1. Introduction
The purpose of life cycle assessment (LCA) can differ across studies, but in most cases the target is to provide information for decision support, aimed at the achievement of a more sustainable society [
1,
2]. Once a decision is made, its implementation requires time and the outcomes will only be evident in the future. Therefore, all decision-oriented LCAs entail, at least to some extent, a future-oriented feature. Nonetheless, to date, the majority of the LCA studies are retrospective in nature and do not explore possible future effects explicitly. A key assumption in this case is that historical trends can be considered as representative for forthcoming situations. But in reality, development is often a non-linear process, including both continuous improvements and disruptive changes. Incorporating a more nuanced analysis of expected future developments in LCA can therefore be essential. Such future-oriented LCAs, referred to as ex-ante LCAs in this review, have received little attention so far.
There are exceptions though: some studies are anticipating possible future states given the rationale that today’s technology performance is not always suitable for assessing future technologies [
3]. Typically, such technologies are in an early stage of development, known as emerging technologies, and they are not ready to enter the market yet. Depending on their technology readiness, they may exist as an industrial pilot, in a lab environment, as a proof-of-concept based on preliminary simulations or even only as a basic idea. Multiple research cycles still have to be run through before such technologies can become available at an industrial scale. Integrating LCA early-on in the technology development process can improve the understanding of the implications of design choices on the environmental performance of a technology [
4]. This is desirable, as such decisions can have a far-reaching influence on the future success of a technology, because they can prevent environmental burdens and unsound investments at a low cost [
5]. In fact, the earlier LCA results can be integrated, the better. A drawback is that only limited data and knowledge on the technology under development is available at this stage. This is also known as the design paradox or Collingridge dilemma [
6]. Moreover, the final goal of an LCA of technologies under development should not be the assessment of the environmental performance of emerging technologies at a lab or pilot scale, but to achieve this for the potential future technology at an industrial scale. An ex-ante approach is indispensable as well for a legitimate comparison of emerging technologies with their more mature counterparts. So, a first relevant application of ex-ante LCA consists of assessing technologies at an early stage of development. Ex-ante LCA can be used (1) to estimate the environmental profile of an emerging technology on an industrial scale and use this information to steer further research efforts and (2) to compare an emerging technology with an incumbent technology on an industrial scale.
The benefits of anticipating future situations are evident and—at least from a conceptual point of view—generally acknowledged in the case of emerging technologies [
4,
7]. However, the concept of ex-ante LCA should not be limited to emerging technologies. Two other applications of ex-ante LCA are worth mentioning. First, all technologies and product systems evolve over time, even if they are mature and abundantly available on an industrial scale for quite some time. It is well-known that technological learning influences both the economic and environmental performance of production systems. For example, the concept of learning (or experience) curves has been developed in order to link empirically observed cost reductions of a technology to key factors such as the cumulative production volume or the installed capacity [
8]. If such cost reductions are triggered by an increase in energy efficiency and/or by an optimised material management, it is likely that these factors will improve the environmental performance as well. More formally, technological learning can be described as a development due to an increase in knowledge that comes from more experience [
8]. Second, once a new technology or product system is launched commercially, its market penetration determines the magnitude of the environmental consequences in the total market mix [
9]. This is also known as technology diffusion, an analysis of the way and the rate that new products, services and ideas spread [
10]. For example, renewable energy technologies are typically considered as a more environmentally sound alternative for fossil fuels, yet for the time being their potential benefits are limited by their low market share [
11].
The previous three applications, i.e., technology development, technological learning and technology diffusion, are relevant for the technology or production system under study, which is also known as the foreground system. Yet, the same logic applies to the entire economic system as well, often referred to as the background system. This yields a fourth application of ex-ante LCA: all background processes ought to be updated systemically when future states are explored [
7,
12]. For example, developments in the electricity sector can have a substantial effect on the environmental profile of an emerging technology that appears to be energy intensive [
13].
