The Environmental Impact of Collecting and Processing Abandoned Shopping Trolleys in the UK

Abstract

Significant numbers of shopping trolleys are taken out of service through abandonment each year, leading to negative environmental, economic, and social impacts. These impacts have been discussed qualitatively in the literature; however, quantitative analysis of the environmental impacts is lacking. We performed a life-cycle assessment (LCA) of the manufacturing and subsequent abandonment, collection through diesel vans, and refurbishment of shopping trolleys in the Canley suburban area of Coventry city, UK. We found that manufacturing one shopping trolley had a global warming impact of 65.14 kg CO₂ equivalent. The impact of collecting one trolley and returning it to the originating supermarket was 0.69 kg CO₂ equivalent. Finally, the impact of transporting and refurbishing one trolley was 5.50 kg CO₂ equivalent. These results show the major benefits of either collection or refurbishment over manufacturing new trolleys to replace abandoned ones. An observation is made that one trolley would have to be collected 93 times to offset its own manufacturing impact. The work also highlights the impact of abandonment. An estimated 520,000 abandoned trolleys in the UK every year lead to 343 tons CO₂ equivalent through van collection only. If 10% of these trolleys required refurbishment, the impact would increase to 652 tons CO₂ equivalent.

1. Introduction

Tackling the issue of shopping trolley abandonment addresses the principles of the circular economy, in that valuable resources are maintained and used for as long as possible. This study seeks to evaluate the environmental impact of collecting and processing abandoned trolleys and subsequently returning them into use. The work focuses on a specific case study in the city of Coventry; however, the findings can be applied more broadly to other suburban areas.

The number of abandoned trolleys is not insignificant. In 2017, 520,000 trolleys were reported as abandoned in the UK [1]. In Sunderland alone, 30,000 trolleys were reported as abandoned between 2020 and 2022 [2]. Likewise, 550 trolleys were collected in a single day in Western Sydney, Australia [3]. It is estimated that a well-maintained shopping trolley could have a lifespan of 15 years; however, with no action taken to recover abandoned trolleys, retailers could have to replace their trolley fleets every five years, leading to significant financial costs [4]. Indeed, it has been reported that, in South Africa, trolley theft has cost one business ZAR 500,000 over three years and another business ZAR 1,000,000 annually [5].

Shopping trolley abandonment has negative environmental, social and economic impacts. Previous studies have identified potential impacts including safety risks to pedestrians and motorists, as well as the contribution trolleys make to flooding by trapping floating debris and waste in canals and rivers [3,6]. Negative social aspects include a loss of visual amenity and loss of safe recreational spaces. The economic aspects include damage to boat propellers in canals, personal injury to pedestrians, and the cost to retailers of collecting and maintaining trolleys.

Supermarkets have explored a range of options to reduce trolley theft with varying degrees of success. Examples of deterrents include coin slots, vertical bars (to stop trolleys leaving the shop floor), wheel-locking mechanisms, and car park wardens.

Williams and Deakin [6] conducted a survey of abandoned trolleys in the River Tawd, Skelmersdale, UK, and correlated the number of abandoned trolleys with the deterrent mechanisms used by local supermarkets. It was found that the most effective mechanism was vertical bars, with the fewest such trolleys observed in the river. The number of abandoned trolleys from supermarkets with car park wardens was observed to increase across the study period, indicating the partial effectiveness of wardens. Coin-release trolleys were also observed abandoned in the river Tawd, indicating the minor deterrent of financial loss to the customer. Finally, most of the trolleys surveyed were associated with a supermarket with simple collection points outside the building. These findings were echoed by a social survey conducted by the authors, with survey respondents viewing physical barriers and parking attendants as the most effective and coin-operation as the least effective. Some research has investigated radio frequency tracking methods for locating abandoned trolleys, although the practical application of this technique is likely to be expensive and impractical [7].

Supermarket trolleys are specifically designed to assist the customer in collecting products in store and transport them to the owner’s vehicle. Why then are trolleys taken out of their designed use phase? A social survey was conducted [6] in 2002 in North-West England to understand the reasons behind trolley abandonment.

The first survey targeted pedestrians using the local supermarkets and consisted of 82 respondents. The survey showed clearly that 8 out of 10 survey respondents did not return trollies to the shop or designated car park return points. The main reason for this was that “that the collection point was too far away from their vehicle after shopping had been loaded” [6]. The survey respondents also identified young people and teenagers as being the main culprits for placing trolleys in the local river.

The second survey was distributed to all year seven students at a local school. Overall, 27% of the school respondents cited “boredom” as the reason for taking trolleys and 13% were stated to use them as playthings.

While such a survey is admittedly limited to a specific time and location, it does give some insight into the varied reasons for trolley theft and abandonment.

Reiersgord [5] gave a socio-economic perspective on trolley abandonment in South Africa. It was shown that trolleys played an important role for business “outside the formal economy in South Africa”, where traders use trolleys to transport goods and can earn between ZAR 100 and ZAR 300 per day for such activities [5]. While not pertinent to the case study presented in this study, it does offer further insight into global socio-economic reasons for trolley theft and abandonment.

With these insights into the effects and reasons for abandonment given, the question remains as to what can be done to address the issue.

In the UK, the Department for Environment, Food and Rural Affairs (DEFRA) has provided guidance for local councils [8], with an emphasis on involving local retailers and trolley providers before enacting powers for collecting and disposing of trolleys. Sunderland City Council has introduced GBP 100 fixed-penalty notices for “anyone caught removing a shopping trolley from retailers’ premises” [2]. In a landmark court ruling, a UK supermarket was fined GBP 31,517 for knowingly allowing trolleys to be dumped in the River Chelmer in 2000 [9].

The UK Environment Protection Act 1990 gives powers to local authorities to collect and charge owners for trolleys’ return. Williams and Deakin [6] criticised the 1990 Act due to a legal loophole in not being able to identify the original owner of the trolley, thus reducing the effectiveness of the deterrent.

The majority of the literature found on shopping trolleys centres around user experience or “smart” trolleys [10,11,12,13,14]. While no studies were found to investigate the collection and refurbishment of trolleys, one study by Val and Lamban [15] compared the environmental impact of the manufacturing and landfill disposal of steel and polymer shopping trolleys.

While detailed modelling of the manufacturing stages was presented, the technological representativeness and sources of information from which these manufacturing stages were modelled were not clear. The study gave an interesting insight, in that the steel trolley manufacturing phase showed a 40% higher environmental impact than the polypropylene trolley manufacturing phase. This was attributed to the material extraction and processing phases. Specifically, Val and Lamban identified the use of chrome-plated steel as contributing the largest environmental impact. Since the steel-processing and chrome-plating processes were amalgamated into one process, the relative contribution of steel processing and chrome plating was not stated. The study presented here aims to address this by segregating the impact of the steel-processing and coating operations.

