Keywords

1 Introduction

The good news is: life cycle assessment (LCA) is approaching mainstream. After many years of method development, case studies, international standardization, database and software development, LCA is mature and robust enough to be used for decision-making—in both private and public organizations. LCA is currently the most accepted tool to assess the environmental performance of products and this basically applies all around the globe and to all stakeholders, e.g. government, industry, non-governmental organizations (NGOs), academia .

Ten years ago, the European Commission testified in their Communication on Integrated Product Policy (IPP), that LCA is the “…best framework for assessing the potential environmental impacts of products currently available” (EU 2003). This statement stood the test of time and is nowadays common sense beyond Europe. However, being ‘the best available method’ does not mean that LCA is ‘perfect’. While the LCA community had to promote LCA uptake for many years, it is now important not to oversell it. A balanced understanding and use of LCA is needed for ensuring sustainable success. We have to avoid going from the ‘LCA-aversion’ of the past straight into a kind of ‘LCA-hype’.

Both, the international standards of LCA and the scientific literature are quite transparent with regard to the gaps and challenges of the method. LCA does not provide the ‘full environmental truth’, at least not just yet. The core standards of LCA (ISO 14040 2006) and (ISO 14044 2006) (see Chap. 3, this volume, entitled ‘The international standards as constitution of LCA: the ISO 14040 series and its offspring’) acknowledge clearly that any LCA study has its limitations. Therefore, the limitations of every study have to be documented in the goal and scope definition (ISO 14040, 5.2.1.2). Moreover, there is a specific chapter on limitations of life cycle impact assessment (LCIA), i.e. ISO 14040 , 5.4.3 Limitations of LCIA: “The LCIA addresses only the environmental issues that are specified in the goal and scope. Therefore, LCIA is not a complete assessment of all environmental issues of the product system under study. LCIA cannot always demonstrate significant differences between impact categories and the related indicator results of alternative product systems. This may be due to

  • limited development of the characterization models, sensitivity analysis and uncertainty analysis for the LCIA phase,

  • limitations of the LCI [life cycle inventory] phase, such as setting the system boundary, that do not encompass all possible unit processes for a product system or do not include all inputs and outputs of every unit process, since there are cut-offs and data gaps,

  • limitations of the LCI phase, such as inadequate LCI data quality which may, for instance, be caused by uncertainties or differences in allocation and aggregation procedures, and

  • limitations in the collection of inventory data appropriate and representative for each impact category” (ISO 14040 2006).

In the scientific literature, an early analysis of drawbacks was performed by Udo de Haes (Udo de Haes 1993). More recent contributions with regard to gaps and research needs include Reap et al. (2008a, b), Finnveden et al. (2009) and Klöpffer and Grahl (2009) . However, based on the understanding of this decade, a comprehensive overview of gaps and challenges is still missing, especially reflecting the more recent developments with regard to carbon footprinting (Finkbeiner 2009) and water footprinting (Berger and Finkbeiner 2010) as well as the less explored aspects like littering or animal well-being.

The following section details the approach and methodology chosen for this review article. Section 3 presents the resulting gaps, challenges and research needs.

2 Methodology

This contribution is based on an extensive desk and literature research by an experienced and interdisciplinary group of LCA scientists and practitioners. Each of the co-authors was responsible for a set of topics, performed the associated literature survey and prepared the necessary background material. Due to the large number of gaps and challenges identified, it was necessary to restrict the gap description and analysis to a high-level summary. More detailed and comprehensive reflections on individual gaps go beyond the scope of this contribution.

The comprehensiveness of the selection of challenges was not stretched to the limit. There are even further issues which could have been included as challenges for LCA. As examples, normalization, definition of system boundaries or the application of cut-off criteria are not discussed, because they were comprehensively covered in previous reviews.

For consistency in the gap descriptions, a common format is used to describe each gap. Four guiding questions were analyzed for every challenge. In order to support readability and easy reference, the presentation in the result section is organized in such a way that—whenever feasible—the guiding questions are discussed in a respective paragraph and according to the following order:

  1. 1.

    What is the topic about?

  2. 2.

    Why is it a gap of or challenge for LCA and in which case is it particularly relevant?

  3. 3.

    What is the state of the art in the scientific literature?

  4. 4.

    What can be done to address the gap or challenge?

Basically, all gap descriptions are self-sufficient. However, to provide a structure for their presentation and for readability purposes they were attributed to one of the following topics which correspond with individual subsections in the results part of the chapter:

  • Inventory aspects (see Sect. 3.1)

  • Impact assessment aspects (see Sect. 3.2)

    • Human health (see Sect. 3.2.1)

    • Ecosystem (see Sect. 3.2.2)

    • Resources (see Sect. 3.2.3)

  • Generic aspects (see Sect. 3.3)

  • Evolving aspects (see Sect. 3.4)

It should be noted that the focus of this contribution is clearly environmental LCA and does not include a discussion of gaps and challenges of social LCA (SLCA) (UNEP/SETAC 2009), life cycle costing (LCC) (Hunkeler et al. 2008; Swarr et al. 2011) or comprehensive life cycle sustainability assessment (LCSA) (Klöpffer 2008; Finkbeiner et al. 2010; UNEP/SETAC 2011b).

3 Results

As a main result of this contribution 34 methodological gaps of and challenges for LCA were identified and grouped according to the structure presented in the section above. Figure 7.1 provides an overview of the topics covered.

Fig. 7.1
figure 1

Structured overview of challenges for LCA (shaded areas indicate crosscutting between areas of protection within impact assessment)

Full size image

In several cases, the attribution of gaps and challenges to one of the overall topics is not straightforward. In such ambiguous cases, the decision for a certain topic is based on our judgment as to which of them represents the most dominant aspect of the gap. As an example, there are definitely still challenges with regard to the impact assessment of ‘water use and consumption ’ (see Sect. 3.1.1), but the currently limiting factor in application of advanced water footprint assessment is the lack of proper inventory data. Therefore, the challenge ‘water use and consumption’ is covered under inventory aspects .

Because of the larger number of challenges and the availability of established areas of protection, the impact assessment aspects are further differentiated into human health , ecosystem and resource topics (Fig. 7.1). Within impact assessment some of the gaps and challenges are clearly crosscutting between areas of protection. This is indicated by the shaded areas in Fig. 7.1. As an example, ‘nanomaterials’, ‘endocrine disruptors ’ and ‘noise’ are attributed and discussed first in the human health section, but they are equally relevant for ecosystems . The following sections provide the individual gap descriptions according to the structure presented in Fig. 7.1.

3.1 Inventory Aspects

This section analyzes gaps and challenges referring to inventory aspects of LCA. This includes a discussion on ‘water use and consumption’, ‘renewable energy’, ‘biogenic carbon’ and ‘delayed emissions’ as well as the inclusion of ‘improbable events’ in LCA. Moreover, inherent challenges like ‘allocation’ and ‘functional unit ’ are discussed.

In Fig. 7.2 gaps and challenges regarding the life cycle inventory are presented and the relation to inputs and outputs of modeled processes in LCA is illustrated.

Fig. 7.2
figure 2

Overview of process inputs and outputs (adapted from Klöpffer and Grahl 2009) and relation to challenges referring to ‘inventory aspects’ of LCA

Full size image

3.1.1 Water Use and Consumption

While water use is the total freshwater input into a product system, water consumption denotes the fraction of water use which is not returned to the originating river basin, mainly due to evapo(transpi)ration or product integration (Bayart et al. 2010).

Some consequences of water use, e.g. eutrophication or human- and ecotoxicity , are sufficiently covered in LCIA by respective impact categories (e.g. Guinée et al. 2002). Additionally, several inventory and impact assessment methods were developed, describing various cause-effect chains on human health , ecosystems, and resources (Berger and Finkbeiner 2010). Since water scarcity is a local phenomenon, impacts of water consumption need to be assessed on a local level, too. Therefore, all impact assessment methods require regionalized inventory flows. Additionally, some methods need information concerning types of water courses, water qualities, and time of consumption. Despite great progress in method development, hardly any of these methods have been applied in practice as inventory requirements are hard to satisfy—especially if complex background systems are involved. So the greatest challenge regarding water use and consumption assessment in LCA is the lack of detailed inventory information and application of existing impact assessment methods. Considering consequences of water consumption is especially relevant for LCAs of agricultural products, like food, natural fibers, or biofuels, as the agricultural sector is responsible for 85 % of global water consumption (Shiklomanov 2003). As energy production is the largest water consumer in industry (Pfister et al. 2011), water consumption should be assessed in energy intense industrial product systems, too.

In order to overcome the gap of lacking regionalized water inventories, Berger et al. (2012) developed a method to regionalize aggregated water consumption figures of LCI databases. In a top-down-approach, the total water consumed in the production of three passenger cars was allocated to manufacturing processes and material groups. Then the material-specific water consumption was assigned to countries based on import mix shares, locations of production sites and suppliers, etc. In terms of water quality, Boulay et al. (2011) developed inventory categories allowing for user-friendly consideration of water quality aspects.

In the short term, it is recommended to collect or estimate local water inventory data as detailed as possible and to apply existing impact assessment methods . In the long term, regionalized water inventory flows and impact assessment methods for water consumption have to be implemented into LCI databases and software tools.

In this context it should be noted that the challenge of lacking regionalized inventory data is also relevant for other impact categories like land use (Sect. 3.2.2.5) or biodiversity (Sect. 3.2.2.2). In these cases similar recommendations on the collection of inventory data and development of databases apply .

3.1.2 Renewable Energy

Renewable energy comes from resources which are continuously replenished, such as biomass, wind, water, earth (geothermal energy), and sun.

Several LCA studies are available for different types of renewable energy systems (Pehnt 2006; Dovì et al. 2009; Varun et al. 2009; Ardente et al. 2005; Burkhardt et al. 2012; Hsu et al. 2012; Ardente et al. 2008; Schleisner 2000; Arvesen and Hertwich 2012; Yee et al. 2009; Cherubini et al. 2009; Dodić et al. 2010). However, if renewable energy is used as an input in other product systems, it is still unclear how it can be modeled in a robust and consistent way. The two main issues are the potential double-counting of renewable energy by modeling specific uses of renewable energy that are considered at the same time in the grid mix and the question whether the production mix or consumption mix of a grid is modeled. Modeling renewable energy is particularly relevant whenever ‘credits’ are given in LCAs or product carbon footprints (PCFs) for the use of renewable energies, as unjustified credits may lead to wrong conclusions and recommendations. For instance, a company that produces solar power at its factory and feeds it into the grid to get the renewable energy subsidies cannot get an additional low carbon electricity credit, because the renewable energy benefit is socialized and part of the grid. The situation in Norway can serve as an example for the consumption/production mix issue, because its electricity production mix contains a high amount of renewable electricity, but the consumption mix often contains relevant shares of imported energy from fossil sources due to high export sales of renewable energy (Ekvall 2002; Curran et al. 2001). This leads to an underestimation of e.g. the global warming potential for goods produced in Norway when the Norwegian electricity production mix is used.

