Assessment of sustainable drainage strategies in urban water systems using urban water metabolism and multi-criteria decision analyses

COD

Chemical oxygen demand

CP

Compromise programming

CSO

Combined sewer overflow

DS

Drainage strategies

U.S. EPA

United States Environmental Protection Agency

GHG

Greenhouse gas emissions

IUWM

Integrated urban water management

KPIs

Key performance indicators

MCDA

Multicriteria decision analysis

PAST

Paleontological statistics

PCA

Principal component analysis

SC

Subcatchment

STO

Storm tank overflow

SUD

Sustainable urban drainage

TCSO

Total of combined sewer overflow

TfrWWTW

Treated outflow from wastewater treatment works

TRUST

Transitions to the Urban Water Services of Tomorrow

TWD

Total water demand

UWS

Urban water systems

WMOST

Watershed Management Optimisation Support Tool

Holistic sustainability assessment is essential for urban water systems (UWSs) to analyse the impact of various external factors and intervention strategies on the overall system performance (Venkatesh et al. 2015). Although the main objective is to improve the UWS performance, different perspectives should be observed since a massive use of resources is needed to change negative impacts on the environment, requiring a long-term multi-objective sustainability application (Behzadian & Kapelan 2015a). According to the United Nations, this approach is essential for sustainable development and accelerating climate change adaptation, for example, by investing in parks and urban agriculture to mitigate floods and heat islands (IPCC 2022).

Sustainable solutions have multiple advantages for UWS (Putri et al. 2023) and, hence, can be considered as the main alternatives to mitigate multiple related problems such as overflows, inundations, and flooding episodes, and promote urban resilience (Kourtis & Tsihrintzis 2021). Integrating existing conventional and sustainable systems is the best alternative, bringing multifunctionality to the system (Browder et al. 2019; Yang & Zhang 2021). Therefore, an unusual approach to performance analysis is necessary, aiming at the urban dynamics and adopting a holistic view to consider different directions (Behzadian & Kapelan 2015a).

Research on the planning process is crucial for decision-makers to enhance performance, standardise the processes, and simultaneously to reduce the dependence on conventional systems. Integrated urban water management (IUWM) is currently a known concept for decision-makers, addressing the impacts of changes on all indicators in UWS components. Implementing combined solutions requires collaborative analyses, as underscored by Mosleh & Negahban-Azar (2021). An integrated approach to UWS management has multiple benefits. Nowadays, a great diversity of available tools can show the efficiency of urban drainage systems. Nevertheless, these tools are usually chosen based on the goals of the decision-makers, making it challenging to standardise the impacts and their respective controlling metrics (Mosleh & Negahban-Azar 2021).

Recent studies have analysed the urban drainage system performance in various indicators for analysing any conventional or sustainable drainage methods. They selected these indicators based mainly on the sustainability criteria in the literature, such as diffuse pollution, environmental impacts, quantity or quality indicators, professional consulting, or even randomly, as demonstrated by Yang & Zhang (2021), Seyedashraf et al. (2021), Ariza et al. (2019), Wang et al. (2017), and Novaes (2016). Several studies have focused on the decision-making between sustainable methods based on the system performance with respect to different key performance indicators (KPIs) (e.g., Santos et al. 2021, Seyedashraf et al. 2021, Yang & Zhang 2021, and others). However, the selection of KPIs has been usually arbitrary among large areas such as social, economic, and environmental aspects.

A dynamic metabolism model was developed to carry out the performance analysis of a sustainability assessment tool in the urban water department that considers the environmental change caused by urbanism and its impact on several fronts – such as water quality, climate change mitigation and adaptation, and environmental life cycle assessment – and also to improve their operations within the water–energy–carbon nexus (Venkatesh et al. 2014). Improvements that change these aspects of economic, social, and sustainable dimensions are mainly required to enhance system's performance (Venkatesh et al. 2014).

Urban metabolism refers to the chemical and biochemical processes that maintain life, which is similar to the human body definition of the word. A water metabolism corresponds to a circulatory process where input is transformed, generating some output, while providing something to the system (Venkatesh & Govindarajan 2014). The metabolism-based approach considers the mass stocks in the system and the inflows and outflows (of materials, energy, chemicals, emissions, and waste generation), considering the life cycle assessment and the urban water dynamics (Venkatesh 2013). The urban water metabolism-based approach behaves in the same manner but focuses on urban water flow and its processes (Venkatesh et al. 2014).