With previous studies mainly focusing on quantifying the number and type of abandoned trolleys, as well as the reasons behind why such abandonment occurs, there is a value to contribute to this field of study by quantifying the environmental impact of handling abandoned trolleys. An understanding of the magnitude of the environmental impacts of trolley abandonment could guide future decision making for driving change in public behaviour.

This study presented here uses a methodology standardised by the International Standardisation Organisation (ISO), known as life-cycle assessment (LCA), to analyse the potential environmental impact of collecting and handling abandoned shopping trolleys within a specific area of the City of Coventry, UK. The study aims to quantify the environmental impact of trolley abandonment, providing inputs to policymakers and shop owners, as well as guidance to the general public.

2. Materials and Methods

The presented LCA study followed the ISO 14040 [16] and ISO 14044 [17] standards. The standards specify four phases: objective and scope definition, the life-cycle inventory (LCI), the life-cycle impact assessment (LCIA), and the interpretation of results. The interpretation phase also consists of a sensitivity check to analyse the sensitivity of results to changing model parameters, such as trolley mass.

2.1. Case Study Definition

The Cannon Park Shopping Centre and surrounding suburban area within the City of Coventry, UK, was chosen as a case study. The shopping centre contains a 6020 m² [18] “superstore” supermarket serving the local populace, as well as the local student population of the neighbouring university.

Retail establishments such as Cannon Park make use of specialist trolley collection services, for example Wanzl TrolleyWise [19] or TMS Collex [20]. These services, henceforth referred to as “commercial collection services” or CCS, offer a downloadable app, which the public can use to notify the CCS drivers of abandoned trolleys. The drivers collect the reported trolleys and return them to the retail establishments, or for refurbishment, if required.

The local university Estates Department also collects abandoned trolleys around campus, stores them in a central holding pen, and notifies the CCS for collection. The distances and transport modes used in this study were obtained through personal communication [21] with parties involved in real-world day-to-day trolley collection, thus giving valuable insight into realistic trolley collection patterns.

Approximately 20 trolleys are collected and stored in the central holding pen each week by Estates, with a weekly mass collection from the holding pen by the CCS. The Warwick University Estates department uses a diesel Transit Tipper 350 RWD 2.0 TDCi with a tail lift to collect the trolleys. The diesel van collects abandoned trolleys as part of its daily movements around campus. The CCS also use a similar vehicle, confirmed by personal communication.

The university has a road network encompassing student accommodation and through-roads of approximately 9.88 km, as measured using an online mapping website [22]. The surrounding suburban area of Canley has a road network of approximately 11.01 km within a 1 km radius of the shopping centre, as measured using the same online mapping website. A radius of 1 km was chosen by the author as a reasonable distance that a member of the public might walk with a shopping trolley. These stated distances are used to model the trolley collection van use phase. It should be acknowledged that further empirical observation is required to validate this assumption, as a certain number of trolleys may be taken beyond the 1 km radius.

The collection van driving phase was modelled using the GaBi 10.7.0.183 [23] process “RER: Mass-induced fuel consumption of automotive part, (NEDC)”. This process considers the use phase of an unspecified automotive part. In this case, the mass of the collection van plus the trolley mass was assigned as a single part to the “unspecified automotive part” as a proxy.

The GaBi process uses the coefficient for the reduction in fuel consumption (CRFC). This determines fuel savings as a function of weight reduction and is essentially the gradient of the relationship between vehicle fuel consumption and vehicle mass.

The CRFC is usually determined according to the fuel consumption of various vehicles by applying their weight on regression curves and is given as x litres/100 km reduction in fuel consumption per 100 kg of vehicle weight saved. The GaBi process utilises a CRFC value of 0.12 litres/(100 kg × 100 km) for diesel fuel. This value corresponds to the findings of Koffler [24].

As mentioned previously, CCS also offer trolley refurbishment services. Wanzl employs their Reviva refurbishment operation [25] on trolleys. The steel trolleys are stripped of coatings until the base material is exposed. The trolleys are then put through repair, cleaning, adhesion, and anti-oxidisation treatments before being made available to supermarkets.

TMS employs their Trollex refurbishment operation [26]. This process is thoroughly detailed; at the process depot, the trolleys are stripped of components (wheels, handlebars, etc.) and the steelwork is repaired (weld repair). The trolleys are put through a pyrolysis process to remove the existing zinc coating and lacquer. Finally, the trolleys are electro-plated with zinc and have a lacquer applied to protect the zinc coating. The present study models the environmental impact of pyrolysis and the reapplication of the protective zinc coating. Due to data uncertainty, the disassembly process and application of lacquer were not included in the model.

A generic 210-litre shopping trolley was assumed in this study, with a construction of mild steel wires spot welded into a basket shape with mild steel tubes forming the chassis, as shown in Figure 1. The wheels and handle are omitted from the considerations due to a lack of data. The trolley weight was approximated to be 22.3 kg [27]. Production occurs in Madrid, Spain [28]. Due to GaBi database constraints, the “EU27” country-based databases were used to model manufacturing processes whenever “ES”-based databases were not available; however, transport distance was modelled to be between Madrid and Coventry.

2.2. Goal and Scope

The goal of the present study was to evaluate the environmental impact of collecting and processing abandoned shopping trolleys around the Canley and Warwick University area of Coventry, UK, by conducting a hot-spot analysis.

The study consisted of the following life-cycle stages:

• Manufacturing stage: upstream processes of raw materials, consisting of zinc-coated mild steel for the fabricated frame and chassis, as well as welding and forming operations.
• Collection stage: use of a collection van, accounting for impacts through diesel fuel consumption during trolley collection around the University and Canley suburban area.
• Transport and refurbishment stage: accounting for the transport of trolleys to and from Coventry city to the refurbishment facility in Tibshelf, where the trolleys undergo pyrolysis and the reapplication of the protective zinc coating.

The function of a shopping trolley is to transport products around a store and to the customer’s vehicle. The reference flow (which is the amount of product required to meet the function) is one shopping trolley.

This study chose to define the functional unit as “one processed shopping trolley”. This choice of functional unit was motivated by an analysis of other transport-related studies and how they differ from the analysis presented in this study. In a study focused on mining trucks, Balboa-Espinoza [29] utilised a functional unit of “one ton of material transported over 1 km”. Additionally, Balboa-Espinoza analysed recent environmental studies of different fuelled vehicles and concluded that the common functional unit was “1 km travelled”. It was determined that these functional units would not sufficiently cover the scope of the current study. Given that it is a function of performance characteristics (the function of the product under consideration), coupled with the fact that the performances of two different products are not compared, the functional unit of “one processed shopping trolley” is logical. An alternative functional unit may have been “kilometres driven per processed trolley”, but this would exclude the manufacturing and refurbishment stages.

