1 Introduction

The concept of circular economy (CE) has emerged as a response to the increasing environmental pressures stemming from the dominant linear “take-make-use-waste” development model (Ghisellini et al. 2016; Helander et al. 2019). It is perceived as a sustainable economic development model that promotes the responsible and cyclical use of resources aiming to decouple economic growth from resource use and, in turn, environmental degradation (Moraga et al. 2019; Corona et al. 2019), while also creating socio-economic co-benefits, such as increased employment and improved resilience of economic systems (Pauliuk 2018; Mayer et al. 2019; Voulvoulis 2022). As such, it has been attracting attention from public authorities, industry, academia, and citizens as a necessary step to achieve climate neutrality (EC 2020) and sustainable development (Corona et al. 2019; Brändström and Eriksson 2022). 

At the urban level, the CE has been gaining traction as a means to alleviate the increasing pressures on resources and the environment due to accelerating urbanization (Sánchez Levoso et al. 2020; Tan et al. 2021). Worldwide, a growing number of municipal governments and local authorities have been embracing the CE concept and have been designing and implementing strategies aiming to facilitate the transition towards CE (Vanhuyse et al. 2021). Such strategies are mainly oriented towards increased resource efficiency and materials circulation (Petit-Boix and Leipold 2018). However, resource efficiency and recirculation can sometimes produce unintended negative consequences like rebound effects and burden shifting (Harris et al. 2021; Chen 2021). For example, recycling degrades material structure requiring energy-intensive processes and additional resources to restore material quality (Cullen 2017). Thus, the implementation of circular strategies may not necessarily provide optimal environmental outcomes.

To ensure that urban CE strategies can contribute positively to environmental sustainability, it is necessary to design strategies based on firm information regarding their environmental implications. This requires identifying what areas have the greatest potential to lead to environmental improvements and holistically assessing the environmental impacts and benefits of circular strategies before their implementation (Kravchenko et al. 2019). Moreover, to ensure that implemented CE strategies generate the intended environmental benefits, it is essential to systematically monitor and evaluate their environmental performance taking a systems perspective (Helander et al. 2019; Rufí-Salís et al. 2021). This type of environmental assessment could be achieved through the coupling of Material and Energy Flow Analysis (MEFA) with Life Cycle Assessment (LCA) under an urban metabolism (UM) perspective (henceforth referred to as the UM-LCA approach).

The UM is a systems-based approach to examine trajectories of resource consumption, waste generation, and associated environmental impacts based on the conceptualization of the urban system as a living organism (Pincetl et al. 2012). It offers a useful framework for analyzing the interactions of natural-human systems based on a wide range of quantitative methods (Pincetl et al. 2012; Goldstein et al. 2013), among which Material Flow Analysis (MFA) is probably the most widely applied (Zhang 2013). MFA is a systematic method to map and quantify material flows into, within, and out of a specific system based on the mass balance principle (Brunner and Rechberger 2017). It can also be applied to analyze material and energy flows in tandem, an approach commonly referred to as Material and Energy Flow Analysis (MEFA) (Derrible et al. 2021). The application of MEFA for UM analysis provides a holistic view of material and energy flows throughout an urban system, helping thereby to identify inefficiencies and track emissions and wastes to their source (Brunner and Rechberger 2017). Nevertheless, MEFA alone cannot comprehensively assess the environmental profile of an urban system (García-Guaita et al. 2018). Its main limitations are (i) it erroneously equates mass to environmental loading disregarding the qualities and environmental burdens of flows and (ii) it ignores impacts from upstream and downstream processes that provide resources and handle wastes, respectively (Giljum and Hubacek 2009; Goldstein et al. 2013; Stephan et al. 2020).

To overcome these limitations, MEFA can be integrated with LCA in the UM-LCA approach (García-Guaita et al. 2018). LCA is a well-established methodology to comprehensively assess the potential environmental impacts of a system from a life cycle perspective (ISO 2006; Finnveden et al. 2009). The object of analysis (system) can include goods and services as well as large-scale systems, including urban areas (Loiseau et al. 2018). When coupled with MEFA, LCA can integrate the inventorying part of MEFA to account for direct, indirect, and supply chain impacts due to the transformation and use of materials and energy in urban areas (Pincetl et al. 2012). Benefits of this approach include (i) the potential to capture embodied environmental impacts of material and energy flows, (ii) the ability to incorporate supply chain flows beyond urban boundaries, and (iii) the aggregation of results into common and easy-to-communicate indicators (Chester et al. 2012; Goldstein et al. 2013).

The UM-LCA approach has been first demonstrated by Goldstein et al. (2013), who assessed the environmental performance of Beijing, Cape Town, Hong Kong, London, and Toronto. In the following years, Dias et al. (2018), García-Guaita et al. (2018), Westin et al. (2019), Rama et al. (2021), and González-García et al. (2021) applied the UM-LCA modeling approach to assess the environmental performance of different urban areas worldwide. In addition, Lopes Silva et al. (2015), Stelwagen et al. (2021), and Dorr et al. (2022) used the approach to perform environmental assessments of specific sectors within urban areas, and Ipsen et al. (2019) used it to assess smart city solutions in Copenhagen.

While these studies have demonstrated the utility of UM-LCA for environmental assessments in urban settings, the UM-LCA approach has not yet been explored as a tool to support the design and evaluation of circular strategies at the urban level. Moreover, the approach has only been applied with a retrospective scope (evaluating urban systems in the past). In the context of CE, applying the UM-LCA retrospectively could provide background information on the environmental profile of urban systems helping to inform the design of new circular strategies and monitor the performance of implemented strategies. However, such an assessment cannot provide all the required information for designing effective circular strategies, as it cannot assess potential environmental implications associated with the implementation of the strategies in the future. Applying the UM-LCA approach prospectively, on the other hand, can provide this type of information. Such an approach could build upon the prospective LCA (pLCA) approach, which aims to assess the environmental performance of systems at a future point in time based on future scenarios (Arvidsson et al. 2018; Steubing and de Koning 2021) and has been principally applied to assess emerging technologies (Villares et al. 2017; Moni et al. 2020). The pLCA has been also applied to assess large-scale systems, such as urban water systems (Loubet et al. 2016) and national energy systems (Martín-Gamboa et al. 2019) but, to our knowledge, it has never been coupled with MEFA under an UM perspective to assess the environmental sustainability of an urban area.

In light of the above, we apply the UM-LCA approach to the Umeå urban area in Sweden, aiming to investigate its utility as a tool to support the design, evaluation, and monitoring of circular strategies at the urban level. We apply the UM-LCA approach to (i) assess retrospectively the environmental profile of Umeå to provide background information for the design of urban circular strategies and (ii) assess prospectively the environmental implications of implementing circular strategies in the future. The contribution of this work is twofold. First, it advances the UM-LCA modeling approach in two ways: (i) by demonstrating its application with a future-oriented scope to enable decision makers (i.e., urban planners and policymakers) to evaluate the potential effects of their future decisions and (ii) by demonstrating its application at the sectoral level to provide a more refined analysis of the environmental performance of the urban system. Second, it contributes to the ongoing discussion regarding designing, assessing, and monitoring circular strategies at the urban level by illustrating the utility of UM-LCA for this purpose.

