Data Model for Product Related Environmental Assessment: SPINE

Life cycle assessment (LCA) is a method for analyzing and assessing the environmental impact of a material, product or service throughout its life cycle, including raw material extraction, processing, transportation, manufacturing, distribution, use, reuse, maintenance, recycling and waste treatment. The procedure for doing an LCA includes goal and scope definition, inventory analysis and impact assessment. The inventory analysis is a flow model of a delimited technical system, comprising energy and mass balances of environmentally relevant flows. Impact assessment includes both modeling the natural systems and cause effect chains there in, and weighting, which takes values held in the social system into account. When doing an LCA of a product or service, waste treatment is usually a part of the technical system. LCA may also be applied on assessment of different waste treatment alternatives. Examples are; treatment of waste generated in production processes in the life cycle, waste treatment of the used product, material recycling. Assessment of recycling schemes often involves modeling on a societal system level, e.g. models of recycling cascades such as food packaging recycled to non food packaging eventually recycled to energy. Life cycle assessment demand large quantities of data, which need to be gathered, documented and manipulated in a transparent and reproducible way. Tools are needed for documentation, storage, communication, manipulation, and administration of data. Database technology facilitates development of such data handling tools. We have developed a data model and a database format for LCA data called SPINE. SPINE allows for reuse of data through storage of data on an arbitrary level of aggregation, i.e. a whole life cycle (flow model of connected processes) may be stored and data for single processes in it may be retrieved and incorporated into another life cycle. SPINE is implemented as relational databases, such as MS Access, Oracle, and Sybase. Commercial LCA calculation software are available. However, the programs are limited in that they do not implement the full potential of the SPINE model. Life cycle assessment Life cycle assessment (LCA) is a method for analyzing and assessing the environmental impact of a material, product or service throughout its life cycle. The life cycle includes raw material extraction, processing, transportation, manufacturing, distribution, use, reuse, maintenance, recycling and waste treatment. The procedure for doing an LCA includes goal and scope definition, inventory analysis and impact assessment (figure 1). The inventory analysis is a flow model of a delimited technical system, comprising energy and mass balances of environmentally relevant flows. Impact assessment includes both modeling the natural systems and cause effect chains there in, and weighting, which takes values held in the social system into account. This paper deals only with inventory analysis (LCI). Figure 1. Illustrations of the flow model of the life cycle inventory analysis and the LCA procedure. LCA and waste management An LCA may be performed in order to assess waste management options, in which case the waste management is the system in focus. On the other hand, waste management is also part of almost any LCA, since most production processes generate waste and residual material and because a product’s life cycle includes waste management. We have found the distinction between foreground and background system useful for LCAs made to support decisions on changes. The foreground system is defined as the collection of processes on which measures may be taken concerning their selection or mode of operation as a result of decisions based on the study. The background system consists of all other modeled processes influenced by measures taken in the foreground system. A sufficient (but not necessary) condition for a process or group of processes to be in the background is that the exchange with the foreground takes place through a homogeneous market (Clift et al 1997). The distinction between foreground and background system has thus to do with in which part of the system measures may be decided on, based on the study. Both the foreground and the background systems are affected by those measures. The distinction between foreground and background systems has nothing to do with the environmental importance of those effects, and therefore should not be confused with requirements on data accuracy regarding the foreground and background systems. The requirements on accuracy are equal to both system types. Effects on environmental loads may be largest in the foreground or the background, and may be judged only from the results of an appropriately performed LCA study. Example: waste water management as foreground system The following example is based on a case study, where concepts very similar to foreground and background systems were arrived at (Tillman et al 1996, Tillman et al 1997). The main question that the LCA was set up to answer was: Which are the environmental consequences of changing the treatment of waste water from households in a well defined area, from sending it to the existing centralized waste water treatment plants to a more local system, with increased degree of recycling of plant nutrients. Three alternative scenarios were compared, with alternative foreground systems, affecting the same background systems to different extents. 1. Existing system with centralized sewage treatment. Heat is recovered from the water before release, through the use of heat pumps, and fed into the district heating system. Sludge is partly used in agriculture. 2. A suggested local system, where the solid fraction of the sewage would be separated immediately outside the houses, to be treated biologically and later recycled to agriculture. The biogas formed would be used in the district heating system. The liquid fraction would be treated on sand filter beds situated in the area. The filter bed acts as a phosphorus trap, and phosphorus may thus be recirculated to farmland. 