Introduction

1.1 History

Life-cycle analysis and subsequent assessment are techniques that have their origin long before these names became used. In economic theory, everything not included in the analysis used to be called "externalities". Reasons for not including certain items in economic analyses were either that they did not lend themselves easily to the monetising considered necessary by the theoretical methodologies used in the past, or that they were inconvenient to include because of their indirect and often uncertain nature. However, at an early stage there were some such externalities that had to be considered in certain contexts, including the risk of severe accidents associated with a range of technological systems, or the supply security for resources physically available only at specific locations. Although the limitation of economic theory to direct costs led to the frequent omission of such "indirect economies", there were instances where they could not be neglected.

An early use of techniques later to become incorporated into the life-cycle analysis (LCA) methodology was in the field of risk analysis. Engineers have always included estimates of risk in their design procedures and at first dealt with such risks by adding safety margins in the design, e.g. by increasing the dimensions of structural beams by a heuristic "safety factor", often quite large. In a few cases it turned out that such safety factors did not avert the risk, because for some materials the thickness is not the proper factor to consider in order to avoid breakage. In other cases there were systemic considerations affecting the risk pattern that could not be dealt with by simple safety factors. Contemporary engineering designs are characterised by more holistic design strategies, but also by reducing costs by keeping safety margins small. Risk analysis is a peculiar business, as it deals with accidents which normally are quite rare, but which in some cases can have very large negative consequences. Average calculations are therefore insufficient, or more precisely, their role in risk assessment has to be discussed and compared to other approaches, such as "worst case" appraisal.

During the late 1960s it was pointed out, notably by Chauncey Starr (1969), that risk analysis could be expanded to include more factors of what we today call externalities. His view on risk assessment was the restrictive one that only average risk counted, that is the direct product of the probability of a given event and the damage it caused. This was a provocative proposition for societies that were accustomed to accepting daily car accidents, but less happy about large airplane accidents and not happy at all about catastrophic nuclear accidents that could make capitals and seats of administration deserted for years, even if the probability was exceedingly small. Indeed, the nuclear accident issue played an important role in the advancement of methodologies to be used in risk assessment [see overview by Sørensen (1979a) and references given therein].

An even more important initiating event for life-cycle analysis and assessment was the new approach to environmental management spurred by Rachel Carson's book "Silent Spring" (1962). It brought knowledge of the threat caused by persistent pesticides to the public, making up with the old approach of keeping "externality" problems away from general attention, to be dealt with by civil servants and expert advisors sworn to professional silence.

From economists came the suggestion that risks and their associated probability of damage should always be seen in relation to the benefits accomplished by the activity in question. The central analysis tool in this "rational" approach was therefore cost–benefit analysis. In principle, such calculations could be performed for impacts other than those expressed in the term of risk, ranging from the factors traditionally incorporated into economic analysis to some of the externalities influencing, for example, the impact of a technological change on society (Rowe, 1974; Pearce, 1974). One could even start to challenge the view that the future could be discounted away simply by applying any positive interest rate to a plan for postponing the clean-up of negative impacts to far into the future. This use of private investor discounting principles to decision making on a national or international scale, rather than distinguishing between commercial interest rates, social interest rates and intergenerational interest rates, was criticised as a "time-displaced irresponsibility" (Sørensen, 1974). Such issues were to occupy an important place in the subsequent theoretical discussion of ingredients to include in a life-cycle analysis and of the best way to deal with positive and negative impacts happening at different points in time.

The notion that damage costs had to be balanced with benefits (or that benefits were required to exceed damage by a specified amount) had been challenged already by Starr (1969). His observation was that people were willing to accept much higher voluntary risks than risks imposed upon them involuntarily, e.g. by a commercial airline or a power plant operator. This raises the important issue of perceived versus physical risk that was to play an important role in extending life-cycle analysis to socially orientated views of the full impacts of complex activities. It should be added that the peculiar risk-perception involved in the voluntary choice of risky activities such as mountain-climbing or motorcar racing for Starr was not an argument against using straight cost–benefit comparisons for purely technical issues such as choosing between two types of power plant. The mixture of objective and subjective factors in the political assessment of given activities is recognised by most recent accounts of risk analysis methodology (e.g. Sprent, 1988).