2 Methods

This section first provides background information on the studied urban area (Sect. 2.1 ). It then outlines the overall modeling approach (Sect. 2.2) and continues by providing an overview of MEFA (Sect. 2.3) and a more detailed description of the retrospective LCA (Sect. 2.4) and prospective LCA (Sect. 2.5).

2.1 Study area

Umeå municipality in northeastern Sweden is the urban system under study. Umeå is the most populated municipality in northern Sweden and the 11th in Sweden. Its population has been steadily growing over the last two decades, from 106,525 citizens in 2002 to 132,235 in 2022 (SCB 2023a), and it is projected to reach 200,000 by 2050 (Umeå municipality 2018). Despite the population growth, Umeå is a sparsely populated municipality. Its total land area is 2317 km2, with 70 km2 being urbanized and the rest being mainly forest (SCB 2023b). Although the urbanized area occupies a very small part of the municipal area, it is assumed that the spatial boundary of the urban system coincides with the administrative boundary of the municipality. The motivation behind this assumption is the following. First, almost 90% of the population in the municipality resides within the boundaries of the urbanized area (SCB 2023b). Second, the most available data are reported for the entire municipality.

One reason for choosing Umeå as the study setting to apply the UM-LCA approach is that the municipal council has made the transition towards the CE a political priority (Umeå municipality 2016) and it intends to develop a circular strategy with sector-specific measures (OECD 2020; Circular Regions 2023). Hence, assessing the environmental profile of Umeå and evaluating the environmental implications of potential circular measures could provide insights that could help local decision makers design an environmentally sound strategy. Another reason is that in a recent study (Papageorgiou et al. 2024), the authors of this article used MEFA to provide a detailed accounting of material and energy flows in Umeå, which can form the quantitative basis for applying the UM-LCA approach. Moreover, Umeå represents a typical Swedish medium-sized urban area, characterized by population growth, high levels of consumption, and dependency on global supply chains, which could help generalize the findings of the study to similar urban settings in Sweden.

2.2 Modeling approach

Figure 1 illustrates a general system definition that integrates the scopes of MEFA and LCA forming the basis of the modeling. MEFA accounts for direct material and energy flows (referred to as metabolic flows) entering and exiting the urban system due to production and consumption processes within its boundaries, thus offering a territorial-based assessment (Athanassiadis et al. 2018). The LCA expands further the scope of the assessment, as it considers all the life cycle stages of a metabolic flow irrespective of whether they take place within the urban system (foreground system) or outside its boundaries (background system). In this way, it accounts for environmental impacts that occur upstream and downstream of the urban system (Yetano Roche et al. 2014).

Fig. 1
figure 1

Conceptual view of the UM-LCA approach as implemented in this study (the use stage of metabolic flows takes place within the urban system, while resource extraction, processing, manufacturing, and end-of-life can take place either within or outside the urban boundaries)

Full size image

The urban system was modeled following a bottom-up approach to analyze the system at the sectoral level. Previous UM-LCA studies (Goldstein et al. 2013; García-Guaita et al. 2018; Dias et al. 2018; González-García et al. 2021; Rama et al. 2021) modeled the urban system by focusing only on key metabolic flows, without considering its internal components (sectors). The present study, in contrast, modeled the urban system as comprising 11 main sectors to provide a more refined analysis (see Fig. 1). The sectors, which were inspired by the NACE classification of economic activities in the European community (Eurostat 2008), are agriculture, forestry, and fishing; manufacturing, energy supply, water supply, mining, and quarrying; construction, transport, households, wastewater treatment, waste management, and service sector. From these sectors, the agriculture, forestry, fishing, manufacturing, and service sectors were excluded from the modeling due to practical constraints, as comprehensive and reliable data for these sectors were not readily available at the time of the study. Due to these exclusions, the results of the model are expected to be underestimated, though not significantly, for two reasons. First, the scale of activity of the agriculture, forestry, fishing, and manufacturing sectors in Umeå is not substantial considering that approximately 1% and 8% of the total workforce is employed in these sectors, respectively (Kolada 2024). Second, products traded but not used by the service sector, such as food, textiles, and electronics, are accounted for in the household sector, where they are finally used.

The modeling was performed following two different approaches: the retrospective and the prospective. With the former, a retrospective MEFA (rMEFA) was first carried out to map and quantify metabolic flows in the urban system and its sectors based on the most recent available data. The quantified flows from the rMEFA were then used to compile the life cycle inventory (LCI) model of the urban system, which is used as a basis of the retrospective LCA (rLCA) to assess the environmental performance of the urban system in its current state. With the prospective approach, the foreground system in the LCI model was modified using data from a prospective MEFA (pMEFA) performed based on future scenarios for the foreground system, and the background system was modified based on data derived from implementing future scenarios for the background system in the premise tool (Sacchi et al. 2022). The derived prospective LCI models were then used as a basis of the pLCA to assess the environmental implications of potential circular strategies in the future.

2.3 Material and energy flow analysis

The rMEFA was carried out in a previous study (Papageorgiou et al. 2024). It was performed following the methodological procedure described by Brunner and Rechberger (2017) with the goal to map and quantify material and energy flows in the urban system of Umeå and its different sectors (as represented in the lower half of Fig. 1) in 2021. The flows were determined based primarily on bottom-up data, i.e., urban-level data from municipal statistics, reports, and articles that are specific to Umeå. The results of the analysis were organized into structured accounts that describe the mass and energy balances per sector. These accounts formed the basis of the LCI of the rLCA.

The pMEFA was conducted with the goal to quantify material and energy flows in the urban system in 2030 as represented by the three future scenarios describing the foreground system (see Sect. 2.5.2). It was performed by modifying the parameters of the model developed for the rMEFA according to the specifications of the foreground scenarios and its results were used to construct the LCI of the pLCA.

2.4 Retrospective life cycle assessment

A process-based bottom-up LCA was conducted using the Brightway2 LCA framework (Mutel 2017) with its graphical interface, the open-source LCA software Activity Browser (Steubing et al. 2020). The life cycle inventory database used is Ecoinvent (v3.8 cut-off system model) (Wernet et al. 2016). In accordance with the ISO 14040 standard (ISO 2006), the LCA in this study comprises four stages: (i) goal and scope definition, (ii) inventory analysis, (iii) impact assessment, and (iv) interpretation. The first three are described in the following sub-sections, while the interpretation of results is presented in Sect. 3.

2.4.1 Goal and scope definition

The goal of the rLCA was to assess the environmental profile of Umeå with its sectors in its current state using the most recent available data from the rMEFA (for 2021).

