Overcoming barriers to an LCAAn IndEco article
By: David Heeney
Life-cycle analysis is increasingly seen as an important tool for incorporating environmental considerations into decision-making. The underlying concept is that the inputs and outputs associated with a product over its entire life-cycle ought to be considered. With this assessment, it will then be possible to compare products, or alternative ways of producing products and to identify the product, process or activity that has the least environmental impact.
However, attempts to undertake life-cycle assessments have often been fraught with difficulty: the task has been deemed impossible, and the results have been considered of questionable validity and utility.
In what follows, I review some of the reasons for these difficulties, and some of the preliminary steps that may be taken to begin to overcome them. Four key issues are addressed:
- the purpose for undertaking life-cycle assessments
- the scope of life-cycle assessments
- sources of data for life-cycle assessments and
- quality issues associated with LCAs.
The purpose of life-cycle assessments is to provide an understanding of the environmental impacts associated with a product, process or activity. This is most useful when these impacts can be compared to the impacts of an alternative product, process or activity (which might include doing without the product).
This reference to the comparison is generally seen as what is important. LCAs are seen as a way of answering questions such as:
- which is the better soft drink container -- refillable or non-refillable glass?
- which is the better hamburger wrap -- polystyrene clamshell or wax paper?
- which is the better package for a baby's bottom -- cloth diapers or disposables?
Unfortunately, these questions point to a major problem with the expectations (and possible the use of) LCAs. Everyone expects an LCA to be able to answer these questions, and many expect the answer to be "scientific" or "objective". Unfortunately, these questions go well beyond the purpose statement outline above, because they implicitly ignore a number of key sub-questions:
- Better for what? There is no single, accepted measure of the "environment", but rather there are multiple dimensions to the environment. There are human health considerations, acid rain, global warming, ozone depletion, persistent toxics in fish, landfill requirements, and so on. The greater the number of environmental "criteria" considered, the greater the likelihood that some of the alternatives will be better on some, and some will be better on the others. There is no objective means for determining which is best overall.
- Better where and when? Different use patterns, and different geographic contexts, for example, will affect the relative advantages and disadvantages of the alternatives. A refillable bottle system is unlikely to make sense where filling is not done locally (e.g. soft drinks in remote areas, foreign wines). Water conservation, for example, is a much more significant issue in some regions than in others.
The life-cycle assessment enables the traditional basis for many of these decisions (i.e. economic considerations) to be extended to encompass environmental and resource related concerns. It can help avoid a blinkered focusing on single environmental issues, like waste generation. However, as governments increase efforts to internalize environmental costs, the need for LCAs is likely to decrease, since traditional economic analyses will capture the benefits and costs that show up in the LCA.
In the meantime, it is important for all stakeholders to make decisions using the best available information, including the information contained within LCAs. The question therefore ought not to be: "Does the LCA prove the 'correct answer'?" but rather "Does the LCA help me to make a better decision?" Life-cycle assessment is not a panacea.
The scope of the LCA is an important determinant of its usefulness, and its feasibility. SETAC describes three stages to life-cycle assessment :
- "Life-cycle inventory-- an objective, data-based process of quantifying energy and raw material requirements, air emissions, waterborne effluents, solid waste, and other environmental releases incurred throughout the life cycle of a product, process or activity.
- "Life-cycle impact analysis -- a technical, quantitative, and/or qualitative process to characterize and assess the effects of the environmental loadings identified in the inventory component. The assessment should address both ecological and human health considerations, as well as other effects such as habitat modification and noise pollution.
- "Life-cycle improvement analysis -- a systematic evaluation of the needs and opportunities to reduce the environmental burden associated with energy and raw materials use and waste emissions throughout the whole life cycle of a product, process or activity. This analysis may include both quantitative and qualitative measures of improvements, such as changes in product design, raw material use, industrial processing, consumer use and waste management."
Historically, much of the attention has been directed at the inventory stage. Collecting accurate and complete data for the inventory has been seen as so daunting that the other two stages were hardly addressed. However, the inventory stage is of very limited usefulness in and of itself, since it is the impact that affects the environment, and the improvement capabilities that determine our ability to reduce the impact. The inventory may be an important input to these other two stages, but should not be seen as an end.
Coupling this view of the reason for the inventory, and the normal financial and practical constraints to undertaking comprehensive analyses has implications on how the inventory ought to be undertaken.