The existing accounting standards and frameworks like ISO 14044 (2006), ISO/TS 14067 (2013) or GHG Protocol (GHG 2011) just address that double counting has to be avoided, but no practical guidance is given on how a consistent yet practical modeling approach could work. Some authors proposed criteria for allowing credits for renewable energy, but these criteria either do not fully solve the double-counting issue or are too restrictive for getting ‘credits’ for the use of renewable energy at all (Grießhammer and Hochfeld 2009).

To deal with renewable energy in LCA modeling in a robust way, a consistent shift in the relevant background databases from electricity production mixes to the electricity consumption mixes including the trade of electricity is needed. In addition, more specific accounting guidelines for modeling renewable energy are required, which avoids double counting without being too restrictive in application.

3.1.3 Biogenic Carbon

Biogenic carbon refers to carbon dioxide stored in or released from biomass. The substitution of fossil carbon with biogenic carbon is one potential solution to abate climate change. The substitution can occur in form of biofuels, renewable fibers in plastics or bio-based chemicals and polymers. In addition, biogenic carbon naturally occurs in wood products or in the pulp and paper sectors (Finkbeiner et al. 2012).

Currently, there is no common practice of accounting biogenic carbon within LCA. Two distinct ways exist how CO2 flows associated with biogenic carbon are modeled. A simplified approach excludes all biogenic CO2 flows from the calculation by assuming that emissions arising directly from biogenic carbon sources are ‘carbon-neutral’. The alternative approach specifically models and accounts for biogenic carbon removal from the atmosphere during plant growth of, e.g., a tree and the future CO2 release of the wood product when disposed (Rabl et al. 2007; van der Voet et al. 2010). Another challenge arises from the modeling of recycled biogenic carbon. The existing LCA literature so far lacks a specific approach to include the recycling of biogenic carbon stored in products, especially with regard to the distribution of greenhouse gas (GHG) removals from the atmosphere. Due to the lack of a consistent approach for the assessment of biogenic carbon, inconsistencies and errors occur frequently within current case studies. This is particularly relevant for product systems containing renewable raw materials such as wood or cellulose-based polymers—especially when secondary materials or recycling are involved .

Several authors (Luo et al. 2009; Rabl et al. 2007) and even industry associations (e.g. Biotechnology Industry Organization) have already argued against the simple exclusion of biogenic carbon from the LCA models and have proposed to include them as individual flows at each stage of the inventory. Some of the reasons given include the better transparency, the fact that biogenic carbon can be transformed into flows other than CO2 (e.g. CH4) or the fact that biogenic carbon can, according to Luo et al. (2009), ‘escape’ co-product allocations (if the biogenic carbon is not modeled explicitly, it cannot be considered in allocation procedures).

To address the challenge to model biogenic carbon properly, it was proposed by Finkbeiner et al. (2012) to do an explicit accounting of inputs (GHG removals) and outputs (GHG releases) of biogenic carbon flows instead of assuming carbon neutrality per se. To deal with the recycling challenge, the same allocation principles have to apply consistently for both burdens (GHG releases) and benefits (GHG removals). That means if burdens are shared between life cycles, also benefits need to be shared between them (see Sect. 3.1.6 on allocation). However, even though a procedure was proposed for modeling biogenic carbon, the approach still allows for different solutions, and further enhancement and harmonization is needed .

3.1.4 Delayed Emissions

Delayed emissions are emissions that are released to the environment with a time delay. The issue of considering delayed emissions in LCA was brought up during the standardization and methodological discussions on carbon footprinting, one impact category within LCA. Therefore, delayed emissions are addressed here for the example of CO2, even though similar considerations apply for other substances as well.

However, so far, modeling of delayed emissions for products containing, for example, biogenic carbon is inconsistent within LCA. As a consequence, different approaches lead to different results, e.g. to different carbon footprints of products. Modeling of delayed emissions is especially relevant whenever there are differences in the timing of the emission release and particularly for products with long lifetimes. One approach to model delayed emissions is to provide a methodological incentive for delayed emissions so that (biogenic) carbon stored in products is supposed to get a better footprint if the product keeps the carbon longer in the technosphere and delays its release back to the atmosphere. The first version of the British carbon footprint specification PAS 2050 (2008) introduced such a method that uses a discounting approach for delayed carbon emissions. The approach proposed gave delayed emissions a lower weight than emissions that occur immediately, and emissions occurring after 100 years were not considered at all. The same discounting approach was proposed for products with a life cycle of more than 10 years. However, such an approach represents a methodological inconsistency with the ISO-standards of LCA, because no time cut-off for delayed emissions is considered there (ISO 14040 2006; ISO 14044 2006) . In addition, such an approach leads to many case study artifacts. For many materials like plastics, the most environmentally preferable end-of-life-treatment option would be a landfill, because there are hardly any emissions occurring within 100 years. Therefore, the more recent carbon footprint standards (ISO/TS 14067 2013; WRI/WBCSD 2004) do not allow a discounting of delayed emissions. They just offer the option to report them separately. Even the PAS 2050 stepped back during its recent revisions and does not allow subtracting delayed emissions from the total emissions anymore (PAS 2050 2011). However, in some studies delayed emissions are still considered due to a lack of awareness or a misinterpretation of the standards.

The solution to this inconsistency is to promote the use of the proper LCA standards as constitution of LCA (Finkbeiner et al. 2012) which do not allow discounting of delayed emissions.

3.1.5 Improbable Events

An improbable event is understood as an event which is unexpected and not likely to happen (OED 2013). It can lead to both positive and negative impacts. The potential impacts of such an event can be evaluated by means of risk assessment (referring to probability and magnitude of the event) (ISO 31000 2009).

Within LCA, currently only steady-state or standard operations are considered. Improbable events—deviations from standard operations or procedures (e.g. an additional extensive cleaning step in a process, a non-routine exchange of a catalyst or machine or an accident)—are not assessed. Hence, improbable events are not covered within the inventory of LCA studies, and their potential impacts are neglected within the impact assessment results . Improbable events (e.g. nuclear meltdown) can significantly influence elementary flows (quantitatively and/or qualitatively) and consequentially affect overall LCA results. For a comprehensive and realistic assessment of potential environmental impacts of e.g. technologies, improbable events need to be taken into account within LCA. Exclusion of the effects of improbable events could lead to wrong conclusions. The degree of impact caused by an improbable event depends on the severity of the event and the type and number of affected elementary flows.

Currently in LCA, risks are considered only within impact assessment (Tukker 2002) where generic risk assessment approaches are used to model human toxicity potential or ecotoxicity potential (Nishioka et al. 2002; Landsiedel and Saling 2002) . On the inventory level, the consideration of additional elementary flows caused by improbable events is missing. The resulting risks are not included in current LCA studies.

Improbable events need to be defined in the goal and scope phase and included in the LCI data to account for all potential environmental impacts associated with products and processes. This could be done by means of scenario analyses, e.g. by considering worst case scenarios. However, it may be difficult to quantify and model the full scope of an improbable event. Furthermore, a respective probability range regarding the deviation from the steady-state/standard procedure needs to be considered for the assessment of related impacts .

3.1.6 Allocation

Allocation is defined as “partitioning the input or output flows of a process or a product system between the product system under study and one or more other product systems” (ISO 14040 2006; ISO 14044 2006). An allocation problem arises due to multifunctional processes as the contribution of individual products/processes to the environmental burden is not obvious .

The handling of allocation is clearly a significant concern within LCA studies. Consistent allocation procedures need to be applied for the same multi-functional process (otherwise, unintended ignorance or unintended double-counting of environmental burdens can occur). This is particularly relevant if co-products from one production system are used in different sectors.

To deal with allocation problems, the international standard ISO 14044 (2006) provides a well-known hierarchy of steps. Whenever possible, allocation should be avoided by means of process subdivision or system expansion. If allocation cannot be avoided, physical relationships should be considered and, if not possible, other relationships may be used. Existing allocation procedures for co-products and recycling are presented in the following two subsections .

  1. 1.

    Allocation procedures for co-products

    Allocation in case of co-/by-products is challenging, as the selection of the allocation procedure (e.g. based on mass, calorific value, price, etc.) is not based on science but on value choices . Hence, there is no right or wrong—only a more or less appropriate solution for the specific case.

    As the choice for the allocation procedure often influences the result of an LCA study significantly, the missing scientific basis and consensus between stakeholders on how to handle allocation can be seen as a gap in LCA . This is especially relevant in LCAs of metals derived from ores in companion with other metals (e.g. copper-zinc-gold ores) and animal husbandry (e.g. cows producing milk, leather, meat, and manure) as the choice of the allocation method has a significant influence on the results.

    With process subdivision practitioners can reduce but hardly eliminate all allocation issues as processes may not consist of physically separable sub-processes (very frequent in chemical reactions leading to several substances) (Reap et al. 2008a; Ekvall and Finnveden 2001). Furthermore, system boundary expansion can be applied: Either the functional unit is expanded (e.g. not only milk, but leather, meat, and manure are considered with the milk) or environmental effects of similar products systems are subtracted from the multi-output system (e.g. environmental effects of gravel production are subtracted from a blast furnace producing steel and slag as a by-product, assuming that slag substitutes gravel in road construction). However, in the first case practitioners may create another allocation problem by including additional processes (Ekvall and Finnveden 2009). In the second case, the risk is that credit is given for a product, which does not reflect the real substitution situation. Both, process subdivision and system boundary expansion, can lead to more data, time, and cost demands, and cause uncertainty (Reap et al. 2008b).

    As allocation and associated challenges can neither be avoided nor solved completely, the best way of dealing with it is to apply different allocation solutions and to analyze the results in a sensitivity analysis as suggested respectively required for comparative studies in ISO 14044 (2006) .

  2. 2.

    Allocation procedures for recycling

    If recycling processes are considered, burdens from primary material production, recycling processes and disposal processes have to be allocated between the life cycles using the material .

    The existing requirements in the international standards of LCA , ISO 14040 (2006) and ISO 14044 (2006), are of generic nature . Consequently, no generally accepted approach exists on how to deal with secondary materials recovered from recycling processes. The assessment of recycling processes can have a decisive influence on the overall results of LCA studies and, thus, is relevant for all product systems using or generating secondary materials.

    ISO states that “reuse and recycling may change the inherent properties of materials” (e.g. down-cycling in case of plastics recycling (Kuswanti et al. 2003)), and that changes in the inherent properties have to be considered (ISO 14040 2006; ISO 14044 2006). The lack of a clear definition of the term “inherent properties” leads to inconsistencies. Depending on whether inherent properties of materials are changed or not, a distinction can be made between closed- and open-loop recycling. Closed-loop describes the return of material to the same product system (real closed loop) or the return to a different system without changes in the inherent properties of the material (quasi closed loop). Open-loop means, that the material is recycled into a different product system and inherent properties are changed (Ekvall and Tillmann 1997; Klöpffer 1996). In case of closed-loop recycling the allocation of the environmental burden of the primary material production can be avoided, e.g. via system expansion, as the use of secondary material displaces the use of virgin (primary) materials.

