Combined material ﬂ ow analysis and life cycle assessment as a support tool for solid waste management decision making

Material ﬂ ow analysis (MFA) and life cycle assessment (LCA) have both widely been applied to support solid waste management (SWM) decision making. However, they are often applied independently rather than conjointly. This paper presents an approach that combines the MFA and LCA methodologies to evaluate large and complex SWM systems from an environmental perspective. The approach was applied to evaluate the environmental performance, focusing on greenhouse gas (GHG) emissions, of a local authority SWM system and to compare it with alternative systems to assess the potential effectiveness of different waste policy measures. The MFA results suggest that national recycling targets are unlikely to be met even if the assessed policies are implemented optimally. It is likely that for the targets to be met, investigated policies would need to be combined with additional policies that target reductions in waste arisings. The LCA results found land ﬁ lling of residual waste to be the dominant source of GHG burdens for the existing system, whilst material reprocessing was found to result in GHG bene ﬁ ts. Overall, each of the alternative systems investigated were found to result in lower GHG impacts compared to the existing system, with the diversion of food waste from the residual waste stream found to be potentially the most effective strategy to reduce GHG emissions. The results of this study demonstrate that the complementary methodologies of MFA and LCA can be used in combination to provide policy and decision makers with valuable information about the environmental performance of SWM systems.


Introduction
Currently, around 1.3 billion tonnes of municipal solid waste are generated annually worldwide and generation levels are projected to almost double by 2025, driven by rapid population growth, urbanisation, and socio economic development in developing countries (Hoornweg and Bhada-Tata, 2012). A substantial proportion of this waste material can be viewed as a resource. As demand for natural resources continues to rise, there is increasing pressure on the world's natural resource base, which is having severe environmental consequences (Hertwich et al., 2010). At a global scale, climate change is a serious international concern and the extraction, processing, and use of natural resources contributes directly to climate change through the burning of fossil fuels, whilst the disposal of materials in landfills contributes through emissions of greenhouse gases (GHG). Improving solid waste management (SWM) by recovering value in the form of material and energy resources can contribute towards enhanced resource efficiency and GHG mitigation efforts (UNEP, 2010).
The waste management sector is under increasing pressure to improve its environmental performance. In the European Union (EU), Member States are legally obligated to formulate and implement regional policy instruments to meet the environmental SWM objectives and targets outlined in a broad international legal framework. Article 4(1) of the EU Waste Framework Directive establishes the "waste hierarchy",afive step priority order of waste management comprising, in descending order of priority, prevention, preparation for reuse, recycling, other recovery (e.g. energy from waste), and disposal (EC, 2008). Under the terms of Article 21(1) of the EU Waste Framework Directive, all waste management decisions must be undertaken in line with the waste hierarchy. The Landfill Directive sets a target for member states to reduce the amount of biodegradable municipal solid waste going to landfill to 35% of 1995 levels by 2016 (EC, 1999). A target of achieving 50% recycling of key household waste materials (paper, glass, metals, and plastics) is established in the Packaging and Packaging Waste Directive (EC, 1994). Moreover, at a broader level the EU is committed to reducing its GHG emissions by at least 20% and 40% of 1990 levels by 2020 (EC, 2009) and 2030(EC, 2014, respectively. Managing resources to maximise environmental sustainability and contribute towards the achievement of national targets set by the EU entails important strategic and investment decisions by local waste managers, who are simultaneously tasked with maintaining a reliable and economical waste removal service to residents under increasing budgetary pressures. To reduce the burden on local waste managers and promote environmental sustainable practices, there is a need for strong waste policies that guide and enable effective local decisions and actions. Analytical tools are required to assist local and national governments in evaluating the environmental performance of potential policy measures and local decisions (Turner et al., 2011). Such tools must be capable of handling the increasing complexity of modern 'integrated solid waste management' systems. Modern SWM encompasses a large number of waste treatment technologies, such as incineration, composting, and anaerobic digestion (AD) that are each designed to manage specific waste streams. Many of these technologies provide additional functions, such as secondary materials production and energy production, that necessitate interaction with other sectors, such as manufacturing, agriculture, and energy production (Giugliano et al., 2011). Furthermore, modern SWM systems comprise a global network of facilities, each with distinct technological facets and different levels of operational performance. It is necessary that analytical tools used to support decision making in complex, interdependent systems, such as SWM, adopt a whole system approach that reflects this complexity (Blengini et al., 2012).
In this paper we apply an approach that combines two systems based methodologies, material flow analysis (MFA) and life cycle assessment (LCA), to quantitatively evaluate a complex, municipality-scale SWM system and use scenario analysis to assess the potential effectiveness of different national waste policy measures. The novel contribution of this paper can be summarised as follows: Application of a combined MFA and LCA approach to evaluate a complex, multi waste stream SWM system at the meso level.
Novel use of publically available waste data to comprehensively model waste flows through the system.
Provision of information to national government regarding the potential effectiveness of waste policy measures.
Assistance to local government in identifying optimal SWM solutions.

