Permeable Pavement in the Northwestern United States: Pollution Source or Treatment Option?

: Permeable pavements can be an effective stormwater mitigation technique, but there are concerns that polluted stormwater may contaminate groundwater as stormwater inﬁltrates through the soil beneath the pavement. This research evaluates the pollutant removal capabilities of pervious pavements using pervious cement concrete (PC) and porous asphalt concrete (PA) cylinders. Stormwater collected from an outfall was used to perform three tests. The inﬂuent and efﬂuent were analyzed for metals, semi-volatile organic compounds (SVOCs), phosphorus, and turbidity. Average percent removal for metals were 37–63% except for zinc, which had an average export of 21% for pervious cement concrete and 52% for porous asphalt concrete. Only 10 of the SVOCs tested had an inﬂuent concentration above detection levels. Complete removal (below detection levels) was observed for benzo(a)anthracene, benzo(a)pyrene, chrysene, and indeno(1,2,3-cd) pyrene. Average removals for benzo(b)ﬂuoranthene, benzo(g,h,i)perlyne, ﬂuoranthene, phenanthrene, pyrene, and bis(2-ethylhexyl)phthalate were 63–96%. No signiﬁcant removal was observed for total phosphorus and reactive phosphate. All contaminant concentrations were below drinking water limits except lead, which would likely be removed in the soil layer below the pavement. This study indicates permeable pavements can effectively remove stormwater contaminants and protect groundwater as a drinking water source.


Introduction
Stormwater mitigation can be challenging in urban areas, particularly in Portland, Oregon, where average annual rainfall is 36.9 inches and mostly occurs during the rainy season (October-May) with little to no rainfall in the period of June-September [1]. Many urban areas have very little permeable surfaces, resulting in significant stormwater runoff [2]. Cities must have efficient drainage systems to keep streets safe for drivers in addition to slowing runoff and reducing discharge of stormwater contaminants to protect receiving waters. Annually, 75% of weather-related crashes occur on wet pavement, resulting in 5700 deaths [3]. To increase safety, streets must drain quickly to prevent standing water on the streets. Standard drainage systems that efficiently minimize standing water can cause erosion of stream banks and beds. Increased impervious areas where stormwater is collected with a curb and gutter system and transported directly to receiving waters in a storm drain have caused a disconnect between groundwater and surface water. In addition, stormwater contaminants, including oils, fertilizers, metals, and pesticides, can negatively impact receiving waters and the aquatic ecosystem [4].
Stormwater management techniques such as green infrastructure and permeable pavements can help reconnect the hydrologic cycle, reduce flooding and erosion, and reduce contaminants in drinking water sources. Maintaining access to clean, safe drinking water is imperative for a sustainable society. The City of Portland has slowed down and water quality parameters, such as SVOCs, need to be tested to ensure the safety of drinking water sources or ecological receptors in receiving waters. Many SVOCs are particularly harmful to human health, and more studies are needed to ensure drinking water derived from treated stormwater is safe for consumption.
This study evaluates removal efficiencies for SVOCs, phosphorus, metals, and turbidity. Two types of permeable pavement were tested: pervious cement concrete and porous asphalt concrete. Stormwater was collected and used to conduct three tests. Influent and effluent samples were taken during each test and analyzed for contaminants that may pose a hazard for groundwater. We hypothesize that permeable pavements will decrease pollutant levels enough to meet drinking water standards and minimize risk of polluting groundwater or receiving water bodies.

Test Specimens
Triplicates of pervious cement concrete and porous asphalt concrete cylinders were made to evaluate water quality. For the pervious cement concrete cylinders, a mix of 76% aggregate, 19% cement, and 5% water was mixed by hand and scooped into 10.2 cm (4-inch)-diameter, 20.3 cm (8-inch)-long cylindrical molds. Aggregate gradation is shown in Table 1. The mix was compacted approximately 10% using a standard tamper, covered, and allowed to cure for seven days. The mix design and procedure for making cylinders were similar to other studies [24]. The porous asphalt concrete cylinders were made following guidelines from the National Asphalt Pavement Association, which specifies a mix of 84% aggregate, 10% mineral filler, and 6% asphalt binder [31]. Asphalt binder (PG 70-22 ER) was obtained from Lakeside Industries (Portland, OR, USA). The aggregate and asphalt binder were heated to 149 • C (300 • F), and all components were mixed in a large container before placing in an aluminum mold to make the cylinders. Both concrete and asphalt cylinders were made in one batch to ensure the consistency of the mix for the replicates. The porous asphalt concrete cylinders were the same size as the pervious cement concrete cylinders. Air voids were 28% and 34% in the pervious cement concrete and porous asphalt concrete, respectively.