A comprehensive analysis of the UWS, which includes the urban drainage system, can be conducted through metabolic analysis, which means the number of resources that the UWS needs and how they can transform the natural conditions of the environment. This could be investment demand, operating costs, materials and chemical products, or change impacts, such as contaminant discharge and general emissions (Behzadian & Kapelan 2015b).

Studies on urban metabolism within the realm of UWS, including their subtypes like urban drainage systems, are conspicuously scarce. Our observation reveals a prevailing trend of limited analyses involving single KPIs and specialised publications focusing on specific facets of water systems, such as water supply or wastewater, urban planning, and management. Few studies have focused on urban drainage systems inside the urban metabolism concept, such as Fenner et al. (2019), Huang & Hsu (2003), and Soto (2019). However, they still have not focused on the urban metabolism performance analysis as a basis to KPI selection for the stormwater system decision-making.

Therefore, this highlights the knowledge gap in sustainable urban drainage methods that include and delimit all these criteria and indicators. As a result, this study sets out to establish a framework for validating the selection of drainage strategies (DSs) and the relevant sustainability-based KPIs for their analysis. This was shown through an appropriate tool to evaluate the effects of the selected strategies on the system performance indicators. This can be used in the decision-making process to identify the best strategies in urban drainage systems.

A recent review of urban system performance analysis assessed the tools and how they are being used to analyse UWS. Notably, two cutting-edge tools emerged as key players in evaluating IUWM: the Watershed Management Optimisation Support Tool (WMOST) and the WaterMet² Model. The WMOST was created in 2013 by the United States Environmental Protection Agency (U.S. EPA) to serve as a public domain and is focused on local and specific watersheds. However, the tool primarily emphasises urban water quantity aspects within sustainable design, not being able yet to calculate water quality aspects and wastewater treatment options. These omitted functions are crucial, making it insufficient for the specific objectives of this study (Detenbeck et al. 2013). WaterMet² was developed in 2014 as part of the EU TRUST project (Behzadian & Kapelan 2015b) as a stand-alone tool, enabling metabolic analyses at both the city and large basin levels (Mosleh & Negahban-Azar 2021).

The WaterMet² software is specific for metabolism analysis and is shown as a potential tool to analyse the KPIs, being capable of measuring diverse system flows (water flow, energy flow, greenhouse gas emission (GHG) flow, acidification/eutrophication flow, chemical flux, pollutant flux, and quantifying the pipeline material flux). The model allows the quantification of many of these indicators, such as the operational and maintenance costs, or any risks and intervention analysis for the components (Behzadian et al. 2014).

This study also seeks to leverage the concept of urban metabolism as a decision-making tool in the planning phase, KPI selection, and adaptation processes. Anchored in urban transformation dynamics and their environmental repercussions, it serves as the underpinning for performance analysis, encompassing and underpinning commonly employed indicators. Our primary objective is to employ the WaterMet² tool to scrutinise the performance of sustainable urban drainage systems (SUDSs) through a metabolic lens, with the ultimate aim of elucidating how this approach can guide decision-making among various SUDS alternatives.

The WaterMet² mechanism uses a water mass balance-based approach to simulate the urban fluxes and quantify the UWS performance. Likewise, it considers the operational dynamics of the system, including system construction, production, and maintenance. It is used to measure quantitative indicators based on the system's metabolic and sustainability performance, as well as other metrics. Its strength lies in the analysis of integrated UWS (Behzadian & Kapelan 2015a). It can analyse different flows, including water, energy, GHG emissions, pipe materials, chemicals, and pollutants. Furthermore, these fluxes are analysed temporally and spatially within the city subsystems (Behzadian et al. 2014). The details of the flows, their functions, equations, datasets, and KPIs can be found in the WaterMet² Report (Behzadian et al. 2014).