The first step in an LCA is often seen as the inventory, and great effort may be expended in trying to assemble a comprehensive inventory. An alternative view is that the first step ought to be a careful consideration of the purpose of the LCA, an assessment of the environmental impacts that are of greatest concern, and then an evaluation of the inventory needs and data requirements. It may be much more useful to have an informed impact assessment than a more comprehensive, and less focused inventory assessment.
Sources of data
There are a wide variety of sources available for use in LCAs, including government data sources, industry associations, research institutes and published documents.  Often these will need to be supplemented with personal communications or primary data collection. The sources drawn upon will depend on the reasons for undertaking the assessment and the available resources. Documents for public distribution are likely to use different sources than documents which will be kept confidential. Similarly, an LCA undertaken with limited resources will need to rely on readily available data sources.
Many data are publicly available. Others are accessible to various degrees. In some cases data need to be aggregated to avoid disclosing data about individual industries before they can be made public. For example, Statistics Canada could combine data from the Census of Manufacturers with data from the energy use surveys to calculate material inputs and energy use per unit of output. Whether this could be reported for individual products by province depends on whether such reporting can be done without violating Statistics Canada's rules for maintaining confidentiality.
However, none of these data were collected specifically for life-cycle analyses, and the data collection effort was not aimed at achieving a comprehensive set of data for an LCA. Consequently, the amount of useful data available and useful to an LCA is limited.
Some of the major sources of data for LCAs of Canadian products include:
- Federal government data -- There is a large body of information available from the federal government, both in terms of regular publications, databases, and individual reports. Several overview publications are available that review or describe these, including the Statistics Canada catalogue, and guides to data sources and databases in Statistics Canada and Environment Canada.
- Provincial government data -- Provincial governments also maintain some databases, or provide data for federal databases. In addition, they may undertake data collection and reporting activities in support of specific policies of programs. For example, Ontario is collecting extensive data on water discharges for its Municipal Industrial Strategy for Abatement (MISA), and Quebec is beginning a similar program. Provincial governments also provide input to the Residual Discharge Inventory System (RDIS) on atmospheric discharges within their jurisdictions.
- Industry associations -- Industry associations collect some information that is useful in LCAs. For example, the Canadian Pulp and Paper Association collects information on energy use in and energy savings measures applicable to the pulp and paper industry. The Canadian Tinplate Recycling Council maintains information on steel can recycling activities.
- Research institutes -- A number of research institutes in Canada collect or produce data of relevance to LCAs. For example, the Centre for Resources Studies at Queen's University has published a comprehensive review of published data on the environmental impact of mining in Canada, and plans to publish an update within a year. The Canadian Energy Research Institute has data on energy requirements for extracting energy products.
- Personal communications -- Much of the information that will be required in LCAs is not published and can only be obtained through personal communications with representatives in the industries involved. Personal communications are also required for data on companies that dominate the Canadian market, since Statistics Canada is unable to release information about these sectors without violating its confidentiality rules. Many of the better known LCAs have relied upon information obtained through personal communications. In some cases, acquiring the data has required complicated multi-party legal agreements to avoid disclosure of confidential data. Unfortunately, where these sorts of agreements have been required, the resulting LCAs are not scrutable; readers of these LCAs must take on faith that the data are accurate, and can make only very limited use of the data, since they are not reported, or are not reported in sufficient detail to be applied in other LCAs.
- International and foreign sources-- Finally there are international and foreign sources of data that may be relevant to LCAs for Canadian packaging materials. Some documents provide a review of industry operations around the world, others present "generic" data that have relevance in the Canadian context.
A common critique of LCAs is that their reliance on aggregated information leads to inaccuracies, and many of the suggestions for improving LCAs involve providing increasingly more specific and disaggregated information.
For example, information can be provided on specific manufacturing facilities used for a product, rather than industry averages, inputs and outputs can be reported by specific chemicals rather than by groups, receptor populations can be identified.
In general, each of these refinements will increase the data requirements and often the data required will only be available from primary sources. A judgement as to whether the refinements are worthy of the additional effort required may be made on the basis of an assessment of the sensitivity of the results, and the variation reported in the 'generic' data (assuming that an indication of variation is provided).
Life-cycle assessments must find the appropriate balance between data that are specific and general. If the data are too general, they will obscure important variations, and will not be relevant in some circumstances; general global data may be completely irrelevant in the Canadian context. On the other hand, if the data are too specific, a life-cycle assessment becomes difficult to undertake and use. Precision for individual situations may be gained, but the hope of practical relevance on a broad scale is lost.
Data quality issues
Data quality can be assessed in terms of three criteria: accuracy, coverage, and variability.