Experiments
The test setup consisted of ring stands to hold the cylinders, separatory funnels, and a container below the cylinders to collect effluent ( Figure 1). Prior to testing, approximately 3 L of stormwater collected from a parking lot at the University of Portland (equivalent to 10 water quality design storms) was applied to the cylinders. The volume of a water quality design storm is 331 mL for a 10.2 cm diameter surface using the City of Portland water quality design storm of 4.1 cm (1.61 inches) [32]. The water quality design storm volume (331 mL) was applied ten times at least two days apart to wet the cylinders and allow for carbonation to develop in the pore spaces of the cylinders before testing to mimic in situ conditions [24]. For the tests, stormwater from the Columbia Slough Outfall 56C was collected, which transports stormwater from North Portland and discharges to the Columbia Slough. The outfall is part of the Oregon Department of Environmental Quality Columbia Slough Sediment Project, an effort to reduce contaminants in the Columbia Slough with the City of Portland, Multnomah County Drainage District, and private parties [33]. This outfall was selected because stormwater is collected from an industrial Three trials were conducted during testing. The experimental flow diagram is shown in Figure 2. During each trial, 1.25 L of stormwater was applied to each cylinder using a separatory funnel to control application rate. Although this is much more than the water quality design storm, the additional volume was needed to complete the water quality analysis. A runoff rate of 10 mL/min was applied to the cylinders. This rate is similar to that used by other studies [24] and represents a rainfall intensity for a permeable pavement road without additional run-on from upstream catchments. Tests were conducted at least two days apart to mimic rainfall patterns and allow for the cylinders to partially dry. Effluent was collected in a polypropylene container, and composite samples were collected from the container after the cylinders stopped dripping. Total phosphorus and phosphate were analyzed in accordance with Standard Methods Section 4000: Inorganic Nonmetallic Constituent [34]. The persulfate method was used to quantify total phosphorus, and the colorimetric method was used to quantify phosphate. Turbidity was quantified using a Hach turbidimeter. Arsenic, cadmium, copper, lead, zinc, and semi-volatile organics were analyzed at the City of Portland s Water Pollution Control Lab (WPCL). A list of the SVOCs analyzed is shown in Table 2. Metals were analyzed using an ICP-MS in accordance with EPA method 200.8. Semi-volatile organics were analyzed using a GCMS in accordance with EPA method 8270. All glassware and sample bottles were acid-washed and rinsed with DI water according to standard methods [34]. Samples were stored at 4 °C, preserved, and analyzed within standard holding times.   Three trials were conducted during testing. The experimental flow diagram is shown in Figure 2. During each trial, 1.25 L of stormwater was applied to each cylinder using a separatory funnel to control application rate. Although this is much more than the water quality design storm, the additional volume was needed to complete the water quality analysis. A runoff rate of 10 mL/min was applied to the cylinders. This rate is similar to that used by other studies [24] and represents a rainfall intensity for a permeable pavement road without additional run-on from upstream catchments. Tests were conducted at least two days apart to mimic rainfall patterns and allow for the cylinders to partially dry. Effluent was collected in a polypropylene container, and composite samples were collected from the container after the cylinders stopped dripping. Total phosphorus and phosphate were analyzed in accordance with Standard Methods Section 4000: Inorganic Nonmetallic Constituent [34]. The persulfate method was used to quantify total phosphorus, and the colorimetric method was used to quantify phosphate. Turbidity was quantified using a Hach turbidimeter. Arsenic, cadmium, copper, lead, zinc, and semi-volatile organics were analyzed at the City of Portland's Water Pollution Control Lab (WPCL). A list of the SVOCs analyzed is shown in Table 2. Metals were analyzed using an ICP-MS in accordance with EPA method 200.8. Semi-volatile organics were analyzed using a GCMS in accordance with EPA method 8270. All glassware and sample bottles were acid-washed and rinsed with DI water according to standard methods [34]. Samples were stored at 4 • C, preserved, and analyzed within standard holding times.

Statistical Analysis
The average and standard deviation of the three replicates were calculated for each trial. The Kruskal-Wallis and Wilcoxon signed-rank test was used to determine whether there was a significant difference between the influent and effluent, different trials, and porous asphalt concrete and pervious cement concrete [35]. This test is commonly used for studies with small sample sizes to determine if different treatments are effective. Differences were considered significant if p < 0.05.

