatly the impact. Furthermore, vulnerabilities of ecosystems also vary considerably in space. To allow for proper impact assessment, regions have to be modelled individually or attributed to classes of defined features (archetypes). Recently, new data sources of relatively high spatial resolution and global coverage have become available for many relevant parameters and their quality is continuously increasing. Furthermore, technological development has led to simplified use of geographic information systems (GIS) and increased computer capacity facilitating integration of spatial differentiation into LCA. These advancements should be used for both, LCI and LCIA. We have developed regionalized models to calculate spatially explicit impact factors with global coverage for more than 10’000 watersheds for 3 different midpoints and applied them to simplified LCA of crop production and other water intensive industries. The results show the relevance of the location of an activity but also raise several questions related to regionalized LCIA, such as the question of scale: For different spatial resolutions, different concepts should be used and values of indicators are not fully comparable between different scales. As literature data often link to a certain scale, it is a challenge to adopt the insights consistently to the model resolution. Furthermore, different parameters are best modeled on different spatial scale and require compromises in the model design. From the data availability perspective, we are facing the issue of variable data quality leading to spatially varying modeling uncertainty: Regarding water consumption, in many of the most vulnerable regions data quality is poor and requires special attention in future research.

Regionalisation of impacts from water use
Anne-Marie Boulay, CIRAIG - École Polytechnique
presentation

Although freshwater only represents 2.5% of the planet’s water resources, and that less then 1% of this water is available for human use, this is still enough to fulfill all current human needs. Most problems are arising from the uneven spatial and temporal distribution of this resource. Through the project «Assessment of Freshwater Use and Consumption within LCA », accepted by the 2nd phase of the Life Cycle Initiative, a conceptual framework to characterize impacts of freshwater use was proposed. The methodology developing from this framework addresses regionalisation issues at several stages of the cause-effect chain leading either to a new mid-point impact category “Water Deprivation for Human uses” or to the use of backup technologies to compensate the lack of water. This paper shows how the impacts of water use vary from a re gion to another around the world and discusses the key parameters influencing the regionalisation.

The geographical location where water is being withdrawn and released is an important factor in characterizing the impacts from water use. This is reflected by six parameters taken into account in the proposed methodology.

1. The scarcity of the region will determine whether or not the use of water will change water availability for other users;
2. The regional distribution of water amongst the different users and
3. the local functionality of water for specific activities such as freshwater fishing or hydropower will affect the degree of competition between users;
4. The socio-economic situation of the country, as monitored by the GDP for example, offers an assessment of the adaptation capacity of users to a change in water availability : a country able to adapt to the lack of water will use backup technologies, which will generate different impacts than deprivation for human uses;
5. The leading desalination technology, which is chosen as backup technology for some of th e human uses;
6. The energy grid-mix, which will influence the impacts from the use of backup technologies. A fictitious example is presented where water use impacts from the same pulp and paper plant are modeled for several localizations around the world.

Integration of LCIA and ERA for the assessment of contaminated sediment remediation options
Michael Ditor, CIRAIG
presentation

Contaminated site management requires a holistic approach that ensures a net positive environmental benefit will be gained from remedial action. Such an approach evaluates both the local primary impacts caused by the site-related contamination and the secondary impacts associated with the remediation activities themselves. LCA has been used successfully to evaluate secondary impacts, however site-related impacts are poorly assessed due to the generic nature of LCIA models. The integration of site-specific environmental risk assessment (ERA) results within LCIA has the potential for improving LCA’s ability to assess contaminated site remediation options.

Taking a contaminated industrial port as a case study, site-specific data as well as ERA results are used to develop a local mass balance model and estimate concentrations over time for the water column and sediment layer, considering processes such as tidal flow, degradation, diffusion and settling. These concentration profiles are used to assess the local ecotoxicity impact as well as the contaminant mass transported to the arm of the Pacific Ocean upon which the site is located. Contaminant impacts within the ocean arm and at the global scale are assessed by means of the LCIA model IMPACT 2002, with an added zone taking parameter values from the site investigation to represent the ocean arm.

An LCA is carried out on the following site remediation options: excavation with secure disposal, capping, and monitored natural attenuation. Each option is evaluated both with and without the integration of ERA results to assess the extent to which site-specific data can modify the conclusions obtained using a generic LCIA method.