Journal of Environmental Management 95 (2012) 114e123

Contents lists available at SciVerse ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Permeability measurement and scan imaging to assess clogging of pervious

concrete pavements in parking lots
Masoud Kayhanian a, *, Dane Anderson a,1, John T. Harvey b, 2, David Jones b, 3, Balasingam Muhunthan c, 4
a

Department of Civil and Environmental Engineering, University of California, Davis, CA 95616, USA
University of California Pavement Research Center, Department of Civil and Environmental Engineering, University of California, Davis, CA 95616, USA
c
Department of Civil and Environmental Engineering, Washington State University, Pullman, WA 99164-2910, USA
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 5 January 2011
Received in revised form
18 July 2011
Accepted 29 September 2011
Available online 13 November 2011

This paper describes a study that used permeability measurement along with physical and hydrological
characteristics of 20 pervious concrete pavements in parking lots throughout California. The permeability
was measured at ve locations: the main entrance, an area with no trafc, and three separate
measurements within a parking space at each parking lot. Hydrological and physical site characteristics
such as trafc ow, erosion, vegetation cover, sediments accumulation, maintenance practice, presence
of cracking, rainfall, and temperature data were also collected for each parking lot. These data were used
to perform detailed statistical analysis to determine factors inuencing changes in permeability and
hence assessing possible cause of clogging. In addition, seven representative core samples were obtained
from four different parking lots with permeability ranging from very low to very high. Porosity proles
produced from CT scanning were used to assess the possible nature and extent of clogging.
Results showed that there is a large variation in permeability within each parking lot and between
different parking lots. In general, the age of the parking lot is the predominant factor inuencing the
permeability. Statistical analysis revealed that ne sediment (particles less than 38 mm) mass is also an
important inuencing factor. Other inuencing factors with lower signicance included number of days
with a temperature greater than 30
C and the amount of vegetation next to the parking lot. The
combined scanned image analysis and porosity prole of the cores showed that most clogging occurs
near the surface of the pavement. While lower porosity generally appeared to be limited to the upper
25 mm, in some core samples evidence of lower porosity was found up to 100 mm below the surface.
2011 Elsevier Ltd. All rights reserved.

Keywords:
Pervious concrete
Parking lots
Permeability
Clogging
Performance
Fine particles
CT scan

1. Introduction
An alternative to managing stormwater runoff from roadways is
low impact development (LID). Among hydromodication
methods, the use of fully permeable pavements in which all surface
water passes through the pavement surface and is kept in the
pavement until it inltrates the subgrade, evaporates or ows out
through a pipe, is gaining popularity as an approach to reduce the
stormwater ow volume and pollutant discharge loading (USEPA,

* Corresponding author. Tel.: 1 (530) 752 8957; fax: 1 (530) 752 7872.
E-mail addresses: mdkayhanian@ucdavis.edu (M. Kayhanian), danederson@
gmail.com (D. Anderson), jtharvey@ucdavis.edu (J.T. Harvey), djjones@ucdavis.edu
(D. Jones), muhuntha@wsu.edu (B. Muhunthan).
1
Tel.: 1 (832) 402 0253.
2
Tel.: 1 (530) 754 6409.
3
Tel.: 1 (530) 754 4421.
4
Tel.: 1 (509) 335 3921.
0301-4797/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2011.09.021

1999; Dierkes et al., 2002; Ferguson, 2005). Both porous asphalt

and pervious concrete pavements (the names generally preferred
by the asphalt and concrete pavement trade organizations,
respectively) have been constructed in the past ten years. Most of
these pavements were constructed in parking lots and their use for
roads under heavy load and heavy trafc has been limited.
Runoff containing a large amount of suspended particles may
cause clogging of pore space during their life time (Nielsen, 2007;
Siriwardene et al., 2007; Yong et al., 2008). Clogging may also
occur due to the inorganic and organic particles from trafc activities and localized vegetation or dust that are often blown onto, or
forced into the roadway (Siriwardene et al., 2007). These particles
and solid materials are pushed into the pores of the permeable
pavement by the weight of the passing vehicles which may also
crush them producing a ner size distribution; sometimes the
debris is brought into the pores by owing water from a storm
event. It is possible that the vehicle weight may push materials into
the pavement at the leading edge of the tire, and the back side of