Results and Discussion
During each trial, all stormwater flowed through the pervious cement concrete and porous asphalt concrete at the rate applied. No ponding was observed. Although this is beneficial for removing stormwater from the street and improving safety during rain events, lower-permeability pavements may remove more pollutants given the tortuous paths and longer residence time in the pavement. Average influent turbidity was 191 NTU, and average effluent was 176 NTU. The effluent was significantly lower than the influent during trials 1 and 3 (p = 0.03) but statistically the same during trial 2 (p = 0.156). Turbidity was statistically the same in the effluent from the pervious cement concrete and porous asphalt concrete cylinders. Results are similar to other studies that showed limited removal of turbidity [25,36]. The pore spaces in permeable pavements were shown to

Statistical Analysis
The average and standard deviation of the three replicates were calculated for each trial. The Kruskal-Wallis and Wilcoxon signed-rank test was used to determine whether there was a significant difference between the influent and effluent, different trials, and porous asphalt concrete and pervious cement concrete [35]. This test is commonly used for studies with small sample sizes to determine if different treatments are effective. Differences were considered significant if p < 0.05.

Results and Discussion
During each trial, all stormwater flowed through the pervious cement concrete and porous asphalt concrete at the rate applied. No ponding was observed. Although this is beneficial for removing stormwater from the street and improving safety during rain events, lower-permeability pavements may remove more pollutants given the tortuous paths and longer residence time in the pavement. Average influent turbidity was 191 NTU, and average effluent was 176 NTU. The effluent was significantly lower than the influent during trials 1 and 3 (p = 0.03) but statistically the same during trial 2 (p = 0.156). Turbidity was statistically the same in the effluent from the pervious cement concrete and porous asphalt concrete cylinders. Results are similar to other studies that showed limited removal of turbidity [25,36]. The pore spaces in permeable pavements were shown to effectively trap suspended solids, but they are not small enough to remove microscopic particles that impact the clarity of water [36]. Table 4 shows the average effluent concentrations for arsenic, cadmium, copper, lead, and zinc during each trial, and Figure 3 shows the influent and effluent concentrations for these metals. Table 3 shows the percent removal from the pervious cement concrete and porous asphalt concrete cylinders during each trial. The effluent for both pervious cement concrete and porous asphalt concrete was significantly lower than the influent for arsenic, cadmium, copper, and lead (p = 0.03-0.04). Zinc concentrations in the effluent were significantly lower than the influent during trial 1 (p = 0.03) and significantly higher than the influent during trials 2 and 3 (p = 0.03). Removal of cadmium and zinc was significantly higher in the pervious cement concrete compared to the porous asphalt concrete cylinders (p = 0.01) but statistically the same for arsenic, copper, and lead (p = 0.25-0.57). effectively trap suspended solids, but they are not small enough to remove microscopic particles that impact the clarity of water [36]. Table 3 shows the average effluent concentrations for arsenic, cadmium, copper, lead, and zinc during each trial, and Figure 3 shows the influent and effluent concentrations for these metals. Table 4 shows the percent removal from the pervious cement concrete and porous asphalt concrete cylinders during each trial. The effluent for both pervious cement concrete and porous asphalt concrete was significantly lower than the influent for arsenic, cadmium, copper, and lead (p = 0.03-0.04). Zinc concentrations in the effluent were significantly lower than the influent during trial 1 (p = 0.03) and significantly higher than the influent during trials 2 and 3 (p = 0.03). Removal of cadmium and zinc was significantly higher in the pervious cement concrete compared to the porous asphalt concrete cylinders (p = 0.01) but statistically the same for arsenic, copper, and lead (p = 0.25-0.57).   Metals removal was similar to other studies that found metals are removed on the order of 50% [8,27], although some studies have shown higher removal rates [24,37]. The Washington Technology Assessment Protocol-Ecology (TAPE) is a protocol used to evaluate stormwater technologies in Oregon and Washington and requires 30% removal of copper and 60% removal of zinc for new stormwater technologies to be approved for use [38]. Average removal of copper was 39% for pervious cement concrete and 37% for porous asphalt concrete, which exceeds TAPE standards. However, 60% removal of zinc was not achieved; zinc was exported in both the pervious cement concrete and porous asphalt concrete. With the exception of cadmium and arsenic, removal was significantly higher during trials 2 and 3 compared to trial 1 (p = 0.03-0.04). Arsenic concentrations in the effluent were statistically the same during trials 1 and 2 (p = 0.06), but removal was significantly higher (p = 0.03) during trial 3 compared to trial 1. Cadmium concentrations in the effluent were statistically the same during all trials (p = 0.06-0.84). Higher removal during trials 2 and 3 could be due to the stormwater that remains in the cylinders between testing; slower mechanisms of removal such as complexation reactions could be occurring between tests.