    The application of existing allocation methods in open-loop recycling systems depends on value judgments, often reflecting the various interests of different stakeholder groups. Thus, the handling of credits and burdens in LCA can lead to under- or overestimation of environmental impacts associated with single life cycles (Reap et al. 2008a).

    Several methods of how to account for recycling were developed. Commonly applied are the two extreme approaches of first and last responsibility, also known as recycled content (secondary material source does not carry any burden, because no credit is given for the primary production) and avoided burden approach (secondary material source carries a burden, because full credit for the primary production burden is given) (Frischknecht 2010; Klöpffer 1996). In addition, several approaches exist in between, e.g. quality-based methods (Azapagic and Clift 1999), economic approaches (Werner and Richter 2000; Guinée et al. 2004) and others (Kim et al. 1997; Wötzel 2007), but no agreement regarding a preferred method has been established so far.

    A possible solution seems hard to find, as no allocation method will be applicable in every case (EPA 1993) and different approaches lead to different incentives. Therefore, Curran (2007) recommends to focus on a macroeconomic point of view instead of staying fixed to single processes: to avoid green washing, credits shall only be granted if recycling of the secondary material actually takes place in reality. Neugebauer and Finkbeiner (2012) developed a so called Multi-Recycling-Approach, focusing on the pool of materials, by building up an environmental profile for a certain material equally including the primary and secondary production route. Additionally an evaluation scheme is proposed to assess conservation respectively changes in the inherent material properties (Neugebauer and Finkbeiner 2012) .

    For consistent modeling of allocation in case of recycling, a consensus for material/product group specific allocation procedures is needed. Until this is achieved, sensitivity analysis should be performed to reflect the influence of the chosen allocation procedure on the results .

3.1.7 Functional Unit

The functional unit is defined as “quantified performance of a product system for use as a reference unit” with the purpose “to provide a reference to which the inputs and outputs are related [and] to ensure comparability of LCA results” (ISO 14040 2006; ISO 14044 2006). According to the International Reference Life Cycle Data system (ILCD) Handbook (EC-JRC 2010b), the functional unit should be defined along the question: ‘what’, ‘how much’, ‘how well’, and ‘for how long’ to support valid comparisons between products.

Due to changing consumption patterns, complex economic systems, and products with multiple functions, the selection of a functional unit is a challenge since different functional units lead to different results for the same product system. This is relevant as a restriction to a strict, functional equivalent may not reflect the reality very well (Hischier and Reichart 2003; Reap et al. 2003; Reap et al. 2008a; Cooper 2003). Realistic modeling is also challenged by issues like lifetime, performance, system dependency, and handling of non-quantifiable or difficult-to-quantify functions (Cooper 2003), e.g. aesthetics of a product. Such information might get lost when defining the functional unit. Moreover, a reasonable straightforward relation of impacts along the life cycle to a functional unit can be questionable, especially when functions are difficult to quantify and effects are included that may be space, time and threshold dependent (ISO 14044 2006; ANEC 2012) .

As parts of environmental aspects may not be treated properly when comparing multifunctional products based on one functional unit, Hischier and Reichart (2003) recommend to apply several approaches to consider the multifunctionality—and thus, to better reflect the reality like e.g. ISO/TR 14049 (2012). With its concept of user acceptance (Lagerstedt et al. 2003), ISO/TR 14049 (2012) provides a possibility to compare multifunctional products which are still considered equivalent by the users. Cooper (2003) suggests specifying the functional unit for comparative analyses, e.g. by differentiating between the functions of systems and subsystems and using different functional units when needed.

The challenges described above refer to the relative approach of LCA as such as the reference to a functional unit is the basis of any LCA study. As a consequence, no clear solution exists for this inherent challenge .

3.2 Impact Assessment Aspects

This section analyzes gaps and challenges regarding impact assessment aspects. Following the established structure of areas of protection, the impact assessment aspects are further differentiated into ‘human health’ (Sect. 3.2.1), ‘ecosystem’ (Sect. 3.2.2), and ‘resources’ (Sect. 3.2.3).

3.2.1 Human Health

This section analyzes gaps and challenges regarding impact assessment methods referring to the area of protection ‘human health’. It focuses on the impact category ‘human toxicity’ and related challenges like ‘completeness and cumulative effects’ or inclusion of ‘direct health effects ’, ‘particulate matter’, ‘nanomaterials’, ‘endocrine disruptors’ and ‘microbiological pollution’. Moreover, challenges related to the impact categories ‘noise’ and ‘odor’ are discussed .

Since the concept of SLCA emerged, there is some debate whether human health issues should be treated as a social or as an environmental aspect. However, human health has traditionally been assessed within LCA and many studies and models are available. Thus, gaps and challenges related to human health aspects are included in this analysis of (environmental) LCA.

A simplified overview of potential impacts on ‘human health’ which are currently insufficiently covered in LCA is shown in Fig. 7.3.

Fig. 7.3
figure 3

Schematic overview of potential impacts on human health caused by toxic substances, odor and noise

Full size image
3.2.1.1 Human Toxicity

Health effects of toxic substances are traditionally covered in LCA and either addressed at midpoint level or aggregated at endpoint level (damage to human health). Currently, the consensus model is USEtox (‘USE’ in the acronym stands for the UNEP/SETAC life cycle initiative) (Henderson et al. 2011; Rosenbaum et al. 2008; Rosenbaum et al. 2011) providing guidance for modeling health effects in LCA based on human exposure and toxicity. However, the discussion of human toxicity assessment is far from being solved. General challenges encompass the absence of regionalized and inventory dependent characterization factors and lacking consistency in fate, exposure and effect evaluation. Beyond that, available models are not complete regarding the chemicals which are potentially relevant for human toxicity and neglect cumulative effects of chemicals . This leads to data asymmetry in comparative LCAs. Further challenges related to human toxicity encompass the modeling of ‘direct health effects’, ‘particulate matter’, ‘nanomaterials’, ‘endocrine disruptors’ , and ‘microbiological pollution’ .

These challenges are addressed in the following sub-sections. As they can be referred to ecosystem as well, they are briefly taken up again in section ‘ecotoxicity’ (see Sect. 3.2.2.1) .

Completeness and cumulative effects. Multitudinous chemicals can cause toxic effects on humans. However, characterization models currently used in LCIA cover only a small fraction of the potentially toxic chemicals. In addition, toxicity is analyzed for individual substances only, and cumulative toxic effects due to exposure to a combination of substances, potentially increasing the toxic effect, cannot be assessed so far. As a consequence, potential toxicity impacts are underrepresented in LCAs of products systems from, e.g., the pharmaceutical sector. Concerning the incompleteness of toxic chemicals covered within LCA, the number of characterization factors has already been increasing steadily from a few hundred in Guinée et al. ( 2002) to more than 3,000 in the current USEtox model (Rosenbaum et al. 2008). Within the on-going research project ‘LC-IMPACT’ further methods and characterization factors for ecotoxicity and human toxicity are developed (Rosenbaum et al. 2012) . The consensus distribution model (multimedia fate model) has been developed for organic chemicals, not for salts, surfactants, acids and bases. For a recent survey of models see Klöpffer ( 2012).

Despite this progress, even more characterization factors for toxic chemicals should be determined and potential interrelations should be assessed to achieve a more complete coverage (especially regarding chemicals with potentially high toxicity and/or relatively high emission levels). Furthermore, current characterization models need to assess ‘unconventional’ toxic effects caused by substances like nanomaterials or endocrine disruptors (see respective sections) . The incompleteness of existing LCIAs is a problem of lacking characterization factors and undefined fate models/impact pathways. Additionally, the assessment of cumulative toxic effects poses a substantial challenge on the inventory level. To assess cumulative effects, inventories need to give insight in which of the chemicals are likely to be emitted together, and inventory dependent characterization factors should be developed for the cumulative effects of these chemicals.

Direct health effects . Direct health effects can occur if humans are exposed directly to toxic substances via the respiratory system, gastrointestinal tract, mucosa or skin. Contrary to existing characterization models for human toxicity that comprise default environmental fate models, direct health effects should be accounted for without prior dilution or decay .

Currently, direct health effects are not considered within LCA. It should be noted that LCA was not developed and is not intended to study such effects in detail. This is the domain of tools like risk assessment. However, neglecting direct impact pathways completely can lead to an underestimation of potential toxic effects of, e.g., paint or toys emitting toxic chemicals (Becker et al. 2010b; Guney and Zagury 2011). The inclusion of some impact pathways that consider direct health effects improves the LCIA—acknowledging that these remain potential impacts below the level of risk assessment results .

Several studies evaluate direct impacts of products on humans. These include possible direct health effects of food intake (Juraske et al. 2009), pacifiers and shampoo (Henderson et al. 2012), flame retardants in impregnated textiles and the assessment of indoor emission of building materials for dwellings (Meijer 2007) or chairs (Skaar and Jørgensen 2013). A new version of USEtox has been announced as an outcome of the LC-IMPACT project, enabling the assessment of indoor emissions by considering average room volumes of houses in an indoor fate model (Ernstoff 2012). This latest developments show a promising way to address (at least some) direct health effects in LCA .

The inclusion of direct health effects leads to additional requirements on the resolution of the inventory, as direct exposure has to be separated from releases to the environment. While effect factors can be used from existing characterization models, the fate models and intake fractions might have to be adjusted.

Particulate matter . Particulate matter (PM) refers to fine particles below 10 µm of particle size, forming aerosols. PM can be either man-made (e.g. from combustion in vehicles and plants) or of natural origin and can have a strong impact on human health (Brunekreef and Holgate 2002). Particles smaller than 10 µm (PM 10) can enter into the bronchi and lungs, particles smaller than 2.5 µm (PM 2.5 ) tend to penetrate into the gas exchange regions of the lung (Delfino et al. 2005 ), and so called ultrafine particles (below 0.1 µm) may even pass through the lungs and cause health effects in other organs (Sioutas et al. 2005).

Even though different LCIA models addressing PM are available (Greco et al. 2007; Spadaro 2004; van Zelm et al. 2008; Humbert et al. 2011), consistency in fate, exposure and effect evaluation (Potting et al. 2007) are lacking and a more comprehensive approach is needed. In most models a linear, no-threshold dose-response relationship is assumed, although Pope et al. (2009) demonstrated a log-linear relationship. Moreover, toxic effects of ultrafine particles and chemicals attached to particulate matter are currently neglected. LCIA models cannot replace detailed risk assessment analysis for these issues, but the simplified: the inclusion of the potential PM impacts in LCA is still relevant, e.g. in the power and heat sector. New combustion technologies are implemented to reduce the carbon emissions but could also result in different PM emissions (Koornneef et al. 2010) and/or changing particle-size distribution. Thus, trade-offs may occur between the intended environmental benefits of the technology and the potential human health damages due to (changed) PM emissions.

Up to now, only a few of the challenges of existing models are tackled, e.g. some research groups try to set up internally consistent intake fraction values (Humbert et al. 2011).

Additional epidemiological studies of PM2.5 and below will help to make the assessment more robust. Chemicals attached to particles have to be included in the inventory, and their toxic effects need to be analyzed in more detail. Finally, case studies are needed to examine data availability to test existing methods and to inspire new methodological developments for a more comprehensive assessment of PM within LCA.

