ABSTRACT
This study investigates the long-term behaviour of a porous asphalt that retrofitted a 36 m² laboratory full-scale street section, in terms of filtration and clogging processes. Throughout 11 experiments with increasing accumulative sediment loads up to 5.5 kg/m², the evolution of surface permeability, water levels inside the asphalt, water flows, turbidity, and TSS concentrations were analyzed. Sediment loads, representing dry-weather buildup, were applied to the asphalt surface followed by simulated 30-minute 80 mm/h rain events. Findings revealed an average permeability reduction rate of 4717 mm/h per kg/m². Surface clogging appeared from a cumulative load of 4.0 kg/m², but the asphalt effectively managed rainfall with only marginal variations in flows and water levels, except for the vicinities of gully pots. The porous asphalt demonstrated a sediment retention efficiency of 93%, with a significant decrease in turbidity and increasing TSS concentrations once the asphalt clogged. Comparison with previous work emphasized the scalability and reliability of small-scale test results for analyzing permeability evolution and retention efficiency, but such tests overlook real-world heterogeneities compromising the representativeness of water and sediment fluxes. The experimental dataset provides novel and valuable data for developing models to accurately simulate permeable pavements towards better urban planning, design, and maintenance.
HIGHLIGHTS
The full-scale model setup was successfully used for monitoring the evolution of filtration and clogging processes in porous asphalt (PA).
PA exhibited strong sediment retention, with a clear correlation between clogging and the reduction in permeability.
Despite surface clogging and increased TSS concentrations, water flow and levels remained stable, demonstrating the asphalt's resilience in managing rainfall.
The heterogeneities present in real-world systems pose a challenge for accurate predictions.
INTRODUCTION
Sustainable drainage systems (SuDSs) have been developed to mitigate the environmental impact of urbanization and the subsequent disruption of natural hydrological processes. Urban expansion often leads to higher volumes of runoff, faster discharge rates, and increased pollutant concentrations. To address these issues, it is crucial to assess SuDS under various scenarios and, importantly, to evaluate its long-term performance (Marchioni & Becciu 2015; Langeveld et al. 2022).
However, although the aging and malfunctioning of SuDS such as green roofs, vegetated swales, infiltration systems, or permeable pavements have been documented and tested in various infrastructures (e.g., Rodriguez-Hernandez et al. 2012; Mullaney & Lucke 2014; Kandra et al. 2020; Sañudo-Fontaneda et al. 2020; D'Ambrosio et al. 2024), data on their long-term performance are limited, as many systems have not yet reached their full operational lifespan. Furthermore, the collection of long-term field data is costly and labour-intensive, making it even more challenging to obtain representative information on their performance over time and ensure their future effectiveness.
Among these systems, permeable pavement stands out as a sustainable and innovative approach to urban infrastructure. Unlike conventional roadways, parking lots, and sidewalks, permeable pavements facilitate rainwater filtration rather than contributing to surface runoff (Kuruppu et al. 2019; Zha et al. 2021). They offer versatility in design and material composition, including porous asphalt (PA), modular interlocking concrete pavers, porous concrete, concrete and plastic gird pavers, and gravel (Scholz & Grabowiecki 2007; Mullaney & Lucke 2014; García-Haba et al. 2022).
Among the different permeable pavement typologies, PA consists of open-graded asphalt that is similar to traditional hot mix asphalt but lacks fine aggregated particles. This design allows water to flow through interconnected voids, which typically comprise 15–35% of the total volume. This feature plays a crucial role in hydrological performance and pollutant removal efficiency during the early service of life (Scholz & Grabowiecki 2007; Mullaney & Lucke 2014; Marchioni & Becciu 2015; Kia et al. 2017). However, its performance can be influenced by local conditions, as sediments from sources like dust, sand, and tyre wear can accumulate and block the pore structure due to runoff and traffic (Hamzah & Hardiman 2005). The clogging processes alter the initial hydrological properties and are crucial factors influencing system durability (Nielsen 2007; Li et al. 2022), impacting permeability reduction and effluent water quality (Cai et al. 2022; García-Haba et al. 2024). Therefore, a deeper exploration and understanding of these complex processes are necessary to improve the design and maintenance of PA (Kia et al. 2017).