The accuracy of data will depend on how current they are and the method used for their collection.
Currency--Depending on the maturity of the industry under consideration, the currency of the data will be of utmost significance.
All industry sectors are affected by changing environmental regulations and the costs associated with these changes. Over recent years, environmental regulations have become much more stringent, with implications on allowable emission and effluent levels, and the costs and technologies for waste disposal. The significance of these changes are greater in some industries than in others. For example, declines in dioxin and furan emissions from the pulp and paper industry are occurring very rapidly, driven by environmental concerns, and the ready availability of relatively simple measures to effect these reductions.
Other factors, such as changing prices and availability of energy and materials, and market demand also limit the usefulness of older data in life-cycle assessments.
Collection methods--Accuracy is also affected by the data collection methods used. Independent monitoring data can be expected to be more reliable than questionnaire responses. The accuracy of questionnaire responses will vary with the size of the sample, the sensitivity of the data, the incentives (or disincentives) for those surveyed to respond accurately, the statutory obligations to respond, and the confidentiality guarantees offered.
Geographic--A wide range of factors can lead to differences in material and energy input requirements, and outputs, and some of these will vary with geography:
- scale and vintage of facilities
- regulatory requirements, and their implications on pollution control equipment installed, operating practices or technological processes adopted
- product mix, including types and specifications to be met, including such things as standard container sizes, demands for 'whiteness'
- availability and costs of inputs, including fuel and material types and their characteristics, for example sulphur content of coal.
These have relevance not only for the applicability of data from other jurisdictions to the Canadian context, but also for variations across Canada.
Incomplete coverage--Despite valiant efforts to collect data relevant to the Canadian context, it is likely that some of the data desired for LCAs of Canadian packaging materials and products will remain elusive. Three approaches for dealing with remaining data gaps are:
- additional primary data collection--In some cases it will be possible to fill gaps in the data required for an LCA with additional effort. Whether these additional data collections efforts are justified will depend on the significance of the data to the final analysis. In others, the data may be inherently unavailable, either because nobody knows, or because nobody will say. In the United States, companies that have undertaken LCAs indicate that they have sometimes been able to get data by entering into confidentiality agreements with providers of the data (Conservation Foundation 1990:15). However, these confidentiality agreements typically mean that the detail underlying a LCA cannot be revealed and the validity of the LCA cannot be assessed by independent analysis. The inability to reveal data may be a particular problem in the Canadian context when production is highly concentrated.
- reliance on data from other jurisdictions--It is unlikely that all the data for a Canadian assessment will be available from Canadian sources, and data from other jurisdictions will need to be used to fill in holes. When foreign data are used as surrogates for domestic values, they should be clearly identified as such. Ideally, the jurisdiction from which the data are taken will have a similar regulatory regime, lifestyles, fuel sources and costs and other characteristics which result in similar types of industry to those that are observed in Canada. In most--but not all--cases, United States data are likely to be more applicable than data from other jurisdictions. The two countries have similar socio-economic conditions and access to resources.
- estimates and sensitivity analyses--In some cases, data may not even be available from other jurisdictions, or data from other jurisdictions may be clearly inappropriate. In such circumstances, it may be necessary and possible to prepare estimates based on adjustments to the experience in other jurisdictions, or theoretical considerations. When these are made, it is appropriate to undertake sensitivity analyses to determine how sensitive the results are to the estimated values.
Measure--It is desirable to have a measure of the variability of any particular data set, rather than just the mean. Use of the mean may obscure significant variations that have implications on the relative advantages of alternatives in particular circumstances. It is probably desirable to know at least the mean and the standard deviation, or the high and low points on the range.
Significance--The significance of the variability will depend not only on how large a variation is observed, but also on the reasons for the variability. If variation is determined by geography, this has implications on the appropriateness of a national assessment. If variations is the result of vintage, this provides an indications of how things might be expected to change over time.
In some cases, it will be appropriate just to use the mean; in other cases it may be desirable to choose particular observations, rather than the mean. For example, one reason for variation in emissions might be the age of equipment. Data from newer equipment, designed and installed with waste minimization in mind, might be more indicative of long term generation rates than would the industry average value. Similarly, data on possible recycling rates, for example, might be chosen for an analysis rather than data for actual recycling rates, particularly if part of the objective is to assess the likely impacts of a new recycling program. The decision on the significance of the variation can only be assessed in light of an understanding of the reasons for the variation.