Nanomaterials . Nanomaterials are manufactured/engineered materials with at least one dimension below 100 nm (Gavankar et al. 2012 ; EC 2012c ). Nanomaterials provide increased strength, chemical reactivity or conductivity and are used in many different sectors with growing production volumes in recent years. Nanomaterials exhibit unique behavior depending not only on chemical composition, structure and shape, but also on interaction with organisms and other pollutants in different environmental media (Gavankar et al. 2012; EC 2012c). According to Birnbaum and Jung (2011) they may have relevant effects on human health, as unintended exposure can result in their presence within the body, with unknown biological consequences, e.g. contribution to lung cancer (Becker et al. 2010a) and other effects on the lungs, brain and blood circulation (EU 2009).

Currently, potential impacts of nanomaterials are not included in LCA case studies, as there are gaps on both inventory and impact assessment level. Challenges on the inventory side encompass, for example, missing data of the specific production processes of nanomaterials and the emissions of the nanomaterials itself (Som et al. 2010; Seager and Linkov 2008). Regarding the impact assessment, different cause-effect chains have to be analyzed for different nanomaterials. However, toxicological characterization models for nanomaterials do not yet exist, and as a consequence no characterization factors are available (Gavankar et al. 2012; Som et al. 2010). The consideration of the potential effects of nanomaterials is relevant and could have a significant influence on the results, in particular with regard to LCAs for sectors using a wide range of nanomaterials, e.g. cosmetics, electronics, textiles, etc. (EC 2012c).

Only a few LCA studies modeled the release of certain nanomaterials (e.g TiN, TiAlN, Ti + TiAlN, carbon nanotubes and others) (Gottschalk et al. 2009; Gavankar et al. 2012; Müller and Nowack 2008), but they remain on the inventory level and are not representative for the multitude of different nanomaterials.

Results from the OECD working group on nanomaterials, e.g. lists of manufactured nanomaterials, exposure measurements or information on safety evaluation and risk assessment (OECD 2013), could be used as a starting point both on the inventory and impact level. The effects of nanomaterials on human health could be included within the existing impact categories, such as human toxicity .

Endocrine disruptors . Endocrine disruptors are hormonally active substances released into the environment that can interact or interfere with hormonal activity (EFSA 2013; EC 2012b). They stimulate or inhibit the endocrine system. Human health can be affected by reproductive development/system disorders, metabolic issues, cardiovascular diseases and gland cancers (US EPA 2012 ; EEA 2012) .

So far, LCA does not fully account for impacts of endocrine disruptors. The information of endocrine disruptors is still limited, and the corresponding causes and effects are complex. Currently, only about 70 substances are listed as main endocrine disruptors and studied internationally (US EPA 2012; EEA 2012). The resulting impacts cannot be taken quantitatively into account in LCIA, as there is currently no epidemiological framework available that covers bioaccumulation, multiple causality, latency, and low doses (US Green Building Council 2008). The assessment of potential effects of the release of endocrine disruptors into the environment is, however, relevant in LCA studies of, for example, pharmacy products, plastics, consumer products or pesticides (Frischknecht et al. 2009), as endocrine disruptors pose a significant concern for human health (Diamanti-Kandarakis et al. 2009).

Some methods in LCA propose an evaluation of the endocrine disruptors in surface or saltwater; for example, the Ecological Scarcity Method. This method uses the estrogenic potential (kg E2-eq/kg) for calculating eco-factors based on yeast estrogenic screening, an accepted method in this research field (Frischknecht et al. 2009). However, since different chemicals can vary widely in their persistence and potency, mass totals are a very crude indication of comparative risk between products.

Due to the novelty of the gap, more inventory and characterization data for endocrine disruptors are required as a starting point for developing a comprehensive solution.

Microbiological pollution . Microbiological pollution refers mainly to the pollution of freshwater by pathogenic microorganisms . Infectious microorganisms can be grouped into bacteria, viruses, protozoa, and helminthes (NRMMC 2006) and cause mainly gastro-related diseases.

In LCA, there are currently neither elementary flows defined on inventory level, nor impact categories available on impact assessment level. Microbiological pollution is particularly relevant for systems characterized by wet conditions, temperature changes, and availability of organic matter. Hence, it should be considered, for example, in LCAs of waste water treatment plants, water dispensers or biogas plants.

To address microbiological pollution, Larsen et al. (2009) proposed a first framework to assess emission of pathogens into recreational water bodies and unintended swallowing of water during bathing measured in disability adjusted life years (DALYs). This framework follows the structure of risk assessment including hazard identification, dose-response analysis, exposure assessment, and risk characterization. With regard to LCA, Larsen et al. (2009) noted that models for dose response presented in NRMMC (2006) may be usable for a simplified assessment in LCA.

In order to address impacts of microbiological pollution on human health in LCA, adequate elementary flows have to be defined providing information on the number and kind of pathogens emitted. Furthermore, additional impact pathways have to be identified, and characterization models have to be developed to assess impacts resulting from microbiological pollution. Such models should include the fate of pathogenic microorganisms comprising distribution through environmental media and survival rates of pathogens. After that, the uptake of microorganisms into the human body (via different uptake routes) has to be considered, and health effects should be predicted by means of clinical dose-response relations .

3.2.1.2 Noise

Noise can be regarded as ‘unwanted sound’ which possibly affects human health, as it may, for example, impair cognitive abilities (Clark et al. 2006), cause sleep disturbance (Griefahn et al. 2006), increase the risk for heart diseases (van Kempen et al. 2002) or lead to hearing impairments.

Although methodologies to assess noise are available and an LCA midpoint category was already suggested by Heijungs et al. (1992) and Guinée et al. (2002), it is so far hardly addressed within LCA case studies. The main reason is the lack of inventory data . Furthermore, the relation of noise of a production process to the functional unit is not straightforward. Neglecting this impact category may lead to incomplete and falsified results in LCAs of transport systems, for example, as potentially significant impacts on human health are disregarded.

Franco et al. (2010) proposed a framework to assess impacts of traffic noise based on the method developed by Müller-Wenk (2002, 2004) where effects of traffic noise are considered, determining the number of annoyed persons attributed to kilometer per vehicles. In addition, the Centre of Environmental Science (CML Leiden, The Netherlands) proposed a method aggregating physical sound levels (Heijungs et al. 1992; Guinée et al. 2002). Further methodologies for assessing noise are available (Althaus et al. 2009; Meijer et al. 2006; Reap et al. 2008b; Cucurachi et al. 2012), but a consistent framework and comprehensive inventory data are still needed.

To include noise in LCA it is recommended in order to conduct more case studies applying, testing and improving available methods.

3.2.1.3 Odor

Odor can be regarded as ‘unwanted and unpleasant smell’, caused by volatile chemical compounds. Many effects of odor are covered within other impact categories (e.g. volatile organic compounds—VOC emissions in human toxicity). Furthermore, odor can affect quality of life and human well-being which could be considered outside the scope of LCA. However, as odor can also have direct effects on human health, e.g. induce headache, sleeplessness, loss of appetite, sickness, nervousness, cough or asthma (Blaisdell 2007), it should be considered within the scope of LCA.

So far, odor is rarely analyzed in LCA due to the lack of both inventory data and robust impact assessment methods . Addressing odor in LCA can be relevant especially with regard to e.g. wastewater treatment systems or biogas technologies, as the impacts of odor on humans can be significant and alter the results of LCA studies.

Several approaches have been suggested for the inclusion of odor into LCA. Heijungs et al. (1992) and Jolliet et al. (2004) developed midpoint indicators. Guinée et al. (2002) proposed to inverse the odor threshold values (OTV) for characterizing odor, so that the smell creation potential (SCP) can be described by dividing the emissions of a substance by the OTV value. Marchand et al. (2012) proposed a site dependent approach for odor assessment in waste management, based on analyzing concentrations, fate using UseTOX (Rosenbaum et al. 2008),—and exposure.

Within the existing impact category and proposals, odor concentration is assessed by determining VOC emissions (kg/m3), but with regard to a complete characterization method additional factors like deposition, evaporation, chemical conversion and meteorological conditions have to be considered. For comprehensive assessment within LCA, inventory data need to be collected and case studies should be performed to develop an applicable impact assessment methodology .

3.2.2 Ecosystem

This section analyzes gaps and challenges regarding impact assessment methods referring to the area of protection ‘ecosystem’. It focuses on the impact categories ‘ecotoxicity’, ‘biodiversity’, ‘biological invasion’, ‘direct non-intended killing of animals’ as well as ‘land use and land-use change’ . The impact category ‘noise’, which was already discussed in Sect. 3.2.1 with regard to human health, is briefly taken up within the section ‘ecotoxicity’ with regard to its potential impacts on animals.

3.2.2.1 Ecotoxicity

In the previous section, methodological gaps concerning the assessment of effects caused by toxic substances on human health have been comprehensively discussed. Obviously many of those challenges and proposed solutions apply for the analysis of ecotoxicity, too. In order to avoid repetition, this section revisits these gaps from an ecotoxicity perspective and considers differences and similarities in the analysis of gaps identified for human toxicity.

Despite the fact that for the assessment of ecotoxicity different fate and effect models have to be used than for human toxicity, challenges regarding ‘completeness and cumulative effects of chemicals’, ‘nanomaterials’, and ‘endocrine disruptors’ are similar (see Sect. 3.2.1) . The lack of detailed toxicological knowledge is the main reason for incomplete or missing characterization factors. With regard to endocrine disruptors, derivation of characterization factors for ecotoxic impacts might be easier, as there are laboratory studies available analyzing effects on animals directly while effects on humans can only be observed indirectly (Matthiessen 2000). On the other hand, recent studies on animals showed that the dose-response relation is not always linear and that endocrine disrupting chemicals may have effects at low doses but no effects at high doses (Vandenberg et al. 2012). This is a significant challenge for the development of characterization models.

Particulate matter might affect ecosystem as well as it can harm plant life and thus animals which are dependent on plants as feed. Moreover, animals can also be affected by breathing in the particulates, similar to humans (HCES 2013). However, as emission of fine dust is mainly relevant in urban areas (except forest fires), ecotoxic effects of particulate matter can be regarded as less important.

Microbiological pollution can be relevant for ecosystems , too, as e.g. animals might be affected by pathogenic microorganisms. Considering that only some species are sensitive to pathogenic pollution, potential characterization models for ecotoxicity would have to focus on target species .

Toxicity impacts caused by direct health effects , e.g. indoor emissions, are relevant for human health but generally not for ecosystems as exposure to indoor pollutants or products emitting toxic chemicals directly into an animal’s body, which is part of the ecosystems, are rather unlikely.

Generally, odor could also have impacts on ecosystems. As no scientific proof could be found whether animals are affected by odors or not, more basic research is needed in this area before discussing potential impact assessment approaches.