Case study
Wales is a constituent country of the UK that covers an area of 20,779 km 2 with an estimated population of 3.1 million in 2014 (ONS, 2015). Wales comprises 22 unitary authorities that are individually responsible for arranging waste collection and disposal. The Welsh Government (WG) has introduced a broad and ambitious sustainable development strategy that aims to make sustainable development the core principle of all national and sub national policy and decision making (HMSO, 2015;WAG, 2009). Two key national targets for 2050 have been set: 1) achieving "zero waste" (i.e. eliminating landfilling as far as possible) and 2) reducing national GHG emissions by at least 80% below 1990 levels (HMSO, 2008;WAG, 2010a,b;. Furthermore, the WG have also set a target of a 3% reduction in national GHG emissions per year until 2050, to which waste management is required to contribute (WAG, 2010a).
The city of Cardiff is the capital of Wales and is located in the south of the country. The city covers an area of 140.3 km 2 , of which around 76 km 2 is considered urban, and has a population of approximately 341,100 and a dwelling stock of 135,796. The city has a reported recycling rate of 52.2%, marginally below the national average of 52.3% (StatsWales, 2015b).
The Council operates an alternate weekly kerbside collection service for household residual waste and dry recyclables. Waste materials collected for recycling include paper, card, aluminium cans, steel cans, mixed plastics, and mixed glass. Garden and food waste are each collected separately on a weekly basis from households, whilst an optional absorbent hygiene products collection service operates fortnightly. A bespoke bulky waste collection service is also offered. There are four household waste recycling centres (HWRC) and 14 bring sites (recycling banks) located across the city. The Council also runs services for the collection of wastes from: commercial organisations, street cleaning, fly tipping, and municipal parks/grounds.

Methodology
A combination of methodologies was applied in this study to quantitatively evaluate the environmental performance of Cardiff's local authority collected waste (LACW) management system and those of alternative systems. LACW comprises all solid waste collected by a local authority. A static MFA approach was applied to assess the mass flows and stocks of LACW into, within, and from the defined systems. MFA is descriptive, systematic approach to assess the metabolism of a defined system and is based on the principle of mass conservation (Brunner and Rechberger, 2004). The freeware STAN v2.5, developed at the Vienna University of Technology, was used to conduct the MFA (Cencic and Rechberger, 2008). STAN allows for visualisation of complex MFA systems, performance of data reconciliation, and consideration of data uncertainties. Mass flow quantities are expressed in terms of mean value and standard deviation, defined as normal distributions. Data reconciliation, based on the method of least squares regression (see Fellner et al., 2011), is performed to enforce mass balance constrains on uncertain, conflicting input data. The extent to which data are reconciled is determined by the initial data uncertainty, which is used as the weighting factor. STAN has been used in many previous MFA studies of SWM systems (e.g. Allegrini et al., 2015;Arena and Di Gregorio, 2014;Boldrin et al., 2011;Mastellone et al., 2009). The environmental impacts of the system described by the MFA were assessed using LCA. LCA has been applied extensively to evaluate SWM systems (see Laurent et al., 2014a). The focus of the LCA undertaken in this study was to evaluate the potential climate change impacts of LACW management and to identify the waste streams and waste treatment processes that contribute most significantly to these impacts. The LCA followed the ISO 14040 and 14044 standards for LCA (ISO, 2006a,b) as far as possible and was performed using the EASETECH software. EASETECH, developed at the Technical University of Denmark, is a LCA model for the assessment of environmental technologies that allows for detailed modelling of complex systems and provides specialist functionality for modelling waste management systems (Clavreul et al., 2014). EASETECH has been used in previous LCA studies of SWM systems (e.g. Butera et al., 2015;Turner et al., 2016;Yang et al., 2014).

Goal definition
The goal of this study was threefold: 1) to quantify the mass balance of waste of the existing LACW management system; 2) to quantitatively evaluate the environmental performance of the existing LACW management system with regards to GHG emissions; and 3) to compare the environmental performance of the existing LACW management system with that of alternative systems designed to represent the implementation of national waste policy measures. The purpose was to support regional policy and decision making, i.e. ILCD (International Reference Life Cycle Data System) decision context situation B, meso level decision support (EC et al., 2010). The study was carried out in cooperation with the WG and Natural Resources Wales (NRW), the regulator and principal advisor to the WG, who, along with Cardiff County Council, were the key audience. Additionally, aspects of the study will be of value to other national policy makers and local governments, as well as the wider SWM MFA and LCA communities.
The quantitative assessment of the systems under investigation was carried out using LCA. The LCA followed an "attributional" approach (EC et al., 2010), with allocation avoided by means of system expansion. The assessment included the potential environmental impacts from SWM activities as well as impacts on processes in other systems that are affected by SWM activities, principally the recovery of materials and energy production (Giugliano et al., 2011). The system comprised two subsystems: (1) a foreground system that includes processes directly engaged in the management of the reference flow (here, LACW); and (2) a background system that interacts with the foreground system by supplying or receiving resources, including avoided primary material and energy production (Clift et al., 2000).
Background data were taken from international life cycle inventory (LCI) databases (ecoinvent v.2.2 and ELCD v.2), the UK GHG conversion factors repository (Defra et al., 2013), and literature (see Appendix A). Foreground system data were derived from various secondary sources and are described in detail in Section 2.3. Following an attributional approach, average data were generally used to model foreground and background system processes. However, based on the recommendations of the ILCD (EC et al., 2010) and Laurent et al. (2014b) for LCA studies aiming to provide meso level decision support, long term marginal process data were used to reflect the large scale consequences on the energy system from avoided material and energy production. This approach, which is similar to that taken by Rigamonti et al. (2010) and Finnveden et al. (2005), was adopted rather than a "pure" attributional approach as it more accurately reflects the large scale implications of SWM activities on other systems. Specifically, the following assumptions were made concerning material and energy substitution: Electricity produced from SWM activities displaced an equivalent amount of electricity generated from combined cycle gas turbines (CCGT), which is considered as being the long term marginal energy source in the UK by the government (DECC, 2014a). Generated electricity is transmitted to the National Grid at an efficiency of 98% (National Grid, 2008).
Heat produced at facilities fitted with combined heat and power (CHP) units would be used internally due to a lack of established district heating networks in the UK (Hawkey, 2012). Hence, no avoided energy production was associated with heat energy production.
Secondary products produced from recovered waste materials replace the production of alternative products, including those produced from primary resources. Avoided materials production was modelled based on the average market situation. However, to reflect large scale consequences on the energy system, avoided material production process electricity use was modelled using the long term marginal electricity source, i.e. CCGT. The extent of the displacement was calculated based a substitution ratio, which was calculated as the product of three parameters: (1) recyclability (the amount of a waste material in a recycled product, considering all process material losses); (2) material quality loss (reflects any diminishment in the inherent technical properties of a waste material incurred through reprocessing); and (3) market substitution rate (the actual amount of an alternative product that is replaced at the market). Material losses and material quality losses are defined for each recycled or reused material. Market substitution rates of 1 were used for all product substitution calculations based on the supposition that production of secondary products would not affect the market situation (Briffaerts et al., 2009;Merrild et al., 2012;Rigamonti et al., 2009).

