Results of EPA Research on Permeable Pavement

Transcription

1 Results of EPA Research on Permeable Pavement Michael Borst Engineer U.S. Environmental Protection Agency Edison, NJ New Jersey Local Technical Assistance Program September 15, 2016 Office of Research and Development National Risk Management Research Laboratory Water Supply and Water Resources Division Urban Watershed Management Branch

2 In talking with practitioners, we heard some recurring themes. It s different where I live. I live in the real world; laboratory-scale results are not convincing to me, my managers, and oversight board. I make very expensive, long-term decisions; I can t rely on short term research results. Uncertainty over maintenance is a killer. 2

3 EPA has ongoing permeable pavement research at three locations Edison, NJ Louisville, KY Fort Riley, KS 3

4 EDISON, NJ 4

5 In order to get measurements at the realworld scale, we need to be involved in the design stage. 5

6 In 2009, EPA designed and built a 1-acre parking lot for facility staff and visitors that was surfaced with three types of permeable pavement. 6

7 Overall, construction took about a year. November 26, 2008 February 26,2009 March 25, 2009 June 1, 2009 August 5, October 6, 2009 October 8, 2009 Green October Infrastructure 28, 2009 Research

8 The design incorporated water quality and hydrologic monitoring capabilities. PICP PC PA Interlocking concrete pavers Pervious concrete Porous asphalt Rain gardens Buried distribution pipes Tree islands Hot mix asphalt Buried well/piezometers Collection tanks Buried WCRs

9 Interlocking concrete paver installation took a little more than a week. 9 EPHenry EcoPavers East Penn Pavement Company

10 Pervious concrete pour took two days followed by a week of covered cure time 10 Weldon Concrete Nova Crete, Inc.

11 Placing the porous asphalt took two days. 11 Stavola, Inc. Stavola, Inc.

12 Vertical cross sections of permeable sections varied slightly from material to material. Concrete Pavers (3.125 in.) AASHTO No. 8 (2 in.) AASHTO No. 57 (4 in.) AASHTO No. 2 RCA (depth varies) EPDM Membrane Existing Subgrade Pervious Concrete (6 in.) AASHTO No. 2 RCA (depth varies) EPDM Membrane Existing Subgrade Porous Asphalt (3 in.) AASHTO No. 2 RCA (depth varies) EPDM Membrane Existing Subgrade 12

13 Four equally-sized and spaced lined sections collect infiltrating water from each monitored permeable surface with the balance infiltrating to the underlying soil. 13

14 Infiltrate drains from the lined sections to 1,500- gallon tanks on the east side of the parking lot where it can be sampled. 14

15 The tanks are designed to collect a 1.5-inch rain event before bypassing. 15

16 Tanks are cleaned before each event. 16

17 Collection tanks are homogenized with a centrifugal pump and sampled while particles are still in suspension. 17

18 Samples of asphalt runoff are collected at the southern end of the parking lot to serve as a control for comparison to infiltrate. 18

19 Rain water is sampled to document stressors from atmospheric deposition. 19

20 SOME WATER QUALITY RESULTS 20

21 All permeable surfaces reduced Suspended Solids Concentration (SSC) to different degrees Mean SSC (mg/l) mg/l Date Range: 1/26/2010 6/4/2013 Events sampled: 42 (PICP, PC, and PA) 28 (Runoff) PICP PC PA Runoff Mean Mean±0.95 Conf. Interval 21 Note: NSQD TSS commercial median 98 mg/l

22 Acidic rainfall is buffered by all pavement surfaces, and PA exfiltrate is surprisingly high Mean ph PICP PC PA Runoff Rain Date Range: 1/26/2010 6/4/2013 Events sampled: 42 (PICP, PC, and PA) 34 (Runoff) 36 (Rain) Mean Mean±0.95 Conf. Interval 22 NSQD commercial 7.4 (n=527)