For building and simulating a WaterMet² model, an input dataset containing weather data and various parameters of the UWS are required, such as flow rates and local characteristics of subcatchments and water demands (Behzadian et al. 2014). For specific details on the dataset and spatial data, references can be found in the Supplementary Material.

For this step, we constructed the structure assuming a single unit for each UWS component, and we utilised the default values provided by the system for the dataset. To simulate an ideal urban area in this analysis, we assumed that half of the area consists of commercial activities, while the remaining half comprises residential areas. This is based on the default data obtained from São Paulo City, as it represents one of the most urbanised areas (IBGE 2010). Other relevant details can be found in the Supplementary Material.

In this phase, we established a simplified structure, considering a singular unit for each UWS component, and adopted the default dataset values provided by the system. In our simulation of an ideal urban area, we assumed an equitable division with half of the area designated for commercial activities and the remaining half allocated for residential purposes. Our selection of default data was grounded in insights gathered from São Paulo City and chosen for its representation of one of the most urbanised regions (IBGE 2010). More comprehensive elaboration can be found in the Supplementary Material.

We selected one KPI from each category, with specific subcategories chosen for the running simulation. These subcategories include water balance, fraction of water demand delivered, sewer system balance, energy, GHG emissions, acidification, eutrophication, contaminant load, and sludge generation. For the plot configuration, we chose to represent values in absolute terms in order to achieve a clearer understanding of the impacts.

Initially, we employed sensitivity analysis to simulate a 100% disturbance in different subcatchments and wastewater variables. These changes resulted in a response in the generated KPIs. In the sensitivity analysis, we examine the cumulative values of these KPIs within a planning horizon of 5 years to assess short-term impacts.

We chose the principal component analysis (PCA) method for this evaluation due to the large number of system variables. PCA is a technique that reduces the complexity of a dataset while retaining its most essential information. It achieves this by generating new variables called principal components (PCs), which are combinations of the original variables that have a significant impact on the process (Ringnér 2008). We selected the PAST (Paleontological Statistics) software because of its user-friendly interface and compatibility with Excel input dataset files. This software simplifies spreadsheet data entry, specifically for the disturbance results, and facilitates the calculation of PCA multivariate statistics (Hammer et al. 2001).

The local administration has established guidelines to address these drainage issues through municipal laws, although the implementation process has been progressing slowly. These guidelines encompass various measures, such as land use policy, sustainable recovery of valley bottoms, containment works, administrative measures for water containment (including the control of land use by basin and the control of works), channel dredging, incentives for cisterns use and retention devices, as well as ‘zero impact’ policies in new neighbourhoods and subdivisions, encouragement of drainage pavement projects, conservation units, and rainwater reuse in existing buildings. Of particular note is the mandatory requirement for rainwater reuse in new buildings that exceed 300 m² (PMB 2017b).

The classification data were structured in a manner that closely resembled the sensitivity test. The area was divided into 13 subcatchments, each corresponding to specific city zones numbered from 1 to 13. For more information on subcatchment details, refer to the Supplementary Material. A summary of the water system scheme for the study is presented in Figure 3. The water supply system is represented by a single surface water abstraction point, with various reservoirs interconnecting before supplying the subcatchments (DAE and HIDROSAN 2014a). On the other hand, the wastewater system operates independently by collecting surface runoff through pipelines and discharging it into receiving water bodies. A similar approach is followed for the storm drainage flows (Ampla 2016a).

Seven DSs were suggested, with the first one closely resembling the conventional system, i.e., do ‘nothing’ or business as usual (BAU). This strategy was the first one used in the case study, as detailed in Table 1. The WaterMet² model is built by setting fixed values and UWS components for water supply and wastewater, as well as storm drainage configurations. Subsequently, we modified and adjusted the strategy characteristics, as outlined below. Detailed instructions on how to complete the WaterMet² forms can be found in the Supplementary Material.

Our primary objective is to provide a robust basis for well-informed decision-making and to facilitate a meaningful comparison of these drainage system concepts. Hence, it is crucial to thoroughly investigate the metabolic implications of each individual model. As a result, in devising our scenarios, we concentrated solely on distinct system types, as delineated and subsequently simulated below.