3. Urine, faeces and grey water, through the use of separating toilets, would be brought out of the houses separately and treated separately. The urine, which contains the larger portion of plant nutrients, would be used in agriculture, as would feaces after biological treatment. The grey water would be treated on local filter beds before release. Figure 2. Principal flowchart for treatment of waste water from households. The foreground system represents those activities on which measures may taken as a result of decisions based on the study, the background system represents all other activities affected by a change of waste water treatment system. For the sake of simplicity, electricity delivered to other parts of the background system is only vaguely indicated. A change in waste water treatment systems affects technical systems outside those of the sewage treatment system. The separating toilets use less water, and thus the production of drinking water is affected. Recycling of plant nutrients reduces the need for other types of fertilizer, if the agricultural production is assumed to be constant. Production of heat energy and biogas reduces the need for other energy sources, again provided that the demand for heat does not change. The useful flows leaving the foreground system were followed to the point where they may be used. The system was then compensated with an alternative production of an equal utility. The alternative production was followed back to its origin in the natural system. The actual use of the useful flows leaving the foreground system, or their alternatives, was not included, since it was considered to be equal between compared alternatives. Thus, the waste water system was defined as the foreground system, whereas all the other technical systems affected by changes in the foreground where defined as background systems. Waste management as background system Waste management processes usually are multi input processes, where waste consisting of many different used products is treated in a common process. When performing an LCA of a specific product, however, the life cycles of all the other inputs are not of interest. Thus, the waste management processes need to be modeled in order to describe the environmental impact associated with the product studied, i.e. the environmental impact from the entire waste management process needs to be allocated between the product studied and the other inputs. Need for reuse of LCI data A product life cycle usually includes a large number of technologies and industrial sectors, all of which need to be thoroughly described. However, it is not feasible to inventory all processes in equal detail in all LCA studies. Thus, access to welldocumented and well-structured inventories of various technical systems such as waste management, production of energyware, and transportation increase the quality of LCAs. It is evident that high quality inventories are best performed by those competent within the specific technology or sector. LCI modeling and levels of aggregation Schematically, the reuse of inventories may be described as in figure 3. Consider the central flow model as being the first one made. While performing a second LCA, the left flow model in figure 3, an analyst may identify a need for data on a technical system supplying a specific product, something which was modeled within the first life cycle inventory. Provided the first study was sufficiently documented at an appropriate level of detail, parts of it may be reused with retained transparency. There is also a need to reuse inventory results on individual processes, as indicated in the flow model to the right in figure 3. Figure 3. Schematic illustration of different types of reuse of life cycle inventory data and documentation. Data and documentation reuse may be done in different ways and may be differently supported by the originator of an inventory. Generally, the documentation of an inventory is done as a technical report. In many cases, both the numerical data and the documentation of the different processes are merged so that the original information cannot be unambiguously reconstructed, i.e. the documentation has not been done transparently. If a report of an inventory is to support reuse of data this is best done if there is a common agreement on how to handle, document and interpret data and documentation. Data modeling We have designed a database format for transparent storage of LCI data and documentation, which supports reuse of data at optional levels of aggregation, in accordance with the LCA needs (Carlson 1995, Steen et al 1995, Carlson et al 1997). The format was designed by the use of data modeling techniques, which resulted in the identification of two crucial and central concepts: activity and flow. Activity is defined as a model of any kind of technical system, such as raw material extraction, production, use, waste treatment or transports. There are no requirements made whether an activity models a small technical system, such as an engine or a boiler, or whether it models a large system, such as a cradle to gate or even a cradle to grave life cycle. Also, an activity may be either indivisible or aggregated. An aggregated activity is made through defining a system boundary surrounding a number of linked activities, aggregated or indivisible. Activities are linked through connecting flows of energy or matter. Flow is defined as the environmentally significant material or energy flows entering or leaving an activity, such as natural resources, raw material, products, energyware, emissions and waste. The flow model of an LCI is constructed as an aggregated activity, in which the processes may be modeled as either aggregated or indivisible activities. They are connected through their exchange of material or energy. Unconnected flows are regarded as exchange with the environment of the aggregated activity, whether surrounding technical systems or the natural environmental system. The concepts are illustrated in figure 4. Figure 4. Schematic representation of the concepts of activity and flow, and their relation to the system environment. The conceptual model provides the means for a structured approach to LCI data handling. However, to enable also structured data documentation, the meta data requirements of LCI was data modeled also. The result from this was a set of meta data categories: identification of the technical system, identification of the procedure to inventory the technical system and methods applied for retrieving or acquiring the numerical data. Computer based LCA data quality Numerical accuracy is still a general dilemma when analyzing large technical systems from an environmental point of view. Basically, this depends on lack of knowledge and underdeveloped traditions in