Managing uncertainty--Uncertainty is an inherent characteristics of the data going into any life-cycle analysis, reflecting variation in sources (often without a good understanding of the reasons for this variation), a lack of data in some areas and inconsistent data in others.
The approach for dealing with uncertainty will depend upon the purpose for which the life-cycle analysis is being undertaken, how it is to be used, and the anticipated outcome. Three approaches are considered.
Additional data collection
To some extent, it may be possible to reduce uncertainty through the collection of additional data, and setting out the data for a life-cycle analysis may point to particular areas where additional data are required, or where there may be conflicting values for specific components from different sources. However, data collection is often very expensive, and for many of the data required in a life cycle assessment, the results of the assessment may not be sensitive to "reasonable" values for the uncertain data.
Sensitivity analyses are a way of testing the implications of the uncertainty on the conclusions reached by the life-cycle assessment. For many of the data going into a life-cycle analysis, the results are unlikely to change with any "reasonable" value. For example, there may be uncertainty about the distance from a supplier to his customers. However, whether the distance is 200 km or 500 km may have no effect on the conclusions reached about the relative merits of one packaging type over another; transportation may be a minor contributor to the overall environmental effect.
The results of a comparison may be very sensitive to the value of a particular parameter. It will be desirable to minimize the uncertainty associated with these parameters as much as possible, or recognize that one can only have limited confidence in the results of the comparison.
Normally, every piece of data will not be used in a sensitivity analysis. IT is common to undertake sensitivity analyses on those data that are of high public interest, such as the recycling rate. Although these analyses may be of explanatory or practical interest, they may do little in helping the assessment and management of uncertainty. For that it is appropriate to undertake sensitivity analyses on those data that are absent, or of questionable validity.
It is also worth paying particular attention to those data whose effect is pervasive: data like mass of packaging per unit product which is a factor in all aspects of the life-cycle analysis.
Although the analysis may be insensitive to values of individual parameters, it is possible for groups of parameters to operate together. In some circumstances, it may be possible to address this with best-case and worst-case analyses. For example, in a comparison of package A with package B, three cases could constructed: a comparison based on a best guess of the values of all uncertain variables, a wort-case scenario for package A in which all uncertain values are set of values which favour package B, and a best-case scenario for package A in which all uncertain values are set to values which favour package A.  In some cases, this "best/worst" analysis may be sufficient to minimize concerns about uncertainty. In other cases, a more sophisticated set of sensitivity analysis may be suggested.
Summary and conclusions
To be most effective, several considerations should be borne in mind when initiating an LCA:
- Be clear about why the LCA is being undertaken, and how the results will be used. LCAs are to be an aid to decision making. If the LCA helps one to make better decisions, then it is useful. If it is too complex or too simple to be useful for decision making, then it is not.
- Think carefully about what aspects of the environment are likely to be affected, and which are most important and focus on those. Resource constraints preclude trying to consider every possible aspect of the environment. A little forethought can help identify the key issues or environmental concerns and focus on these.
- Be aware of the available sources of information, and make use of readily available information. There is a lot of information already out there. Collecting information is expensive, and available sources should be made use of.
- Carefully evaluate the quality of data being used, and the implications of less than perfect data. Blind acceptance of available data will not lead to good decisions; neither will their rejection while "perfect" data are sought.
Once the LCA is done, each of these four translate into four additional actions:
- Take actions, using the best available information, and an understanding of its limitations and the consequences of these limitations.
- Monitor the effectiveness of the improvements adopted.
- Publish the results of the LCA. This increases the credibility of the process, and may lead to the identification of new data sources or methodologies. Although confidentiality is always an issue, the level of precision that determines competitive advantage may be irrelevant to the LCA. For example, competitive advantage may be determined by a difference of one per cent in energy requirements; the LCA may be insensitive to differences of less than five per cent.
- Continually review whether the LCA (and the actions it is used to support) remain valid, or whether data need to be revised or updated. This is just responsible decision making.
 Fava, J.A.. R. Denison, M.A. Curran, B. Vignon, S. Selke and J. Barnum, eds. 1991. A technical framework for life-cycle assessments. Washington D.C.: Society of Environmental Toxicology and Chemistry and SETAC Foundation for Environmental Education, Inc.
 For a review of available data sources relevant to Canadian LCAs, see VHB Research & Consulting Inc. 1992. Sources of data for life-cycle analyses of Canadian packaging products. Prepared for the National Packaging Task Force, Environment Canada.
 Presumably if package A is preferred in the worst-case comparison for A relative to B, then there may be little value in performing the best-case comparison.