Despite the growing awareness of the importance of incorporating possible future evolutions in LCA, the current implementation of ex-ante LCA is still rather limited and non-uniform. This non-uniformity has resulted in the creation of a variety of definitions, concepts, approaches and procedures, each of them targeting different future-oriented aspects. Therefore, the aim of this study is (1) to clarify current definitions, terminology and concepts related to ex-ante LCA, (2) to develop a generic theoretical framework that fits all four relevant applications of ex-ante LCA, (3) to categorise practical methods and procedures on how to incorporate future-oriented features in LCA, based on a broad selection of case studies, and discuss their position within the framework, and finally (4) to highlight some limitations and research opportunities. This study builds on and updates prior research efforts of Arvidsson et al. [
7] and Cucurachi et al. [
4]. However, a more comprehensive review was necessary to go beyond the analysis of emerging technologies only.
The main focus of this review is to identify the existing methods and concepts that include the effect of technological and socio-economic changes in LCA. In this context, the three most relevant steps as defined in the ISO 14040 series [
14,
15] are: a goal and scope definition, a life cycle inventory (LCI) and an interpretation. A detailed analysis of the life cycle impact assessment (LCIA) step is beyond the scope of this review. Yet, for the impact assessment, two important topics can be observed in an ex-ante context, namely the time-dependent evaluation of the assessed environmental performance [
16] and the lack of required data or relevant impact categories in case disruptive emerging technologies are to be assessed, e.g., data on toxicity for nanotechnology [
17]. More information about these topics can be found in [
18,
19].
2. Methods
2.1. Taxonomy
Over the years, many names for several modes of LCA have been coined, such as attributional LCA (ALCA), backcasting LCA (BLCA), consequential LCA (CLCA), dynamic LCA (DLCA), environmental LCA (ELCA), etc., [
20]. Some of these modes have seen their names changed over the years, which has probably increased confusion. One of the most persistent debates in the LCA community is about attributional and consequential LCA [
21]. In the early days, consequential LCA was sometimes referred to as prospective LCA, despite the fact that both a retrospective and a prospective consequential modelling approach are possible [
22]. The former assesses observed changes following a prior decision, while the latter aims at exploring how a current decision will change future flows. Still, to date, some studies define themselves as prospective LCAs when they mean to say they follow a consequential modelling approach [
23,
24]. A similar reasoning applies to dynamic LCA, a mode of LCA that is often associated with future states as well. The latter are studies “in which the temporal dimension is taken into account in one way or another” [
25]. Such studies are not future-oriented by definition, since both empirically observed evolutions as well as expected temporal effects can be considered.
For modes of LCA that do account explicitly for possible future states, the terminology is also far from homogeneous. First, a prospective LCA aims at exploring potential future life cycle environmental impacts, typically using scenarios [
26,
27,
28]. Second, an anticipatory LCA is defined by Wender et al. as “a forward-looking, non-predictive tool that increases model uncertainty through inclusion of prospective modelling tools and multiple social perspectives” [
26]. The difference, they state, is that now stakeholder values and/or multiple social perspectives are to be included in order to achieve responsible research and innovation, while this is not a prerequisite for the aforementioned prospective LCA. However, many studies that define themselves as prospective LCA do include multiple socio-economic perspectives too [
12,
27,
29]. A third mode is ex-ante LCA which “explores the future by assessing a range of possible scenarios that define the space in which the technology may operate” [
4]. But Cucurachi et al. [
4] put forward a narrower definition, based on the application of the study in question. They considered an LCA to be ex-ante only when the upscaling of an emerging technology was assessed or when an emerging technology was compared with the evolved incumbent technology at an industrial scale. Note that the definition of the term emerging technology is subject to a similar ambiguity as the taxonomy in LCA. It is used for technologies that have entered the market recently, technologies of which the market share is increasing rapidly [
30], as well as for technologies that are still under development and are not ready yet to enter the market [
31]. In this review, the last interpretation was adopted. A more detailed discussion can be found in [
32].
So, subtle but not consistent differences can be found between prospective, anticipatory and ex-ante LCAs, mainly related to whether a social perspective is included or not. This review aims at including as many relevant approaches, methods and techniques to explore and incorporate possible future states in LCA as possible. Therefore, the broadest interpretation is adopted and all these types of modes will be referred to as ex-ante LCA in this study. The term prospective LCA is not used, so as to avoid confusion with the early definition of consequential LCA.