To achieve this goal, several studies have been conducted in both field conditions and controlled laboratory environments (e.g., Bean et al. 2007; Nichols et al. 2015; Andrés-Valeri et al. 2016; Nichols & Lucke 2017). These studies utilized a range of methodologies that account for factors such as the type of porous surface, accumulation dynamics, type of clogging material used, rainfall characteristics, and other relevant variables.
Field campaigns offer reliable insights into the long-term analysis of permeable pavements under real maintenance conditions. For instance, Sañudo-Fontaneda et al. (2018) performed permeability tests using variable head measurements to evaluate a parking lot constructed with permeable mixtures and interlocking concrete block pavements (ICBPs). Their results showed that, after 10 years of operation without maintenance, the bays made with permeable mixtures were entirely clogged, reaching the end of their functional life after 9 years; in contrast, the ICBP surfaces maintained high infiltration rates, demonstrating greater resistance to sediment blockage. Similar studies have reported variability in infiltration rates based on the type of porous surface evaluated. These rates are significantly influenced by factors such as the characteristics of the measurement zone, traffic, erosion, vegetation cover, sediment loads, or maintenance practices, as documented by Bean et al. (2007), Sañudo-Fontaneda et al. (2013), Vancura et al. (2012), and Winston et al. (2016). However, field studies require long-term experimental campaigns on existing pavements, and the significant uncertainties inherent in external variables such as sediment build-up load, sediment characteristics, or rainfall conditions make it challenging to conduct detailed studies of clogging phenomena. These uncertainties hinder the identification of the specific effects of various factors and consequently make it difficult to generalize results under these uncontrolled environmental conditions.
Among the early approaches in small-scale laboratory tests that served as the foundation for developing methodologies to evaluate the infiltration of permeable pavements under rain and clogging conditions in controlled environments are those developed, for instance, by Davies et al. (2002), Rodríguez et al. (2005), or Shackel & Pamudji (1997). These methodologies use accelerated methods with simulated rainfall to provide a practical alternative for analysing the long-term hydraulic performance of permeable pavements, enabling relatively rapid testing of various slabs and optimal assessment and development of new and existing permeable solutions (Nichols et al. 2015).
Moreover, controlled laboratory conditions facilitate a better understanding of the variables influencing the clogging process. For example, Hashim et al. (2022) employed a rainfall simulator to assess the behaviour of diverse surface concrete block patterns under different rainfall intensities and slopes. Alsubih et al. (2017) and Hernández-Crespo et al. (2019) investigated the effects of rainfall and pollution build-up on the hydrological response of permeable pavements, respectively. García-Haba et al. (2023) used sediment collected from different urban areas to assess the influence of sediment characteristics on clogging processes. This study is similar to Nichols et al. (2015), who concluded that surface infiltration capacity can be influenced by the type and size of sediment, even with low sediment loads, as highlighted by Andrés-Valeri et al. (2016). The challenge in this type of testing lies in transferring the results to real-world scenarios and understanding how heterogeneities and particularities found in urban catchments, such as slopes, curbs, or gully pots, can influence the initial and long-term behaviour of permeable pavement solutions.
In this context, the present study thoroughly analyses the filtration and clogging processes of PA in the retrofitting of a large-scale installation covering a 36 m2 street section under controlled laboratory conditions and simulated rain. Novel results obtained in this experimental laboratory campaign have enabled the analysis of the long-term behaviour of a PA layer at the full scale, considering the spatial heterogeneities found in real urban catchments. This has generated valuable knowledge and data crucial for developing models towards accurate simulation of permeable pavements, helping in the planning, design, and maintenance of such solutions. Moreover, the facility has previously been used for hydrological and sediment transport studies using the impermeable concrete layer supporting the PA in these experiments (Naves et al. 2020a, b). Additionally, a layer of the same PA was previously tested at a small scale by García-Haba et al. (2023). Thus, the results presented here also aim to provide insights into the effect of installing a permeable layer on a conventional urban catchment and the representativeness of small-scale laboratory tests.
The remainder of the paper proceeds as follows: Section 2 outlines the experimental facility, PA, and the experimental procedure; Section 3 presents, analyses, and discusses the results in terms of water flow, water levels inside the asphalt, and water quality at the outlets, and finally, Section 4 provides the general conclusions.
MATERIALS AND METHODS
Experimental setup
General view and schematic representation of the street section physical model.