23 TN concentrations in PICP, PC, and Runoff were not significantly different, but all three were significantly less than PA. 1.2 A Average EMC (mg/l) B B B A C B B C A D A A A A B C B C D C B C C D NH3 NO2 NO3 TON TN Stressor PICP PC PA RO RAIN Date Range: 10/28/2010 9/7/2011 Events sampled: 13 (PICP, PC, and PA) 12 (Runoff and Rain)

24 Total phosphate concentrations in PICP and PC were significantly larger than in runoff, and all were significantly larger than PA Average EMC (mg/l) A A A AB B B C D D T-PO4 F-PO4 Stressor C PICP PC PA RO RAIN Date Range: 10/28/2010 9/7/2011 Events sampled: 13 (PICP, PC, and PA) 12 (Runoff and Rain)

25 It is easiest to consider the concentration of metals in the infiltrates as groups. Exceeds GEL Yes No Sources Differ Yes Al, Fe Ba, Cu, Cr, Mn, Ni, Zn No As, Cd, Pb, Sb Exceeds GEL = less than 95% meet the Groundwater Effluent Limit Analysis by EPA (ICP-AES) generated results for 22 metals, Of the 22 metals, 12 have established GEL concentrations. 25

26 The copper concentration varied by surface but are uniformly less than the GEL. Concentration (mg/l) log -scale Total Copper PICP PC PA RO Rain Concentration (mg/l) log -scale Total Copper Exceedance Probability 26

27 Calcium chloride (with no sand) and a rubber edged plow blade were used to manage snow and ice. 27

28 After winter salt application, chloride concentration decreases throughout the remainder of the year. 10,000 1,600 LS Mean Chloride Conc. (mg/l) 1, Acute (860 mg/l) Chronic (230 mg/l) MDL (5 mg/l) 1,400 1,200 1, Cumulative Depth (mm) 28 PICP PC PA Snow (by season) Rain (since previous snow event) Bars represent 95% confidence intervals; snowfall data from NJ State Climatologist.

29 The reported concentrations are the values leaving the storage gallery. Column studies suggest that there is significant sorption analytes passing though the soil column. 29

30 SURFACE INFILTRATION & SURFACE CLOGGING 30

31 We monitored infiltration in the four permeable sections. PC-N PICP PC PA Interlocking concrete pavers Pervious concrete Porous asphalt Rain gardens Buried distribution pipes Tree islands Hot mix asphalt Buried well/piezometers Collection tanks Buried WCRs

32 We measured surface infiltration rates using a modified version of ASTM C1701 at roughly monthly intervals for about three years. 32 Modifications were (1) how the seal was achieved between the ring and the surface (2) added temperature measurements of surface and water.

33 We made many infiltration measurements. Solid circles < 1000 cm/hr < 500 cm/hr < 1000 cm/hr < 50 cm/hr PA PC PC-N PICP 33

34 The infiltration rate varies among the four tested surfaces, but all surfaces are sufficient to handle maximum expected direct rainfall rates m (6-in.) thick 100-year, 5-minute rainfall intensity 0.20-m (8-in.) thick Edison, NJ 20.8 cm/hr (8.2 in/hr) 34

35 Mean Monthly Infiltration Rate (cm/h) 3,000 2,500 2,000 1,500 1,000 Infiltration decreased with age for the three surfaces that received run-on from driving lane Number of Months Since Opening PICP Linear (PICP) 35 Bars represent standard error. y = x R² = p =

36 Inspection suggests there is progressive clogging upgradient to down gradient. 36

37 The progressive accumulation was also visible in the PICP section. No sediment Sediment 37

38 We developed a working hypothesis of the mechanics of the infiltration processes. Flow Small response Large response 38

39 As gaps fill with sediment, the location of the primary infiltration area moved downgradient. Flow Large response Medium response 39

40 The surface clogging progression has varied slightly because of microtopography. High resolution LASER survey PICP row 40

41 wikipedia.com LOUISVILLE, KY 41

42 The test basin is the 17 acre drainage basin CSO130 in the Butchertown section of Louisville. Work being done under a consent decree. Metric Current Target Reduction AAOV (MG/Yr) % Overflows % 42