2.2. Case Study Selection and Evaluation
In order to develop a generic theoretical framework that is able to incorporate the four applications identified in the first section, both case studies and review papers were analysed. Fifty-two case studies that define themselves as ex-ante, prospective or anticipatory LCA, complemented with LCA studies that focus on technological change, were selected. Web of Science was used as bibliographic database. After a first screening on the keywords ‘ex-ante LCA’, ‘prospective LCA’ and ‘anticipatory LCA’ only case studies targeting the broad domain of environmental assessments were retained for further analysis. Additionally, thirteen review papers were screened, together with a smaller number of studies that analyse technological learning and technology diffusion. In the latter case, only highly cited papers were included, given their importance in their respective research domains. Incorporating technological learning and technology diffusion in LCA is currently still in its infancy, and although the reviewed studies do not always have a direct link with LCA, they highlight interesting research opportunities.
Based on an initial screening, the following criteria are therefore selected to classify and evaluate the case studies. A first criterion is the level of technological maturity of the studied subjects. Two commonly used scales for estimating this technology maturity are the technology readiness level (TRL) scale and the manufacturing readiness level (MRL) scale [
33]. The first scale focuses on functional readiness, while the second includes the maturity of components or subsystems from a manufacturing perspective as well. Both scales can be directly linked, since technology readiness is a prerequisite for manufacturing readiness [
34]. The TRL scale is a more widely applied system and will be used in this study. Once a technology or a production system has entered the market, further developments and optimisations will take place, due to learning-by-doing and learning-by-researching. To facilitate comparison with technologies under development, technologies available on the market are assigned a TRL of 9+ to emphasise their continuous improvements.
The second criterion relates to whether changes in the foreground and/or background system are taken into account. In this review, the ‘management’ perspective on foreground and background systems of the ILCD handbook is applied, which focuses on whether processes can be directly controlled or influenced by the decision-maker [
1]. The foreground system can be directly affected, the background system cannot. The analysis of feedback loops from the foreground to the background system is an extension of the second criterion.
The rationale of the system model forms the third criterion. In short, this can be summarised as the choice between an attributional and a consequential modelling approach.
And finally, the exact modelling techniques of how technological changes and future states are explored are the fourth criterion. These techniques are subdivided into three subcategories: technology development (TRL < 9), technological learning (TRL 9+) and technology diffusion (TRL 9+).
Please note that this review will examine the case studies from a methodological point of view, in order to gain insight into how future states are to be explored and to better understand the current state-of-the-art of ex-ante LCA. It is important to note that this analysis is by no means a judgement on the overall quality or the relevance of the reviewed case studies.
4. Discussion and Conclusion
There is increasing attention being given to ex-ante LCA, mainly addressing emerging technologies. However, this trend is accompanied by a proliferation of definitions, concepts and methods. Therefore, the purpose of this study is to clarify terminology and to structure current practice by introducing a general framework that fits recent research efforts. Unlike previous work on ex-ante LCA, the proposed framework includes the entire technology life cycle, from the early design phase up to continuous improvements of mature technologies, including their market penetration. This approach not only facilitates the comparison of different studies, but it also allows for a categorisation and evaluation of the applied techniques, concepts and procedures in order to assess technology development, technological learning and technology diffusion. Additionally, it can assist in the determination of eventual knowledge gaps or research opportunities.
It is generally acknowledged in the literature that integrating LCA early-on in the technology developing process can steer design choices towards a maximisation of the final environmental performance of a technology. However, to date, only a few studies focus on integrating environmental aspects at low TRLs. The ideal system baseline approach is the only observed procedure that is applicable from the formulation of the first ideas on, albeit with only a few applications. There is room for improving this approach to include modelling techniques that estimate more realistic process conditions, by combining for instance molecular dynamics with macro-scale kinetics, and mass and heat transfer calculations. Yet, this requires further study, and the link between various techniques used and reduced uncertainty is unclear to date. Additionally, the explorative approach as proposed by Villares et al. [
75] looks promising for examining possibilities for low TRLs as well, at least from a conceptual point of view, but it lacks implementation. Most studies focus on very narrowly defined system boundaries to exclude uncertain aspect, but as a result, unexpected yet promising research opportunities might be overlooked. Relevant in this context is the story-telling approach of Tan et al. [
17], who included multiple research iterations of a single technology: they carried out two simulated and two experimental stages. The storyline provides insight into how practical problems were tackled in each iteration from the very start and how this affected further research efforts.