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

the tire may create a suction to remove some of the solids. Under
certain conditions and depending on the type and size of solids,
once in the pores, the debris may become embedded and begin to
accumulate, clogging the pores. When a signicant percentage of
the pores in the permeable pavement are clogged, the permeability
of the pavement is reduced and water begins to pool on the
pavement surface. This defeats the purpose of permeable pavements and the clogging materials must be removed to restore the
permeability (Cooley, 1999).
Most previous studies of porous pavement clogging were performed under laboratory condition or the eld investigation was
limited to few parking lots. At present, the combined use of eld
permeability, statistical evaluation and CT scanning image analysis
for the investigation of clogging is not widely encountered in the
literature. This study was undertaken with the following specic
objectives:
(i) Measure the permeability of pavement surfaces at ve
different locations within each parking lot for 20 pervious
concrete parking lots throughout California,
(ii) Collect data on hydrologic and physical site characteristics.
(iii) Perform statistical analysis of measured permeability and site
characteristics data to determine the primary factors inuencing permeability, and
(iv) Utilize porosity proles developed from computed tomography (CT) scanning images to assess pavement clogging.

2. Methods
2.1. Parking lots site selection
A list of over 100 pervious concrete parking lots throughout the
California was obtained. Several criteria including age, climate
region, and parking lot size were used to create a nal list of 20
pervious concrete parking lots for this study (Table 1). California
has wide variety of climate regions for pavement design, including
three coastal regions, two mountain regions, Inland Valley, High
Desert and Desert. After grouping the parking lots, it was apparent
that more parking lots were constructed in the Inland Valley
climate region, making it impossible to achieve uniform age range
Table 1
Selected permeable concrete parking lots by their location, climate region, age, and
size.
Site ID

Location

Climate region

Date
constructed

Age (yr)a

Size (m2)

PL1
PL2
PL3
PL4
PL5
PL6
PL7
PL8
PL9
PL10
PL11
PL12
PL13
PL14
PL15
PL16
PL17
PL18
PL19
PL20

Auburn
Red Bluff
Fresno
Manteca
Davis
Vacaville
Morgan Hill
Elk Grove
Elk Grove
Fair Oaks
Fair Oaks
Fair Oaks
Fort Bragg
Eureka
Palo Alto
Santa Cruz
Menlo Park
Oxnard
Petaluma
San Bernardino

Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
Inland valley
North coast
North coast
Central coast
Central coast
Central coast
South coast
Low mountain
South mountain

Jul-08
Mar-08
Apr-06
Oct-05
Sep-04
Sep-04
Jun-04
Dec-06
Aug-06
Aug-02
Jul-02
Jan-01
Apr-08
Sep-07
Jan-09
Aug-07
Nov-05
Jan-06
May-05
May-05

0.67
1.0
3.0
3.0
4.0
4.0
4.5
2.0
2.5
6.5
6.5
8.0
0.92
1.5
0.17
1.58
3.0
3.0
3.5
3.5

418
279
557
4645
4088
2787
557
1858
372
2323
2323
1524
492
186
465
465
279
4645
743
111

Age of pervious concrete parking lot when permeability was measured.

115

within other climate regions such as the mountains. In addition,

as expected, the use of pervious parking lots in the low rainfall
desert climate regions was entirely absent. To ensure adequate
coverage of the different geographic and climatic regions of the
state, 37 parking lots were initially selected. The nal list was
prepared by balancing size, rainfall and temperature characteristics
between them.
2.2. Field permeability measurement
The permeability at the 20 selected locations was measured
using an NCAT eld permeameter (Maupin, 2000; Cooley, 1999). A
photo view of the NCAT permeameter during eld measurement is
shown in Fig. 1a. To ensure accurate eld permeability measurement using the NCAT permeameter we made extra effort to prevent
water leakage between the base of the permeameter and the
pavement surface. To prevent leakage, the permeameter was sealed
to the pavement using Ecoex 5 silicone sealant by Smooth-On. To
correctly seal the permeameter to the pavement, a caulking gun
was used to ll a ring at the base of the permeameter and the
permeameter was quickly turned over and placed at on the
pavement. Small weights were placed around the base of the permeameter to help create a uniform seal. After roughly 5 min,
a quick check was performed by tapping the base of the permeameter to see if it would move. If the permeameter did not
move, the sealant was ready and water could be poured into the
cylinder. An O-ring of Ecoex sealant with no indication of water
leaking is shown in Fig. 1b. This method has been found to be more
effective than pluming putty that may not completely seal the base,
particularly with pervious concrete with rough edges.
The permeability test is performed using the falling head
method by lling the cylinder with water and measuring the time
required for water to fall the desired depth within permeameter
tiers. If water leaks were present for any reason, the test was
terminated. The data were then used to calculate the coefcient of
permeability, K, also known as the saturated hydraulic conductivity
using the following relationship:

Ks

 
al
h
ln 1
h2
At

(1)

where, Ks saturated hydraulic conductivity (cm/s), a inside

cross-sectional area of inlet standpipe (cm2), l thickness of
parking lot surface pavement (normally 10e15 cm), A crosssectional area of tested pavement, (cm2), t average elapsed time
of water ow between timing marks, (s), h1 hydraulic head on
specimen at time t1 (cm), h2 hydraulic head on specimen at time
t2 (cm).
It is important to note that currently a standard method (C1701/
C 1701 M-09) is available to measure the eld permeability of
pervious concrete. This standard method became available in
September 2009 and by that time we had completed our eld
permeability measurement using NCAT device. This NCAT permeameter is commercially available and several studies were performed to compare its measurement performance against other
design under eld condition and with laboratory measurement
(Maupin, 2000; Prowell and Dudley, 2002; Gogula et al., 2003;
Williams, 2008, Kayhanian et al., 2008; Ballestero et al., 2009)
and found to produce reliable and fairly accurate repeatable results.
Both NCAT and the new standard method uses Darcys law principal
and falling head method to measure eld permeability. The bottom
opening of the falling head pipe in standard method is slightly
larger and this slight difference in diameter may produce different
results that need to be investigated. However, it is important to
realize that the focus of our study was the relative permeability

116

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

Fig. 1. Photo view of NCAT permeameter and Ecoex sealant during eld permeability measurement.

measurement rather precision in measurement. We used permeability as among multiple other parameters to investigate clogging
of pervious concrete parking lots. Since we used the same method
and measured multiple permeability measurement within each
parking lot, the small difference compared with a presumed
accurate value would not change the outcome of our investigation.
A total of ve locations was tested within each parking lot (see
Fig. 2). These locations were selected to demonstrate the effects of
varying levels of trafc on the pavement permeability and to see
the variability of permeability within each parking lot. The rst
location was in the wheel path of the parking lot entrance, assumed
to have experienced the most trafc. The second and third locations
were located in each wheel path of a representative parking space
in the parking lot, and the fourth was located in the center of the
same parking space, but shifted toward the back of the parking
space to minimize interference from consecutive permeability
tests. The fth testing location was located in a paved area that
receives no trafc.
2.3. Statistical method
A regression analysis was performed to determine the inuence
of various parameters on measured eld permeability. To perform
this statistical analysis, various hydrologic and physical site characteristics data were collected including: climate region, pavement
age, trafc volume, erosion, vegetation cover, cracks, maintenance,
annual average rainfall, annual maximum rainfall, annual rainfall
>0.5 in (12.7 mm), annual average temperature, annual maximum
temperature, number of days with temperature >30
C, number of

Five field permeability measurement spots

ME = Parking entrance within the traffic wheel path

days with temperature <10
C, total mass of sediment per parking

lot area, total mass of sediment particles larger than 250 mm
diameter, and total mass of sediment particles less than 38 mm
diameter.
Some of these data were obtained from observations during site
visits. Rainfall and temperature data were obtained from the
weatherbase.com website and the annual average data were averaged over a period of approximately 50 years. Representative
sediments mass from each parking lot was collected using a DeWalt
DC500 portable vacuum, model. The collected sediments were
sieved in the lab based on the following sieve size fraction: >1000,
1000e650, 650e300, 300e150, 150e75, 75e38 and <38 mm. Only
sediments less than 1000 mm were used for statistical analysis.
Sediment mass less than 38 mm was further analyzed for particle
size distribution using a Laser In-Situ Scattering and Transmissometry (LISST) portable particle size analyzer. To further
determine the nature of sediment as organic or inorganic material,
the sediment less than 38 mm was tested for total volatile solids
following the APHA Standard (1989) method 2540E by igniting the
dry solids at 550
C.
Because there were numerous potential predictors relative to
the number of permeability measurements, it was not practical
(both computationally and conceptually) to look at all predictors
simultaneously. For this reason, the list of predictors was reduced to
those that were the most promising before running the regression
analyses. To produce the most reasonable predictor variables,
a three-pronged approach was used. The rst part of the analysis
was to use Classication and Regression Trees (CART) to produce
a subset of the original predictors that are most promising. CART is

NT
LWP

RWP

LWP = Left wheel path of parking space

RWP = Left wheel path of parking space
BWP = Between wheel path of parking space

BWP

NT = Pavement area with no traffic

ME

Fig. 2. Schematic drawing of eld permeability measurement locations within each parking lot.