Defining the functional unit when applying the UM-LCA approach is challenging, given that urban systems are complex systems that provide multiple functions to their citizens (Goldstein et al. 2013). In line with previous studies (Goldstein et al. 2013; García-Guaita et al. 2018; Dias et al. 2018; González-García et al. 2021), an explicit traditional LCA functional unit was not defined and instead, the gross annual environmental impacts of the urban system were normalized by population (per capita) to represent the impacts of a conceptual average citizen of the urban area. As highlighted by González-García et al. (2021), this practice facilitates the comparison of different urban areas with different populations, though it does not allow to consider quality of life aspects (Goldstein et al. 2013).

The analyzed system consists of the foreground and the background system (Fig. 1). The foreground system includes the urban sectors defined in Sect. 2.2. The background system consists of all the upstream processes that supply materials to the urban system and all the downstream processes that handle residuals and are located outside the urban boundaries. The system boundary of the rLCA includes the eight sectors of the foreground system that were included in rMEFA, along with the related background processes. To determine the metabolic flows included in the system boundary, the concept of “total responsibility” (García-Guaita et al. 2018) was applied, according to which metabolic flows associated with foreground activities that are required to satisfy the demand of the urban system are accounted for, regardless of whether these activities take place within or outside the urban boundaries. For example, all emissions due to air travel by Umeå’s citizens are accounted for, even though most of the emissions are released outside the urban boundaries. In a few cases, minor flows associated with foreground processes were cut off from the system boundary, as there were no representative datasets in Ecoinvent, e.g., pharmaceuticals and sports goods in households. Moreover, following Liljenström (2021), the system boundary does not include capital goods (e.g., buildings, machinery) that accumulated in the urban stock in previous years, as the scope covers annual environmental impacts.

For the system modeling, the attributional approach was applied. This approach was chosen, as the rLCA in this study is of descriptive nature, i.e., its main purpose is not to directly support decisions, but rather to describe the environmentally relevant physical flows to and from the system (Finnveden et al. 2009) This is in line with the International Reference Life Cycle Data System (ILCD) handbook that recommends the attributional approach for this decision context (situation C—accounting) (Bjørn et al. 2018b). Thus, the LCI model was compiled based on the Ecoinvent cut-off system model database (v-3.8) (Wernet et al. 2016), which is compliant with the attributional approach, and, accordingly, the “cut-off approach” was used to resolve multi-functionality of end-of-life (EoL) processes. This means that recyclable materials are available burden-free (cut-off) to their users and do not provide any credit to their producers (Bjørn et al. 2018a). Likewise, non-waste by-products of waste treatment processes, such as electricity and heat from waste incineration, are burden-free and the incineration process does not receive any credit for their provision.

A drawback of the used cut-off approach is that it gives no incentive to improve recycling processes, as it rewards the use of recycled materials rather than the recovery of materials for recycling (Hermansson et al. 2022). By contrast, the alternative “EoL recycling approach” allocates credits, plus the burdens of recycling, to the producer of recyclable materials, thus rewarding recycling processes (Corona et al. 2019). Nevertheless, the EoL recycling approach is not fully consistent with the cut-off Ecoinvent database and may lead to double counting of recycling benefits (Nordelöf et al. 2019). Hence, the cut-off approach was preferred to ensure that multifunctionality is handled consistently both in the foreground system and the background system (i.e., Ecoinvent database).

2.4.2 Inventory analysis

The LCI was compiled by combining data from the rMEFA and datasets from Ecoinvent. The key modeling choices and assumptions made for the inventory analysis are discussed below. Additional information regarding the chosen datasets is provided in the Supplementary Information.

In constructing the LCI model, the metabolic flows included in the scope of the model were matched to the best available representative datasets in Ecoinvent. In this regard, datasets that represent end products close to the consumer, such as cars and laptops, were prioritized to better capture consumer-level impacts. In cases where such datasets were not available, datasets representing earlier stages in a product’s supply chain were selected (e.g., textile fabrics instead of finished clothing) following a similar approach to Dias et al. (2018). For any products lacking a suitable Ecoinvent dataset, custom-built datasets were created using information sourced from the literature or other reliable databases, or proxy datasets were used assuming that they represent the characteristics of the products (e.g., the dataset of tomato, processing grade, was used for frozen and canned vegetables assuming that a large share of canned vegetables is tomato sauce). Moreover, as the level of granularity of MEFA data was generally higher than the available LCI data in Ecoinvent, certain flows were aggregated into categories matching available Ecoinvent datasets (e.g., the dataset of breadcrumbs was used for all bread and bakery products).

It should also be noted that for the LCI, the market datasets of Ecoinvent were preferred instead of the cradle-to-gate datasets. The market datasets represent the consumption mix for a given product and region and include its transport from the producer to the consumer (Wernet et al. 2016). Therefore, they were selected to represent the regional or global supply chains of products used in Umeå, with regional datasets (e.g., European) being prioritized over global ones whenever both were available. This choice entails some double counting, as impacts from the transportation of goods within the urban boundaries (i.e., the last mile) are also accounted for in the transport sector. This is a limitation of the study. Nevertheless, it was preferred to use the market datasets to ensure that upstream impacts from the transportation of goods, which account on average for about 5–6% of the impacts in Ecoinvent’s market datasets (Steubing et al. 2016), are not disregarded.

Energy supply

The sector supplies grid electricity and heat, which is distributed through its district heating system. As all the produced heat is consumed locally to satisfy the demand within the urban area, all the used fuels and generated emissions related to district heat production were included in the LCI. For the electricity, it was assumed that the sector supplies all the electricity consumed in Umeå. This assumption builds on the results of MEFA that showed that the produced electricity by power plants located within the urban area is higher than the demand (3.1 TWh produced and 1.7 TWh consumed in 2021). Although, in reality, it is not possible to determine the exact sources of consumed electricity from the grid; this assumption was made to better understand what environmental pressures are exerted from the urban energy system. Because of this choice, only the fuel inputs and emissions associated with the production of the amount of electricity consumed in Umeå (1.7 TWh) were included in the LCI, as the rest does not satisfy the demand of the urban system. Moreover, to avoid double counting, the electricity and heat use was excluded from datasets representing activities in the foreground system.

Transport sector

The sector includes Umeå’s public transport system (buses) and taxis, road freight transport by vehicles registered in Umeå, and air travel by Umeå’s citizens. For the modeling of road transport activities, Ecoinvent datasets that represent activities complying with the EURO 5 emissions standard were chosen. Additionally, new datasets were created for means of transport that are not included in the database, e.g., hybrid vehicles.

Households

The inventory of this sector includes all the material inputs to households, which were categorized into food products, electronics and electrical appliances, textiles, packaging materials, and other goods. Since packaging was modeled as a separate category, it was excluded from other datasets of household goods to avoid double counting. In addition to the abovementioned materials, the household’s inventory includes flows related to private transport (i.e., vehicle flows and fuels) and fuels used by standalone heating systems.