The street surface consists of a tiled sidewalk and a PA layer atop the impermeable concrete surface of the model. It has an approximate transversal slope of 2% up to the curb and a longitudinal slope of 0.5% leading to a lateral outflow channel. These values represent typical drainage slopes found in real roads and streets. They ensure that rainwater filtered through the PA drains into two gully pots (GP1 and GP2) measuring 0.3 × 0.3 m, as well as into a lateral outlet channel that directs the flow to a system outlet. The outflows from the gully pots and the outlet are directed into underground tanks, where flow rates are measured using a pre-calibrated ultrasound sensor that monitors water levels over a triangular weir (further explained in Section 2.3). For a more comprehensive description of the rainfall simulator setup and the physical model, refer to Naves et al. (2020a, c).
PA and distributed sediment characteristics
PSD of the PA (dash line) and the distributed sediment during clogging tests (solid line).
PSD of the PA (dash line) and the distributed sediment during clogging tests (solid line).
The PSD of the sediment used for assessing the long-term performance of the PA was measured by a laser coulter counter (Beckam-Coulter LS 13 320) and is also presented in Figure 2. This sediment was previously obtained from road-deposited material collected from the parking lots of the University of A Coruña (UDC) campus. It was calcined at 550 °C and classified using blind sieves of 63, 125, 250 and 500 μm, with a maximum size of 1,000 μm. The calcination aims to remove organic matter (2.2%) reducing the uncertainty that cohesivity may introduce in the sediment preparation and distribution process. Combining these fractions in proportions of 10, 15, 20, 25, and 30% resulted in a continuous granulometry (d50 = 282 μm, ρs = 2,929 ± 4 kg/m3) that represents a realistically graded road deposit sediment. This sediment was previously used in Goya et al. (2022) and García-Haba et al. (2023) and is consistent with ranges observed in the literature (Viklander 1998; Vaze & Chiew 2002; Deletic & Orr 2005; Zafra et al. 2008). It is also similar to findings from other permeable pavement investigations (Nichols et al. 2015; Hernández-Crespo et al. 2019).
Tests procedure
To analyse the impact of PA clogging on long-term hydrological performance, a total of 11 tests were conducted. Each experiment involved measuring the cumulative distribution of sediment over a 5 m long and 0.75 m wide strip adjacent to the curb, representing sediment build-up during dry-weather periods. The experiments also included simulating a constant rainfall of 80 mm/h for 30 min and measuring water flows, water levels, water quality, and surface permeability for each test. Table 1 includes the configuration of the tests conducted. The first test (CT_01) represents the initial condition of the PA with no sediment load. Subsequent tests involved the addition of sediment over the asphalt surface to analyse the influence of clogging on measured variables as the sediment cumulative load increased from 0.5 kg/m2 (CT_02) to a maximum value of 5.5 kg/m2 (CT_10). The remaining solids on the asphalt surface were then recovered using a 1,600 W industrial vacuum until no further particles could be suctioned. The collected surface sediment was weighed, and its PSD was determined by a Laser Coulter particle size analyzer (Beckam-Coulter LS I3 320). Finally, the last test (CT_11) was conducted without additional sediment load to evaluate the potential permeability recovery of the PA due to the vacuuming.
Configuration of all tests conducted during the experimental campaign, including details on rain characteristics, cumulative sediment loads, the number of manual water samples, permeability measurement points, and whether vacuuming was performed before each test
Test ID . | Rain intensity (mm/h) . | Duration (min) . | Cumulative sediment load (kg/m2) . | No. of manual samples . | Water level measuring points . | Permeability measuring points . | Surface vacuumed . |
---|---|---|---|---|---|---|---|
CT_01 | 80 | 30 | 0.0 | 0 | WL1–WL8 | P1–P6 | No |
CT_02 | 80 | 30 | 0.5 | 58 | WL1–WL8 | – | No |
CT_03 | 80 | 30 | 1.0 | 58 | WL1–WL8 | P1, P3, P5 | No |