ST02: the constructed wetlands are recommended for stormwater treatment, mainly the surface water flow wetlands (Fletcher et al. 2020), which is chosen for this study. This type of wetland contains a specific type of vegetation that slowly filters the water as it passes through, degrading and absorbing pollutants (Fletcher et al. 2020). Since constructed wetlands for runoff water treatment require a significant amount of space due to the need for a sedimentation basin and an outlet structure (Lopez et al. 2018), this study focuses on their application only in park zones in the study area. The park areas are transformed into constructed wetlands, accounting for 50% of the total area. These wetlands are designed based on a model from an external study, which has a size of 1,910 m² and a depth of 2 m (including 10 cm of free edge). This design allows for the treatment of a total of 3,629 m³ of treated runoff (Lopez et al. 2018).

ST03: rain barrels are implemented in all the residences and businesses in the region. The idea is aligned with certain government programmes that have shown progress in providing incentives to install cisterns in residences located in semi-arid areas to meet the water needs (Secretaria Nacional de Desenvolvimento Regional e Urbano 2019; ASA 2022) that are 20 L per capita according to the World Health Organization (Herkenhoff 2020).

ST04: infiltration trenches are also called filter drains, which are considered an application adjacent to roads in their centre square and in the side bed. The trenches are designed with 1 m depth, which also aids in pollutant removal (Ballard et al. 2015).

ST05: the selected permeable pavements in this strategy consist of porous asphalt, which is applied to 50% of all pavements. Even when compared to cases where the sole function is flood attenuation through infiltration, it still has a positive impact. This is primarily due to its capacity for pollutant removal. In this case, the total infiltration type involves direct infiltration into underground waters, without the need for an additional drainage system (Ballard et al. 2015).

ST06: green spaces are assumed in this strategy in all the available areas (Figure 4), and a method involving the creation of ditches with appropriate depths is applied with infiltration directly to groundwaters. In this strategy, the water will infiltrate, and this type of SUD can be applied in many places due to the variety of applicable surface materials (Ballard et al. 2015).

ST07: this strategy includes a centralised runoff reservoir similar to those proposed for storage tanks and rain barrels. Since the centralised system usually has a significant volume of water and storage size, a water pump is used to return this water back to the water system. The water pump for this simulation is selected based on the ones used for system correction in the city, with a consumption rate of 0.014 kWh/m³ (PMB 2019), which can be used for non-potable public uses.

The BAU is simulated in WaterMet² using values from the water supply, wastewater (including the drainage system), and subcatchment datasets. Subsequently, based on the sensitivity results, each of the six strategies is simulated individually by modifying the specific settings of relevant strategies. The variables used in the tool comprise aspects of the drainage system, wastewater data (specifically related to treatment and drainage system types), water supply data (pertaining to water reuse), and subcatchment data (relevant to infiltration or structural changes). This range of modifications and input methods is aligned with the objectives of this study. The input system characteristics of different drainage system types provide metabolism performance of the UWS due to the storm drainage alterations.

Both the case study and sensitivity tests are conducted with a short-term planning horizon of a period of 5 years (SMDU 2012). While using an annual time step for analysing the KPI, the comparisons between strategies are made using only the accumulated values for the entire planning horizon. We also select those UWS components to ensure a comprehensive evaluation of the entire urban system and generate accurate results.

The values generated by the WaterMet² enable us to quantify the impact of each strategy on the KPIs. Subsequently, we utilise a multicriteria decision analysis (MCDA) to rank these strategies based on their performance. The KPIs are distributed evenly based on the fluxes, considering the sub-criteria outlined in Table 2.

The compromise programming (CP) method involves measuring the distance of proposed alternatives from an ideal point in order to achieve an accurate solution that aligns with the preferences of decision-makers (Behzadian & Kapelan 2015a; Cecagno et al. 2019). Given the progressive nature of objective attainment and the inherent challenge of reaching the ideal point, this work used the nearest one known as the ‘goal point’ (Cecagno et al. 2019).

After running the model simulation in WaterMet², results are generated with a minimal number of inputs, demonstrating its capability to simulate UWS of any size using a streamlined dataset. Additional information and graphics can be found in the Supplementary Material.