Scope definition
The function of the system under investigation is to manage LACW from Cardiff. Hence, the functional unit was defined as the management of the total amount of collected LACW in Cardiff between April 2012 and March 2013. The amount is 168,526 tonnes. LACW is categorised into five primary waste streams (commingled materials, source segregated food waste, source segregated garden waste, source segregated dry recyclables, and residual waste) that are derived from nine waste collection sources (household kerbside collection, bulky waste collection, HWRCs, bring sites, street cleaning, fly tipped materials, non household collections, voluntary sector/community collection, and municipal parks/grounds waste collection). All activities required to manage the LACW from the point of collection to its ultimate disposal, reprocessing, or reuse were considered. Hence, the processes of collection, transportation, intermediate processing (sorting and handling), thermal treatment, biological treatment, material reprocessing and reuse, and landfill were individually quantitatively analysed in terms of their resource consumption and emissions.

System boundaries
The spatial boundary of the system was defined by the administrative boundary of Cardiff County Council, i.e. only waste generated within this defined spatial boundary was considered in the analysis. The temporal boundary was the April 2012 to March 2013 fiscal year. In this study, the "zero burden assumption" was adopted, whereby the environmental impacts from upstream life cycle stages prior to the collection of LACW were not included (Ekvall et al., 2007). Environmental impacts from capital goods were also excluded, an approach consistent with other LCA studies of waste management systems (Cleary, 2009). Due to difficulties in data collection, environmental burdens from operation of HWRCs were also excluded.
Processes included in the assessed system comprise collection and transport of LACW, waste management facilities (including those for the treatment of residuals, such as incinerator bottom ash or fly ash), and utilisation of secondary products (e.g. digestate or compost). Waste management facilities included in the assessed system were identified from Cardiff's WasteDataFlow 1 Question 100 2 returns for the four quarters of the fiscal year 2012e2013 (WasteDataFlow, 2014), which include details of facilities that handled collected LACW. The Cardiff LACW management system comprised 158 individual facilities located across nine different countries. 139 facilities are located in the UK (41 in Wales and 98 in England), five were located in Europe, and 13 were located in Asia. Facilities was categorised into one of 11 different waste treatment technology types (e.g. AD, incineration, material reprocessing, etc.), based on their description in WasteDataFlow, and were allocated individual identification codes that reflected the facility type and location (domestic or foreign). In addition, 15 generic processes were included to manage collected waste where no waste treatment facility was specified in WasteDataFlow. Details of the facilities (unit processes) included in the assessed system are provided in Appendix B. Facilities were modelled in the MFA and LCA as unit processes.

Impact coverage
The quantitative potential environmental impact assessment of this study focused on GHG emissions. Hence, climate change was the only potential impact category included in the life cycle impact assessment. This decision was made for two principal reasons: Climate change is recognised as being a significant global environmental problem of foremost importance (IPCC, 2014; ISWA, 2015; WAG, 2010a); and Climate change impact has been identified as being a good proxy for the overall environmental impacts of SWM (Defra, 2011).
The GHGs considered in the assessment were carbon dioxide (CO 2 ) (fossil and biogenic), methane (CH 4 ), and nitrous oxide (N 2 O). Emissions of these GHG represent more than 90% of total GHG emissions from SWM (Bogner et al., 2007).

Scenarios
Four scenarios are compared in this study. The first scenario represents the existing LACW management system, whilst the other scenarios reflect different systems based on national waste policies. Scenarios are compared on the basis of functional equivalence e i.e. it was assumed that waste generation was not influenced by the different scenario assumptions. S1 Baseline scenario Baseline scenario representing the existing LACW management system for Cardiff in April 2012 to March 2013. S2 Enhanced food waste capture Wales' Municipal sector plan (WAG, 2011) states that in order for Wales' 70% recycling target for 2025 to be met, it is likely that local authorities will need to separately collect for biological treatment at least 80% of biodegradable waste. To evaluate the potential GHG impacts of enhanced biodegradable waste capture, in this scenario an 80% capture rate of food waste via household kerbside food waste collection is assumed, with food waste diverted from the household residual waste stream. All food waste that is collected is sent to AD, the priority destination for food waste in Wales (WAG, 2010b). S3 Enhanced incineration The Towards Zero Waste (WAG, 2010b) waste strategy for Wales sets a target for a maximum level of incineration (with energy recovery) of MSW of 30% by 2025. This scenario was constructed to evaluate the impact of enhanced incineration of LACW. Based on the total quantity of collected LACW for Cardiff in 2012e2013 (168,426 t), the 30% limit set by WG allows for up to 50,530 t of LACW to be incinerated. In this scenario, 50,000 t of residual waste originally collected and sent to landfill is instead sent for incineration. S4 Enhanced dry recycling A target for a minimum level of recycling and biological treatment of municipal solid waste of 70% by 2025 was set by the WG in Towards Zero Waste (WAG, 2010b). To evaluate the impact of an enhanced dry recycling strategy, in this scenario optimum household kerbside collection recycling rates (taken from Eunomia et al., 2011), are achieved for key recyclable waste materials (glass, paper, cardboard, steel cans, aluminium cans, and plastics), with materials diverted from the household kerbside collected residual waste stream.