CT_04 | 80 | 30 | 2.0 | 57 | WL1–WL8 | P2, P4, P6 | No |
CT_05 | 80 | 30 | 3.0 | 58 | WL1–WL8 | P1, P3, P5 | No |
CT_06 | 80 | 30 | 3.5 | 60 | WL1–WL8 | – | No |
CT_07 | 80 | 30 | 4.0 | 60 | WL1–WL8 | P2, P4, P6 | No |
CT_08 | 80 | 30 | 4.5 | 60 | WL1–WL8 | – | No |
CT_09 | 80 | 30 | 5.0 | 59 | WL1–WL8 | P1, P3, P5 | No |
CT_10 | 80 | 30 | 5.5 | 58 | WL1–WL8 | P2, P4, P6 | No |
CT_11 | 80 | 30 | 5.5 | 58 | WL1–WL8 | P1–P6 | Yes |
Test ID . | Rain intensity (mm/h) . | Duration (min) . | Cumulative sediment load (kg/m2) . | No. of manual samples . | Water level measuring points . | Permeability measuring points . | Surface vacuumed . |
---|---|---|---|---|---|---|---|
CT_01 | 80 | 30 | 0.0 | 0 | WL1–WL8 | P1–P6 | No |
CT_02 | 80 | 30 | 0.5 | 58 | WL1–WL8 | – | No |
CT_03 | 80 | 30 | 1.0 | 58 | WL1–WL8 | P1, P3, P5 | No |
CT_04 | 80 | 30 | 2.0 | 57 | WL1–WL8 | P2, P4, P6 | No |
CT_05 | 80 | 30 | 3.0 | 58 | WL1–WL8 | P1, P3, P5 | No |
CT_06 | 80 | 30 | 3.5 | 60 | WL1–WL8 | – | No |
CT_07 | 80 | 30 | 4.0 | 60 | WL1–WL8 | P2, P4, P6 | No |
CT_08 | 80 | 30 | 4.5 | 60 | WL1–WL8 | – | No |
CT_09 | 80 | 30 | 5.0 | 59 | WL1–WL8 | P1, P3, P5 | No |
CT_10 | 80 | 30 | 5.5 | 58 | WL1–WL8 | P2, P4, P6 | No |
CT_11 | 80 | 30 | 5.5 | 58 | WL1–WL8 | P1–P6 | Yes |
Spatial homogeneity of each sediment load was ensured by dividing the application area into 11 sectors, with six doses of the sediment distributed in each sector using a tailored sieve. The 80 mm/h rainfall for 30 min was chosen to facilitate sediment penetration into the pavement and reach steady-flow conditions in each test, as it falls within the ranges reported in the literature (e.g., Sañudo-fontaneda 2014; Hernández-Crespo et al. 2019; Brugin et al. 2020; Fernández-Gonzalvo et al. 2021).
(a) Measuring points for clogging tests and (b) a general image of the experimental setup during the sediment distribution procedure over the model surface, including sensors installed for registering depths (WL1–WL8) and permeability measuring points (P1–P6).
(a) Measuring points for clogging tests and (b) a general image of the experimental setup during the sediment distribution procedure over the model surface, including sensors installed for registering depths (WL1–WL8) and permeability measuring points (P1–P6).
RESULTS AND DISCUSSION
This section presents the variability of punctual permeabilities, water flows, water levels, and water quality as the cumulative sediment load was increased in the PA, leading to clogging.
PA permeability
The initial permeability values of the entire PA layer, measured in a 1 m grid, ranged between 11,089 and 37,575 mm/h, with a mean value of 22,822 mm/h and a standard deviation of 7,099 mm/h. This falls within the common ranges reported for PA (7,500–50,000 mm/h) by Fwa et al. (2015), Mullaney & Lucke (2014), or Kayhanian et al. (2012). Similarly, the initial permeability at the six control points located within the sediment distribution area (Figure 3) ranged between 15,314 and 34,886 mm/h, with a mean value of 2,5625 mm/h and a standard deviation of 6,743 mm/h, all within these reported ranges. The permeability heterogeneity and the slightly elevated values observed, though within the literature range, can potentially be attributed to minor variations in compaction levels during the asphalt laying (Pieralisi et al. 2017). These variations likely arose because compaction was performed with a vibratory plate compactor instead of heavier machinery due to operability constraints inside the laboratory.
(a) Punctual surface permeabilities of PA at six control points (P1–P6), with PA Street representing the mean values and PA Slab representing results from García-Haba et al. (2023) of the same PA in a small test stand. (b) Surface and cross-sections images of the PA with no sediment added (left), at 5.5 kg/m2 of sediment (middle), and after surface vacuuming (right).
(a) Punctual surface permeabilities of PA at six control points (P1–P6), with PA Street representing the mean values and PA Slab representing results from García-Haba et al. (2023) of the same PA in a small test stand. (b) Surface and cross-sections images of the PA with no sediment added (left), at 5.5 kg/m2 of sediment (middle), and after surface vacuuming (right).