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

a model-building method that searches recursively for predictors

and threshold values that separate responses in different ranges.
For example, at the rst step of a regression tree, the algorithm
looks for a single predictor and cutoff value that best separates
between the largest and the smallest responses.
With CART, the selected variables are considered potentially
important and promising for use in the regression analysis. The
second part of the analysis consisted of performing a standard
(stepwise) regression analysis to identify a second subset of
predictors from the ones obtained from CART that are promising as
linear predictors of the response. An important part of this analysis
was to check the residual errors to validate the model assumptions,
that the errors are normally distributed with a constant residual
variance. If not, then the dependent variable was transformed, and/
or outliers were dealt with, by Winsorizing the data (a technique
for handling or adjusting those extreme observations without
deleting them). The third part of the analysis included checking the
assumption that the outcome measurements are linear in the
predictors by running a generalized additive model (GAM). This is
a method that (typically) uses spline methods having a relationship
between a predictor and the outcome, using a model of the form:

Y f1 X1 f2 X2 //

(2)

If the tted spline functions f1 are essentially linear, then the

stepwise linear model is one that can be defended as a good choice.
If not, the models will need to incorporate the nonlinearity of the
response into the predictors. CART models were calculated using
the statistical package R. All other analyses were done using SAS
software version 9.2 (Schlotzhauer and Littell, 1997).

2.4. Pavement coring procedure

The purpose of pavement coring was to obtain eld samples for
full depth CT scanning. Core samples were made using a portable
air-cooled coring machine with a 102 mm (4 inch) inner-diameter
drill bit. Air-cooled coring was used to eliminate the possible
clogging of the cores with water/sediment slurry associated with
water cooled coring. To further reduce the possibility of cooling air
blowing drill bit and coring residue through the voids in the
sample, a non permeable paper patch was glued to the pavement
surface over the cored area so that the cooling air could not pass
through the cored sample. For coring, four pervious concrete
parking lots were selected based on their age and the average
permeability ranging from low, medium, high, and very high.
Generally, two representative core samples (except PL4) were obtained from four parking lots. The core samples were collected from
the same areas where permeability was measured. A summary
description of the seven core samples and related information such
as permeability, age, and sediment mass is summarized in Table 2.

117

2.5. CT Scanning image analysis

Each core sample was scanned to produce a porosity prole
including the entire core depth. The overall processes involved with
X-ray CT scanning are shown in Fig. 3. The X-ray CT scan setup
involves two X-ray sources that are capable of generating a 420 keV
and 225 keV voltages respectively. The 420 keV source is used for
relatively bigger samples and the 225 keV is used to attain
enhanced resolution and is best suited for smaller samples. These
X-ray sources are networked to a central work station, a processing
platform that consists of four parallel computing processors and
software that control and execute the scanning process.
CT scanning of core samples was performed at the Washington
State University High-Resolution X-Ray Computed Tomography
(WAX-CT) Laboratory. Samples were initiated with the FlashCT DAQ
program which outputs raw data, as header les. The header les
are processed with a program called FlashCT DPS which gives
a reconstructed image of the scanned slices. The reconstructed
images of the slices are converted into a 3-Dimensional image with
a third software called FlashCT VIZ. Finally, the processed image is
analyzed with MFC software to get the XY, XZ and YZ-sliced image
formats through image processing software such as Mat LAB and
Image Pro Plus.
3. Results and discussion
3.1. Field permeability results
The results of eld permeability measurements for ve locations
within each parking lot for 20 pervious concrete parking lots are
summarized in Table 3. As shown, the eld permeability within
a single parking lot varied widely with testing location and within
each site. For example, in parking lot PL7 there is a difference of
three orders of magnitude between the no-trafc and the main
entryway eld permeability. This shows a large permeability range
within a single parking lot, which can be explained by the difference in trafc volume experienced at the two different locations.
However, site PL7 is an extreme case. Many of the other sites had
comparable results for the no-trafc and main entryway locations.
This is may be due to the fact that the only available no-trafc
locations were often located next to the edge of the parking lot and
most often exposed to erosion of soils from surrounding landscaped areas.
The range of permeability at different locations within a single
site is comparable to the range of permeability at different sites
throughout the state. For example, site PL11 yielded consistently
low permeability values, on the order of 103 cm/s, while site PL17
yielded consistently high permeability values, on the order of
100 cm/s. The difference between the permeability of these sites is
three orders of magnitude. Field permeability testing of these

Table 2
Description of seven core samples and pertinent parking lot characteristics.
Core sample IDa

PL2-1
PL2-2
PL4-1
PL6-2
PL6-3
PL12-1
PL12-3
a

Description of core sample

First core sample from parking lot 2

Duplicate core sample from parking lot 2
Single core sample from parking Lot 4
First core sample from parking lot 6
Duplicate core sample from parking lot 6
First core sample from parking lot 12
Duplicate core sample from parking lot 12

For parking lot geographic location refer to Table 1.