Life cycle inventory 2.3.1. Reference flow characterisation
A static MFA approach was applied to characterise the mass flows and stocks of LACW into, within, and from the defined system. Due to its complexity and size, the system was divided into four subsystems, representing the management of primary waste streams (residual waste, commingled materials, source segregated dry recyclables, and source segregated food and garden wastes). The import flows into each system were the collected LACW primary waste streams, and the export flows comprised secondary products and emissions. To account for flow multidimensionality, secondary waste types (e.g. paper, glass, food waste) were characterised as fractions of the primary waste stream.
Due to the functional complexity of many processes and the profusion of functionally equivalent processes in the system (i.e. SWM facilities engaged in the same activity), it was necessary to divide many processes into subsystems that contain subprocesses and, in some cases, further subsystems. The four defined systems for the baseline scenario (S1) are, collectively, composed of 548 processes and 1042 flows.
Mass flows of waste into the LACW management system and to and between processes were estimated by aggregating Cardiff's WasteDataFlow Question 100 returns for the four quarters of the fiscal year 2012e2013. Where data were not available from WasteDataFlow (e.g. mass flows of wastes and secondary products from treatment facilities and reprocessing), process transfer coefficients were derived from the LCI process inventory models (detailed below). All data used in this study were single value and were inputted into the MFA as mean values and an associated uncertainty, expressed as a percentage. Details of the data reconciliation approach are provided in Appendix C.

Waste composition
The overall waste composition for the investigated system is presented in Fig. 1. Further details are provided in Appendix D.

Collection and transport
Details of the modelling approach for collection and transport are presented in Appendix E.

Transfer station
Transfer station process electricity and diesel consumptions of 4 kWh/t and 1 l/t, respectively, were assumed, based on measured data from UK based transfer stations reported by Eunomia et al. .

Anaerobic digestion (food waste only)
The AD of food waste process was modelled based on dry, mesophilic (one stage, one phase) technology, as described by Møller et al. (2009). Details of process inputs and parameter values are presented in Table 1. During waste reception, reject rates of 8%, 50%, and 95% were applied for food waste and fine material, other biodegradable waste, and non biodegradable waste, respectively (Bernstad and la Cour Jansen, 2011). Biogas was assumed to be utilised on site in a CHP engine -the most common biogas utilisation technology in the UK (Horne et al., 2013) -with gross electrical and heat energy generation efficiencies of 31% and 49%, respectively (DECC, 2014b). Heat energy was assumed to be used internally. CHP unit process emissions, produced through incomplete combustion, of 434 mg CH 4 /MJ biogas and 1.6 mg N 2 O/MJ biogas were assumed (Nielsen et al., 2010). Digestate application was modelled based on the average situation in the UK in 2012, with 91% applied to agricultural land, 6% used in horticulture/gardening, and 3% used as alternative landfill daily cover (Horne et al., 2013). See Section 2.3.8 for details.

Composting
Two composting processes were modelled: in vessel composting (IVC) and open windrow composting (OWC). During waste reception, reject rates of 8% and 95% were applied for biodegradable waste and non biodegradable waste, respectively (Bernstad and la Cour Jansen, 2011). Details of process inputs and parameter values are presented in Table 2. The IVC process was assumed to be equipped with a biofilter. Biofilter CH 4 oxidation efficiency was modelled based on the mean of reported values in the literature, whilst the biofilter was assumed to have no effect on N 2 O emissions due to a lack of consensus in the literature. A biofilter was not included in the OWC process as it is an open technology. Application of compost produced through IVC and OWC was modelled based on the average situation in the UK in 2012 as follows: for IVC derived compost, 68% to agricultural landspreading, 29% in horticulture/gardening, 2% as alternative landfill daily cover, and 1% for incineration; and for OWC derived compost, 73% to agricultural landspreading, 21% in horticulture/gardening, 5% as alternative landfill daily cover, and 1% for incineration (Horne et al., 2013). See Section 2.3.8 for details.

Mechanical biological treatment
Mechanical biological treatment (MBT) involves a combination of mechanical separation and biological treatment (either bio stabilisation/bio drying, IVC, or AD) (Defra, 2013b). The MBT facility in the investigated system comprised integrated IVC technology.
Based on primary data from a UK based MBT facility, process diesel and electricity consumption was assumed to be 0.32 l/t and 7.1 kWh/t, respectively. During mechanical separation, waste is sorted and classified to separate a fine, compostable fraction and two coarse fractions: one for refuse derived fuel (RDF) preparation and the other for disposal in landfill. Ferrous and non ferrous metals are recovered by overband magnets and eddy current separators, respectively, whilst plastics are recovered by near infra red optical separation. Waste transfer coefficients for the mechanical separation stage were derived from facility specific data from WasteDataFlow. For biodegradable waste, 28% is sent for composting, 64% is used as RDF, and 8% is sent for disposal in landfill. 100% of metals and plastics are recovered for recycling. 5% of non biodegradable waste is sent for composting and 95% is sent for disposal in landfill (Bernstad and la Cour Jansen, 2011). The biological treatment stage was modelled based the IVC process model described in Section 2.3.6. The MBT process produces a solid, residue called compost like output. Compost like output application was modelled based on the average situation in the UK in 2012, with 3% applied to agricultural land, 80% used as alternative landfill daily cover, and 17% disposed of in landfill (Horne et al., 2013). See Section 2.3.8 for details.