2.3. Life Cycle Inventory

The Life-Cycle Inventory (LCI) consists of an inventory of all relevant flows in the system boundary. Use was made of the GaBi life-cycle inventory, which contains data on energy production, use, and distribution. Other data include raw material extraction and transport. The GaBi software provides supporting documentation to its datasets and references to the source of the data. The GaBi database inventories are mainly based on industry data and are therefore considered up to date and technologically representative. The steel wire and seamless pipe data inventories were collected on site by steel industry experts. Energy-grid mixes are also stated to be country specific, providing a more representative impact calculation from energy use in manufacturing and transport. Data that were not available through the GaBi database were collected from the literature.

The LCI model was divided into the three stages shown in Figure 2. All processes were based on the GaBi databases, with the following exceptions: transport distances were based on online secondary sources [22,28], the pyrolysis energy requirements were based on online secondary sources [30], and the number of abandoned trolleys was established through personal communications with the University Estates department.

The foreground system, encompassing primary data collected for the study, consisted of driving cycles and the numbers of trolleys collected. The background system consisted of raw material extraction and processing (including galvanising), manufacturing operations (bending and welding), and diesel fuel production.

It was determined that the production and use of a shopping trolley has no secondary functions relevant to other systems of the technosphere with which the defined system interacts. Therefore, the modelling framework did not account for joint production or multifunctional processes.

2.4. Cut-Off Criteria

Cut-off rules are given to justify omitting less relevant processes from the LCA study. The cut-off rules in GaBi are such that the coverage of the processes are “of at least 95% of mass and energy of the input and output flows, and 98% of their environmental relevance (according to expert judgement)” [23].

2.5. Life Cycle Impact Assessment

The Life-Cycle Impact Assessment (LCIA) was informed by previous transport-related LCA studies [29,31,32]. The ReCiPe Midpoint method [33] was selected due to its coverage of midpoint indicators. A midpoint method evaluates environmental impacts earlier along the cause–effect chain than endpoint methods. Endpoints methods are generally considered to have a higher level of statistical uncertainty than midpoint methods [34]. The hierarchist perspective for the characterisation model was selected, using a 100-year time horizon for climate change that falls between the Individualist (20-year time horizon) and Egalitarian (1000-year time horizon) perspectives.

While it is acknowledged that the environmental impact of the trolley extends beyond the selected impact categories detailed below, the focus of the study is on the impact of trolley collection. It is the authors’ opinion that aligning the stated impact categories to previous transport-related studies would provide enhanced focus on the study aims.

The impact categories selected for this investigation include the well-known Climate Change (characterised as Global Warming Potential (GWP)) category, used in transport related studies [31]. Fossil Depletion (FD) and Metal Depletion (MD) capture the impact of extracting minerals and fossil fuels. Fine Particulate Matter Formation (FPMF) captures the impact of air pollutants through vehicle use and is used in transport-related studies [35]. Photochemical Ozone Formation (POF) assesses the impact of photochemical ozone pollution through vehicle use.

The impact of water pollution was accounted for through Freshwater Eutrophication (FE) and Marine Eutrophication (ME).

An alternative to the proposed methodology would be to use the endpoint method, which indicates damage in only three categories; Damage to human health, Ecosystem damage and, finally, Resource damage. Presenting results in these three damage categories would provide an easy-to-understand “big picture” of the environmental damage caused by trolley abandonment, especially if presented to non-experts such as the public and industry. A case could thus be made that the endpoint method would aid in public education about trolley abandonment. The aim of the present work, however, is to conduct a hot-spot analysis and dive deeper into the specific causes of environmental impacts. Therefore, the present study uses the midpoint method but may be adjusted in future to present results using the endpoint method for public stakeholder engagement.

2.6. Environmental Sensitivity

The environmental sensitivity of the defined systems was evaluated by modelling an increase in trolley mass of 10%, 30%, 50%, and 100%. The range of theoretical mass increase was chosen by the author. An arbitrary range of mass increase was selected with the main purpose of establishing whether the resulting impacts showed a linear or non-linear response to mass variation.

The resulting relationships between environmental impacts and an increasing trolley mass gave a linear relationship for all environmental impacts. Having established a linear relationship, further results were evaluated by modelling an arbitrary representative deviation of ±10% in trolley mass. The variation in trolley mass was used as a proxy to investigate the impact of collecting more or fewer trolleys (vans transporting a different weight) as well as number of trolleys manufactured.

2.7. Modelling Assumptions and Limitations

The steel wire and pipe material were modelled using the “EU: Steel wire rod worldsteel” and “EU: Steel seamless pipe worldsteel” processes. Both databases are based on data collected by industry experts on site. The data contained in these processes are based on the average site-specific data of steel producers in the EU. Therefore, they are not specifically country representative, but nevertheless they are based on European data and are therefore deemed relevant.

Trolley manufacture is included in the model. A shopping trolley is made from an array of 16 wires positioned orthogonally to 28 wires. This array is then spot welded, as shown in Figure 3. GaBi utilises a “resistance spot welding” process, which is modelled on creating a 1 mm spot weld. The welding process is scaled to create 448 welds. The utilised “GLO: Steel sheet (1 mm) spot welding Sphera <u-so>” process is based on a single electrode creating a spot weld, whereas Figure 3 indicates electrodes with more surface area used. It is beyond the scope of the current investigation to identify the environmental effect of this difference. The process is globally representative and is therefore not specific to the European region of application.

The 3.3 mm diameter mild steel wires are approximated to cover a flat area of 0.0015 m². This area approximation was used to model the metal bending operation in GaBi, which utilises a “1 m² sheet metal bending” operation. This process was thus scaled to account for the trolley’s material dimensions. It should be acknowledged that the geometrical difference between a continuous sheet and separated wires would affect the necessary input energy to provide the bending operation. Capturing the effect of this deviation is beyond the scope of the current investigation.

Due to the uncertainty of the data, manufacturing aspects such as bending operations of the tubular chassis and joining the chassis to the basket were omitted. The fabrication and maintenance of buildings and machines were omitted due to data uncertainty.

Fuel use for truck transport was captured through use of the German-based “DE: Diesel mix at filling station Sphera” and British-based “GB: Diesel mix at filling station Sphera” for the manufacturing and collection life-cycle phases, respectively, thus giving greater geographical representativeness to each use phase. It should be acknowledged that the manufacturing phase is based in Spain and thus the German-based fuel database deviates. Nevertheless, it is still based in Europe and keeps a relatively good representativeness.