As previously mentioned, accumulation rates are significantly influenced by various factors, but they can also vary widely due to the nonlinear nature of the process and its temporal and spatial variability. For instance, sediments often concentrate in specific areas of the pavement such as curbs (Sartor & Boyd 1972; Vaze & Chiew 2002; Pitt et al. 2005). Considering linear daily sediment accumulation rates ranging from 0.2 to 2 g/m2/day, a preliminary estimate for the service life of a 4.0 kg/m2 load is approximately from 5 years to over 50 years. However, the wide range of these values reflects the uncertainties and fluctuations inherent in the sediment accumulation process, which can cause substantial variations in the actual service life.
The permeability measured at the control points after vacuuming (test CT_11) showed a 3% increase, indicating a limited recovery of the PA (Figure 4). Although, as noted by Razzaghmanesh & Beecham (2018), the obstruction caused by the sediment is mainly concentrated within the first 2 cm of depth, the industrial vacuum was ineffective in removing the particles trapped by the PA. This is illustrated in Figure 4(b), where both surface (first row) and cross-sectional (second row) views of the PA after the experimental campaign reveal the formation of a sediment layer up to 2 cm thick that produces clogging. The vacuum cleaner's capacity to remove this layer was notably insufficient, as it only cleared surface particles. This observation aligns with findings in Vancura et al. (2012), indicating that maintenance vacuuming effectively removes particles only within approximately 3 mm of the pavement surface. Hence, maintenance strategies should be tailored to cleaning routines based on expected sediment contributions to prevent clogging from reaching a point where maintenance efforts become ineffective (Henderson & Tighe 2012; Drake & Bradford 2013). Furthermore, surface vacuuming did not significantly restore permeability at the monitored sites, aligning with previous research that suggests PA responds best to treatments that penetrate deeper into the surface, such as high-pressure water or air combined with high-suction methods (Danz et al. 2020). In fact, Winston et al. (2016) demonstrated that milling to a depth of 2.5 cm almost completely restored the surface infiltration rates of 21-year-old PA pavement to conditions similar to those of new pavement. Therefore, developing a comprehensive maintenance plan to regularly restore infiltration capacity is essential for the long-term performance of permeable asphalt. Incorporating these strategies into the design and planning phases is critical to ensure the sustained effectiveness and durability of permeable pavement systems.
In Figure 4(a), we also present the permeability results obtained from García-Haba et al. (2023) who tested a 0.18 m2 slab of the same PA using a similar methodology in a small rainfall simulator. The behaviour of the slab in terms of initial permeability and the sediment load at which the slab completely clogs are consistent with the values observed at control point P2. This control point presented the lowest initial permeability, probably due to slightly higher compaction in that area of the PA, which may be similar to that of the slab. Additionally, the permeability values obtained from the present work and the slab tests after vacuum-cleaning are also similar (1,783 and 1,246 mm/h, respectively). Thus, the comparison indicates that employing a small-scale test setup enables the reliable assessment of the long-term permeability and recovery capacity of PA. However, it is crucial to consider the heterogeneities present in real-world applications for proper interpretation and transferability of the results obtained.
Water levels inside the PA layer
Water levels inside the PA (a) with the street curb boundary influence: WL1, WL2, WL3, and WL4 and (b) with the gully pot and the lateral outflow channel influence: WL5, WL6, WL7, and WL8. Solid lines represent the initial PA conditions (test CT_01), and dashed lines represents those for a total cumulative load of 5.5 kg/m2 (test CT_10).
Water levels inside the PA (a) with the street curb boundary influence: WL1, WL2, WL3, and WL4 and (b) with the gully pot and the lateral outflow channel influence: WL5, WL6, WL7, and WL8. Solid lines represent the initial PA conditions (test CT_01), and dashed lines represents those for a total cumulative load of 5.5 kg/m2 (test CT_10).
In the first scenario (Figure 5(a)), points aligned transversally to the curb and not directly influenced by gully pots are considered (WL1, WL2, WL3, and WL4). Here, it is evident that water tends to accumulate in the vicinity of the curb, following the main slope of the impermeable concrete used as structural support for the PA layer, with higher water levels observed nearer to the curb. The comparison between initial and clogged conditions reveals that clogging did not significantly affect water levels. Given that clogging occurred primarily within the first 2 cm (Figure 4(b)), it appears that the remaining section of PA ranging from 0.03 to 0.058 m thickness is adequate for internal transport of the rainwater generated without significantly modifying water levels. Likewise, surface vacuuming during test CT_11 showed no notable variations in the measured levels.