Noise can affect animals and cause auditory damage due to high noise levels or physiological changes, e.g. increased heart rate, problems with respiration, and behavioral changes, e.g. changes in migration patterns (Cornman 2003). Currently, that is not addressed in LCA. Challenges and possible first steps to integrate noise effects on animals in LCA can be considered similar to those discussed in Sect. 7.3.2.1.2.

3.2.2.2 Biodiversity

Biodiversity is the “variability among living organisms from all sources including, inter alia, terrestrial, marine and other aquatic ecosystems and the ecological complexes of which they are parts: this includes diversity within species, between species and of the ecosystems” (UN 1992). Biodiversity is important for the stability of habitats and ecosystems and highly sensitive to the alteration of all environmental aspects, e.g. quantitative and qualitative changes in air, water and soil (Gontier et al. 2006).

The already existing concepts to include biodiversity in LCA as midpoint or as endpoint category have been regarded as too limited and vague to be applied, and not equally suitable and valid to all kinds of ecosystems (differences between terrestrial, freshwater and marine needs to be considered) (Noss 1990). Thus, an agreed overall method is still missing within LCA. Over the last decades a strong decline in biological diversity occurred (Curran et al. 2011). Human activities can affect biodiversity directly e.g. by agricultural production leading to land use (Michelsen et al. 2012), transport processes (e.g. ships) causing biological invasion (Schenck 2001), plastic packaging risking littering (Wong et al. 1974) . They can also affect biodiversity indirectly, e.g. by pharmaceutical products causing the release of nanomaterials (Som et al. 2010) or endocrine disruptors (Frischknecht et al. 2009) . Therefore, the respective impacts need to be considered within LCA. Consideration of effects on biodiversity is relevant e.g. for LCA studies on agricultural and food products, as direct nature occupation occurs in this connection (Knudsen and Halberg 2007).

In LCA, biodiversity besides ecosystem functions is often used to assess ecosystem quality (Henzen 2008) and discussed both within midpoint approaches (Guinée et al. 2002) and endpoint approaches to LCIA (Curran et al. 2001). Curran et al. (2011) and Watson et al. (2005) discuss endpoints of biodiversity in LCA and name five drivers for biodiversity loss. Three of them with regard to habitat change, climate change and pollution are represented in current impacts categories (e.g. land use, eutrophication, and acidification). The other two drivers for biodiversity loss, invasive species and overexploitation of ecosystems , especially overfishing/by-catch (Sects. 3.2.2.4 and 3.2.3.2), are so far hardly represented in any impact category (Curran et al. 2011; Watson et al. 2005). Several indicators (like species richness, variety or number of species, species vulnerability) and approaches for the assessment of biodiversity were proposed (Milà i Canals et al. 2007a; Schenck 2001; Souza et al. 2013; UNEP 2010b), but none of them provides a comprehensive solution to include biodiversity in LCA. Researchers currently analyze the multiple aspects of biodiversity loss on a mostly local level with the help of ecosystem services (UNEP 2010b).

To address this gap, a deeper understanding of the biological, geophysical, and geochemical processes and intensive research is needed (Balmford et al. 2005). Furthermore, already existing indicators, inventory data (Jeffery et al. 2010) and approaches should be revised and extended to achieve regionalized data capturing a representative sample of the diverse terrestrial, freshwater and marine habitats (Curran et al. 2011) . Exemplarily, one relevant indicator to describe biodiversity, namely biological invasion, is further discussed in Sect. 3.2.2.3. Another approach is to include biodiversity in existing impact categories within LCA, e.g. combining biodiversity loss with land use (e.g. by including data from the Geographic Information System—GIS in LCA) or global warming (Geyer et al. 2010; Knudsen and Halberg 2007) .

3.2.2.3 Biological Invasion

Biological invasion, or invasion of foreign species into ecosystems, can cause damages in local ecosystems if indigenous species are displaced by invaders.

So far, biological invasion is not considered within LCA. Biological invasion is relevant for processes like transports, especially for shipping: First, channels used for shipping connect ecosystems and allow for a spreading of species. Second, ballast water used to stabilize container ships during empty running is a potential source for biological invasion as alien species, such as jellyfish or mussels, can be transported over long distances and are then released into non-indigenous ecosystems.

In an LCA context, biological invasion is proposed as an indicator describing impact pathways on biodiversity, by means of, e.g., “percent coverage of invasive species within protected areas” (Schenck 2001). However, this can be regarded as an inventory indicator, and a deeper analysis of impacts is missing. Recently, Narščius et al. (2012) published a method which allows for the assessment of biological invasion impacts. The model determines a biological pollution level by relating a classified abundance and spreading range of invasive species to the magnitude of their consequences on communities, habitats and ecosystem functioning.

To include biological invasion in LCA, further research is needed both on an inventory and on an impact assessment level. Regarding the latter, cause-effect mechanisms described in the method proposed by Narščius et al. (2012) might be transferred into a characterization model for LCIA.

3.2.2.4 Direct Non-Intended Killing of Animals

Direct non-intended killing of animals includes the non-intended mechanical, electrical or thermal causes of direct death of animals by human activities, e.g. impacts due to road traffic, wind or water power plants, oil spills (and other man-caused accidents) or industrial fishing (by-catch). Some of these impacts are related to accidents, but the shredding of fish in water power plants or the by-catch in industrial fishing are not really accidental, they are a non-intended part of standard operation. The term ‘direct’ is used in this context to differentiate the resulting impacts from the indirect effects on animal health due to other impact categories like ecotoxicity or water use.

The direct non-intended killing of animals is not addressed within LCIA so far, as existing impact categories like ecotoxicity, acidification and eutrophication (Cleuvers 2003; Goedkoop et al. 2009; Jolliet et al. 2003; Rosenbaum et al. 2008; EPA 2012; Jolliet et al. 2004; EC-JRC 2010a) address potentially lethal impacts on species only indirectly, e.g. in the endpoint models for eutrophication or ecotoxicity. The inclusion of the direct non-intended killing of animals and possible consequences like injury, mortality and even declines in species seem particularly relevant for the following examples: fish mortality due to hydropower plants, mortality of bats due to collision with offshore oil and gas platforms and mortality and injuries of birds caused by electrocutions from power lines (Banks 1979; BPA 2013; NYSERDA 2009; Watts 2010) or by collision with turbine rotors of wind farms (Cole 2011; Ferrer et al. 2012). In addition, direct non-intended killing of animals can also be caused by littering , by-catch (e.g. fish, dolphins), oil spills (and other man-caused accidents) (e.g. affecting birds), and human constructions like swimming pools and building pits (e.g. affecting hedgehogs) (Ambrose et al. 2005; Burger 1993; Erickson et al. 2005; Fleet et al. 2009; Garrity and Levings 1993; Pro Igel 2012).

The direct non-intended killing of animals is already addressed by risk assessment and environmental impact assessment (EIA). Within risk assessment models for example, killing of birds during the operation phase of wind power plants has been displayed to predict, assess and possibly reduce bird mortality (Ferrer et al. 2012). Furthermore, EIA studies exist, which evaluate the killing of birds and bats in connection with the construction of wind farms (Arnett et al. 2007; Smallwood et al. 2007). Although the direct mortality rates vary widely, depending, for example, on the location, the risk for birds and bats can be seen as proven (Cole 2011; Ferrer et al. 2012; Kuvlesky et al. 2007; Smallwood et al. 2007). Besides that, mortality caused by other human constructions is hardly addressed within case studies so far.

To include direct non-intended killing of animals in LCA models, inventory data and indicators need to be developed. As a possible starting point the existing risk and EIA studies can be used. The developed EIA indicators, like total discounted bird years (Cole 2011), can be used, improved and translated into LCA, e.g. inclusion of animals and/or species mortality by means of years of animal life lost (YOALL). Within the development process of indicators, also threshold values shall be included that consider endangered species or specific limitations to avoid population damages.

3.2.2.5 Land Use and Land Use Change

Since clear definitions of land use (LU) and land use change (LUC) cannot be found in literature, a clear distinction between both is not a straightforward task. For some authors, occupation and transformation are equally responsible for land use and land use change (EC-JRC 2010a; Saad et al. 2011); others consider land use to be related to occupation and land-use change to transformation (Mattila et al. 2011). Some authors consider only biodiversity when assessing land use (Vogtländer et al. 2004), others consider also changes in soil quality, (Milà i Canals et al. 2007b) changes in GHG emissions (Anderson-Teixeira et al. 2011) or the effects on fresh water (EC-JRC 2010a; Saad et al. 2011). Within this section, the two terms are discussed separately based on their development within the LCA community for the last few years. We use Milà i Canals (2007) definition for LU as damage to ecosystems due to the effects of land occupation (of a certain area during a certain time) and transformation (of a certain area) . On the other hand, LUC is a general term for the alteration of one land use category to another (Mattila et al. 2012). Assessing LU and LUC is especially important when considering the production of land-intensive products from agriculture, mining or infrastructure.

Land Use

The main consequence of land use is the loss of biodiversity (Millennium Ecosystem Assessment 2005). In addition, impacts on ecological functions like changes in soil quality and ecosystem services like biomass production occur (Koellner et al. 2013).

Currently, the impact category land use is hardly applied in LCA. One reason is the lack of a broadly agreed impact assessment method (Mattila et al. 2012). Most of the existing methods are only adding up square meters (in general or based on hemeroby classes), which do not adequately reflect losses of biodiversity or soil quality aspects (Milà i Canals 2007). Lack of sufficient inventory data is another problem as for most of the existing methods, spatial inventory data is needed. Even though background data on square meter level is partly existing, databases are far from being complete (Milà i Canals et al. 2007a; Vogtländer et al. 2004; Mattila et al. 2012).

The ILCD Handbook (EC-JRC 2011) recommends to assess land use with the midpoint indicator soil organic matter (Milà i Canals 2003), which can only reflect some changes in soil quality, but not biodiversity as such (Milà i Canals et al. 2007b). Assessing land use at endpoint level is suggested in Koellner and Scholz (2008) and Goedkoop et al. (2009).

Koellner et al. (2013) introduce guidelines to develop a land use impact assessment for biodiversity and ecosystem services and also recommend specific methods and characterization factors for some of these aspects.

Following Mattila et al. (2012), further steps could be the application of existing midpoint- and endpoint-methods to test their feasibility. As a crucial prerequisite for broader application, inventory data needs to be collected.

Land Use Change

LUC is often divided into direct LUC (dLUC) and indirect LUC (iLUC). dLUC occurs when new agricultural land is taken for production and feedstock purposes and therefore, displaces e.g. a forest (Sanchez et al. 2012). iLUC occurs when land currently used for feed or food crops is changed to the production of a different product and the demand for the previous land use remains (Piemonte and Gironi 2011).

So far, no globally accepted and used methodology to assess LUC exists in LCA (Mattila et al. 2012). A challenge in LUC when creating an LCA model is the precise definition of the system. It is tricky to assume whether the burdens will occur immediately or over multiple years of production. Almost every LCA study assumes LUC will occur immediately. However, it is expected that in reality LUC effects occur over many years (especially for iLUC) (Anderson-Teixeira et al. 2011). Another important challenge that occurs when dealing with land transformation is the estimation of indirect emissions. There is currently no consensus on how this is to be performed in LCA/carbon footprint (CF) studies (Finkbeiner 2013). Considering LUC in LCA is relevant as LUC is appearing to be one of the challenges in LCAs of renewable raw materials and biofuels, “often determining whether or not a biofuel meets GHG reduction thresholds” (Anderson-Teixeira et al. 2011).