For the collection van use phase, we took the approach of calculating the effect of trolley mass on the fuel use of the diesel vans, as previously discussed with regards to CRFC in Section 2.1. This aspect of the analysis was performed using the GaBi “RER: Mass-induced fuel consumption of automotive part (NEDC)” process. The process documentation states that “impacts are based on the fuel consumption of the entire car scaled to the mass of the respective part”, with the “part” being one shopping trolley plus the van.

The use of this process does introduce some approximation, in that a passenger car is modelled and not a collection van. Diesel fuel is, however, used as it correlates to real-world trolley vans. Additionally, the mass of the collection van (3500 kg) is included in the use-phase modelling.

This process is modelled on the vehicle driving 150,000 km; hence, all results had to be scaled to appropriate distances driven around the university and suburban areas.

The van use phase is approximated as follows: the university Estates van travels 9.88 km daily for six days a week, collecting 3.33 trolleys per day, and then it returns the trolleys to a central holding pen. These values were based on a personal communication with Warwick University Estates.

The commercial collection service (CCS) van has two use phases; the first is concerned with driving around the suburban area around Cannon Park (11.01 km twice a week) and collecting five trolleys per trip. The second commercial collection service (CCS) van use phase is a single weekly trip from Cannon Park to the university Estates holding pen and back again to Cannon Park (5.66 km total distance, transporting 20 trolleys).

For the refurbishment phase, it is assumed that trolleys are transported in a CCS van from Coventry to Tibshelf, and back again, twice per year, with 50 trolleys per trip, equating to a total distance of 220 km.

The above driving scenarios are illustrated visually in Figure 4 below, with each line style representing a specific driving scenario.

The GaBi database did not contain relevant galvanising datasets. In accordance with ISO 14044, subdivision was used to separate the environmental impact of steel production and the galvanisation process. The two processes shown in the bullet points below are equivalent in that they are cradle-to-gate processes, accounting for raw material extraction and the main production steps in an integrated steel plant.

The environmental impact of hot-dipped galvanisation was thus calculated by subtracting the environmental impact of the “uncoated steel sheet production” process from that of the “hot-dipped galvanised steel sheet production” process.

• Uncoated steel sheet production: “DE: Steel sheet cold rolled coil 1.5 mm”.
• Hot-dipped galvanised steel sheet production: “DE: Steel sheet 1.5 mm HDG”.

The GaBi database did not contain relevant pyrolysis data, thus requiring a custom process to be created. The thermal properties of zinc were obtained from the website of a supplier of high-purity metals for research and development applications [30]. The latent heat of fusion (100.9 kJ/kg [30]) and the latent heat of vaporisation (1782 kJ/kg [30]) of zinc was used to approximate the energy required to remove the galvanised coating. A mean galvanised coating thickness of 70 μm was assumed, in accordance with the relevant ISO standard [37]. A limitation of this approach is that the energy required for initial furnace heating and subsequent heat loss is not accounted for. The effect of this may be that real-world environmental effects are greater than those reported in this study.

Table A1. List of processes used in GaBi modelling software.

Life-Cycle Stage	Process Name	Function/Purpose
Manufacturing	EU: Steel seamless pipe world steel	Trolley chassis
	EU: Steel wire rod world steel	Trolley wire basket
	DE: Steel sheet deep drawing (adjustable) Sphera <u-so>	Bending process approximation
	GLO: Steel sheet (1 mm) spot welding Sphera <u-so>	Welding operation
	ES: Electricity grid mix Sphera	Power input for manufacturing
	DE: Diesel mix at filling station Sphera	Fuel for truck transport
	GLO: Truck, Euro 6 A-C, 7.5 t–12 t gross weight/5 t payload capacity Sphera <u-so>	Truck transport from Spain to UK
CCS Collection	GB: Diesel mix at filling station Sphera	Fuel for van transport
	RER: Mass-induced fuel consumption of automotive part (NEDC) Sphera <u-so>	CCS van transporting trolleys within Coventry city
Refurbishment	GB: Diesel mix at filling station Sphera	Fuel for van transport to refurbishment facility (Tibshelf) and back to Coventry city
	RER: Mass-induced fuel consumption of automotive part (NEDC) Sphera <u-so>	Truck transport of trolleys to refurbishment facility and back to Coventry city
	GB: Electricity grid mix Sphera	Power input for pyrolysis
	DE: Steel cold rolled coil 1.5 mm	Process 1 of 2 for subdivision
	DE: Steel sheet 1.5 mm HDG	Process 2 of 2 for subdivision

in Appendix A lists all the relevant processes used for software modelling.

3. Results

3.1. Manufacturing Phase

The environmental impact of the trolley manufacturing phase, including transport from the Spanish manufacturing facility to the UK, is shown in Figure 5. The error bars indicate the effect of a ±10% in trolley mass, with the lower error bar indicating a 10% decrease and the upper error bar indicating a 10% increase in trolley mass. The hot-dip galvanisation data does not include error bars, due to lack of data. A more quantitative interpretation of the results is provided in

Table 1. Percentage contribution of each manufacturing process to the respective environmental impact categories.

		% Contribution to Total Respective Impact Category
	Total	Chassis	Steel Basket	DE: Electricity Grid Mix	Hot Dip Galvanisation	Diesel Mix at Filling Station	Truck Transport (Euro 6)
Climate change	66.43 kg CO2 eq	14%	59%	6%	7%	2%	12%
Photochemical ozone formation	0.08 kg NOx eq	14%	56%	5%	9%	5%	11%
Fine particulate matter formation	0.04 kg PM2.5 eq	12%	71%	3%	8%	3%	3%
Fossil depletion	17.81 kg oil eq	14%	55%	7%	8%	17%	0%
Metal depletion	1.60 kg Cu eq	16%	67%	1%	16%	1%	0%
Freshwater eutrophication	8.01 × 10−5 kg P. eq	5%	36%	27%	4%	28%	0%
Marine eutrophication	1.36 × 10−3 kg N. eq	9%	69%	12%	3%	8%	0%

, showing the percentage contribution of the total environmental impact of each stage of the manufacturing phase.

It is clear that the steel wire basket dominates across all impact categories in Figure 5a–g. The steel pipe chassis contribute the next largest amount in terms of GWP, FPMF, and POF. A possible reason for such large relative contributions is that the chassis and wire basket LCIs cover raw material extraction (mining) as well as final product processing. Given that the CO₂ footprint of primary production steel can range from 1.6–1.85 kg/kg [38], a trolley would result in 35–41 kg/kg embodied CO₂ from primary production, which correlates with the presented results.

The “electricity grid mix” results capture the impacts of the wire bending and welding operations and tend to show a relatively minor contribution (less than 8% of total impacts), except for freshwater and marine eutrophication, showing a contribution of 27% and 12%, respectively.