Metals
Results indicate pervious cement concrete and porous asphalt concrete was not effective at removing zinc from the stormwater. Removal occurred during the first trial, but then, export occurred during trials 2 and 3. Other studies have shown 0.55-101% export of zinc [6,25,26]. This may be due to zinc saturation of the cylinders, which is then released during subsequent storm events. Further research is needed to determine why zinc is exported from permeable pavements and how it can be minimized. Table 5 shows the average effluent concentrations of select SVOCs during each trial, and Figure 4 shows influent and effluent concentrations. Table 6 shows percent removal from the pervious cement concrete and porous asphalt concrete for each trial. The remaining SVOCs were below detection limits in both the influent and effluent during each trial. Complete removal was observed for chrysene, benzo(a)pyrene, benzo(a)anthracene, and indeno(1,2,3-cd)pyrene during all three trials; effluent concentrations were below the detection limit. Effluent concentrations were significantly lower than influent for the SVOCs that had influent concentrations above the detection limit (p = 0.03-0.04). Effluent from the pervious cement concrete and porous asphalt concrete cylinders were statistically the same (p = 0.06-0.40), and there was no difference between trials (p = 0.06-0.83). Results are similar to other studies that showed removal of PAHs, diesel fuel, and motor oil from both pervious cement concrete and porous asphalt concrete [7,25,30,39]. For porous asphalt concrete, Brattebo and Booth (2003) observed complete removal of diesel fuel and motor oil [7], and Charlesworth et al. (2017) observed 99.9% removal [30]. This indicates both porous asphalt concrete and pervious cement concrete can effectively remove common organic pollutants. Removal of SVOCs may occur in the upper portion of the permeable pavement, similar to Charlesworth et al. (2017), who observed oil removal in the top 10 cm of pavement [30]. A microbial biofilm was found on the surface layer of the pavement, which likely degraded the oil. Additional studies would be needed to confirm whether biological removal is the main removal mechanism for SVOCs. pervious cement concrete and porous asphalt concrete cylinders were statistically the same (p = 0.06-0.40), and there was no difference between trials (p = 0.06-0.83).     Table 7 shows the average effluent concentrations of total phosphorus and phosphate during each trial, and Figure 5 shows influent and effluent concentrations. Table 8 shows the percent removal from the pervious cement concrete and porous asphalt concrete for total phosphorus and phosphate. The influent and effluent were statistically the same for total phosphorus during trials 1 and 2 (p = 0.44-0.59) but significantly lower (p = 0.03) during trial 3. Influent and effluent phosphate concentrations were statistically the same during all trials (p = 0.53-1.0), which indicates phosphate is not effectively removed in pervious cement concrete or porous asphalt concrete. For discharge to receiving waters, EPA recommends an effluent limit of 0.05 mg/L for discharge to streams entering lakes and 0.1 mg/l for streams with no reservoirs or lakes [40]. The effluent in this study was much higher (greater than 0.5 mg/L) than the recommended limit. Stormwater flowing through permeable pavements will likely flow through the aggregate base layers and soil before reaching receiving waters, where some of the phosphorus could be retained. Percent removal was low for most trials, with export occurring during some of the trials. This is in contrast to the significant total phosphorus removal observed by Jayasuriya et al. (2007) and does not meet the TAPE standards of 50% total phosphorus removal [6,38]. However, other studies have shown export of phosphorus [39]. Differences in removal may be due to the main removal mechanism that is likely removing phosphorus in the permeable pavement. Total phosphorus is typically associated with soils and has been observed to be transported with sediment [41]. Permeable pavements with larger pore sizes may be less effective at removing phosphorus via physical filtration compared to pavements with smaller pore sizes. Additional studies are needed to determine the mechanisms for phosphorus removal and how to optimize removal in pervious cement concrete and porous asphalt concrete. pavements with smaller pore sizes. Additional studies are needed to determine the mechanisms for phosphorus removal and how to optimize removal in pervious cement concrete and porous asphalt concrete.    Table 9 shows average results compared to maximum contaminant levels (MCLs) set by the Oregon Health Authority and EPA (OAR-333-061-0030). Only contaminants with limits for drinking water are listed; many of the contaminants tested in this study do not have MCLs for drinking water. Both copper and lead have an action level, not an actual MCL. Public water systems must take action if copper concentrations are higher than 1.3 mg/L and/or lead concentrations are higher than 0.015 mg/L. Zinc does not have an enforceable MCL, but secondary guidelines recommend an MCL of 5 mg/L.    Table 9 shows average results compared to maximum contaminant levels (MCLs) set by the Oregon Health Authority and EPA (OAR-333-061-0030). Only contaminants with limits for drinking water are listed; many of the contaminants tested in this study do not have MCLs for drinking water. Both copper and lead have an action level, not an actual MCL. Public water systems must take action if copper concentrations are higher than 1.3 mg/L and/or lead concentrations are higher than 0.015 mg/L. Zinc does not have an enforceable MCL, but secondary guidelines recommend an MCL of 5 mg/L. All contaminants are below the MCL or action levels for drinking water except for lead. Lead has an action level of 0.015 mg/L for drinking water, and average lead concentrations in this study were 0.018 for pervious cement concrete and 0.017 mg/L for porous asphalt concrete. Although there was 48-65% removal through both the pervious cement concrete and porous asphalt concrete, effluent concentrations were slightly higher than the action level. Additional lead will likely be removed in the soil layer beneath the pervious pavement. Legret et al. (1996) found a 79% decrease in lead concentrations through pervious pavement and the underlying soil layer compared to a catchment with no treatment [37]. Metals were observed to accumulate on the surface of the pervious asphalt as well as the underlying soil and geotextile layer. Contaminant transport modeling performed by the City of Portland to support their Underground Injection Control (UIC) Program for stormwater management indicated that lead was removed in the soil layers to levels below the action level [42]. In addition, bioretention studies have shown that 92-99% of lead is removed in 2 ft of soil [43]. If 90% removal is assumed in the soil layer, lead concentrations would be 0.002 mg/L, which is significantly below the action level. Thus, the stormwater flowing through pervious cement concrete and porous asphalt concrete likely meets drinking water standards and will not contaminate groundwater. Additional field studies would be needed to confirm.