Various methods have been proposed to include LUC in LCA and show “extremely different results” (Flysjo et al. 2012). Thus, the iLUC modeling process is not yet mature or sufficiently advanced to be used as policy-making tool (Plevin et al. 2010). Several other studies observe different options to lower the uncertainty when evaluating LUC emissions (Anderson-Teixeira et al. 2011), especially the indirect ones (Sanchez et al. 2012; Jannick H Schmidt 2012). There is also the fundamental issue why indirect market effects are supposed to be included in LUC at all, as no other indirect market effects are included in LCA. In that sense, iLUC is methodologically inconsistent with the basic principles of LCA. It might be consistent with the approach of consequential LCA, but fully depends on uncertain market predictions.

LCA community needs a deeper analysis and discussion, if and how to estimate LUC emissions (Flysjo et al. 2012). If further research achieves a better understanding of LUC, scientifically robust and consistent LUC quantification, factors might be reconsidered as a potential element to be included in LCA and CF. Until then, LUC should be reported along with LCIA results .

3.2.3 Resources

This section discusses gaps and challenges in impact assessment methods referring to the area of protection ‘resources’. Besides ‘abiotic resources’ and ‘biotic resources’, challenges in impact assessment schemes for ‘changes in soil quality’, ‘desertification’ and ‘salinization’ are analyzed .

3.2.3.1 Abiotic Resources

Abiotic resources comprise all non-living resources such as minerals, fossil fuels, water and are a relevant input to many products and production processes.

By extracting resources, the concentration in the earth’s crust is changed. However, if and to what extent biogeochemical cycles are affected or changed and if environmental changes occur due to resource depletion is not clear. Thus, resource scarcity is not always seen as a true ‘environmental impact’ (UNEP 2010a). Still, existing LCA characterization models, like abiotic depletion potential (Guinée et al. 2002) or surplus energy (Goedkoop and Spriensma 2001), aim at assessing the environmental dimension of resource depletion. As these models usually consider geologic abundance of resources, they are in practice often used to analyze resource availability for production processes. Yet, when analyzing resource scarcity, an assessment of geologic availability is not enough; limiting socio-economic factors such as country concentration of reserves or monopolistic trade structures need to be taken into account. This is particularly relevant for, e.g., materials needed in future technologies, which are commonly perceived as scarce such as rare earth metals. So far, within LCA studies, these materials do not contribute to geologic depletion in a noticeable manner. Existing indicators deliver no decision support and can even lead to wrong conclusions concerning the actual availability of resources.

A method for considering anthropogenic material stocks in addition to geologic resources was developed (Schneider et al. 2011a) reflecting physical resource availability more realistically. However, for a comprehensive evaluation resource assessment has to go beyond physical considerations. In terms of socio-economic resource availability, several frameworks have been developed (Graedel et al. 2012; EC 2010; Erdmann and Graedel 2011; National Research Council 2008), but only limited research is available regarding an integration or combination with LCA.

To promote a more comprehensive and realistic assessment of resource use and availability within LCA, a method for resource scarcity analysis in LCA, which transfers the socio-economic indicators into impact categories including characterization models is proposed (Schneider et al. 2011b; Schneider et al. 2012; Schneider et al. 2013).

3.2.3.2 Biotic Resources

Biotic resources, like natural forest or fish, can reproduce and are regarded as living (Guinée et al. 2002). Agricultural products, such as crops, farm animals (including fish-farms) or grown wood, are not considered as biotic resources, but as products derived from resources such as land, water, solar energy, and nutrients.

While impacts associated with the use of biotic resources like land use change, water consumption or global warming are addressed in the respective impact categories, the depletion of biotic resources is often neglected. Depletion of biotic resources occurs if their use exceeds their renewability rates. Hence, it should be assessed in LCA to avoid incomplete assessments of, e.g., fishery or products containing tropical wood.

A potential solution to assess biotic resource depletion, considering stocks in relation to production and renewability rates, is given by Guinée et al. (2002).

However, similar to abiotic resources (see Sect. 3.2.3.1), the availability of biotic resources can be limited by socio-economic factors as well (Schneider et al. 2011b; Schneider et al. 2012). Additionally, factors like vulnerability to natural disasters (e.g. forest fires or pest infestation) and logistic constraints, like storage stability, may affect biotic resource availability (VDI 2013). Consequently, such aspects need to be analyzed in more detail and transferred into characterization models leading to a comprehensive physical and socio-economic assessment of biotic resources.

3.2.3.3 Changes in Soil Quality

Existing soil quality definitions can be grouped in two main categories: (a) definitions which focus on ecological services like filter, buffer or storage, and (b) definitions which focus on soil uses in terms of soil fertility such as agricultural production. Soil quality might be influenced by soil texture (particle-size distribution), soil structure (framework and position of soil particles), pH value, and presence and availability of nutrients, weeds, pathogens, salts, organic matter, and contaminants (Cowell and Clift 2000).

Even though a number of studies and approaches assess soil quality in LCA, there are still gaps both on inventory and on impact assessment level. So far, no impact assessment method exists which combines numerous interrelated soil characteristics. Furthermore, agricultural practices are influencing soil quality and are not yet considered within LCA (Garrigues et al. 2012). The assessment of soil quality within LCA is relevant for products and services affecting soil quality or use soil, e.g. agriculture, forestry or civil engineering.

Potential impacts caused by heavy metals and pesticides are already considered in the impact category ‘terrestrial toxicity’. However, further impact categories on soil quality are still under development, for example soil compaction (Garrigues et al. 2013) and soil erosion (Núnez et al. 2012). For the calculation of inventory data for soil erosion, Wischmeier and Smith (1978) proposed the ‘Universal Soil Loss Equation’ (USLE); for the soil erosion impact assessment, Bindraban et al. (2012) and Garrigues et al. (2012) suggest using the characterization factors from the ‘International Soil Reference and Information Centre’ (ISRIC). Mila i Canals (2003) proposed to use soil organic matter as an indicator for soil quality. However, these approaches only consider single soil properties and therefore are not suitable to depict the whole complexity of soil quality. Some approaches consider multiple criteria: Cowell and Clift (2000) combine soil erosion, change in organic matter and soil compaction and the ‘Swiss agriculture life cycle assessment—soil quality’ (SALCA-SQ) method contains nine physical, biological and chemical properties for Swiss conditions (Garrigues et al. 2012).

To derive a single method for soil quality impact assessment, robust and globally valid impact indicators for changes in soil parameters need to be developed and subsequently combined (Garrigues et al. 2012). As a large number of parameters contribute to soil quality, the development of a single soil quality impact method is challenging. Modifications of even a few soil parameters could have severe consequences in certain regions. Thus, it is suggested to assess desertification and salinization separately, as these processes dominate impacts on soil quality for certain regions in the world . Therefore, methods to assess desertification and salinization are described in separate sections (see Sects. 3.2.3.4 and 3.2.3.5) while other factors like erosion and organic matter are summarized under soil quality. The relation between the impact categories is described in Fig. 7.4 .

3.2.3.4 Desertification

Desertification describes the degradation of land due to drastic changes in soil properties that lead to limitations in soil functions such as the supply of nutrients to plants or the water holding-capacity. Factors like climate variations or human activities (inappropriate farming practices, water management, etc.) might cause desertification. In particular arid, semi-arid and dry sub-humid areas are affected (UN 2011).

Even though desertification is a severe problem, only one approach exists to assess desertification in LCIA (Núñez et al. 2010). Desertification is usually neglected in current LCA studies. Assessing desertification is especially relevant for agricultural products from regions vulnerable to desertification, e.g. cotton produced in Egypt.

The existing methodology by Núñez et al. (2010) is based on a set of four variables (aridity, erosion, aquifer exploitation, and fire risk). Characterization factors for desertification are provided for the main terrestrial ecological regions (Núñez et al. 2010). However, there is limited case study experience to test scientific robustness.

In order to start considering desertification impacts in LCA, the method of Núñez et al. (2010) should be tested in case studies of products providing the risk of desertification. Even though this might be challenging as spatially explicit inventory data is required, case studies are needed to validate and, if necessary, improve the characterization model .

3.2.3.5 Salinization

Salinization is defined as the accumulation of water-soluble salts in soil or water bodies caused by human activities or natural processes . It may reach to levels that may cause impacts to the environment and humans (Podmore 2009). High levels of salinity can influence the water absorption of plants, may lead to soil degradation by changing the soil texture, affect surface and groundwater properties and could, to some extent, lead to distraction of habitats and biodiversity. Moreover, high salt levels in water can affect infrastructure (e.g. corrosion) (Podmore 2009; Leske and Buckley 2003).

In LCA, salinization is mentioned in relation to land use or freshwater depletion (Jolliet et al. 2004) as well as biodiversity (Amores-Barrero et al. 2013). However, consistent frameworks are missing, and so far, salinization is mostly neglected in LCA case studies (EC-JRC 2010c). Anthropogenic sources for soil salinization are mainly irrigation processes in agriculture. Salinization of water bodies is caused by over-fertilization, (industrial) wastewater, leachate from landfills or by high amounts of road salt (Podmore 2009; Leske and Buckley 2003). Thus, considering salinization is especially relevant for LCAs on agriculture or waste (water) management.

Feitz and Lundie (2002) developed an indicator for soil salinization and irrigated salinity, the so called salinization potential (SP), and proposed a preliminary soil salinization impact model as an indicator for land degradation from poor irrigation practices. Besides soil, Leske and Buckley (2003) also consider salinization of water bodies and proposed a new impact category for salinization including a separate characterization model. The main limitations of these models refer to the restricted scope—e.g. limitation to irrigation in Feitz and Lundie (2002)—and to the detailed inventory data requirements such as soil composition. Moreover, site specific fate model parameters were so far only developed for Australia (Feitz and Lundie 2002) and South Africa (Leske and Buckley 2004) .

To include salinization in LCA, further clarification is needed, whether salinization can be addressed within existing impact categories (e.g. soil quality) as indicated in Fig. 7.4 or a separate midpoint category is required (Jolliet et al. 2004; Reap et al. 2008b). In any case, local or regional salinization potentials would be needed for the assessment of individual regions (Feitz and Lundie 2002; Leske and Buckley 2004).

A simplified overview of the relation between the impact categories ‘land use and land use change’ discussed in Sect. 3.2.2.5 and ‘changes in soil quality’, ‘desertification’ and ‘salinization’ described in the previous sections is shown in Fig. 7.4 .

Fig. 7.4
figure 4

Schematic relation of land use and land use change, changes in soil quality, desertification and salinization

Full size image

3.3 Generic Aspects

This section analyzes gaps and challenges referring to overarching aspects of LCA including ‘data quality analysis’, ‘uncertainty analysis’ and ‘weighting’. Moreover, recent developments aiming at a more comprehensive assessment of potential environmental interventions are discussed. These include ‘macroeconomic scale-up’ and ‘consequential modeling approach’ as well as the consideration of ‘rebound effects’ in LCA.