This eutrophication result is associated with the environmental impact of powering the forming and welding equipment using the Spanish (ES) electricity grid mix, with the energy sources shown in

Table 2. Gross production of ES electricity (2019) from various energy sources [23].

Electricity Source	% Gross Production for 2019 (Valid Until 2025)
Natural gas	30.65
Nuclear	21.36
Wind	20.37
Hydro	9.85
Fuel oil	4.75
Hard coal	4.37
Photovoltaics	3.45
Solar thermal	2.08
Biomass (solid)	1.42
Waste-to-energy	0.64
Coal gases	0.41
Biogas	0.34
Lignite	0.34

. As shown in Table 2, natural gas makes up a significant proportion of ES energy generation. Energy generation from gas generally has greater environmental impacts than other options, except for coal-powered energy generation [39]. The disposal of drilling waste from gas extraction was identified as a significant eutrophication factor, due to drilling waste containing phosphorous [39].

The hot-dip galvanisation process shows a relatively minor contribution throughout all categories, being less than 10% in all categories except MD, with a 16% contribution. Arguillarena et al. [40] stated that the abiotic depletion (which may be correlated with MD) of hot-dip galvanisation can be attributed to the impact of silver in zinc production, with silver being a by-product of zinc ore processing.

Finally, the transport truck use shows the largest impact in the areas of GWP and POF, linked to tailpipe emissions. The well-to-pump Diesel Mix at Filling Station results show the largest impact in FD and FE. FD may be explained by oil extraction. Li et al. [41] reported that the NH₄ emissions from diesel fuel production had a positive correlation with an increase in eutrophication.

As mentioned previously, a sensitivity study was performed by increasing the trolley mass by 0%, 10%, 30%, 50%, and 100%. The total environmental impact magnitudes (for example, the total value of the kg CO₂ equivalent for the manufacturing phase) were plotted against the trolley masses (22.3 kg, 24.5 kg, 29.0 kg, 33.5 kg, 44.6 kg). The resulting lines of best fit are shown in

Table 3. Lines of best fit between environmental impact categories and increasing trolley mass.

	Line of Best Fit	R2 Value
Climate change (kg CO2 eq.)	y = 2.92x + 0.05	1
Photochemical ozone formation (kg NOx eq.)	y = 0.34 × 10−2x + 6.00 × 10−5	1
Fine particulate matter formation (kg PM2.5 eq.)	y = 0.18 × 10−2x + 3.00 × 10−5	1
Metal depletion (kg Cu eq.)	y = 0.07x + 0.12 × 10−2	1
Fossil depletion (kg oil eq.)	y = 0.80x + 0.02	1
Freshwater Eutrophication (kg P. eq.)	y = 3 × 10−6x + 5 × 10−8	1
Marine Eutrophication (kg N. eq.)	y = 6 × 10−5x + 1 × 10−6	1

. Focusing on the gradient values, it is clear that a variation in mass, and by proxy, in the number of trolleys manufactured, would result in climate change showing the biggest change. This is followed by fossil depletion and metal depletion.

3.2. Collection Phase

The environmental impacts of collecting and transporting abandoned trolleys using diesel collection vans around the University and Canley suburban area are shown in Figure 6. The data in Figure 6 represent one trolley being transported by the distances for each respective driving scenario, as indicated by the normalised data in

Table 4. Real-world and normalised collection van data.

	Real-World Distances and Number of Trolleys		Normalised Distances and Number of Trolleys
	Weekly Distance (km)	Trolleys Collected per Week	Normalised Weekly Distance	Normalised Trolleys Collected per Week
University Estates van	59.28	20.00	2.96	1.00
CCS: suburban driving	22.02	10.00	2.20	1.00
CCS: holding pen	5.66	20.00	0.28	1.00

.

In Figure 6, each driving scenario shows the same relative position in each of the environmental impacts, with the highest being the university Estates van and the lowest being the CCS holding pen. This consistent relative position can be attributed to the distances driven and the number of trolleys collected in each driving scenario.

Figure 6a indicates that van use makes the largest contribution compared to the well-to-pump model of the diesel fuel in terms of GWP. In all other impact categories, as shown in Figure 6b–g, the diesel fuel has the largest impact, which can be attributed to the fact that the well-to-pump model includes well drilling, crude oil production and processing, and transportation from the refinery to the filling station.

As mentioned previously, a sensitivity study was conducted by varying the trolley mass from the baseline 22.3 kg to an increase of 10%, 30%, 50%, and 100%. Trendlines were calculated for each environmental impact.

Table 5. Line of best fit of environmental impact value (y) against percentage change in trolley mass (x).

	Line of Best Fit	R2 Value
Climate change (kg CO2 eq.)	y = 4.00 × 10−5x +0.13	1
Photochemical ozone Formation (kg NOx eq.)	y = 1.00 × 10−8x + 5.00 × 10−5	1
Fine particulate matter formation (kg PM2.5 eq.)	y = 4.00 × 10−9x + 1.00 × 10−5	1
Metal depletion (kg Cu eq.)	y = 3.00 × 10−9x + 1.00 × 10−5	1
Fossil depletion (kg oil eq.)	y = 1.00 × 10−5x + 0.04	1
Freshwater Eutrophication (kg P. eq.)	y = 7.00 × 10−12x + 2.00 × 10−8	1
Marine Eutrophication (kg N. eq.)	y = 5.00 × 10−11x + 2.00 × 10−7	1

shows the calculated trendlines for all environmental impacts, with “y” indicating the relevant environmental impact value and “x” indicating the percentage change in trolley mass. The sensitivity analysis was performed for the 1 km driven by the diesel van. It is clear that Climate Change and Fossil Depletion show the largest gradient and thus higher sensitivity to increases in trolley mass.

3.3. Transport and Refurbishment Phase

Figure 7 shows the environmental impacts of transporting trolleys from Coventry City to the refurbishment facility in Tibshelf, removing the zinc coating through pyrolysis, the reapplication of zinc coating through hot-dip galvanisation, and then transporting the trolleys back to Coventry City.

The Tibshelf distances and number of trolleys collected are calculated as follows. It is estimated that trolleys are taken to the refurbishment facility in Tibshelf twice per year, with 50 trolleys per trip. The total yearly distance is thus 440 km. In

Table 6. Real-world and normalised collection van data for the “transport and refurbishment” phase.

	Real-World Distances and Number of Trolleys		Normalised Distances and Number of Trolleys
	Weekly Distance (km)	Trolleys Collected per Week	Normalised Weekly Distance	Normalised Trolleys Collected per Week
Transport to and from the Tibshelf refurbishment facility	8.46	1.92	4.40	1.00

, this is represented as equivalent weekly distances and the number of collected trolleys.