In the second scenario (Figure 5(b)), we examine the impact of the gully pot and the lateral outfall channel as free discharge outlets on the water levels at points WL5 to WL8. These points are positioned at the same distances from the curb as in the previous scenario, but this time they are aligned with gully pot 2. Points WL5 and WL6, situated outside the sediment distribution area, experienced a roughly 20% reduction in water levels compared to the corresponding points in the first scenario. Notably, the influence of the gully pot was most pronounced at point WL7, where there was a 70% reduction in water level compared to point WL2. This point also exhibited the most significant impact of clogging, evidenced by a 6 mm increase in water level for the maximum cumulative load compared to the initial condition. This finding suggests that clogging is diminishing the extent of the gully pot's influence in terms of water levels.
The evolution of the water sheet across the measured profiles before and after clogging is provided in the Supplementary Material, where it is evident that the profiles are influenced by both the longitudinal and transverse development of the flow.
Flow discharges
(a) Comparison of flow discharges in GP1 and GP2 from the initial test (CT_01) and the final test after surface vacuuming (CT_11), with those obtained from Naves et al. (2020b) over the impermeable concrete surface. (b) Variation in the steady-flow rate according to the cumulative sediment load tested.
(a) Comparison of flow discharges in GP1 and GP2 from the initial test (CT_01) and the final test after surface vacuuming (CT_11), with those obtained from Naves et al. (2020b) over the impermeable concrete surface. (b) Variation in the steady-flow rate according to the cumulative sediment load tested.
A comparison of discharged flows with those obtained from Naves et al. (2020b) under the same rainfall intensity and over the impermeable concrete surface previous to PA retrofitting reveals that the PA layer induced a more homogeneous distribution of flows among the three outflows. Additionally, as can be seen in Figure 6, the PA delayed the runoff peak by more than 15 min and reduced the summarized peak flow in both gully pots by roughly 75% at 5 min, which is the duration of the rain in case of impervious experiments. This behaviour was not shown in the small-scale experimental campaign developed by García-Haba et al. (2023), where no significant differences were observed between the PA and an equivalent impermeable surface. Thus, large-scale experiments are crucial for accurately assessing the lamination effect of permeable pavements in catchment hydrology.
Deposited sediment mobilization
Maximum and event mean concentration (EMC) values for TSS and turbidity from manual grab samples at the entrance of gully pots GP1 and GP2.
Maximum and event mean concentration (EMC) values for TSS and turbidity from manual grab samples at the entrance of gully pots GP1 and GP2.
As the accumulated sediment increases, turbidity values tend to decrease, possibly due to the filtration effect of sediment accumulation in the PA. With increased sediment load, the pores of the PA tend to close, enhancing the retention capacity of smaller particles associated with turbidity, as well as other pollutants like Chemical Oxygen Demand (COD), Total Nitrogen (TN), and Total Phosphorus (TP). Similar patterns were observed in PA slab experiments and other studies at small and real scales (Mata & Leming 2012; Kia et al. 2017; Fernandez-Gonzalvo et al. 2020; García-Haba et al. 2023).
In contrast, TSS values tend to increase from a sediment cumulative load of 4.5 kg/m2, corresponding to the test where permeability reaches minimum values for some control points and the clogging of the PA begins to be evident. Since this does not correspond to turbidity behaviour, it may be hypothesized that the solids contributing to these higher concentrations originate from coarser sediment fractions that are unable to penetrate the asphalt and are transported towards the gully pots. There is no significant change in turbidity values after vacuuming; however, comparing EMC values obtained in the last test before vacuuming (CT_10), TSS values were notably reduced, consistent with the hypothesis drawn.
Finally, the sediment vacuumed before test CT_11 was weighted, resulting in 0.11 kg, while the total sediment washed through the gully pots and the lateral outlet channel was estimated at 0.20 kg from the EMC and total stormwater volume. Considering that the total amount of sediment distributed is 4.5 kg, it is estimated that the sediment that remains inside the PA is roughly about 4.0 kg. The retention efficiency, calculated as the percentage of sediment accumulated on the surface and retained in the asphalt, is about 94% This underscores the effectiveness of the PA in mitigating sediment runoff and improving water quality. The result is conditioned by the methodology used in terms of sediment characteristics and load and rainfall properties, making reliable comparisons with other studies or even with the impervious surface prior to the installation of the PA layer. However, similar results to the 98% (85% inside and 13% over the asphalt) retention efficiency obtained from García-Haba et al. (2023) indicate the feasibility of small-scale slab tests to assess the retention capacity of PAs and their representativeness at larger scales.