3.3.1 Data Quality Analysis

According to ISO 14044 (2006) , data quality is the “characteristics of data that relate to their ability to satisfy stated requirements” and “should be characterized by both quantitative and qualitative aspects, as well as by the methods used to collect and integrate those data”. The requirements for data quality should be specified in the goal and scope and address e.g. information on time-related or geographical coverage, representativeness, sources, data variability as well as information on uncertainty of information, including data, models, assumption (ISO 14044 2006). Data quality can be influenced by a lack of data, wrong and ambiguous data, inaccurate measurements and model assumptions (Baker and Lepech 2009; Ciroth et al. 2004; Heijungs and Huijbregts 2004; Huijbregts 1998).

However, there is a lack of consensus regarding the systematic methodology of how data quality can be assessed. Therefore, it is not adopted in a consistent and constant way in LCA studies (May and Brennan 2003). Additionally, overlaps exist between data quality analysis and uncertainty analysis (Sect. 3.3.2): according to the definitions in ISO 14044 (2006), data quality analysis also includes uncertainty information, and uncertainty analysis also includes data variability. Data quality analysis is relevant for every LCA study, for both background and foreground data, as systematic errors deriving, for example, from wrong data or assumptions, as well as statistical errors from data variability, can significantly influence LCA results. It is particularly important for judging the significance of differences in comparative studies (Canada Mortgage and Housing Corporation 2004; Huijbregts 1998; Notten and Petrie 2003; Sonnemann et al. 2003).

In literature several approaches for data quality analysis exist. Weidema and Wesnaes (1997), for example, introduced a ‘pedigree matrix’ for a semi-quantitative evaluation of data quality, considering reliability, completeness and temporal as well as geographical correlation. Kennedy et al. (1996) developed a methodology to convert deterministic LCA models into stochastic models to quantify the effects of data quality uncertainty on the result of an LCA. May and Brennan (2003) recommend a quantitative assessment in combination with a separate qualitative assessment on data quality. However, none of these approaches represents a broadly accepted data quality analysis scheme.

Because of the limited robustness of data quality analysis, particular caution has to be paid when drawing conclusions from a study.

3.3.2 Uncertainty Analysis

According to Huijbregts (1998), Sonnemann et al. (2003) and Heijungs and Huijbregts (2004) uncertainties within LCA consist of parameter, scenario and model uncertainties. ISO 14044 (2006) defines uncertainty analysis as a “systematic procedure to quantify the uncertainty introduced in the results of a life cycle inventory analysis due to the cumulative effects of model imprecision, input uncertainty and data variability” .

However, terminology is not yet standardized: the terms ‘uncertainty’, ‘variability’ and ‘sensitivity’ are not clearly defined within the LCA community (Heijungs and Huijbregts 2004; Baker and Lepech 2009). Moreover, the relation between data quality analysis (Sect. 3.3.1) and uncertainty analysis is not always clear. The ISO 14044 (2006) definition, for example, includes data variability and moreover describes uncertainty analysis as “additional method for LCIA data quality analysis”. Currently, a systematic methodology on how to assess uncertainty is lacking. Uncertainty analysis is relevant in the LCA evaluation/interpretation phase, as uncertainties, e.g. deriving from statistical errors like data variability or model imprecision, can significantly influence the LCA result .

Specific approaches like the Monte Carlo analysis, fuzzy set methods, Bayesien methods or pedigree matrices (Jolliet et al. 2009) are used to determine the uncertainties of the phases within LCA (Sonnemann et al. 2003; Heijungs and Huijbregts 2004; Lloyd and Ries 2007; Lo et al. 2004). But it is still difficult to define, classify and assess the whole range of uncertainties (Benetto and Dujet 2003). Some generic LCI databases like ecoinvent (ecoinvent 2013) already provide information about distributions and data quality to calculate statistical values (Heijungs 2010; Jolliet et al. 2009; Sonnemann et al. 2003). However, systematic errors, like methodological choices concerning cut-off criteria or allocation procedures, are often more relevant than statistical errors due to random scattering of data.

Consequently, the existing uncertainty analysis methods do not really help in defining the significance of results. Furthermore, the need for software and databases which support these approaches is undeniable. To address this challenge, further scientific developments and consensus on proper uncertainty analysis methods are needed to ensure that the uncertainty of uncertainty analysis methods is smaller than the uncertainty of the LCA study itself.

3.3.3 Weighting

According to ISO 14044 (2006) , weighting is an optional element which can be used to convert the results of the different impact categories into one single score indicator by using numerical factors based on social, ethical and political value choices from one or more stakeholder groups; “it shall not be used in LCA studies intended to be used in comparative assertions intended to be disclosed to the public” ISO 14044 (ISO 14044 2006) .

As different individuals, organizations and societies may have different preferences, it is possible that different parties will reach different weighting results based on the same indicator results or normalized indicator results (ISO 14044 2006), potentially leading to different conclusions. Regularly, LCA results are communicated to LCA nonprofessionals who may prefer single-score results due to clear and easy interpretation. However, whenever single-score indicators are used, existing gaps and shortcomings of weighting methods, which might affect conclusions and tradeoffs between impact categories, would be hidden.

The three most commonly used weighting methods are the panel method, the distance-to-target method and the monetary method. For example, the eco-indicator 99 (Goedkoop and Spriensma 2001) and ReCiPe (Goedkoop et al. 2009) methods, which are based on the panel method, include three different cultural perspectives with different views on nature that can be chosen depending on the involved stakeholders (Huppes and Oers 2011) . Those three views are only representative for very few stakeholders and therefore, not generally applicable (Goedkoop et al. 2000). The distance-to-target approach, as applied in the ecoscarcity method (Frischknecht et al. 2009), is based on the current performance of a country in relation to aspirated standards, laws or goals within the society (Finnveden 1999; Howard and Kneppers 2011). Those targets vary between countries since they are foremost politically based, and therefore results are dependent on the context. The monetary approach, e.g. the environmental priority strategies in product design method (Steen 1999), can be expressed through the willingness-to-pay principle, where a monetary value is assigned to the potential damage to environmental goods and services (Finnveden 1999). Since ways of calculating these real market prices are not available, the willingness-to-pay of different stakeholders is the basis for the weighting (Howard and Kneppers 2011).

An alternative approach to subjectively weighted single-score results could be the consistent presentation of un-weighted results in addition to weighted scores. Moreover, different weighting methods should be applied in case studies in order to analyze sensitivity of value choices taken .

3.3.4 Macroeconomic Scale-Up

Macroeconomic scale-up describes the expansion from a product or company perspective towards a country or regional perspective, including dynamic effects of production structures (e.g. production of one lithium battery vs. one million lithium batteries). Macroeconomic scale-up approaches aim to link environmental burdens on the micro level (e.g. product, process data) with information on the macro level (e.g. national, regional or sector data) (Reimann et al. 2010).

Process-based LCA is traditionally used to evaluate the environmental impacts of a specific process or product, but hardly addressing macroeconomic effects. As an example, production structures (mass production vs. niche production) and industrial dynamics (e.g. technological progress) are typically not considered within process-based LCA so far (Risku-Norja and Mäenpää 2007; Zhai and Williams 2010). The analysis of environmental impacts by means of process-based LCA may neglect significant parts of the overall environmental burden by ignoring macroeconomic infrastructures (e.g. road construction, machinery) (Frischknecht et al. 2007). The implementation of a macro level perspective can also be relevant for biofuels, bioplastics and other agricultural products, as the total amount of products might be limited due to limited availability of natural resources (Bringezu et al. 2009; Piemonte and Gironi 2012).

The two methods mainly used when considering macroeconomic perspective in connection with LCA are the hybrid LCA and the Basket-of-Products approach. Hybrid LCA combines a bottom-up approach based on facility-micro-level data (process-based LCA) with a top-down economic input-output (EIO) model to account for unavailable indirect macro-level data (Deng et al. 2011; Rugani et al. 2012; Strømman and Hertwich 2004; Suh et al. 2004; Suh and Nakamura 2007). However, the application of hybrid LCA is not obvious, as miscalculations may lead to double counting or leaked emissions (UNEP/SETAC 2011a). Even though the model aims at creating a more complete system by using additional inventory data and combining the macro and micro level data (Jeswani et al. 2010), hybrid LCA fails in including macroeconomic market dynamics . The Basket-of-Products method has been developed by the European Commission Joint Research Centre (JRC) for analyzing environmental burdens associated with a representative product of one sector. The approach matches macro level data on private consumption per capita with LCI micro level data for single products consumed. It reflects the environmental impact and the resource use associated with the final consumption of an average citizen in the EU-27 over the entire life cycle of goods and services. Direct environmental effects of consumption behavior in the corresponding sectors are displayed, based on apparent domestic final consumption and demand (EC 2012a). However, so far this approach has only been tested within one pilot study for five main sectors (nutrition, shelter, consumer goods, mobility, and services) in Germany and does not yet deliver a comprehensive solution to address macro-scale-up from a global perspective. The Basket-of-Products approach can serve as a basis to include the macro perspective into process LCA, even though industrial dynamics are not included. In addition, this approach could be complemented by macro level tools like EIO (environmental input-output)/hybrid LCA, MFA (material flow analysis) or related indicators like EMC (environmentally weighted material consumption) (Guinée et al. 2006; Risku-Norja and Mäenpää 2007; van der Voet et al. 2009).

3.3.5 Modeling Approach: Consequential LCA

LCA studies can be categorized into two general types: attributional (ALCA) and consequential (CLCA). The large majority of the LCAs today use the attributional modeling focusing on the total emissions during the life cycle of a product. The effects of changes within a life cycle are not considered. CLCA seeks to assess these changes (either positive or negative) in total emissions which result from a marginal change in the level of output of a product (EC-JRC 2010c) to inform about the consequence of decisions (e.g. the effects of an increase in milk production on soybean production) (Thomassen et al. 2008). CLCA tries to model the consequence of one additional unit of output rather than the average consequences of a product.

However, CLCA also adds uncertainty on top of existing gaps in LCA, as the effects of changes assessed in CLCA depend on economic mechanisms, complex economic models and market predictions representing relationships between demand for inputs, market effects, etc. (Ekvall 2002). Furthermore, due to the limited application in case studies, practitioners lack a common understanding of how to model CLCAs, and diverse understandings and procedures exist. Awareness and consideration of these current shortcomings and challenges are relevant as increasing attention is paid to consequential modeling by LCA practitioners.

Some case studies aim at doing CLCAs and assessing the effects of changes in a system (e.g. Thomassen et al. 2008; Schmidt and Weidema 2008; Dalgaard et al. 2008; Reinhard and Zah 2009; Ekvall and Andrae 2006). However, existing case studies and used models rarely provide high levels of accuracy, completeness or precision. Moreover, case studies just reflect a few effects, as CLCAs cannot describe the full consequences of a change (Ekvall 2002). Additionally, significant double counting of emissions can occur (Brander et al. 2009) as the scope of different CLCAs case studies may overlap, and the same emissions may be accounted for in multiple CLCAs.