The results show that the total monthly energy consumption ranges from 138,000 to 143,400 kWh. Notably, electrical energy accounted for 80% of the total energy, and the total GHG emissions exhibit a similar pattern, with 96% of GHG emissions originating from electricity and embodied energy. This is often a result of the use of chemicals in the processes and pipeline materials.

It is evident that the major contribution of energy and GHG emissions in the UWS is from electricity consumption, which highlights the need for concentrated efforts in this section. To put it into perspective, the average monthly electricity demand of 140,70 kWh and GHG emissions of 29.50 tonnes CO₂-eq are equivalent to the annual electricity consumption of seven households and the burning of 26.29 gallons of diesel fuel, respectively (EPA 2022).

Based on these results, for the next analysis, we only considered the total values of eutrophication, acidification, GHG emissions, energy consumption and contaminants. We also looked at the single result of sludge generation, as well as the multiple results of total water demands, total leakage, fraction of water demand delivered, total combined sewer overflows (TCSO), and total storm tank overflow (STO).

The short-term sensitivity simulation results for each KPI under disturbance were analysed using the PAST software to identify the PC. Following the PCA quantification, it was found that PCs 1, 2, and 3 collectively accounted for 98.20% of the cumulative variance proportion in the data, indicating their significant influence as the most influential variables.

The variables that most affected PC1 were irrigation and other water demand, frost tapping, industrial and commercial demand, rehabilitation, electricity, fossil fuel usage, treatment capacity, and storage capacity. In the case of PC2, the most influential variables included the number of properties, indoor water demand, occupancy, infiltration coefficient, and exfiltration. On the other hand, for PC3, the total area emerged as the predominant factor. Notably, the transmission capacity was the only variable that did not influence the principal component and, consequently, the KPIs. This underscores the importance of all variables in the process. Further details can be found in the Supplementary Material. Therefore, in our case study analysis, we meticulously considered all variables in accordance with the analysed strategies.

Table 3 displays the WaterMet² results for the case study, which corresponds to the UAB. This table presents the absolute values of the KPIs. These results have been also validated by comparing them with corresponding measurements of the same components.

In 2013, the Water and Wastewater Department (DAE) of Bauru City published a report containing annual data for the water supply and wastewater systems. According to this data, the department produced 77,38 m³/day of water out of a total consumption of 190,28 m³/day by the catchment. This production represents 40% of the total water demand delivered by the department.

Additionally, according to both private and public well maps, it was observed that the suburban areas of the city are mainly supplied by the public department, while central areas depend on other sources (which were not considered in this study). This implies that the proportion supplied by the public department may be even smaller (DAE and HIDROSAN 2014b), possibly corresponding to around 30% of the total water demands delivered from public sources calculated in WaterMet². According to the DAE document, the total water demand (190,28 m³/day) is equivalent to 500 L/d per capita. This figure can be compared to the 788.8 m³ per capita over 5 years, which is equivalent to 440 L/d per capita as calculated by WaterMet². This calculation not only validates the total water demand but also ensures a water balance within the system.

The water balance values fell within the normal range, and the total water demand amounted to 440 L/d per capita. This is considered normal because Bauru's UWS experiences a loss of nearly 50% of the produced water.

According to the city's sanitation plan, a wastewater treatment facility is currently under construction. The facility is intended to treat 100% of the city's effluent, but it is not yet operational. The basin's sewage is being discharged directly into the Bauru River (PMB 2017a), which makes it challenging to determine whether the values of wastewater overflow are within normal limits. The plan also reports an energy consumption of 1.10 kWh/m³ for water treatment (PMB 2017b) and a consumption of 0.30 kWh/m³, totalling 1.40 kWh/m³ of electricity consumption (direct energy). Compared with the 2.40 kWh/m³ calculated by WaterMet², the difference can be explained by the embodied energy (indirect energy) from the chemicals used in wastewater treatment.

GHG emission values in WaterMet² are calculated in units of KPIs per capita to facilitate a more precise comparison. The calculated rate is 0.288 tonnes of CO₂-eq per capita. The IPCC value for the average emissions in Brazil is 0.275 tonnes of CO₂-eq per capita (Minx et al. 2021), indicating only a slight margin of error.