During the 1970s, components of what constituted the "indirect economics" were gradually identified. These included resource depletion, environmental impact, lifetime energy inputs, type of interest rate (see above) used in economic evaluations, economy of scale and degree of decentralisation, impact on foreign payments balance and on employment, and questions of global equity (Sørensen, 1979b). It became clear that such precursor life-cycle analyses could be made for individual products, for generic technologies and for entire regional systems such as energy supply chains. Lists of concerns to be investigated were produced and the first attempts at quantifying positive and negative impacts were made, leaving subjective estimations of suitable "indicators" as an option in cases where uncertainty or poorly defined quantities made concrete numbers less meaningful. A checklist of concerns, with which a system's compatibility or lack of compatibility could be estimated, could look like this (Sørensen, 1981, 1982):

1. Satisfaction of basic biological needs

2. Acceptable health risks

3. Ensuring individual security

4. Facilitating meaningful social relations

5. Facilitating meaningful work activities

6. Acceptable accident risks

7. Small impacts on the physical environment, specifically

7a. Impacts on climate

7b. Impacts on air, water and soil

7c. Impact of availability of mineral resources

7d. Impacts on biota and ecosystems

8. Positive contribution to work and mental environments

9. Compatible with agreed goals of society, for instance

9a. A competitive society

9b. A society based on equity

9c. A society based on solidarity

9d. A highly stratified type of society

9e. A highly traditional society

9f. A highly pluralistic society

10. Encouraging democratic participation in technology choice

11. Avoiding redundant institutionalisation and infrastructure

12. Avoiding formation of monopolies and power concentration

13. Contributing to high material standards

14. Encouraging high non-material standards

15. The system having an acceptable share in the overall economy

16. Having acceptable cost uncertainties

17. The system being resilient to changing conditions

18. The system being resilient to legal and political changes

19. Keeping future options open

20. Minimising the risk of conflicts and war

21. Improving international relations

22. Not restricting development options for the poor world

23. Being insensitive to uncertainty of the impact analysis

One purpose of lists such as this was to make decision makers aware of the many non-technical aspects of technology choices, and to suggest that even if they could not be quantified, they still had to be somehow included in the decision process. As an intermediary method between quantitative and qualitative analysis, the use of indicators was proposed as a less pretentious semi-quantification of qualitative considerations, e.g. on a course scale from -1 to +1 (in relation to the list above meaning non-compatibility to compatibility). An example of using this idea is given in Section 3.1.

The 1980s saw increasing use of components of life-cycle analysis, such as the "total energy analysis" aimed at including the energy spent in raw materials and manufacture of energy conversion equipment, the operational energy use and finally the energy used at final disposal or recycling of decommissioned equipment. This constituted a complete life cycle ("cradle-to-grave"), but only energy usage (i.e. conversion to lower quality) was counted, not other impacts such as environmental damage. The purpose of such calculations was to be able to make a fair comparison between "energy pay-back times" for different kinds of energy, in other words the time that the energy production of the equipment takes to match the initial outlays of energy in establishing the system. Without such an effort, the comparison between renewable energy, nuclear and fossil fuel-based systems could not be meaningful, as noted already in the 1970s (Roberts, 1978).

The environmental dimension was included in many of the studies carried out during the 1980s and 1990s, using at first names such as "integrated impact assessment" or "full cycle analysis", and later the "cradle-to-grave" or "lifecycle analysis" terminology. The Environmental Impact Assessment methodology developed by the US Environmental Protection Agency (US EPA, 1978) was soon copied and made a legal requirement in several countries, under the name of "Environmental Impact Statement". Some of the studies included more than energy and environmental impacts, typically also occupational or air-pollution induced health impacts. United Nations organisations were central in these developments, with UNEP sponsoring several early environmental studies (see e.g. El-Hinnawi and Biswas, 1981) and IAEA looking specifically into health impacts (see e.g. IAEA, 1982). UNEP published an entire series of Environmental Impact Reports (1979–86), mixing casual data from different sources in a little convincing way. For example, the volume on environmental impacts from fossil energy use managed to forget about greenhouse gases and global warming (El-Hinnawi, 1981).