Figure 7 shows that the hot-dip galvanisation has the largest environmental impact across all considered impact categories. The transport activity shows the second largest impacts in the Climate Change and Fossil Depletion categories. The pyrolysis activity shows a slightly higher impact in freshwater and marine eutrophication. The electricity used to power the pyrolysis process may be associated with the higher eutrophication results.

3.4. Comparison of All Life-Cycle Phases

Table 7. Environmental impact of the three life-cycle phases.

	Manufacturing Stage (Normalized)	Collection Stage (Normalized)	Transport and Refurbishment Stage (Normalized)
Climate change	1.0 × 10+0	1.1 × 10−2	8.4 × 10−2
Photochemical ozone formation, ecosystem	1.0 × 10+0	3.7 × 10−3	9.9 × 10−2
Fine particulate matter formation	1.0 × 10+0	2.0 × 10−3	8.0 × 10−2
Metal depletion	1.0 × 10+0	3.5 × 10−5	1.6 × 10−1
Fossil depletion	1.0 × 10+0	1.2 × 10−2	8.6 × 10−2
Freshwater eutrophication	1.0 × 10+0	2.2 × 10−3	5.7 × 10−2
Marine eutrophication	1.0 × 10+0	8.2 × 10−4	3.3 × 10−2

shows the sum of the environmental impacts for each life-cycle stage (the manufacturing stage, the collection stage, and finally the transport and refurbishment stage). All data have been normalised to the manufacturing phase for each environmental impact. It is clear that the manufacturing phase dominates by orders of magnitude across all environmental impacts. The “transport and refurbishment” phase is consistently the life-cycle phase with the second largest impact. The collection stage consistently shows the least impact across all categories.

4. Discussion

Three life-cycle stages were modelled: manufacturing, van use for trolley collection, and the “transport and refurbishment” stage. Overall, the manufacturing phase showed the highest impact across all life cycle stages, with the steel wire basket contributing the largest amount to manufacturing environmental impact.

The relative magnitude of the impact of steel may be attributed to mining and processing being included in the model. It should be acknowledged that the current study did not explicitly model a circular loop for the use of recycled steel to manufacture trolleys. This was mainly due to a lack of data on trolley recycling. A discussion of the current decarbonisation pathways for steel manufacturing and their potential influences on the presented results is thus necessary.

The steel processing databases used in this study are based on the World Steel Association Life-Cycle Inventory Methodology [42], which considers both the blast furnace and electric arc furnace processing routes. While the blast furnace route utilises up to 30% steel scrap [43], the remainder is primary ore and results in approximately 1.8 t_CO2/t_steel [44]. Electric arc furnaces can be charged with up to 100% steel scrap, resulting in approximately 0.9 t_CO2/t_steel [44]. The blast furnace production route is dominant in Europe, accounting for 57% of European primary steel [44]. The dominance of blast furnace production is captured in the GaBi database and explains the reported environmental impacts of steel relative to other manufacturing impacts.

The European Green Deal has set a clear target for achieving a circular economy in Europe, which favours higher outputs from secondary steel production from recycled materials. A current constraint on this decarbonisation pathway is the availability of high-quality scrap to be used in electric arc furnace production. Nevertheless, the current focus on steel decarbonisation and the implementation of a circular economy could mean that the impacts of steel production could fall in the future. As well as European decarbonisation drivers, the UK government is supporting two major steel companies in switching from blast furnace to electric arc furnace production [45], reinforcing the observed trend of driving decarbonisation for steel production. This trend therefore suggests that the manufacturing phase of trolley production may decrease over the next few decades.

The collection phase showed the least impact compared to the manufacturing and “transport and refurbishment” phases. As mentioned previously, a modelling approximation was necessary to use a diesel passenger vehicle instead of a van; however, the appropriate fuel and vehicle mass were incorporated into the model. Given that the overall use-phase results were two to five orders of magnitude lower than those for the manufacturing phase, a clear ranking of the magnitude of impact for the use phase can be established.

The “transport and refurbishment” phase was the second highest contributor to environmental impacts across the three life cycle stages, with hot-dip galvanisation being the main contributor across all environmental impacts.

A more quantified perspective of the above discussion can be obtained by relating the results from Table 7 to the previously discussed number of reported abandoned trolleys (520,000 trolleys in the UK in 2017 [1]). Focussing on GWP as an example, Table 7 shows that the van use (collection) stage is ~1% of the manufacturing impact and “transport and refurbishment” is ~10% of the manufacturing impact. The sum of all the manufacturing phases in Table 1 is 66.43 kg CO₂ eq. These quantities can be used to explore two different scenarios: one with all 520,000 trolleys collected and returned to the supermarket, and the other with 10% of the 520,000 trolleys requiring refurbishment and the remaining 90% collected and returned to the supermarket.

In the first scenario, trolley abandonment alone (i.e., a van collecting and returning trolleys to the supermarket) results in 343,200 kg CO₂ eq. In the second scenario, trolley collection and refurbishment results in 652,080 kg CO₂ eq. The above discussion shows that, theoretically, if 10% of trolleys were deemed to be too damaged to directly return into service and thus diverted for refurbishment, the GWP would increase 1.9 times. This reinforces the benefits of directly collecting abandoned trolleys as soon as possible, before damage requiring refurbishment occurs.

Another interesting perspective on the previously presented results pertains to break-even points. The GWP of the collection stage (University and Canley suburban area trolley collection) was 0.69 Kg CO₂ equivalent and the GWP of the manufacturing stage was 65.14 kg CO₂ equivalent.

Based on these data, one trolley would have to be collected 93 times to offset its own manufacturing impact. This view further reinforces the benefits of keeping trolleys in use for an extended time and avoiding manufacturing new trolleys to replace abandoned ones. It is prudent to contextualise the findings of this study against those in the wider literature. As previously stated, only one other study was found to investigate shopping trolley production and end of life [15]. Val and Lamban [15] used the eco-indicator 99 methodology with normalisation and weighting applied to the results. A direct comparison of impact in terms of the kg CO₂ equivalent or the kg Nox equivalent is therefore not possible. Nevertheless, interesting observations can be made. Val and Lamban [15] showed three material stages in the manufacturing phase, consisting of “Aluminium production mix”, “Chromium steel”, and “Polypropylene granulate”, with steel contributing 92% of the total material impact. This dominance of steel to environmental impact correlates with the results presented in the current study.

Val and Lamdan presented results in terms of 11 impact categories (climate change, ecotoxicity, land use, ozone, etc.) and the percentage contribution of the manufacturing phase and landfill phase to each impact category.

It was shown that the steel trolley manufacturing phase contributed more than 99% to each impact category, except “carcinogens”. On an equivalent basis, this means that the landfill phase contributed less than 1% to each impact category, except “carcinogens”, where a 9.33% contribution was calculated. This contribution to carcinogens was associated with the chrome-plating corrosion protection of the steel trolley.