Application to land
Application to agricultural land was assumed to require use of a manure spreader, consuming 0.3 l/t and 0.45 l/t of diesel for compost/compost like output and digestate, respectively (Bernstad and la Cour Jansen, 2011;Boldrin et al., 2009b). The degradation and fate of carbon and nitrogen in solid residues after they have been applied to land was modelled using proxy data from Denmark (Bruun et al., 2006;Hansen, 2006b) and Europe (Smith et al., 2001), presented in Table 3. Solid residues applied to agricultural land were assumed to partially substitute for commercial fertilisers, with the amounts substituted calculated based on the solid residue nutrient content and a substitution rate. Solid residues used in gardening/horticulture were assumed to substitute for peat (Smith et al., 2001). Details of fertiliser and peat substitution rates applied are presented in Table 3.

Material recovery facility
Two material recovery facility (MRF) processes were modelled: single stream commingled materials MRF and residual waste MRF. The single stream commingled materials MRF process comprises a series of mechanical separation activities designed to recover the following product streams: mixed container glass, paper, card, high density polyethylene bottles, polyethylene terephthalate bottles, mixed plastics, plastic film, steel cans, and aluminium cans. A residual waste stream is also produced and sent for landfilling. Electricity and diesel consumption of 35 kWh/t and 2 kg/t, respectively, was assumed for the single stream commingled materials MRF process, based on measured data from UK based MRFs reported by Eunomia (2011). Details of product stream material composition are provided in Appendix D. Due to a lack of specific data, electricity consumption for the residual waste MRF process was assumed, based on Pressley et al. (2015), to be approximately 25% more than that of a single stream commingled materials MRF. Hence, an electricity consumption of 44 kWh/t was assumed. Diesel consumption was based on the single stream commingled materials MRF process. The mass of materials recovered at residual waste MRFs was modelled using facility specific data from WasteDataFlow.

Refuse derived fuel production
The RDF production process was modelled based on a theoretical facility with a gas fired drying stage. Electricity consumption of  40 kWh/t of RDF produced was assumed (Burnley et al., 2011). Process RDF yield was based on facility specific data from WasteDataFlow.

Merchant/exporter
The merchant and exporter processes were modelled using the transfer station process as a proxy.

Incineration
The incineration process was modelled based on an average grate furnace incinerator e the most common incinerator type worldwide (Defra, 2013a) e with dry flue gas scrubbing and de NO x technology, as described by Boesch et al. (2014). Details of energy and material inputs and waste characteristic transfer coefficients for the incineration process are presented in Table 4 and Table 5, respectively. Thermal energy released during the incineration process is recovered via a boiler and a fully condensing turbine and transformed into electrical energy by a generator at an assumed gross generation efficiency of 22% (EC, 2006;Defra, 2013a). A 2% electrical energy loss was included to account for part load operation and heat transfer losses (Consonni and Vigan o, 2011).
Non combustible solid residues (i.e. bottom ash and fly ash) are discharged from the base of the furnace and cooled. Fly ash was assumed to be utilised as a backfilling material, with electricity and diesel consumption of 29 kWh/t and 1.5 l/t, respectively, assumed for the process (Fruergaard et al., 2010). Bottom ash is sent for de scrapping to recover metals for recycling. The de scrapping process was modelled using data from Bosech et al. (2014). 97% of incinerator bottom ash is sent for de scrapping, with the remaining 3% sent to landfill. Recovery rates of 30%, 80%, and 30% were assumed for aluminium, steel, and copper, respectively, with non recovered material sent to landfill. 3 kWh/t of electricity is required for de scrapping process.

Landfill
The non hazardous landfill process (also used as a proxy for inert landfill) was modelled based on an average UK medium sized conventional landfill with gas utilisation. Emissions were modelled over a 100 year time period. Waste material decay rates were taken from IPCC (2006). Energy and material inputs are presented in Table 6. Details of the gas management system and its performance over time are presented in Table 7. Landfill gas recovery efficiencies were averaged over the 100 year time period to better reflect an average tonne of waste deposited, rather than the first mass of waste deposited (see Table 8). Collected landfill gas is utilised in an internal combustion engine to generate electricity, with a gross efficiency of 32% assumed (Patterson et al., 2011). A 1% fugitive CH 4 emissions rate was specified for gas utilisation processes (US EPA, 2011).

Material reprocessing
Reprocessing of different dry recyclables was modelled using the process models and LCI data presented in Turner et al. (in press).

Books and bric a brac.
Books and bric a brac were each assumed to be resold in charity shops. An electricity input at the charity shop of 357 kWh/t was assumed (James et al., 2011). No material loss or avoided production was assumed.