PSD for the distributed and vacuumed sediment is collected in comparison with the results from García-Haba et al. (2023) for vacuumed and washed sediment.
PSD for the distributed and vacuumed sediment is collected in comparison with the results from García-Haba et al. (2023) for vacuumed and washed sediment.
CONCLUSIONS
In this study, we investigated the processes of filtration and clogging in a large-scale installation of a 36 m2 street section retrofitted with PA. Under controlled laboratory conditions, we measured the evolution of surface permeability, water levels inside the PA layer, flow discharges, turbidity, and TSS concentrations while increasing the cumulative sediment load over the asphalt surface. The main conclusions drawn from our experimental campaign are as follows:
Initial permeabilities of the PA ranged between 15,000 and 35,000 mm/h due to spatial heterogeneities inherent in real-world construction. Consequently, the control points on the asphalt required different sediment loads to become completely clogged, ranging between 4.0 and more than 5.5 kg/m2, which was the maximum cumulative load applied within the tests. However, the experimental campaign quantified the permeability reduction rate and its variability with precise control over the variables involved, finding that the rate of permeability reduction was consistent across all measured points on the asphalt independent from the corresponding initial permeability or the load at which complete clogging occurred, averaging 5,422 mm/h per kg/m2 of sediment.
Sediment primarily accumulated in the top 2 cm of the PA layer, leading to surface clogging. However, the remaining section remained free, allowing water to flow through the asphalt. This study characterized this flow by precisely measuring water levels within the pavement and flows towards the gully pots and their evolution due to clogging. Results showed that only water levels close to gully pots have shown a non-negligible maximum increase of 6 mm due to clogging, with low impact on water levels and flows. Moreover, the water retention capacity of the PA layer was demonstrated with a peak flow reduction of 50% and a delay of 3 min compared to an equivalent impervious surface.
Analysis of event mean concentrations of turbidity and TSS at the gully pots showed contrasting behaviours in direct response to clogging. Turbidity decreased, indicating that smaller sediment fractions were filtered as the pores in the PA diminished in size due to sediment accumulation. Conversely, the increase in TSS concentration once clogging was initiated arose from coarser sediment fractions that could not penetrate the asphalt and were instead transported towards the gully pots. The overall sediment retention efficiency of the PA was found to be 93%, underscoring the effectiveness of the PA in mitigating sediment runoff and improving water quality.
Additionally, our study confirms that the results on permeability evolution and sediment retention capacity obtained from small-scale tests on asphalt slabs are both scalable and reliable. Test stands serve thus as effective tools for analysing and evaluating long-term permeable pavements in a cost-efficient manner and for optimizing maintenance strategies. However, such tests may not fully account for boundary conditions and heterogeneities present in real applications, making it challenging to scale the effects of PA in terms of flow rates and sediment fluxes. Therefore, the novel experimental campaign provides also valuable data openly available at https://zenodo.org/records/12758177 for developing models to accurately simulate permeable pavements in urban basins, thereby enhancing the planning, design, and maintenance of such solutions.
ACKNOWLEDGEMENTS
This work was partially funded by the EU under the Horizon 2020 programme through a contract for Integrating Activities for Starting Communities (Co-UDlabs project, GA No.101008626) and by the projects POREDRAIN (grant number RTI2018-094217-B-C33) and SUDSlong-LCG (grant number PID2021-122946OB-C31) funded by MCIN/AEI/10.13039/501100011033 and by ‘ERDF A way of making Europe’. The experiments were carried out with the support of the research infrastructure at the Centre of Technological Innovation in Construction and Civil Engineering (CITEEC) at the University of A Coruña.
AUTHOR CONTRIBUTION
A.G.-H. performed the experiments with the supervision of J.N., J.S., and J.A., who conceptualized the study. A.G.-H. and J.N. prepared the original draft. All the authors critically revised the final version.
DATA AVAILABILITY STATEMENT
All relevant data are available from https://zenodo.org/records/12758177.
CONFLICT OF INTEREST
The authors declare there is no conflict.