The polypropylene trolley exhibited a more varied contribution between the manufacturing phase and landfill phase across all impact categories; however, the landfill contribution to “carcinogens” clearly dominated and was calculated to be 92.65%. This dominant contribution was associated with polypropylene breaking down in landfill.

The findings produced by Val and Lamdan present an interesting perspective on using alternative materials for shopping trolleys and how this might affect their environmental impact. Val and Lamdan stated that less material was required for the polypropylene trolleys due to the lower material density, thus showing a reduced environmental impact. Val and Lamdan also stated that polymer trolleys do not require environmental protection coatings, as steel does. These two aspects help to reduce polymer trolleys’ environmental impact. However, once the trolley goes into landfill (or the landfill equivalent through abandonment), a significant environmental impact occurs through carcinogen release as the trolley breaks down.

Arguillarena [40] analysed the environmental impact of hot-dip galvanisation processes in Spain, including the initial steel production process. It was found that the steel production exhibited a larger impact than the galvanisation process in terms of global warming potential, the abiotic depletion of fossil fuels, and photochemical ozone creation. This corresponds to the findings presented in this study.

It was reported that galvanisation was more strongly influenced by the abiotic depletion of elements, toxicity categories, and the ozone layer depletion potential. It was stated that the abiotic depletion of elements was associated with silver’s impact on zinc production, due to silver being a by-product of zinc production. The toxicity impacts of galvanisation stemmed from factors such as the emission of heavy metals to fresh water, and emissions of cadmium, copper, and lead to air [40]. The current study did not take toxicity impacts into account and would be enhanced by future work that does.

Zackrisson [46] investigated the environmental impact of various corrosion protection maintenance options for 8 mm thick hot-dip-galvanised steel structures. The author presented an interesting calculation regarding the extension of service life required to be environmentally beneficial compared to replacing the steel structure entirely. The author first considered the ratio between the environmental impact of maintenance and the environmental impact of new production. It was stated that, if the service life of the maintained structure is greater than the service life of the new structure multiplied by the previously mentioned ratio, then it can be considered environmentally beneficial to refurbish. This calculation might be useful for policy makers and supermarkets to quantify the benefits (or drawbacks) of refurbishing abandoned trolleys.

A comparison was made between thermally sprayed zinc and various painting options (zinc epoxy, zinc ethyl silicate, etc.), with zinc being either a virgin material or recycled. The pre-treatment was performed using either wet abrasive blasting or laser cleaning, which makes an interesting comparison with the pyrolysis method chosen in this study.

The results presented by Zackrisson can be compared with the results presented in this study. The functional unit of Zackrisson’s study was “corrosion protection for steel infrastructure per m² and year”. The steel trolley basket in this study consisted of wires 3.3 mm in diameter, with 16 wires of 1785 mm in length in the y-direction and 28 wires of length 2285 mm in the x-direction (established through physical measurements conducted by the author of trolleys in a supermarket). This equates to approximately 1 m² of the surface area of the trolley basket.

Zackrisson reported that the baseline scenario of replacing an old steel structure with a new galvanised structure showed a GWP impact of approximately 52 kg CO₂ eq. Figure 5a) shows a GWP of 45 kg CO₂ eq. if the “steel basket” and “hot-dip galvanisation” results are summed together, thus correlating with Zackrisson’s results. All refurbishment options explored by Zachrisson showed a GWP impact between 1 and 5 kg CO₂ eq. Zackrisson reported that refurbishment operations utilizing virgin zinc and wet abrasive blasting had the highest environmental impact, whereas recycled zinc and laser cleaning had the lowest impact. Figure 7a) correlates with Zachrisson’s results by also showing a GWP impact of 5 kg CO₂ eq. for the re-application of hot-dip galvanic coatings. An interesting alternative to refurbishment can be explored in utilising selective laser cleaning and the selective re-application of a galvanic coating, thus potentially reducing the energy and resources required to refurbish the entire trolley. Use can be made of Zackrisson’s calculation method by policymakers and supermarkets if they wish to establish whether refurbishment is environmentally beneficial.

Beyond economic deterrents such as fines or physical barriers such as wheel locking mechanisms, other solutions to mitigating trolley abandonment can be explored, such as behavioural interventions and monitoring. The “watching eyes” effect has been studied in terms of its effectiveness in crime reduction [47,48]. This effect stems from our evolutionary response to feeling watched and the ways in which we adapted our behaviour to avoid predators and, later, to protect our reputation in social groups. Nettle [48] conducted a study on the effectiveness of displaying images of “watching eyes” with text stating “Cycle Thieves: We Are Watching You”. The study reported a 62% decrease in cycle thefts at the signage areas. However, the non-signed controlled areas experienced a 65% increase, indicating that theft was simply displaced. Nevertheless, the interesting possibility of utilising the psychology of surveillance is presented, while being more cost-efficient than traditional CCTV. Indeed, Dear [47] performed a meta-analysis of the effectiveness of “watching eyes” and reported a decrease of 35% in antisocial behaviour, whereas CCTV showed reduction in crime of 16%.

Further use of currently installed CCTV surveillance could be considered, whereby technological concepts such as facial recognition or gunshot detection technology [49] could be adapted for alerting supermarket security of trolleys being taken outside of designated areas.

While the direct impacts of the trolley life cycle have been presented, a broader view of the implications for the circular economy would add further value. One aspect for consideration is the utilisation of electric delivery vans. Giordano [50] reported that replacing diesel vans with electric vans in cities with relatively clean energy mixes could result in a reduction in GHG emissions of 44–53%. The same analysis for coal-reliant cities reported a reduction of 10–25%. The economic implications of replacing diesel vans with electric vans was also considered. It was reported that, as the BEV lifetime increases, the annualised costs are better than for the diesel van (in the order of a few thousand EUR per year) in cities with good incentives (such as registration tax, subsidies, ownership taxes, etc.). Without incentives, the BEV vehicle only showed better annualised costs in the minority of usage/lifetime scenarios. While a clear benefit can be seen in utilising electrically powered vans, the right incentives must be in place for economic viability.

Another interesting aspect of the circular economy to consider is using alternative materials to manufacture trolleys. Val and Lamban [15] considered the environmental impact of replacing steel shopping trolleys with polypropylene trolleys. Polypropylene trolleys exhibited better environmental impacts in the manufacturing phase as they use less material (associated with reduced energy consumption during manufacturing), as well as a reduced carcinogenic impact due to not requiring a protective coating. In the end-of-life phase, the situation was reversed, with the polypropylene trolley showing the highest carcinogenic impact due to decomposing in landfill.