Total acidification and eutrophication are indirectly determined for all types of GHG emissions, considering their associated chemicals, materials, energy sources, and emissions from by-products (Behzadian et al. 2014). The total eutrophication was higher than the acidification due to the high amount of phosphorus and nitrogen compounds (nitrate) in this eutrophication process, with most of them coming from wastewater systems. The elevated levels of contaminants, such as nitrate and phosphorus, also explain the significance of chemical oxygen demand (COD), which is typically directly proportional to the concentration of contaminants.

According to the DAE document, the excess water and the wastewater generated by the water treatment are discharged into the river without any additional measures (DAE and HIDROSAN 2014c). As a result, it becomes impossible to validate the sludge generation as calculated by WaterMet².

Each strategy was simulated in WaterMet², and the results of the analysing KPIs for each strategy is shown in Table 4. In the wetland's strategy, it can be seen that KPIs related to water balance, energy consumption, GHG emissions, total acidification, contaminants, and sludge generation remained unchanged. There is also a minor reduction in the TCSO, although it is insignificant. However, the STO increased by 161,210 m³. This increase could be attributed to the utilisation of impermeable wetlands, a strategy applied to previous areas, which, in turn, increases runoff and, consequently, the volume of collected water.

The installation of rain barrels on all properties had a negligible effect on the TCSO. However, it had a favourable impact on the STO, reducing the need for public STO by 19,327 m³. When compared to the planned construction for the catchment, this reduction corresponds to a 30% decrease in the size of the larger centralised stormwater tank or even the possibility of eliminating the smaller one (HIDROSTUDIO engenharia 2016).

The implementation of infiltration trenches in roadside beds would result in a slight reduction in contaminant concentrations. Moreover, this method exhibits significant improvements in infiltration performance and the redirection of water towards groundwater sources.

Likewise, the use of permeable pavement demonstrated significant improvements in reducing STO, with a reduction of 49,373 and 845,860 m³ in total combined sewer and STO, respectively. This results in a decrease of 27 m³/day in the combined sewer overflow and 464 m³/day in the STO over a period of 5 years. Additionally, this method effectively reduces contaminant levels, with only a negligible impact on eutrophication.

The implementation of DS with green spaces has a larger impact in terms of area coverage, but it has a minor effect to offset the number of construction zones that are needed. Nevertheless, it still demonstrates a favourable impact when compared to conventional systems. This strategy has a significant effect on reducing the TCSO and the STO, resulting in a positive impact on pollutant removal.

In summary, several alternatives demonstrated positive impacts on the KPIs, bringing them closer to the ideal point. For instance, infiltration trenches, permeable pavement, and green spaces have a positive impact on total eutrophication, STO volume, and TCSO. Rain barrels are also found to reduce TCSO. However, some alternatives have negative impacts, particularly constructed wetlands on STO and overall eutrophication levels.

The use of permeable pavements is also found to be the most effective strategy in terms of its positive impact on the system. However, it is important to note that these strategies did not significantly influence other KPIs, such as pump usage, fossil fuel consumption, and other forms of water treatment. This underscores the need for further studies to explore various contextual combinations and their potential impacts.

Figure 5 displays the normalisation and comparison of results based on Equation (1). The normalised outcomes fall within the range defined by the minimum and maximum alterations. Each bar on the graph represents a range closer to the ideal solution, without the use of weights. Consequently, it becomes evident that sustainable DS ST05 (permeable pavements) and ST06 (green spaces) have a higher impact on the KPIs, while ST02 (constructed wetlands) has the lowest impact. Strategies ST05 and ST06 have a significant influence on almost all KPIs and exhibit the highest unitary disturbance values among the alternatives, making them the most impactful choices.

Notably, it is important to highlight the sensitivity of the software to their incomes, given the extensive application of these systems. The DS with a more pronounced impact on the KPIs are those applied across an extensive physical area, highlighting the tool's sensitivity. These findings align with the study's objectives, which focus on evaluating the capacity to select appropriate DSs for specific subcatchment characteristics.

On the other hand, the strategy ST02 yields the least favourable results in terms of KPI impact, affecting only total GHG emissions, acidification, and eutrophication. This can be explained by the system's application in areas with permeable surfaces, which reduces infiltration zones, slows down the water flow, and retains it in impermeable locations during treatment.