Many research efforts focus on learning curves of cost reductions, and some of them have already observed a correlation between cost reduction and improvement of the environmental performance of a technology. Nonetheless, a research opportunity is to structurally incorporate technological learning in ex-ante LCA in order to investigate the system wide environmental consequences of technological developments. It is essential to model the total observed cost reduction of a technology, as a result of the combined learning effect related to the direct inputs to a technology such as materials, energy and labour [
98]. Such a component learning approach is relevant to identifying the learning effect of each input, which is useful for complementing the cost estimates with environmental impact estimations. For example, Bergesen and Suh [
42] decompose technological learning into subcomponents and then identify changes in (1) the quantities of direct inputs required, (2) the direct value added, (3) the quantities of intermediate inputs required by products upstream in the supply chain and (4) the supply value added. The first and the third points are especially of interest for the assessment of the environmental performance. To streamline such research efforts, Huenteler et al. [
101] distinguish between two categories of products that define the learning potential. For mass-produced products and commodities (e.g., PV panels), early product innovations are followed relatively quickly by a surge of process innovation with learning-by-doing as the main driver. Complex products and systems on the other hand (e.g., windmills) have a longer initial period with competing product architectures, while afterwards the focus of innovation shifts to different parts of the product, rather than from the product to process innovations. In this case, learning-by-researching is the main driver. Both Bergesen and Suh [
42] and Huenteler et al. [
101] focus on decomposing technological learning, which is essential to assess environmental issues.
The lack of transparency is a general issue in many LCAs, but it is even more pronounced in an ex-ante context. Considering future situations, many assumptions are needed to explore potential developments. One would expect that all assumptions made are clearly described and justified. Yet in many studies, not all information on how a technology was scaled up is disclosed, for reasons of confidentiality [
17,
27,
33]. The same remark applies when expert judgements are used. These can be a valuable input, yet without being able to discuss or validate such information a study becomes more or less a black box with little added value. Understanding the insights gained during technology development and validating the applied methods and concepts are often more important than obtaining the final numeric results. For example, disclosing information about (partially) failed research iterations instead of presenting only the final results would certainly be useful in the field of ex-ante LCA, as demonstrated by Tan et al. [
17]. There is a clear need for transparent data presentation, which is definitely possible as, for example, is shown by Muñoz et al. [
24], Caduff et al. [
35,
36] and Villares et al. [
74,
75].
Three other research opportunities deserve to be mentioned here: the combination of different aspects of the framework, a diversification of studied topics and the incorporation of innovative technology into the corresponding business models. Up to now, few studies have combined changes in both the foreground and the background system or have explored potential feedback mechanisms. The authors also do not know of any ex-ante LCA study that covers all three aspects of the proposed framework, namely technology development, technological learning and technology diffusion. Regarding the assessed topics, the technology development of energy systems, electric vehicles and nanotechnology is often assessed, just as technological learning is in energy systems. An opportunity would be to broaden the field of application of ex-ante LCA. Finally, in this work the focus is on technological evolutions over time. Therefore, a last research opportunity would be to dig deeper into how such technology is expected to be used in the future. As a result of the growing attention for the circular economy concept, alternative business models relying on sale-and-take-back, lease and pay-per-use contracts could be explored in combination with the expected technological evolutions [
102].
To conclude, ex-ante LCA has the potential to influence technology development from the very start and to lead research efforts towards maximisation of the final environmental performance. Also, ex-ante LCA should not be limited to technologies under development alone, as mature technologies may still improve over time as well, sometimes with changing market shares as well. Additionally, aiming to include ex-ante LCA at lower TRLs can provide interesting design opportunities. This review has tried to present an overview of the different aspects of ex-ante LCA by introducing a generic framework and has linked this framework to an overview of practical concepts and procedures. However, further research will have to focus on refining these methods and procedures, with special attention for validation and transparent communication of the results and the underlying assumptions.