Waste management

The sector includes waste collection, food waste management through anaerobic digestion, sorting of recyclables, green waste composting, incineration, landfilling, and management of waste electrical and electronic equipment (WEEE), end-of-life vehicles (ELVs), and hazardous waste. Since this sector includes the impacts of managing waste generated in Umeå, the end-of-life stage of products used within the urban area was excluded from their datasets to avoid double counting. Furthermore, for the modeling of recycling, the ratios of sorted materials that are recycled were assumed to be the same as in the rest of Sweden (Avfall Sverige 2020).

Other sectors

The environmental impacts due to drinking water supply and wastewater treatment in Umeå were considered in the corresponding sectors. Thus, the impacts due to the production of drinking water and handling of wastewater were excluded from datasets representing activities of other sectors to avoid double counting. Similarly, impacts associated with aggregates used in construction activities were excluded from the construction sector, as they were considered in the mining and quarrying sectors.

2.4.3 Life cycle impact assessment method

The ReCiPe 2008 Midpoint (H) impact assessment method (Goedkoop et al. 2009) was applied using a 100-year time horizon to convert the LCI data into impact categories. This method was chosen, as it includes a wide variety of midpoint impact categories and incorporates three different cultural perspectives to assess impacts; Hierarchist, Individualist, and Egalitarian (Goldstein et al. 2013; González-García et al. 2021). The Hierarchist perspective was assumed in this study, as it takes a middle ground regarding future environmental risk and is the most widely used approach of the method (Goldstein et al. 2013). Another reason for selecting ReCiPe is that it has been applied by previous UM-LCA studies (Goldstein et al. 2013; García-Guaita et al. 2018; González-García et al. 2021), thus facilitating comparison of results. Twelve midpoint impact categories were chosen for the assessment (see Table 1), considering their relevance to previous studies and aiming to provide a comprehensive view of the environmental pressures exerted by the urban system.

Table 1 The chosen impact categories from the ReCiPe impact assessment method and their abbreviations
Full size table

2.5 Prospective life cycle assessment

2.5.1 Goal and scope definition

The goal of the pLCA was to assess the environmental implications of implementing circular strategies in Umeå in the near future (2030). For this purpose, different future scenarios, both for the foreground and background systems, were defined. The three foreground scenarios represent the implementation (or non-implementation) of future circular strategies in Umeå, while the two background scenarios describe different developments in the background system. As with the rLCA, instead of defining an explicit functional unit for the foreground scenarios, the gross annual environmental impacts of the urban system were normalized at per capita units. To ensure that the three foreground scenarios are functionally equivalent, the “equal basket of benefits” approach (Vandermeersch et al. 2014; Barrera et al. 2016; Goronovski et al. 2018), which is based on the system expansion concept, was applied (see Sect. 2.5.3).

2.5.2 Scenarios

A significant challenge in pLCA is to ensure temporal consistency between the foreground and the background systems, which requires considering future developments in both systems (Arvidsson et al. 2018; Georgiades et al. 2023). According to Arvidsson et al. (2018), there are two main approaches to modeling future developments in the foreground and background system: (i) predictive scenarios that represent environmental impacts considering the most likely development and (ii) scenario ranges that illustrate potential environmental impacts, including extreme scenarios that represent minimum and maximum environmental impacts. In this study, the latter approach was used both for the foreground and for the background system. A schematic illustration of the analyzed scenarios is shown in Fig. 2. In the foreground system, three different scenarios were developed to evaluate environmental implications associated with the implementation (or non-implementation) of circular strategies in Umeå. In the background system, two different scenarios were implemented to examine how future developments in the background system could affect its environmental performance.

Fig. 2
figure 2

Description of the foreground scenarios (FS) and background scenarios (BS) and their possible combinations

Full size image

The foreground scenarios (FSs) include a reference scenario (FS1), where no circular strategy is implemented, and two scenarios (FS2, FS3), where circular strategies are in place in Umeå in 2030. FS2 and FS3 are focused on the construction sector and households, considering the results of rLCA that indicated these sectors as major drivers of environmental impacts in Umeå (see Sect. 3.1), and on waste management, considering its fundamental role in the CE. Moreover, the two scenarios were developed considering the 4Rs (reduce, reuse, recycle, recover) principles (Kirchherr et al. 2017) and the operational principles of CE (Suárez-Eiroa et al. 2019), which highlight not only the value of recycling and recovery in the CE, but also the importance of reducing the total inputs to the system. In this regard, FS2 represents a circular strategy emphasizing reduced material consumption, while FS3 represents a strategy emphasizing recycling. The FSs are as follows:

FS1 Baseline scenario

This scenario considers the urban system without the implementation of a circular strategy. Apart from population growth, no other change in the urban system is assumed until 2030 (i.e., the average metabolic flows per capita are the same as in 2021), and thus only the background system will change by 2030. Although this assumption is not realistic, as resource use patterns are ever-changing, it enables the use of S1 as a reference scenario, allowing to evaluate the effects of implemented circular strategies in S2 and S3 and understand the influence of changes in the background system.

FS2 Promoting measures to reduce material consumption

This scenario assumes that a circular strategy with measures primarily aimed at reducing the inflow of materials to Umeå is in place in 2030. The measures target households and the construction sector and promote reductions in the use of those materials that contribute to municipal solid waste (MSW) and construction and demolition (C&D) waste generation (i.e., no measures for fuels and vehicles are included). Measures for households may include developing platforms for food sharing, organizing campaigns promoting less wasteful food consumption habits, and creating second-hand markets and repair workshops for household items (e.g., furniture, electronics) (Petit-Boix and Leipold 2018; Suárez-Eiroa et al. 2019). Measures for the construction sector may include developing exchange platforms for excess construction materials, organizing circular construction capacity-building workshops, and introducing requirements to developers to reuse materials and reduce C&D waste (Petit-Boix and Leipold 2018; Williams 2019). By virtue of the implementation of the strategy, the levels of material use in households and the construction sector in 2030 are assumed to be 10% lower than in 2021.

FS3 Promoting measures to increase recycling. This scenario assumes that a circular strategy with measures aiming to increase the recycling of MSW and C&D waste, thereby diverting waste from landfilling and incineration, is in place in 2030. Measures to promote the recycling of MSW may include organizing awareness-raising campaigns to promote household recycling, installing more recycling bins, enhancing the infrastructure for separate waste collection (e.g., food waste, packaging), and optimizing sorting systems (Petit-Boix and Leipold 2018). Potential measures for the construction sector may include promoting the use of material passports, setting specific recycling targets for developers, creating dedicated areas for temporary storage and sorting of C&D waste, and introducing requirements to construction companies to apply selective demolition techniques (OECD 2020; Tirado et al. 2022). For MSW, the implementation of the strategy is assumed to lead to (i) an increase in the proportion of MSW collected for recycling (including biological treatment) from 45% in 2021 (Avfall Sverige 2022) to 65% in 2030 and (ii) an increase in the percentage of biologically treated food waste from 37% in 2021 (Avfall Sverige 2022) to 65% in 2030. These improvements reflect targets in the existing municipal waste plan (Umeå region 2020). For C&D waste, the implementation of the circular strategy is assumed to result in an increase in the waste collected for recycling and recovery from 73% in 2021 (Papageorgiou et al. 2024) to 85% in 2030.