Conclusions
This study indicates the following:

1.
Pervious cement concrete and porous asphalt concrete can effectively remove most contaminants of concern for drinking water to below MCLs. Lead concentrations were slightly higher than the MCL, but further removal will likely occur in the soil layer beneath the pervious pavement; 2.
Zinc was exported from the permeable pavements, but concentrations were still well below the secondary MCL. Total phosphorus and phosphate were either exported or minimally removed, but this does not pose a threat to groundwater in terms of drinking water safety; 3.
Significant removal was observed for SVOCs, which indicates pervious cement concrete and porous asphalt concrete can effectively be used as a treatment method for organic contaminant removal; 4.
Permeable pavements minimize the risk of infiltrating stormwater polluting groundwater and provide a method of removing pollutants before reaching the groundwater table.

Study Limitations
This study was limited to testing two types of permeable pavements in the laboratory: pervious cement concrete and porous asphalt concrete. Further removal in the soil layer beneath the pervious pavement was not evaluated. Additional testing would be needed to verify whether further removal occurs in the soil layer beneath the pervious pavement to levels below the MCL for lead. The study also did not evaluate long-term performance or possible accumulation in the pavements or underlying soils. Removal capabilities may vary seasonally or over time in the field as the pavements weather, experience wetting/drying cycles, and roads are maintained (e.g., cleaning and sweeping).

Future Work
Field studies are needed to verify these results would occur on a larger scale and on a long-term basis. Field studies could determine how changes in temperature, rainfall patterns, wetting/drying cycles, and other factors that are limited in laboratory studies impact water quality below installed permeable pavements. Long-term studies over at least 12-24 months to capture seasonally changes are also needed to evaluate changes in pavements due to weathering, pollutant accumulation in the pavement over time, and potential leaching. Pervious cement concrete and porous asphalt concrete can be a potentially powerful tool for increasing traffic safety, reducing flooding, lowering noise from roadways, reducing the heat-island effect, and protecting groundwater from stormwater contaminants. Increased use of permeable pavements can improve the sustainability of urban areas and protect valuable drinking water sources. Municipalities should consider using permeable pavements for new or replacement pavement projects more extensively where green infrastructure is not feasible.
Author Contributions: Conceptualization and methodology, C.P., R.S. and R.G.; writing-original draft preparation, J.K. and C.P.; writing-review and editing, C.P., R.S. and R.G.; supervision, C.P.; funding acquisition, R.G. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the City of Portland (CTL001657).

Data Availability Statement:
Water quality data are available on request from the corresponding author.