End-of-life in South African product life cycle assessment

Abstract

Life cycle assessment (LCA) is a tool specifically developed for quantifying and assessing the environmental burden of a product across its entire life cycle, thus providing powerful support for sustainable product design. There exists a geographical imbalance in the adoption and distribution of LCA studies, with a notably poor penetration into developing countries, resulting from a lack of technical expertise, reliable data, and an inability to engage with the key issues of developing countries. These challenges are particularly prevalent in waste management. The limitations in current LCA capacity for representing product end-of-life, coupled to the disparity in waste management practices between developed and developing countries means that LCA is currently unable to accurately model product end-of-life in South Africa. This means that, for imported products designed on the basis of LCA, the upstream impacts may be accurate, while the end-of-life is not. Therefore, to improve the use of LCA as a tool to support sustainable product design, there is a need to develop life cycle datasets and methods that accurately reflect the realities of waste management in developing countries. The objectives of this dissertation are to (i) identify the current shortcomings of existing LCA datasets in representing the end-of-life stage of general waste in a South African context, and (ii) propose modifications to existing datasets to better reflect the realities of waste management in a South African context and extract lessons from this for use elsewhere. To meet these objectives, research was undertaken in three main stages, with the outcome of each stage used to inform the development of each subsequent stage. The first stage aimed to establish the status quo with regards to general waste management in South Africa. This investigation was informed through a desktop review of government and other publicly available reports, supplemented by field work and stakeholder engagements. These results formed the basis for the second stage: a review of LCA capacity for representing product end-of-life in the South African context. The review of datasets was limited to those contained within SimaPro v8.3 and was undertaken with the aim of understanding the extent to which current datasets are capable of representing South African waste management practices. Finally, three cases of existing LCA datasets were explored. This included testing modifications that could be made in an attempt to improve their applicability to the South African reality. In South Africa, a major limitation in developing a quantified mapping of waste flows lies in the paucity of reliable waste data and the exclusion of the contribution of the informal sector in existing waste data repositories. It was estimated that South Africa generates approximately 12.7 million tonnes of domestic waste per annum, of which an estimated 29% is not collected or treated via formal management options. For both formal and informal general waste, disposal to land (landfill and dumping) represents the most utilised waste management option. Landfill conditions in South Africa range from well-managed sanitary landfills to open dumps. Considering only licensed landfill facilities, it is estimated that large and medium landfill sites accept the majority of South Africa’s general waste (54% and 31% respectively), while the balance is managed in small (12%) and communal (3%) sites. Considering the quantity of informal domestic waste enables a crude estimation of household waste distribution between different landfill classes. In this instance, while the majority of waste (40%) is still managed in large formal landfill sites, an appreciable quantity (26%) is managed in private dumps. Within SimaPro v8.3 landfill disposal is best represented by the sanitary landfill datasets contained within the ecoinvent v3.3 database. SimaPro preserves the modular construction of the ecoinvent dataset, meaning that various generic modifications to these datasets can be made, such as the elimination or addition of burdens, redefinition of the value of a burden, or substitution of a linked dataset. Practically, such modifications are limited to process-specific burdens. However, wastespecific burdens are of greater significance in the life cycle impact assessment (LCIA) result of a landfill process. Waste-specific emissions are generated using the underlying ecoinvent landfill emission model. The current model structure allows for the parametrisation of waste composition in addition to landfill gas (LFG) capture and utilisation efficiencies. However, besides the incorporation of a methane correction factor to account for the effect that various site conditions have on the waste degradation environment, the extent to which the existing model can be adapted to represent alternative landfill conditions is limited. This is particularly true in the case of leachate generation and release. Although adaptation that incorporates the effect of climatic conditions on waste degradability and emission release is possible, this requires a high level of country-specific data and modelling expertise. Thus, the practicality of such a modification within the skills set of most LCA practitioners is questionable. Further limitations in the existing modelling framework include its inability to quantify the potential impacts of practices characteristic of unmanaged sites such as open-burning, waste scavenging, and the presence of vermin and other animal vectors for disease. Analysis of the LCIA results for different landfill scenarios showed that regardless of either the deposited material or the specific landfill conditions modelled, the time frame considered had the most pronounced effect on the normalised potential impacts. Regardless of landfill conditions, when long-term leachate emissions are considered, freshwater and marine ecotoxicity impacts dominate the overall potential impacts of the site. This result implies that if landfill disposal is modelled over the long-term, the potential impacts of the process has less to do with site-specific conditions than it does to do with the intrinsic properties of the material itself. Given the ensuing extent of degradation that occurs over the time frame considered, the practise of very long-term modelling can equalise landfills that differ strongly in the short-term. In terms of product design on the basis of LCA, the choice of material can be more strongly influenced by the time frame considered than the specific landfill scenario. From a short-term perspective, for fast degrading materials the impacts incurred from leachate emissions and their subsequent treatment are of lesser importance than those arising from LFG. From a long-term perspective by contrast, leachate emissions have a significant effect on the LCIA result. Investigation into the effect of reduced precipitation on the LCIA result showed that the exclusion of leachate emissions lowers the potential impacts of a number of impact categories, with the most substantial quantified reduction observed in the freshwater and marine ecotoxicity impact categories. This result implies that for dry climates, the long-term impacts of landfilling could be significantly lower than when compared to landfill under temperate conditions, with the potential impacts of the waste remaining locked-up in the landfill. Given quantified findings on South Africa’s dependence on both formal and informal disposal, and the variation in landfill conditions across the country, it can be concluded that LCA results for the impacts of products originating from global supply chains, but consumed and disposed of in South Africa, will be inaccurate for the end-of-life stage if modifications to end-of-life modelling are not made. The findings from this dissertation provide the basis for i) a crude estimate of ‘market shares’ of different disposal practises and ii) guidelines for parameterisation of material specific emission factors, in particular for shorter term emissions, focused on LFG and leachate emissions.