The background scenarios (BSs) were derived from the premise tool (Sacchi et al. 2022), which generates prospective LCI (pLCI) databases through the combination of integrated assessment models (IAMs) with the Ecoinvent database (Wernet et al. 2016). IAMs are broad models that describe the complex relationships between human systems (e.g., food system, energy system) and the natural environment (Mendoza Beltran et al. 2020) using globally consistent future scenarios that illustrate possible socioeconomic, technological, and environmental developments (Steubing and de Koning 2021). Based on projections provided by IAM models, the premise tool performs transformations on energy-intensive activities available in Ecoinvent (e.g., electricity generation, cement, and steel production) to integrate the IAM scenarios into the background LCI database.

The pLCI databases in this study were generated by integrating data from implementing two future scenarios in the REMIND IAM (Baumstark et al. 2021) with the 3.8 cut-off version of Ecoinvent. The two background scenarios were chosen to represent two very different future pathways (a less optimistic and a more optimistic), thereby allowing us to explore how different potential developments in the background system affect the results of pLCA. More specifically, the scenarios represent the “Middle-of-the-road” shared socioeconomic pathway (SSP2) (O’Neill et al. 2014), which describes future socioeconomic and technological developments extrapolated from historical developments, and two representative concentration pathways (RCPs), which describe possible trajectories for radiative forcing from GHGs by the end of the century (van Vuuren et al. 2011). The less optimistic first background scenario (BS1) (the “baseline” scenario) represents the SSP2 without any climate change mitigation policies, which corresponds to RCP 6.0, in which the global radiative forcing reaches 6 W/m2 (global mean temperature increase 3–4 °C) by 2100. The more optimistic second scenario (BS2) represents the SSP2 complying with the more ambitious RCP 1.9, which is consistent with the Paris Agreement target to not exceed the remaining carbon budget of 500 Gt CO2-eq for limiting radiative forcing to 1.9 W/m2 (global mean temperature increase less than 2 °C).

2.5.3 Inventory analysis

For the background system, two pLCI databases were derived from the premise tool, one for each BS, representing the year 2030 (details regarding transformations on the Ecoinvent database performed by the tool are provided in the Supplementary Information). Next, the superstructure approach (Steubing and de Koning 2021) was used to convert the two individual pLCI databases into a single superstructure database, which is a regular LCI database that contains all the activities and exchanges (elementary and intermediate flows) of the two pLCI databases. The superstructure database was then used in the Activity Browser software (Steubing et al. 2020) together with a scenario difference file, which stores differences between the two pLCI databases and can be used to convert the superstructure database into one of the databases in a fast and practical way. This allows to link the foreground system to a single LCI database (the superstructure database) representing different background systems instead of multiple background databases, thereby simplifying the modeling process (Steubing and de Koning 2021).

For the foreground scenarios, the population growth in Umeå from 2021 to 2030 was assumed to be 10% in accordance with projections by (Umeå municipality 2022a). For FS1, no other changes were assumed, and thus the LCI was compiled based on processes of the rLCA model without changing the per capita metabolic flows. For FS2, the material inputs to households and the construction sector were reduced by 10%, excluding fuel and vehicle inputs to households, which are not targeted by the strategy (note though that indirect reductions in fuel consumption in the construction sector because of reduced usage of machinery and vehicles are accounted for). Furthermore, a linear relationship between material inputs and waste outputs is assumed, and thus the generated amounts of MSW and C&D waste are also reduced by 10% by 2030. For FS3, the quantities of different MSW and C&D waste fractions sent for recycling and recovery were adjusted in accordance with the assumed increases in this scenario (for more details see Supplementary Information).

Due to different assumptions behind the foreground scenarios, the outputs from multifunctional processes in each scenario are different. The multifunctional processes in the foreground system are the waste management processes sorting for recycling, waste incineration, and anaerobic digestion that handle waste and, at the same time, provide useful by-products, i.e., recyclables, electricity and heat, and biogas and stabilized sludge, respectively. As these processes produce different by-products (benefits) in the three foreground scenarios, the “equal basket of benefits” approach was applied to ensure that all scenarios deliver the same benefits. Following this approach, when a multifunctional process produced less or no by-products, it was assumed that the traditional supply chain (TSC) supplements the necessary products to complete the market demand (Barrera et al. 2016). In this way, the scenarios were made functionally equivalent and thus comparable. Table 2 shows how the “equal basket of benefits” (i.e., system expansion) approach was performed. More details regarding the estimated amounts can be found in Supplementary Information.

Table 2 Benefits from material and energy recovery and additional products from TSC (TSC: traditional supply chain) to achieve an equal basket of benefits in each scenario
Full size table

2.5.4 Life cycle impact assessment method

The ReCiPe 2008 Midpoint (H) impact assessment method was applied to convert the LCI data into the same 12 impact categories as in the rLCA.

3 Results and analysis

This section presents the interpretation of the results of the rLCA (Sect. 3.1) and pLCA (Sect. 3.2).

3.1 Interpretation of the rLCA results

3.1.1 System overview

The application of MEFA in a prior study (Papageorgiou et al. 2024) indicated that the construction sector and households are critical sectors in Umeå’s UM. The construction sector is the largest consumer of materials (i.e., construction materials) and the largest producer of solid wastes (i.e., C&D waste). The households sector is the largest consumer of energy and drinking water, the largest producer of fossil greenhouse gas (GHG) emissions and wastewater, the second largest consumer of materials, and the second largest producer of solid waste (MSW). These findings gave a first indication that the construction sector and households should be among the priority areas in a circular strategy, as they offer great opportunities to apply CE principles to optimize resource usage and reduce waste and emissions.

The application of UM-LCA backed these findings. Figure 3 indicates that households are the major drivers of environmental impact in Umeå, as they are the largest contributors to 11 out of 12 impact categories (ALO, CC, FD, FET, FE, HTT, MD, PMF, TA, TET, and WD). The construction sector, which MEFA identified as a key sector in terms of material consumption and waste generation, contributes considerably to most environmental impact categories, but not to the same extent as households. Apart from these two sectors, other sectors that play an important role in the overall environmental impacts of the urban system are the energy, transport, and waste management sectors, with the waste management sector being the largest contributor to OD. As regards the water supply, mining and quarrying, and wastewater treatment sectors, their impacts are negligible compared to the other sectors, except for water supply that accounts for about 20% of the total impact in WD due to the extraction of groundwater.