43 Within CSO 130, MSD decided that the preferred stormwater control is permeable paver strips installed as parking lanes near the catch basins. MSD made a policy decision to assume responsibility for the ongoing maintenance. This led them to place the controls on publically-owned property. The limited publically-owned property and generally narrow sidewalks forces the controls into the streets. 43

44 The sewer district selected a Permeable Articulating Concrete Block/Mat installed in the parking lanes. 44

45 Street slopes (longitudinal and transverse) and curbs (height and condition) are important considerations when placing the controls in a curb and gutter system. Curb / Sidewalk Crown of road Short curbs by design (or resulting from multiple resurfacing without milling) will limit the working flow width of the control. Steeper slopes concentrate flow and use less surface area. 45

46 The important consideration is working surface area, not surface area. It is not obvious what to use as a working width (e.g., mean / median calculated 5-minute average flow width). 100% 80% 60% 40% The current ICPI guidance is a 20% ratio of control area to drainage area. 0% CPDF S c =2.81% S o =1.30% n = Louisville Control 19G minute average flow Width (ft) Y2004 Y2005 Y2006 Y2007 Y2008 Y2009 Y2010 Y2011 Y % of 5-minute average flow widths (excluding infiltration) are 0.75 ft or less

47 We expect flow to concentrate along the curb with much smaller flows from the road crown and direct rainfall entering elsewhere. Curb Edge x= 22.9 m x= 12.2 m x= 6.1-m x= 2.29 m x= 0.76 m 0.15 m 1.07 m 1.07 m 0.15 m 47 Street Edge = Assumed Flow Direction = TDR Location x Not to Scale

48 We expect the concentrated flow to transport and deposit sediment from the drainage area. Curb Edge x= 22.9 m x= 12.2 m x= 6.1-m x= 2.29 m x= 0.76 m 0.15 m 1.07 m 1.07 m 0.15 m Street Edge = Assumed Flow Direction 48 Not to Scale = WCR TDR Location x

49 Water Content Reflectometers (WCRs) were installed 40 cm below the paver surface in the aggregate to measure surface clogging progression. 49

50 The foundational research 1 linking the WCR response to moisture content was done in mineral soils, not gravel so the output value is the Relative Volumetric Water Content (RVWC) Topp, G. C., Davis, J. L., and Annan, A. P. (1980). Electromagnetic determination of soil water content: Measurements in coaxial transmission lines. Water Resources Research, 16,

51 The curb edge TDR responses support the predicted clogging progression. 51 Data before 1 st Maintenance Control 19G Length: 36.6 m

52 Distance of Curb Edge TDR from Upgradient Edge (m) The TDR response can be used to determine the control s longitudinal clogging rate as a function of rainfall. y = 0.135x R² = y = 0.111x R² = First RVWC > 0.10 Last RVWC > Cumulative Rain Since Installation (mm) 52 Control Louisville 19G Response threshold 0.10 RVWC The initial clogging rate was about m per mm (10 ft per inch) of rain.

53 If this model is accurate, the particle size distribution of up gradient and down gradient sediment should differ. Flow Large response Medium response 53

54 100 Particle size distribution of the clogging sediment collected from a control in Louisville Percent Passing Down gradient 19 G-A 19 G-B 19 G-C 19 G-D Flow Particle Size (mm) Based in Research by Amir Ehsaei, University of Louisville

55 Removing pavers at selected locations shows how clogging advances with gaps filled and small debris loading. 55

56 With aggregate between the pavers, most of the fines are trapped in the top 20 mm. 56

57 When pavers are installed without interstitial gravel, sediment progresses the full depth of the paver. 57

58 Some meso-scale tests Ratio of impervious to pervious: 20:1 Sediment in each experiment ~ 1.3 kg 20% organics, 80% inorganics by mass 50.8 cm cumulative rainfall PSD of sediment matched Louisville samples Varied Slope Gap width Presence / absence of interstitial stone 58

59 The flume tests gave an opportunity to observe the progression of the wetting and clogging front. 59

60 Post-test examination shoes the same clogging pattern that we observed in the field. 60