regards of the accounting of environmentally significant flows. There are great differences in accuracy between different data. Due to coincidental relations with economical accounting, environmental legislation and measurement techniques, some data may be very accurate while the deviation from a probable level might exceed hundreds of percents for other data. This holds both for single numerical data, as well as for entire data sets on a technical system, i.e. an activity. Another aspect constituting the accuracy dilemma is the difficulty with quality checking the accuracy. Which norms should be used? By definition a norm should be an accurate measure of something which is well defined, and an item tested against this norm should be equally well defined. However, the accuracy dilemma is true both for any potential norm as well as for the item being tested: there are no confidently accurate data. What can be tested, however, are the methods and procedures used for acquiring and retrieving data: one single data or a set of data may be regarded as accurate if it has been retrieved or acquired using appropriate methods. SPINE was modeled to support the idea of implicating data accuracy from assessment of the acquisition methods. Any single numerical data or any data set may be supplied with a detailed description of the methods or procedures applied to the data; thereby enabling a serious handling of environmental data on technical systems. In addition, SPINE also allows for statistical information on data. Any figure may be supplied with a range, a minimum and a maximum, in accordance to a statistical distribution. It is also possible to describe a standard deviation. Most data today, however, are not supplied with any such statistical information. Due to the difficulties with finding accurate and statistically described data, it needs to be stressed that there is a risk associated with data aggregation: error propagation will always increase the uncertainty of the final result. Data aggregation as supported in SPINE, however, enables not only numerical data to be aggregated, but also enables undistorted aggregation of all documentation: the result of an inventory includes all input data with all of its documentation. SPINE The result of the data modeling was implemented as a relational database structure, entitled SPINE (Sustainable Product Information Network for the Environment) (Carlson 1995, Steen et al 1995). Since SPINE was published, it has been implemented as an LCA data storage facility within many different companies and other organizations. Among these are CPM (Centre for Environmental Assessment of Product and Material Systems), at competence center at Chalmers University of Technology, in which approximately 15 of the largest Swedish industrial corporations participate. These participants represent a diversity of industrial sectors, e.g. automotive, electronics, pulp and paper, chemicals, energy, white goods and heavy industrial appliance. Most of the CPM participants use SPINE for their internal LCA data storage and communication. The CPM group (companies and university departments) share their common data through a common SPINE database, which is placed at Chalmers. Agreements on how to use, interpret and further develop the SPINE structure are also made with CPM as a platform. The data administration at CPM educates, coordinates and provides support to the CPM companies and to other organizations using SPINE. SPINE is a freely available format, and it may be employed by anyone. For example, commercial LCA calculation and analysis software has been developed, which supports parts of the SPINE capabilities. Also CPM provides non commercial software for LCA data documentation and review. But there are further needs for more general software and software for different specific applications. Further developments CPM and SMS (svensk Materialoch Mekanstandard), has suggested a new ISO workgroup with aim to reach an international standard for an LCA data documentation format. This suggestion has been sent to the technical ISO committee ISO/TC 207/SC 5 within ISO 14041. The SPINE report was enclosed with the suggestion. However, the result from the standardization procedure might come to deviate from the present SPINE model, as published in the SPINE report (Carlson 1995, Steen et al 1995). In addition to the expected changes implied by the standardization procedure, there also is a need for extension of the SPINE model. Today every model assumes linear behavior between inputs and outputs. But this may be improved by adding consistent algorithmic descriptions of other types of systems models. Examples are automatic routines for allocation based on economical data, simulation of non-linear processes, simulation of the non-linear environmental effects of the speed of ships, etc. SPINE was modeled not only to handle data on LCI data, but also to hold compatible lifecycle impact assessment data, such as data describing models of environmental causeeffect chains and models of social systems attitudes and weighting of environmentalimpact. This modeling, however, needs to be further developed. ReferencesTillman, A-M., Lundström, H. och Svingby, M (1996). " Livscykelanalys av alternativaavloppssystem i Bergsjön och Hamburgsund. Delrapport från ECOGUIDE projektet".Rapport 1996:1, Teknisk Miljöplanering, CTH, Göteborg. In Swedish, short englishversion available in Tillman et al 1997. Tillman, A-M., Svingby, H. and Lundström, H (1997). "Life Cycle Assessment ofMunicipal Waste Water Systems A Case Study". Submitted to Int J LCA.Clift, R, Frischknecht, R, Huppes, G, Tillman, A-M and Weidema, B (1997). "Towards acoherent approach to life cycle inventory analysis". Report from the SETAC WorkingGroup on The Enhancement of Life Cycle Inventory Analysis, in preparation. Carlson R (1994). "Design and implementation of a database for use in the life cycleinventory stage of life cycle assessments". M.Sc. thesis. Department of ComputingScience and Technical Environmental Planning, Chalmers University of Technology,Göteborg. B Steen, R Carlson & G Löfgren (1995). "SPINE. A relation database structure for lifecycle assessments". IVL Report L 95/196. Swedish Environmental Research Institute,Chalmers University of Technology and Chalmers Industriteknik, Göteborg 1995.Carlson, R., Steen, B., Tillman, A-M., and Löfgren, G (1997)."LCI data modelling and adatabase design". Accepted for publication in Int J LCA.