Fig. 3
figure 3

Contributions of the urban sectors to the per capita life cycle environmental impacts of Umeå (for abbreviations and impact categories, refer to Table 1)

Full size image

Considering the above, it can be suggested that a circular strategy should be focused on the households and construction sectors. In this regard, it is important to understand which flows and processes in these sectors should be prioritized in the strategy. For this purpose, an analysis of the main sources of environmental impacts in the households and construction sectors is performed in the next section to identify the most impactful flows and processes. Furthermore, given the fundamental role of the waste management sector in the CE, an analysis of the environmental pressures exerted by the waste management sector is also performed to identify potential areas that could be improved through the implementation of a circular strategy.

3.1.2 Sectoral analysis

Figure 4a shows the distribution of environmental impacts from households. Food inputs represent the most important source of impacts, as they are the largest contributors to eight impact categories (ALO, CC, FE, HTT, PMF, TA, TET, and WD), while they also contribute considerably to the other four (FET, FD, MD, and OD). Among food inputs, livestock-based products (meat, dairy, and eggs) contribute to more than half of the total impacts in nine impact categories (ALO, CC, FD, FET, HTT, MD, PMF, TA, and TET) (Fig. 4b). In the other impact categories, fruits and vegetables are the largest contributor to FE (mainly due to wastewater from vegetable oil production) and WD (due to irrigation), and fish products to OD (primarily due to the use of HCFC-22 refrigerant on fishing vessels). Besides food inputs, other notable sources of impacts from households are electronics and electrical appliances, private transport, and textiles. Electronics and electrical appliances contribute to almost 50% of environmental burdens in MD, 40% in FET, 25% in FE, and 20% in HTT. Private transport represents approximately 40% of impacts in FD, 35% in OD, and 30% in CC, mainly due to the use of fossil fuels, and 25% in FET, due to the inflows of new vehicles. Textiles are the second largest contributor to WD after food, representing almost 30% of the total impact. Overall, the findings above highlight that a municipal circular strategy should not only include measures to reduce food inputs, which are the largest flows to households, but also measures targeting smaller flows like electronics and textiles that cause considerable environmental burdens.

Fig. 4
figure 4

Distribution of environmental impacts from a households, b food inputs of households, c construction sector, and d waste management

Full size image

Figure 4c presents the environmental impacts of the construction sector. Asphalt and steel account together for more than 65% of environmental burdens in nine impact categories (CC, FD, FET, FE, HTT, MD, OD, PMF, and TA), with steel being the largest contributor to six of these categories (CC, FET, FE, HTT, MD, and PMF). Asphalt is also responsible for more than 40% of all burdens in WD, followed by concrete, which accounts for almost 30% of all burdens. From the other materials used in the construction sector, wood and wood products are responsible for almost 75% of impacts in ALO and paints for 25% of impacts in TET. As regards construction activities, they are the largest contributor to TET, mainly due to brake wear emissions from vehicle operation, while they also contribute considerably to CC, FD, OD, PMF, and TA due to fuel use in vehicles and machinery. It is, therefore, important for a circular strategy to promote efficient construction practices to reduce material consumption in the sector, given that the use of construction materials contributes to a range of environmental impacts, both directly and indirectly (e.g., through fuel use in construction activities).

Figure 5d displays the distribution of environmental impacts among different waste management processes. Waste incineration is the dominant source of impacts (approximately 80%) in CC, primarily due to the combustion of fossil-based waste materials, such as plastics. Moreover, it is the largest contributor to TA and WD and the second largest to FD, FET, FE, HTT, MD, PMF, and TET. The management of WEEE, ELVs, and hazardous waste is responsible for almost all the burdens in OD (due to handling of domestic refrigerators), 70% of the impact in FET (mainly due to incineration of residues from mechanical treatment of WEEE) and 50% of impact in FE (primarily due to the incineration of hazardous waste). Waste collection is the largest source of burdens in ALO and TET and the second largest in WD (chiefly due to the use of biofuels), and sorting is the largest source in FD and PMF (due to use of fossil fuels) and in MD (mainly due to usage of steal for maintenance of facilities). Finally, landfilling is responsible for more than half of the impacts in HTT due to the leaching of contaminants in groundwater. These findings indicate that even though Umeå has a well-developed waste management system, which mainly relies on recycling and energy recovery from waste, there are still environmental pressures associated with waste management activities, especially incineration. Thus, a circular strategy should prioritize waste prevention and reuse to reduce the amounts of waste collected for handling.

Fig. 5
figure 5

Environmental impacts of all scenario combinations normalized to FS1_BS1 scenario (FS1_BS1 = 100%)

Full size image

3.2 Interpretation of the pLCA results

Figure 5 shows the results of the pLCA from the implementation of the foreground scenarios FS1-FS3 based on BS1. It indicates that a circular strategy focused on promoting reduced material consumption (FS2) could lead to reductions in all impact categories, ranging from 4.3% for FD to 8.6% for ALO. A circular strategy focused only on recycling and recovery (FS3) could also lead to reduced impacts, albeit smaller ones (ranging from 0.2% for TA to 1.2% for HTT). The most notable reductions in HTT (1.2%) and CC (1.1%) are mainly attributed to reduced emissions from diverting materials from incineration. This indicates that although increased recycling leads to reduced energy from waste, compensating for this energy loss by using other energy sources (as done based on the “equal basket of benefits” approach), it does not create nearly the same level of impact as waste incineration. It is also noteworthy that a strategy focused on recycling and recovery causes a slight increase in TET (0.1%) (see also Table 3) due to higher consumption of biofuels (rape seed oil) for transporting increased amounts of recyclables after sorting.

Table 3 The environmental impacts of the assessed scenarios in absolute values (the bold numbers indicate the best-performing scenario, and the values in the parentheses represent additional impacts due to system expansion)
Full size table

The figure also presents the results of the FSs based on BS2. The relative performance of these three scenarios under BS2 is similar to that under BS1, with FS2 providing higher reductions than FS3. This does not mean that efforts towards recycling and recovery should diminish. It rather emphasizes the need to develop circular strategies with initiatives aiming to reduce material use. This is in line with circular principles, like the 4Rs (reduce, reuse, recycle, recover) principles (Kirchherr et al. 2017) and the operational principles of CE (Suárez-Eiroa et al. 2019), which prioritize reductions in resource consumption.

In addition, the comparison of all scenario combinations emphasizes the influence of the background system in the results. Between the two background scenarios, only the optimistic BS2 entails ambitious policies that could lead to transformations contributing to climate change mitigation (see Supplementary Information). Such transformations could lead to reductions in fossil fuel use and create environmental co-benefits, e.g., less flows of phosphorus from lignite mining. By virtue of these changes in the background system, the environmental burdens in CC, FD, FET, FE, HTT, MD, OD, PMF, and TA for the three foreground scenarios are lower in BS2, compared to the “baseline” BS1 (see also Table 3). At the same time, these changes entail trade-offs, as they cause increased impacts in TET, ALO, and WD, mainly due to increased biofuel consumption for transport activities in the background. These findings highlight that the environmental performance of urban areas is strongly dependent on the broader socio-economic context, indicating that a broad perspective is necessary when performing this type of assessment.