Paint.
No energy or material inputs were included for the reuse of waste paint process. Post reuse, the steel container (200 kg/t) was assumed to be sent for reprocessing. Reused waste paint was assumed to substitute for primary paint production. No material quality loss was assumed. Data for the production of primary paints were adapted for the UK situation from a study undertaken in Abu Dhabi (Nayak and Kumar, 2008). Assuming a paint density of 1 kg/litre, inputs of 33 kWh/t of electricity and 200 kg/t of steel (Saft, 2007) were assumed for the primary paint production process.

Textiles and footwear.
The reuse of textiles and footwear process comprises three stages: 1) sorting; 2) transport to end markets; and 3) reuse. Data for the sorting process were taken from the a UK based textiles recovery plant (EA, 2010). Based on Bartlett et al. (2013), it was assumed that 37% of sorted textiles and footwear are reused domestically, whilst 63% are exported for reuse abroad. Based on the market situation in the year 2008, we assumed that 50% of textiles and footwear exported for reuse abroad are sent to Eastern Europe, 17% to South Asia, and 33% to Sub Saharan Africa (McGill, 2009). Textiles and footwear sent for reuse domestically were assumed to be resold in charity shops (see Section 2.3.15.1). No energy or material inputs were included where textiles and footwear are exported. Reused textiles and footwear were assumed to substitute for the production of textiles from and 0.68 t m À3 , respectively (Smith et al., 2001), and an assumed substitution rate of 1m 3 peat m À3 compost/CLO/digestate . virgin fibres (for details, see Appendix A). Due to their comparatively shorter lifetimes and older fashions, market substitution rates of 60%, 75%, and 85% were assumed for textiles and footwear reused in the UK, Eastern Europe, and South Asia and Sub Saharan Africa, respectively (Farrant et al., 2010;McGill, 2009).

Large domestic appliances.
Waste electrical and electronic equipment large domestic appliances (LDA) were assumed to be reused domestically. However, based on James et al. (2011),i tw a s assumed that 35% of LDA send for reuse are actually recycled, due to insufficient quality. No energy or material inputs were included for the LDA reuse process. Reused LDA were assumed to substitute for the production of primary LDA (for details, see Appendix A). Due to the comparatively shorter lifetimes of reused LDA compared to new LCA, a market substitution rate of 50% was assumed (James et al., 2011).

Life cycle impact assessment
GHG emissions were characterised by Global Warming Potential (GWP) using a 100 year time period (expressed as t CO 2 e), with characterisation factors taken from the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (Forster et al., 2007). The emissions "savings" from biogenic carbon sequestration in landfills or soils (see Christensen et al., 2009) were excluded. This assumption was investigated as part of the sensitivity analysis.

Material flow analysis
Comprehensive material flow diagrams for the LACW management systems of each scenario (i.e. the results of the mass balance of waste) are presented in Appendix C. For S1, the baseline scenario, the dominant primary waste stream was residual waste (82,079 t), followed by commingled materials (42,065 t). 63,518 t of waste from the residual waste stream was sent directly to landfill. Of the remaining waste, 18,562 t was sent to one of three residual waste MRFs, with 9,829 t sorted and sent for reprocessing (the remainder was sent to landfill), of which 7,843 t was reprocessed into a secondary product; an overall recycling rate of 42.3% (note that rates varied between facilities). By comparison, all commingled materials were sent to single stream commingled MRFs, with 34,871 t recovered and sent for    reprocessing and 27,920 t ultimately reprocessed into a secondary product. This represents an overall single stream commingled MRF recycling rate of 80.0% (note, again, that rates varied between facilities). These results support those of Pressley et al. (2015), who found that recovery rates at single stream commingled MRFs were higher than those at residual waste MRFs. This suggest that policies aiming to increase recycling rates should focus on improving sorting efficiencies at the waste producer level (e.g. households), rather than on recovering materials from the residual waste stream. Table 9 displays a selection of key indicator results from the MFAs for each scenario. For S1, a reuse/recycling/composting/AD (henceforth, recycling rate) of 49.8% was calculated for the baseline scenario, marginally higher than the 52.8% value reported by Cardiff County Council (StatsWales, 2015b). However, where rejects from reprocessing and biological treatment are included the final (adjusted) recycling rate was estimated as being 45.2%, highlighting the importance of considering downstream reject rates when calculating recycling rates. The recycling rate was highest for S2 (enhanced food waste capture), where high rates of food waste diversion were achieved, whilst the recycling was also higher for S4 (enhanced dry recycling) compared to S1 due to the increased recovery of dry recyclables. For S1, over half of LACW was ultimately landfilled, with 37.7% of LACW sent directly to landfill and 16.4% rejected downstream (Table 9). S3 (enhanced incineration) was found to result in the highest heat and power recovery rate, whilst it was the most successful scenario in terms of diverting material from landfill.
The results of the MFA suggest that it may not be possible for Wales to achieve its recycling rate target of 70% by 2025 given current waste arisings, even if optimum dry recycling and food waste diversion rates are achieved. Rather, it is likely that significant reductions in residual waste arisings will be required, as well as improvements in recycling rates (Timlett and Williams, 2011). Wales have already made progress in this respect, with a steady rise in the national LACW recycling rate from 37.5% in 2008e2009 to 52.3% in 2012e2013, accompanied by a 9.9% reduction in LACW arisings over the same period (StatsWales, 2015a). However, reductions in total waste arisings have stagnated in recent years. In order for Wales to continue progressing towards its recycling targets, further reductions in waste arisings are likely to be required. Fig. 2 shows the total GHG emissions per scenario and the contribution by primary waste stream. S1, the baseline scenario, was found to result in a net GHG burden (8009 t CO 2 e). Note that positive values represent an environmental load (i.e. GHG burden), whilst negative values represent an environmental saving (i.e. GHG benefit). GHG benefits were observed for the management of commingled materials (À9410 t CO 2 e), source segregated dry recyclables (À3212 t CO 2 e), and source segregated food waste (À2596 t CO 2 e). However these benefits were heavily outweighed by the GHG burdens from residual waste management (21,778 t CO 2 e).