Pertinent parking lot characteristics

K cm/s (in/h)

Age (yr)

Total
sediment (g)

Fine sediment
(<38 mm)
fraction (%)

0.82
0.82
0.02
0.002
0.002
0.0007
0.0007

1
1
3
4
4
8
8

6.15
6.15
51.25
68.16
68.16
16.99
16.99

2.01
2.01
8.14
2.92
2.92
4.19
4.19

(1162)
(1162)
(28)
(2.8)
(2.8)
(z1)
(z1)

118

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

Fig. 3. X-ray FlashCT setup and associated processes.

pervious concrete sites yields a large range of results because each

test only measures a small area of the pavement, and any inconsistency in the construction of the pavement or any damage to the
pavement in the testing area greatly affects the test results. For this
reason, great care was taken when selecting locations to test within
each parking lot to achieve a test section representative of the
whole parking lot.
To compare the results of permeability from parking lots, only
the averages of three permeability measurement from the parking
space (LWP, RWP, and BWP) and the average of all permeability
measurements was used. These permeability results were then
ranked in descending order. The results showed that: (i) older
parking lots generally had lower permeability, (ii) the highest
permeability was measured in PL17 even though the parking lot

Table 3
Field permeability constant within ve locations for 20 pervious concrete parking
lots.
Site ID

PL1
PL2
PL3
PL4
PL5
PL6
PL7
PL8
PL9
PL10
PL11
PL12
PL13
PL14
PL15
PL16
PL17
PL18
PL19
PL20
a
b

Permeability constant (K), cm/s

Main entryway

LWP

RWP

BWP

(1)

(2)

(3)

(4)

(5)

0.912
0.653
0.313

0.016
0.692
0.165
0.057
0.009
0.002
0.035
0.397
0.0002
0.0002
0.0015
0.0009
0.365
0.5002
0.214
0.058
1.494
1.350
0.060
0.742

0.188
0.910
0.137
0.007
0.0013
0.0045
0.012
0.426
0.0002
0.0003
0.0002
0.0005
0.212
0.704
0.122
0.050
1.986
1.270
0.189
0.058

0.473
0.963
0.207
0.004
0.006
0.002
0.063
0.483
0.0002
0.009
0.0002
0.0009
0.851
0.690
0.224
0.004
1.971
1.962
0.277
0.134

0.068
0.878
0.327
0.0012
0.005

0.004
0.0002
0.001
0.238
a

0.042
0.002
0.0007
a
a

0.017
0.028
a
a

0.004
a

The entryway was not permeable.

Not available because no-trafc zone in this parking lot.

No trafc

1.355
b

was 3 years old, and (iii) large difference in average permeability

between parking lots with of similar age. The large variation in
permeability measurements is not uncommon as has been reported
by other researchers (Ferguson, 2005; Montes and Haselbach,
2006). In addition, earlier studies performed by St. John et al.
(1997) and Wei (1986) showed that the eld permeability
decreased by 90% after ve years of operation.
3.2. Factors inuencing eld permeability measurement
The factors inuencing the eld permeability measurement
were evaluated through statistical analysis. To perform this statistical analysis, the permeability measurements were considered as
the dependent variables:
KME permeability at main entry to the lot,
KWPs permeability at average of the left and right wheel paths
of a parking space,
KBWP permeability at between the wheel paths in the same
space, and
KNT permeability at an area of the parking lot where there was
no trafc.
From the initial linear regression models it became apparent
that the dependent variables were better suited for analysis after
being log-transformed. The CART and stepwise regression analysis
were run on both raw and log-transformed dependent variables,
but the analysis from that point on was limited to the log-

0.026
0.002
b
b

0.189
1.320
0.066
b
b

0.429
b

Table 4
Dependent variables based on regression and CART.
Permeability

ln(KME)
ln(KWPs)
ln(KBWP)
ln(K NT)

Dependent variables based on statistical method

Regression

CART

Temperature
Pavement age, days 30
Pavement age
Pavement age

Pavement age
tsm38, tsm/lot area
tsm38, days 30
Vegetation

Note:tsm38 total sediment mass of particles less than 38 mm.

tsm/lot area total sediment mass per parking lot area.
Days 30 number of days that the temperature is >30
C.