4 Discussion

4.1 Towards holistic circular economy assessments in urban areas

The application of the UM-LCA approach to Umeå demonstrates that the coupling of MEFA with LCA under the UM perspective can provide decision makers with a useful tool to support the design, evaluation, and monitoring of environmentally sound circular strategies.

The metabolic perspective facilitates a better understanding of how the urban system functions and metabolizes material and energy inputs into outputs (products and residuals). This perspective can help decision makers understand that a sustainable CE transition requires to design holistic strategies that not only aim to reduce the outputs of the system through the recirculation of materials, but also to reduce its inputs through reduced and more efficient use of resources. Moreover, the conceptualization of the urban system based on UM offers a metaphorical framework to study the system from a systems perspective enabling its modeling as comprising different interacting sectors. This can enhance the comprehensibility of analysis and ease the communication of the results (Schwab et al. 2017). Moreover, it could help decision makers focus on specific sectors and design sector-specific circular measures, similar to the European Commission’s Circular Economy Action Plan (EC 2020), which prioritizes key sectors of the economy.

The application of MEFA provides quantification of materials and energy flows, which not only functions as the basis of the LCI, but also helps identify key metabolic actors and flows within the urban system. For Umeå, the application of MEFA (see Papageorgiou et al. 2024) showed that the households and construction sectors are key metabolic actors, indicating that these two sectors should be among the priority areas in a circular strategy. However, although MEFA can be useful in identifying key metabolic flows and actors, thereby indicating opportunities for CE, it is not sufficient to support the design of environmentally sound circular strategies, as it only accounts for metabolic flows in mass and energy terms without considering embodied environmental burdens.

The use of LCA addresses this limitation, as it allows to assess environmental burdens associated with metabolic flows and activities in the urban system from a life cycle perspective. In this way, it helps moving beyond mass and energy as proxies of environmental impacts (Goldstein et al. 2013), enabling the assessment of sectors and flows based on environmental pressures they exert. For Umeå, the application of LCA indicated that even though households metabolize less materials than the construction sector, they are the main drivers of environmental pressures from the urban system, as they use materials with higher embodied environmental burdens. It also revealed that besides food inputs, which are the largest material inputs (excluding drinking water) to households, other smaller material inputs, such as electronics and electrical appliances and textiles, also contribute considerably to environmental impacts from this sector. These findings indicate the importance of designing a circular strategy with a holistic scope that includes measures targeting not only the largest flows, but also the flows causing environmental pressures.

The application of LCA with a future-oriented scope based on future scenarios enhances further this potential, as it allows to assess potential environmental implications of different circular strategies in the future, helping to identify the most environmentally beneficial options. For example, the pLCA in this study highlighted that a circular strategy promoting reduced material use (FS2) can offer higher environmental benefits than a strategy focusing only on recycling (FS3), thus corroborating the importance of reducing consumption. Importantly, the pLCA also shows that increased recycling in FS3 could even introduce more impacts than the baseline (e.g., TET). This indicates the critical role of LCA in identifying such trade-offs, underscoring its potential to support decision makers in designing circular strategies by providing them with comprehensive information.

Furthermore, the analysis of the background system based on the two BSs revealed that the future environmental performance of the urban system strongly depends on changes in the background system, indicating that decision makers need to define the level of ambition of a circular strategy while also considering the broader systems context. For example, although a reduction in material use levels in households and the construction sector by 2030 (FS2) may bring reductions in most environmental impact categories, it may not significantly improve the environmental performance of Umeå, if the global socioeconomic system develops in accordance with BS1. In this case, perhaps more ambitious reduction targets in material use may be needed, considering also the ambitious environmental goals of Umeå, e.g., to become climate neutral by 2040 (Umeå municipality 2022b).

Besides providing decision makers with a useful tool to support the design of municipal circular strategies, the UM-LCA approach could also function as a tool for monitoring the performance of implemented strategies. MEFA can help track temporal shifts in material and energy flows from the implementation of a circular strategy and measure progress toward defined targets. The coupling of MEFA with LCA can expand the monitoring scope by including environmental indicators (e.g., midpoint impact categories) that allow to also track temporal shifts in environmental impacts of the urban system. In this way, the UM-LCA approach can provide comprehensive information that could help decision makers holistically evaluate the efficacy of implemented circular strategies and decide on potential amendments.

As a final note, it is important to highlight that the UM-LCA offers opportunities to examine synergies between CE and climate neutrality. According to EC’s Circular Economy Action Plan (EC 2020), there is a need for appropriate modeling tools to assess the potential benefits of the CE on GHG emission reductions. As demonstrated in this study, the UM-LCA offers a comprehensive modeling framework to account for GHG emissions associated with urban flows and activities taking a life cycle perspective. Thus, it could serve as the basis of a tool to analyze how the implementation of the CE at the urban level can contribute to achieving climate neutrality.

4.2 Limitations

An inherent limitation of the UM-LCA approach is that it is highly dependent on data availability. The UM-LCA is a bottom-up approach to model urban systems that requires extensive data on material and energy flows at the urban level. However, this type of data is not always readily available, which can complicate the application of UM-LCA and may make it necessary to limit the scope of the analysis. For example, in the case of Umeå, the lack of data did not permit the inclusion of agriculture, forestry, fishing, manufacturing, and service sectors in the model, reducing the completeness of the assessment. Furthermore, the process-based modeling approach of UM-LCA requires the use of LCI datasets that adequately represent the environmental impacts of all flows. Nevertheless, considering the large number of metabolic flows included in a UM-LCA study, it is practically impossible to model all of them using fully representative datasets. For example, even though Ecoinvent is one of the most comprehensive LCI databases available, it does not include fully representative datasets for all the metabolic flows determined through MEFA in this study. This necessitated the use of proxy datasets (for more details see Sect. 2.4.2 and Supplementary Information), which contributed to uncertainties.

Another limitation of the UM-LCA approach is that it cannot spatialize the metabolic flows in the global economy to the same extent as multiregional input–output approaches do (see for example Bahers & Rosado 2023). Due to this limitation, UM-LCA cannot provide a detailed description of the metabolic relationships of the urban system with its hinterlands, making it difficult to understand the extent of environmental pressures exerted by the urban system outside its boundaries. Thus, it cannot support the design of circular strategies with specific measures and targets aiming to reduce the externalized impacts of metabolic flows.