Evaluation of the existing LACW management system
Contribution to total primary waste stream GHG impacts by process type for each scenario is shown in Fig. 3. For S1, the dominant contribution in terms of GHG burdens was landfilling of residual waste (25,866 t CO 2 e). Material reprocessing & reuse was the only process type that consistently resulted in GHG benefits. For S1, the most significant source of GHG benefits was from reprocessing of commingled materials (À19,219 t CO 2 e). The contribution of biological treatment to total GHG impacts was found to be negative for the source segregated food waste stream (i.e. GHG benefit) but positive for the source segregated garden waste stream (i.e. GHG burden). This is due to the different types of biological treatment used to manage each waste stream (see Appendix C). Food waste was treated predominately by AD, a process that utilises produced biogas to generate electricity that is exported to the National Grid and substitutes for electricity generated from CCGT, resulting in GHG benefits. By comparison, garden waste was largely treated by composting (either IVC or OWC), through which energy cannot be recovered. This finding correlates with those of Yoshida et al. (2012) and Bernstad and la Cour Jansen (2011) who each found that AD of organic waste resulted in a greater GHG benefit compared to composting.
Overall, transport was identified as the third largest contributor to the total GHG impacts for S1 ( Fig. 3; 7296 t CO 2 e; 10% contribution). However, its relative contribution to the GHG impacts of each primary waste stream management system was highly variable. Whilst the contribution of transport to the GHG impacts from the residual waste (1492 t CO 2 e; 4% contribution) and source segregated dry recyclables (160 t CO 2 e; 3% contribution) management systems was minor, transport contributed substantially to the GHG impacts from source segregated garden waste management (565 t CO 2 e; 24% contribution). This is likely due to the considerable distances that garden waste was transported for treatment. For example, 16,780 t of collected garden waste was transported between 78 and 181 km to an IVC facility, resulting in substantial GHG burdens. The contribution of transport to the total GHG impacts of commingled materials management was also found to be large (4543 t CO 2 e; 16% contribution). This is likely due to the large quantity of recyclate that was exported overseas for reprocessing. Of the 42,055 t of commingled materials collected, 11,334 t Table 9 Comparison of MFA indicator results for the four scenarios. Note that S1, S2, S3, and S4 refer to Scenarios 1e4, respectively.

Indicator
Description Scenario S1 S2 S3 S4 (predominately paper) was exported to reprocessing facilities in Asia (see Appendix C). These results suggests that the calculated GHG impacts from materials recycling in this situation are more greatly affected by transport distance compared to results from previous studies Merrild et al., 2012;Salhofer et al., 2007). The formulation and implementation of policy measures that promote domestic reprocessing of waste materials may help to reduce the GHG impacts of transport and, consequently, enhance the overall GHG benefits from materials recycling. Table 10 presents the contributions of wastes from different collection sources to the total GHG impacts for each scenario. For S1, the largest contribution to the total GHG impacts was from household kerbside collected waste, particularly residual waste, which was the most significant source of GHG burdens, and commingled materials, the largest source of GHG benefits. Overall GHG impacts from bulky waste collection, fly tipped materials, the voluntary sector, and municipal parks/grounds waste collection were found to be minor, with each contributing less than 1% to the total GHG impacts. An interesting result from a policy perspective concerns the contribution of street cleaning waste management to the total GHG impacts. Street cleaning waste management was found to contribute 8.8% (751 t CO 2 e) of total GHG impacts for S1. Waste from street cleaning comprised both residual waste, the management of which incurred a GHG burden of 2269 t CO 2 e, and commingled materials, the management of which resulted in a GHG saving of À1518 t CO 2 e. Diversion of street cleaning waste from the residual waste stream by promoting recycling is an area not commonly targeted by policy makers, but could be an effective source of potential GHG benefits.

Comparison of scenarios
Compared to S1, each of the three alternative scenarios were found to result in lower overall GHG burdens (Fig. 2). S2 was found to be the best performer with an overall GHG impact of À1930 t CO 2 e, the only scenario for which a net GHG benefit was found. The contribution of household kerbside collection residual waste to total GHG impacts was substantially reduced in S2 compared to S1 (Table 10), with a large portion of the GHG burdens of residual waste management replaced by GHG benefits from the management of source separated food waste. This highlights the considerable influence of the organic content of residual waste on total GHG burdens (see Section 3.2.3.2).
After S2, S3 was the next best performer (1046 t CO 2 e; Fig. 2). S3 showed the greatest reduction in GHG burdens from landfill compared to S1 (Fig. 3). For S3, the substantial GHG burdens from landfill were replaced by comparatively lower GHG burdens from incineration (Fig. 3), suggesting that incineration represents a better option with regards to GHG impacts compared to landfill. Table 10 LCA results (GHG emissions, t CO 2 e) and contribution analysis (%t CO 2 e=FU) of different scenarios per collection source and primary waste stream. Note that S1, S2, S3, and S4 refer to Scenarios 1e4, respectively. This compares favourably with the findings of previous studies (e.g. Arena et al., 2003;Fern andez-Nava et al., 2014;Massarutto et al., 2011). The GHG benefits of S3 would likely be further enhanced if district heating networks were more widely established in the UK. This would create markets for thermal energy recovered through incineration, leading to further GHG benefits through avoided thermal energy production (Giugliano et al., 2011;Turconi et al., 2011). Of the three alternative scenarios, S4 was found to result in the improvement on the results of the baseline scenario, with a total GHG impact of 5548 t CO 2 e ( Fig. 2). This is likely due to the relatively high dry recyclables sorting efficiencies that are already achieved in Cardiff, particularly for mixed glass, paper, and dense plastics. Considering the results for S2, this finding also suggests that waste policies that target residual waste prevention may be more effective in achieving GHG emissions reductions compared to those that target greater recovery of dry recyclables.