It is worth noting that all strategies have an impact on certain KPIs, highlighting the effectiveness of WaterMet² in achieving the specific goals of the decision-maker. Furthermore, the most consistently influenced KPIs across all strategies are total GHG emissions, demonstrating significant variability. Approximately 85% of the strategies affect total GHG emissions, STO, total acidification, and total eutrophication. This indicates that these KPIs are strongly influenced by the characteristics of the strategies.

By applying Equation (1) to these values, we are able to collect data to calculate the Euclidean distance of the strategies from the ideal solution, as described in Equation (2), and subsequently rank them which are given in Table 5.

After assigning equal weights to the criteria, we consistently observe that permeable pavement and green spaces are the top-ranking strategies with respect to their influence on the KPIs. This holds true even when weights are not taken into account, as evidenced by the normalised results. However, the ranking for the worst (i.e., the lowest rank) strategy changed, which can illustrate the significant impact of the decision-maker's preferences on the best option. In summary, for this case study, permeable pavements have proven to be the most individually efficient drainage alternative, particularly due to their extensive application area. This aligns with findings from other studies that have also identified permeable pavements and green spaces as the best options for high-density urban areas (Putri et al. 2023). These conclusions are based on the equal weights that were applied. If the decision-maker prioritises different flows or KPIs, the best option may change accordingly.

The KPIs used in this study are aligned with recent research, which has identified water quantity and quality modelling as some of the most studied indicators. Furthermore, there has been a growing emphasis on reducing spatial scales in research, aiming for more precise multi-objective analyses. However, it is worth noting that the majority of studies have been carried out at the city and catchment levels. However, WaterMet² can incorporate household and neighbourhood data and unify the effects at the city scale that all contribute to integrated urban analysis (Ferrans et al. 2022).

We also examine the impact of implementing urban DSs in catchments and neighbourhoods on city-scale metabolism performance indicators. The results highlight the impact of these urban DSs on various KPIs, allowing for the selection of the most suitable options or their combination. The ranking of these strategies can assist decision-makers in selecting these systems or devising combinations to improve the metabolism-based performance of UWS.

This study aims to introduce the urban metabolism concept as a suitable approach to support decision-makers in the urban drainage assessment process in a comprehensive way. These correspond to the characterisation of the strategy and the simulation in a specific tool. Thus, this proposal aims to assist the simulation process by providing alternative results, indicating which sustainable strategy could be the most effective for the final solution.

The PCA showed that all variables analysed were important to the process and had a significant impact on at least one KPI. This demonstrates a direct effect on the results and operation, making it ideal to combine with other design instruments.

The case study analysis illustrated that the WaterMet² tool can generate metabolism-based performance indicators of UWS that closely resemble measured indicators. It measures the catchment water metabolism characteristics and enables decision-makers to quantify the conditions of urban dynamics. This tool was also sensitive to design variations in the properties that were typically used to represent the characteristics of urban drainage systems, which makes it a valuable decision-assistant tool. Moreover, it allows for the standardisation of KPI selection for assessing urban drainage systems and determining the best alternative, which, in this case study, was the permeable pavement. However, the weight used in the analysis and the importance of specific KPIs can strongly influence the ranking and selection of strategies.

In conclusion, this study aims to fill the gap in continuous temporal resolution that exists in short-term analyses. This is achieved through the use of the WaterMet² tool. It also examines the impacts of applying drainage system concepts at the catchment scale and evaluates how these variations influence the performance of urban metabolism at the city-scale simulations.

For future works, it is important to diversify the alternatives by associating or modifying the energy consumption using high-impact treatment SUD and other types of systems to observe their impacts on KPIs such as energy and GHG emissions. Additionally, the case study should consider the local climate characteristics and analyse how the localisation and the climate variations can influence the results. Furthermore, when studying the specific contaminants that can be carried out by the urban drainage systems, it is important to track the forever chemicals such as the per- and poly-fluoroalkyl substances and other emerging organic and inorganic materials in the air, microplastics, and others, instead of solely relying on conventional contaminants.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.