A third limitation is that the approach to normalize the gross annual environmental impacts at the per capita level, instead of defining a traditional functional unit (see Sect. 2.4.1), does not allow the consideration of social and economic aspects in the assessment. More specifically, it does not allow to consider differences in the quality of life, income, diets, and consumption habits of urban residents (Goldstein et al. 2013; Albertí et al. 2019). This limits the ability to identify drivers of environmental pressures from different social and economic groups within the urban system. Moreover, it only provides an understanding of how implementing circular strategies can contribute to environmental sustainability, but not to economic and social sustainability. To address such issues Albertí et al. (2019) proposed to use the City Prosperity Index (CPI) when defining the functional unit. Although this approach can provide a more holistic perspective, its application is, for the moment, difficult due to a lack of sufficient data, as the CPI is only available for a limited number of urban areas worldwide and the most recent data are from 2016.

Finally, a limitation of the pLCA as performed in this study is that it does not consider in the modeling of the foreground scenarios how specific measures to promote reduced consumption and recycling could be deployed in the urban system. For example, it does not consider the increased requirements for materials and energy from developing and operating sharing platforms, second-hand shops, and repair workshops or from installing more recycling bins. This limits the ability of the model to capture potential burden shifting between life cycle stages or environmental impacts, which in turn may lead to overestimating the potential environmental benefits of circular strategies. Modeling thus the deployment of such measures in the urban system would enhance the comprehensiveness of the assessment but, at the time of the study, it was not possible due to a lack of relevant data.

4.3 Uncertainties

The UM-LCA approach is based on the coupling of MEFA with LCA. Uncertainties are unavoidable both in MFA (Mehta et al. 2022) and in LCA models (Rosenbaum et al. 2018). As discussed in our previous study (Papageorgiou et al. 2024), the MEFA model suffers from uncertainties, especially model uncertainty, due to simplifications and assumptions, and parameter uncertainty, due to uncertainty of input data. The coupling of MEFA with LCA increases further the degree of these uncertainties considering the additional data, assumptions, and simplifications that are required for the modeling and the fact that some of the Ecoinvent datasets used to compile the LCI may not be fully representative of the metabolic flows. Additionally, LCA requires normative choices associated with allocation, functional unit, system boundaries, inventory data, and characterization methods that lead to scenario uncertainty (Björklund 2002; Huijbregts et al. 2003), while its application with a future-oriented scope causes epistemological uncertainty, as it entails modeling systems in the future, which is inherently uncertain (Björklund 2002).

To deal with epistemological uncertainty in the pLCA, this study uses future scenarios representing possible futures in the foreground and background systems. According to Mendoza Beltran et al. (2020), this approach helps to acknowledge rather than to reduce epistemological uncertainty by exploring future pathways and associated impacts instead of predicting them. In this way, the results of the assessment can outline and better inform directions for action in policy-making, which was the purpose of the current study.

While the use of future scenarios helped to portray epistemological uncertainty, it did not reduce this type of uncertainty, nor parameter, model, and scenario uncertainties. Thus, the results of the assessment should not be considered absolute, but rather indicative and they should be interpreted cautiously. Nevertheless, it should be highlighted that, despite the uncertainties in the results, the value of UM-LCA is indisputable, as it can provide a comprehensive understanding of the environmental performance of the urban system.

4.4 Comparison with previous studies

Table 4 compares the results of the present study with those of earlier studies (Goldstein et al. 2013; García-Guaita et al. 2018; González-García et al. 2021) that used the UM-LCA approach to assess the environmental performance of different urban areas worldwide. The comparison is based on the results of the rLCA, as previous studies only applied the approach retrospectively, and is only performed at the urban system level (i.e., without comparing sectors), as previous studies did not model the urban areas at the sectoral level. The included impact categories are the common ones between the studies, i.e., CC, FET, FE, HTT, OD, and TA of the ReCiPe method.

Table 4 Comparison of estimated per capita environmental impacts in Umeå from the rLCA with environmental impacts reported in previous studies
Full size table

The table shows that the estimated per capita CC impact in Umeå is comparable to the impacts in most of the other studied urban areas (i.e., Santiago de Compostela, Madrid, Hong Kong, Cape Town, and London), with Umeå having the second lowest impact among them. For TA and FE, the per capita impacts in Umeå are relatively close to the ones reported for Santiago de Compostela and Madrid, while, for FET, the impacts in Umeå and Madrid are almost equal. By contrast, larger differences are notable in OD and HTT between Umeå and Santiago de Compostela and Madrid, and in FET between Umeå and Hong Kong, Cape Town, London, Beijing, and Toronto. These differences can be attributed to variations in socio-economic conditions, lifestyles, infrastructure, and technological development in the studied areas, and to variations in modeling choices (e.g., system boundaries, assumptions) and data quality. Despite these differences, the results of the current study are generally within the same order of magnitude as the results of the previous studies indicating the validity of the UM-LCA model developed in this study.

5 Conclusion and outlook

In this study, we apply the UM-LCA approach to Umeå aiming to explore its potential to support the design and monitoring of circular strategies at the urban level. More specifically, we combine MEFA with LCA under the perspective of UM to assess retrospectively the environmental performance of Umeå to inform the design of circular strategies and to evaluate prospectively environmental implications of implementing circular strategies in the future.

The main conclusion of this study is that the UM-LCA approach has great potential to support decision makers in their efforts to design, evaluate, and monitor circular strategies at the urban level. The combination of two core industrial ecology tools, MEFA and LCA, provides relevant and complementary information, as MEFA allows to map and quantify metabolic flows throughout the system with its sectors, while LCA allows to assess environmental burdens associated with these flows from a life cycle perspective. This type of information can help decision makers identify key metabolic flows and sectors allowing them to design circular strategies with flow- and sector-specific measures and targets. Furthermore, the application of UM-LCA with a future-oriented can provide enhanced insight into how potential future circular strategies could alter the overall environmental performance of the urban system, thereby helping to identify the most environmentally sound strategies. At the same time, the coupling of MEFA with LCA provides decision makers with a valuable monitoring tool to track temporal shifts in metabolic flows and environmental impacts, which could help them evaluate the efficacy of implemented circular strategies.

Nevertheless, our application of UM-LCA to Umeå indicated also certain limitations of this approach to study urban systems in the context of the CE. More specifically, it emphasized its strong dependence on data availability, which is often poor, its limited potential to comprehensively analyze the interactions of the urban system with its hinterlands, and its inability to consider social and economic aspects in the functional unit. Future research should aim at exploring how the availability and accessibility of urban-level data can be improved through, for example, standardized classification of material and energy flows and data collection procedures. It should also explore how the UM-LCA can be integrated with multiregional input–output approaches and how social and economic aspects can be included in the functional unit. Furthermore, future research could investigate how the environmental indicators analyzed in LCA (e.g., midpoint impact categories) could be complemented with indicators that can help assess aspects related to the CE, such as waste management performance, urban governance, and innovation, to fully capture the multiple dimensions of the CE transition.