Scenario analysis
3.2.3.1. Carbon sequestration. A sensitivity analysis was performed to investigate the impact of including carbon sequestration on the LCA results through scenario analysis. For each scenario, the amount of carbon sequestered was calculated as the amount of biogenic carbon that, after 100 years, remains in a landfill or, in the case landspreading of organic solid residues from biological treatment, is bound to soil. Fig. 4 shows the total GHG impact for each scenario and the contribution by process type. The inclusion of carbon sequestration had a substantial impact on the calculated results. Overall, net GHG benefits were observed for each of the four scenarios. Landfilling was found to result in overall GHG benefits, effectively rendering landfills a "carbon sink". This contradicts the findings of the UK's national GHG inventory return for GHG emissions from SWM (Salisbury et al., 2015), despite the fact that carbon sequestration is also considered in the national inventory analysis (Hogg et al., 2011).
The ranking of scenarios in terms of their environmental performance was affected by the choice of including carbon sequestration. Whilst S2 remained the best option and S4 was still found to result in additional GHG benefits compared to S1, the overall GHG burdens for S3 were found to be considerably higher than S1 (Fig. 4). This suggests that when carbon sequestration is included in the calculations, landfilling is generally a better option compared to incineration in terms of GHG impact, which contradicts the results of previous studies (e.g. Arena et al., 2003;Fern andez-Nava et al., 2014;Massarutto et al., 2011). Given the significant effects of including carbon sequestration in the LCA on the calculated results, these authors suggest that further research into carbon degradation rates for waste materials in UK landfills is needed.

Residual waste composition.
A sensitivity analysis using scenario analysis was performed to test the influence of the modelled organic content of the residual waste stream on the overall LCA results. In this study, residual waste composition data were taken from a national waste composition study (WRAP, 2010) that predates the introduction of separate household food waste collection in Cardiff. The use of these data may, therefore, result in an overestimation of GHG burdens as the total organic content of the input waste may be overestimated. Two scenarios were analysed, with the proportion of organic material in residual waste increased and decreased by 10% compared to the baseline values for each scenario. Compared to the total GHG impact of S1 (8008 t CO 2 e), the 10% increase in organic content resulted in a þ15.5% variation (result: 9254 t CO 2 e), whilst the 10% decrease resulted in a À9.4% variation (result: 6826 t CO 2 e). The sensitivity analysis shows that the overall LCA results are highly sensitive to the organic content of residual waste.

Conclusions and recommendations
In this paper we have presented an approach that combines the systematic methodologies of MFA and LCA to quantitatively evaluate the environmental performance of large and complex SWM systems. The approach was applied to evaluate the GHG emissions performance of a local authority SWM system and compare its performance with those of alternative systems to assess the potential effectiveness of different waste policy measures. The results of the MFA show that S2 (enhanced food waste capture) resulted in the highest recycling rate of the investigated scenarios, whilst S3 (enhanced incineration) resulted in the highest rate of waste diversion from landfill. Overall, our results suggest that, in this case, the national government is unlikely to achieve its recycling targets, even if each of the assessed policy measures are implemented optimally. It is likely that in order for the recycling targets to be achieved, these policy measures need to be combined with further policies focused on waste prevention. Fig. 4. Sensitivity analysis (SA1) results (GHG emissions, t CO 2 e/FU) of different scenarios and the contribution by process type e the inclusion of carbon sequestration. Note that SA1.1, SA1.2, SA1.3, and SA1.4 refer to Scenarios 1e4, respectively. GHG, greenhouse gas; FU, functional unit.
The GHG impacts of the existing SWM system were evaluated through LCA, focusing on potential climate change impacts. The results showed that the dominant source of GHG burdens was from residual waste management, principally through landfilling, which was the most significant source of GHG burdens. Material reprocessing and AD were found to be the greatest sources of GHG benefits. GHG burdens from transport were found to be generally minor, except with regards to the management of source segregated garden waste and commingled materials.
Compared to the baseline scenario, each of the three alternative scenarios were found to perform better in terms of GHG impact. Overall, the best performing scenario was S2 (enhanced food waste capture), which was the only scenario that resulted in an overall GHG saving. However, the results of the LCA were found to be highly sensitive to the choice of excluding GHG benefits from carbon sequestration. When carbon sequestration was included, all four scenarios were found to result in net GHG benefits. The order of the four scenarios in terms of overall performance was also found to be affected by the choice, with the GHG burdens for S3 (enhanced incineration) found to be greater than those of the baseline scenario. The results were also found to be sensitive to the organic content of the residual waste stream.
Overall, this paper has demonstrated that the complementary methodologies of MFA and LCA can be used in combination to provide valuable information about the environmental performance of a SWM system. The approach can be applied to assist national governments in appraising existing and possible waste policies and to support local waste managers in identifying environmentally optimal SWM strategies.