Consultation of other methods can improve CLCA models: dynamic optimization models can improve knowledge on marginal effects; partial equilibrium models can improve knowledge of what product flows are affected by a change; general equilibrium models can give insight on rebound effects (Ekvall 2002). However, current application of CLCA can lead to suboptimal systems as the use of marginal data can lead to wrong incentives. As the future is inherently uncertain and significant limits exist to comprehensively describe future consequences of a change (Ekvall 2002), it is proposed to assess changes by means of a baseline using ALCA and different scenarios instead of using CLCA until more robust and consistent methods and case study experience are available.

3.3.6 Rebound Effects

Rebound effects refer to the change of environmental impacts when the implementation of an improvement option liberates or binds a scarce production or consumption factor (e.g. money, time, space) which can offset the effect of the measures (Schmidt and Weidema 2008; Weidema et al. 2008; Hertwich 2005; Schettkat 2009; Spielmann et al. 2008). Rebound effects can occur due to changes in production and consumption, e.g. when energy savings resulting from a more energy efficient option are cancelled out by increasing overall energy demand (Khazzoom 1980) or when the time saved by using faster and more efficient trains is used to travel further (Spielmann et al. 2008).

Although rebound effects are already considered in some LCA case studies, it is far from being common practice. Assessing and quantifying the changes in production and consumption can become very complex and challenging as, for example, a regional perspective (Spielmann et al. 2008) or personal behavior have to be included. Ignoring rebound effects leads to either under- or over-estimations of the impacts of products (Spielmann et al. 2008). Hence, for a realistic assessment of the environmental impacts related to a decision, rebound effects should ideally be included in LCA. This modeling of consequences of a decision is important for including a macro level perspective into LCA as well as to assess potential positive effects of products/processes.

Several approaches exist for the consideration of rebound effects within LCA. Finnveden et al. (2009) and Ibenholt (2002) proposed to use general equilibrium models, which can provide insight in rebound effects, and Ou et al. (2010) included scenario analysis in an attributional LCA case study. Furthermore, also the CLCA approach aims at considering rebound effects. However, several publications (e.g. Ekvall and Weidema 2004; Erikson et al. 2007; Frees 2008; Halleux et al. 2008) lead to the conclusion that the existing CLCA modeling (marginal or affected technology) does not address rebound effects properly .

As a good option for practitioners to capture the complexity of production and consumption, the use of (future) scenarios within attributional LCA is recommended by the European Commission (2010b) until robust and comprehensive macroeconomic scale up models are available that include rebound effects.

The main characteristics of gaps and challenges referring to ‘generic aspects’, described above, and ‘evolving aspects’, discussed in the following section, are summarized in Fig. 7.5.

Fig. 7.5
figure 5

Characteristics of ‘generic’ and ‘evolving’ challenges for LCA

Full size image

3.4 Evolving Aspects

This section discusses aspects which are not yet been fully discussed in the context of LCA, such as the integration of ‘positive impacts’, ‘animal well-being’ and ‘littering’ in LCA .

3.4.1 Positive Impacts

Positive impacts refer to desirable or beneficial consequences or outcomes of product systems.

Current characterization models in LCA only consider negative environmental impacts, even though certain emissions can have a positive effect on the environment as well. One example is sulfur dioxide which reacts to sulfates, reducing global warming by increasing cloud generation (Brakkee et al. 2008) and by increased albedo (IPCC 2001). This example shows the relevance of including positive effects in LCA.

A systematic approach (the so called yin-yang concept of LCA) to identify positive (yin) and negative (yang) environmental impacts of substances and to include these impacts into the characterization models for each impact category, is proposed in Ackermann et al. (2009). The inclusion of potential positive effects of, e.g., substances was tested for the impact category human toxicity (Ackermann et al. 2010) based on data from a pharmaceutical database from Germany (ABDA 2012).

A broader set of potential positive impacts has to be identified and analyzed. It has to be explored in more detail how they can be taken into account in characterization models to enable comprehensive consideration in future LCAs.

3.4.2 Animal Well-Being

Existing definitions of animal well-being focus on how the animal is coping with its environment or stress, the fundamental behavioral needs that must be satisfied, or how animals should live according to their nature (Hewson 2003b). Many aspects of animal welfare are still unknown, and no broadly accepted definition of animal well-being exists (Hewson 2003b; BMELV 2012). When addressing animal well-being, two different types of ecosystems have to be distinguished: the ecosystem that is naturally available (e.g. natural forests) , and the ecosystems created by humans (e.g. animal farm, managed forests). Regarding natural ecosystems, animal well-being could address the impacts that products and processes have on animals in the ecosystem, beyond existing impact categories like ecotoxicity or land use. With regard to ‘agricultural ecosystems’, animals can be considered as products, and animal well-being would refer to impacts on the products as such, in contrast to other categories that ‘only’ consider impacts of products on the environment.

So far, aspects influencing animal well-being are not included in the inventory or impact assessment within LCA. Generally, it could be considered as being outside the scope of LCA and rather being part of other methods, such as EIA or—with regard to the increasing consumer’s interest in adequate animal husbandry—even SLCA. However, it could be interesting and beneficial to also include animal well-being into the (environmental) LCA discussion. LCA studies comparing conventional with organic animal farming mostly evaluate conventional systems as more efficient and environmentally friendly (per unit of output), but differences in animal treatment (between organic and conventional animal farming) are not included within the assessment (Hospido et al. 2003; Thomassen et al. 2009).

Some studies focusing on the development of indicators and frameworks to assess animal well-being related to farming are already available (Hewson 2003a; Deimel et al. 2010; de Vries et al. 2011; Hofmann et al. 2000; Häusler and Scherer-Lorenzen 2002). However, integration or combination with LCA was not yet proposed. For an inclusion of animal well-being in LCA, indicator development should consider the two aforementioned ecosystem classifications. Due to an increasing number of agricultural LCAs, indicator development for animal farming could be a priority. Indicators available in existing studies (e.g. to assess space or naturalness of feed) can be used as starting point for indicator development regarding animal well-being in LCA. For instance, weighting factors based on the distance-to-target principle could be developed. By relating the current situation of, for example, space per animal or open-air access to target values derived from organic farming standards, exceedance of animal well-being thresholds could be quantified. To establish animal well-being as a new impact category, further research is needed to clearly define the impact pathways, to enhance existing indicators, to better understand feasibility of these indicators, and to possibly also broaden the scope to address both ecosystems related to animal well-being .

3.4.3 Littering

Littering describes the disposal of products directly into the environment without any waste treatment. Consequences of littering comprise, for example, leaching of chemicals from electronic devices and batteries, cause of fires by glass or cigarettes, killing of animals and fish when eating plastics (Wong et al. 1974), and aesthetic disturbance of the landscape. Littering is caused by human behavior, e.g. due to the lack of proper education, as well as the absence of adequate regulations in many countries (Gamarra and Salhofer 2007).

As LCA has historically considered the intended way of disposal only; littering has not been addressed on the LCI or on the LCIA level. Hence, LCA neglects the fact that some products are more likely to be littered (e.g. beverage cartons) than others (e.g. deposit bottles) and that consequences of littering might differ between products (e.g. glass and plastic bottles).

In order to address this shortcoming, littering could be assessed by determining the percentage of a product being disposed of improperly. The consequences of littering could be evaluated by means of existing or new impact categories. For instance, leaching of chemicals from a battery could be assessed by means of human- and ecotoxicity impact categories. Forest fires caused by glass or cigarette littering could be assessed by land use and global warming characterization models. The killing of animals due to swallowing of plastic waste could be evaluated by means of a new impact category ‘direct non-intended killing of animals’ as proposed in Sect. 7.3.2.2.4.

4 Conclusion

This chapter summarizes the content, relevance, state of the art literature, and potential solutions of 34 gaps and challenges of LCA identified. They encompass ‘inventory aspects’ (see Sect. 3.1), ‘impact assessment aspects’ (see Sect. 3.2), ‘generic aspects’ (see Sect. 3.3), and ‘evolving aspects’ (see Sect. 3.4). Despite the large number and the broad range of these challenges, the overall scientific robustness achieved in LCA needs to be acknowledged first .

LCA can assist in identifying opportunities to improve the environmental performance of products at various points of their life cycle. Moreover, it enables fact-based decision support for industry, government or non-government organizations (e.g. for the purpose of strategic planning, priority setting, product or process design, environmental hot spot analysis, and optimization). Despite the large number of gaps and challenges identified, LCA is still the “…best framework for assessing the potential environmental impacts of products currently available” (EU 2003).

However, the methodological gaps and challenges can have a significant influence on the results of LCA studies, even though not every individual case study suffers from all 34 gaps. Some of the challenges discussed above, e.g. ‘nanomaterials’, ‘animal well-being’ or ‘littering’, are only relevant for particular applications or products. As a consequence, they do not represent severe issues for most case studies. However, if LCAs for plastic packaging or livestock production are performed, the absence of a proper coverage of ‘littering’ or ‘animal well-being’ excludes potentially significant issues from the assessment.

A number of challenges, e.g. ‘allocation’, ‘functional unit’ or ‘uncertainty analysis’, is inherent to the LCA method as such. While many of the challenges identified above can be addressed by future scientific work and progress, these fundamental challenges may inherently require value choices . These choices can be scientifically informed, but they remain value choices. On a scientific level, it can only be checked as to whether these value choices are made consistently throughout a study and particularly between alternatives.

The conclusion of this study for the scientific community is rather self-explanatory. The challenges described can be used as a research agenda for LCA. Most gap descriptions include some specific proposals for further research. They are intended to motivate the scientific LCA community for tackling these challenges and to inspire the development of scientifically robust solutions. A recurrent topic for many challenges identified is the need for additional, robust and relevant data. This is a task for stakeholders, not only for science.

Until these solutions are developed, users and decision makers face a slightly more complex situation, as the relevance of the gaps depends on the products studied and the intended application of the LCA. As a consequence, each case study should include a check, if and how the identified gaps influence or even limit the conclusions of a particular LCA study. For this purpose, Fig. 7.1 could be used as a kind of checklist for case studies. If the gaps summarized in Fig. 7.1 are screened against a specific case study context, the scope definition of this study is better informed. For each case study, some gaps will turn out to be irrelevant, some will be of minor relevance and some might be significant. The potentially significant issues should be either documented as limitation of the scope of the study, or they should be tackled with complementary tools like, e.g., risk assessment, environmental impact assessment, material flow analysis or input-output analyses.

Decision-makers in both private and public organizations need to appreciate the benefits of LCA. However, a robust, sustainable and credible use of LCA requires a proper consideration of its gaps and limitations. LCAs should be seen as one relevant element of environmentally motivated decision making, but as ISO 14044 (2006) puts it: “An LCIA shall not provide the sole basis of…overall environmental superiority or equivalence, as additional information will be necessary to overcome some of the inherent limitations in the LCIA.”