While the study [15] did not consider the impact of collecting trolleys, the polypropylene trolley was estimated to weigh 3 kg less than the steel trolley. The combination of their reduced manufacturing impact and lighter weight indicates that polymer trolleys are a promising alternative, especially if the collection phase avoids the carcinogenic impacts associated with landfill. One must also consider the longevity of steel and polymer trolleys. Polymer trolleys are not likely to be economical to repair, whereas steel trolleys can be economically repaired through welding and re-galvanisation. Therefore, as long as polymer trolleys can be kept in a usable condition, it is possible that their environmental impact would be reduced compared to steel trolleys.

In terms of future research directions, it would be interesting to investigate the effect of variations in retail policies on consumer behaviour. Potential insights may be gained into the effectiveness of directing company resources directly to the root of the problem, compared to the financial and resource costs of surveillance or tracking solutions. The current research could be extended to focus on consumer behaviour by applying Nettle’s [48] “watching eyes” experiment to trolleys.

Behavioural interventions could also be used to inform retail policies. Preventative measures such as surveillance and fines may help to reduce the number of trolleys being taken out of service; however, changing customer behaviours may help to target the root cause. Thogersen [51] stated that the public are faced with a trade-off between sustainability and other valued outcomes and hence require a goal if they are to change their behaviour to be more sustainable. It was postulated that, by strategically manipulating situational stimuli, the importance of sustainable goals can be emphasised, thus making them more attractive to the public. Three goal types were defined: hedonic (instant gratification), gain (maintaining/improving the person’s resources), and normative (doing the “right” thing). It was found that, when study participants were primed with a normative goal, more importance was placed on the sustainable choice, rather than cost or instant gratification. “Normative” goal priming had a similar effect; however, “gain” goal priming had limited influence, most likely due to cost being naturally important to the public. Thogersen concluded that it was possible to “promote pro-environmental choices by activating a normative goal frame”.

Policymakers should thus emphasise normative goals in their communications to the public, while keeping hedonic and gain goals in the background.

Borg [52] considered how social norms could influence sustainability choices and behaviours. It was stated that an individual’s behaviour is influenced by descriptive norms (i.e., they are influenced by what others are doing) and injunctive norms (i.e., they are influenced by what others approve/disapprove of). An individual’s behaviour, in aspects such as littering, recycling, and energy consumption, was found to be modified by informing them about others’ behaviours. Thus, individuals are more likely to engage in sustainable practices/behaviours if they perceive them to be “normal”. Borg concluded that social norm messaging by policy makers becomes more effective if a particular practice, such as placing trolleys back in place, becomes more normal.

5. Conclusions

This study presents a case study of the environmental impact of collecting abandoned trolleys in the Canley suburban area of Coventry city. While focused on a specific suburban area, the results are easily transferable to other similar cities within and outside the UK.

The LCA results showed that the environmental impacts of manufacturing shopping trolleys dominate all other life-cycle stages. It may be concluded that any environmental impacts caused by collecting and refurbishing abandoned trolleys are minimal compared to the impact of losing trolleys through abandonment, which may result in new trolleys being manufactured to replace them.

There is scope to reduce this environmental impact by reducing the number of abandoned trolleys. As discussed previously, options include coin slots, vertical bars (to stop trolleys leaving the shop floor), wheel locking mechanisms, and car park wardens. These do have limited success and public education of the holistic view of environmental impact of trolleys might add to these deterrents. Potential previously discussed options include utilising social norm messaging and normative goal framing. The analysis presented in this study was conducted using the ReCiPe midpoint method. To engage stakeholders and the general public, it may be more prudent to present the results in broader end-point categories.

It is acknowledged that the sustainability aspect might not fully provide incentives for supermarkets and retailers to prioritise trolley management. It is therefore likely that a multitude of factors, such as financial and regulatory incentives, will also play a role in the action taken by these stakeholders. For example, the cost to replace a single lost trolley ranges from GBP 150–400 [53] and, as mentioned previously, local councils have begun to issue fixed penalties to those taking trolleys off premises [2].

It is possible that retailers will not engage in trolley collection and refurbishment if it is ultimately deemed too expensive. For future research, we recommend performing a cost–benefit analysis to gain further insight into the economic feasibility of trolley sustainability practices, which may also be used to influence policies.

The presented study can be used to suggest policy recommendations for the implementation of the findings. The quantified benefit of keeping trolleys in service, and thus avoiding their manufacture or refurbishment, could be used as part of the company’s carbon accounting, if participating in the Science-Based Targets initiative (SBTi) [54]. Active participation in SBTi would aid in any applications for sustainability-linked financial loans or participation in carbon trading schemes. At the time of writing, most major UK supermarkets are signed up to the SBTi. Indeed, an in-depth case study of the benefits to Tesco of joining the SBTi was presented [55]; however, their focus was mainly on electricity consumption and supply chain considerations, such as the carbon footprints of agriculture and food manufacturing. The conclusions of the presented study could be used to expand the benefits that companies may obtain from engaging in carbon accounting.

While the environmental impact of the trolley life cycle has been quantified, within the context of the broader economy, its contribution is relatively minor. In terms of CO₂ production, the UK transport sector, electricity sector, and industrial sector contributed 35%, 22%, and 10%, respectively, to the 309 Mt of total UK CO₂ emissions in 2022 [56]. The sources of these CO₂ emissions in the UK from fuel combustion are 47% from oil and 47% from natural gas. Further efforts in the field of renewable energies are thus required to address these dominant emissions sources.

The current study makes suggestions for policymakers to address trolley abandonment. First, the environmental benefit of keeping trolleys in service and avoiding manufacturing and refurbishment has been quantified. Social norm messaging and utilising “normative goals” in communications with the public might help to reduce trolley abandonment. By making it known that actively avoiding trolley abandonment is socially “normal”, the public might be influenced to reduce such practices. Furthermore, the study identifies interesting possible future research directions, such as the further exploration of utilising electric collection vehicles (made economically feasible if the right incentives exist), the exploration of selective refurbishment (through selective laser cleaning and re-galvanising), and reducing trolley abandonment economically through “watching eyes” deterrents, or perhaps making use of already-installed CCTV with enhanced technological upgrades such as facial recognition or gunshot detection. In terms of alternative materials, polymer trolleys show promise, with reduced density compared to steel, and not requiring environmental protective coatings. The crucial thing is to stop such trolleys being placed in landfill (or equivalently abandoned) and leeching carcinogens. The repairability of polymer trolleys must also be considered.

It may be prudent to consider how the methodology presented in this study could be applied to other sustainability challenges. Worldwide, the economy is based on producing products, using them, and then finally disposing of them. The presented methodology could be applied to other sustainability challenges to contextualise the relative impact of producing products versus keeping products in use, thus quantifying and reinforcing efforts to create a circular economy.