The state of California was the first to require a life-cycle analysis and assessment as a part of the approval process for new industrial products and facilities. Indeed, both the proposing party and the administration would designate a consultant to make life-cycle impact reports, and the final decision would be using both. In several cases, there were orders of magnitude differences between the two reports. Not just because one was representing the industry involved and the other a public entity, but in most cases due to serious disagreement on the principles of performing life-cycle analyses. The consulting business organisations saw this as a serious impediment for what otherwise could be a very profitable business area. Early LCA studies dealt with recycling of soft drink bottles and selling milk in bottles or cardboard boxes (see e.g. Tellus, 1992). A pressure followed for establishing standards for performing life-cycle analyses, e.g. through organisations such as SETAC (1993; see also Fava et al., 1992; Consoli et al., 1993) or the US EPA (1995), and subsequently through the international standardisation procedures (norms for performing LCA first published in 1997; latest update in ISO, 2006). As a result, guidelines were procured that would make the reports of different consultants rather similar. This, however, does not guarantee that the common results are the correct ones, as already demonstrated by the benchmark studies on nuclear accident probabilities that flourished during the late 1980s, in the wake of the Chernobyl accident (Sørensen, 1987). Concern over this state of affairs was duly expressed, e.g. by Ayres (1994) and by Krozer and Vis (1998).

It is always a problem when science is turned into pseudo-science in order to satisfy some prescribed commercial or political purpose. One wonders if the US energy industry could have influenced a large government-supported externality study of different energy supply options that came out with the surprising conclusion that externalities were minute and unnecessary to consider (ORNL/ RfF, 1992–95). To arrive at this conclusion, items like long-range transport of atmospheric emissions from coal-fired power plants were omitted, despite their early identification as key causes of negative impacts (Rodhe et al., 1972), highlighted at the UN Environment Conference in Stockholm in 1972 and incidentally giving rise to the formation of UNEP. The European Commission's Framework Programme saw an interest in transferring these results to Europe and established a joint US–EU collaborative research project. How- ever, the EU side of the externality project, called ExternE, soon found that real externalities, especially for coal-fired power plants, were far greater than found by the US side of the study (see ETSU/IER, 1995; European Commission, 1995). Because fossil externalities in the operating phase were so dominant, suggestions to extend ExternE to complete life-cycle analyses were never whole-heartedly followed, and the French project executioner decided to freeze the methodology and conduct a large number of country implementations using the fixed methodology (e.g. Curtiss et al., 1995; Schleisner and Nielsen, 1997). Quite recently, the European Commission is trying again to become active in the externality and life-cycle area, through a database project performed in collaboration with European industry, but with the restricted ISO-type interpretation of life-cycle inventories and assessment (EC/JRC, 2010a).

SETAC and UNEP have recently joined forces, cautiously and with lots of disclaimers (see e.g. UNEP, 2009) and the European Union is trying to expand its life-cycle database project to global applicability. Still, most of the market for performing run-of-the-mill environmental life-cycle assessments has already been captured by private consultants, notably the software enterprises GaBi (2010) and Pre' (2010). What they offer is easy-to-use software producing final reports more or less in agreement with the ISO standards, based on input regarding the user's product or installation, but falling back on a generic impact database if no specific data are supplied by each user. Commercial generic databases are available (e.g. Ecoinvent, 2010) and although not encouraged by the ISO prescriptions, the software packages tend to offer to perform a weighting of the individual (and often incommensurable) impact results, so that the decision maker is faced with a clear ranking of the solutions studied, e.g. based on "eco-points" (that is, indicators equivalent to monetised impact values, $, &8364;, etc., translated from physical impacts in different units by the software company, often in a little transparent way). Faced with a graph of the type shown in Figure 1.1, the decision maker or politically elected parliamentarian would appear quite superfluous. The message is that policy decisions can safely be left to bureaucrats in possession of the right commercial software.