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

lnKNT 6:18  0:57  age

Table 5
Model R2 and p-values for dependent variables.
Dependent

Model R2 (%)

Predictor parameter

p-Value*

ln(KME)

61

ln(KWPs)

85

Pavement age
Temperature
Pavement age
f1 (pavement age)
tsm38
f2(tsm38)
Pavement age
tsm38
f3(tsm38)
Pavement age

0.0013
0.0316
0.0213
0.0188
0.0845
0.0083
0.1093
0.0082
0.0002
0.0858

ln(KBWP)

ln(KNT)

77

23

119

where,

f1 x 0:99 1:65x  1:16  2 0:30  3;

1:09  0:62x 0:05x2 ;

1:23 28:77x  3:19x2  34:85x3

0:47 < x < 0:6

52:59  10:20x  84:06x 36:60x

(3)

lnKWPs 7:03  2:55  age  1:22  tsm38 f1 age

(4)

(5)

0:60 < x < 0:87

56:92  28:46x 139:73x2  57:69x3

f3 x

lnKBWP 7:06  0:33  age  1:88  tsm38 f3 tsm38

x>3:5

f2 x 0:46 7:61x  2:07x2  4:32x3 x < 0:47

transformed variables. From the stepwise regression results

dependent variables were identied as either being statistically
signicant or being marginally signicant. For the log-transformed
analyses, the variables that were identied as promising are
summarized in Table 4.
In the next phase of the analysis, using the GAM model, the
dependent variables were further rened because of collinearity
with some of the other predictors or removed from a composite
model for lack of signicance. When these predictors were
combined in GAM models, attempts were made to eliminate
extraneous predictors and express them as linear predictors, rather
than tting with splines, whenever possible. From the above
combined analysis, the predictor variables that best t the current
permeability data are presented below

f2 tsm38

x < 2:0

2:0 < x < 3:5

0:65 0:10x  0:005x2 0:0008x3

*p value shows the signicance; lower the p value the higher the signicant impact.

lnKME 9:25  0:87  age 1:03  temp

(6)

1:66 11:12x  1:96x2  13:95x3

2

x>0:87

x < 0:47
3

7:29 11:60x  10:26x  14:18x

0:47 < x < 0:70

45:65  8:11x  72:13x2 29:37x3

0:70 < x < 0:87

77:97  29:29x 180:01x  76:21x3

x>0:87

x in f1(x), f2(x), and f3(x) age, tsm38 and tsm38, respectively.

The signicance of each inuencing parameter along with the
overall model R2 is presented in Table 5. As shown, the pavement
age was found to be a major factor that signicantly inuences the
pavement permeability and hence lower permeability measurement in older parking lots can be assumed to be partially related to
clogging. However, lower permeability measurement alone may
not be used as a basis to evaluate surface pavement clogging. For
instance, it can be seen that several newly constructed parking lots
had lower permeability compared to the older parking lots and
their low permeability is not necessarily an indication of clogging.
The lower permeability in newer parking lots may partially be due
to improvement in pavement design in which the void ratio are
reduced to improve the structural capacity of the pavement to
handle heavier loads and higher speed while still meeting the
inltration capacity. This evolution of pervious concrete mix design
is due to the fact that permeabilities that are more than 1000 times
greater than normal rainfall intensity may not be required and
lower permeability up to 500 times greater may be sufcient to

Table 6
Particle size distribution for sediment collected from each permeable parking lot site.
Site

Measured sediment mass

% sediment mass fraction less than 1000 micron by particle size range (mm)

<1000 mm mass
in each PL(kg)a

<1000 mm mass/PL
area (g/m2)

600e1000

425e600

250e425

125e250

75e125

38e75

<38

0.9
22.1
22.0
11.0
54.0
24.5
187.8
1.8
24.8
34.7

53.96
38.61
57.55
31.22
42.88
37.30
18.29
39.06
25.74
29.95

16.73
16.69
21.34
13.08
18.63
16.19
12.11
17.18
13.42
15.32

10.97
18.25
7.63
14.91
15.35
16.91
19.36
17.65
17.86
19.27

8.45
14.28
5.92
13.63
10.52
13.27
21.46
13.24
19.85
18.14

3.60
6.25
3.12
8.63
4.02
6.84
12.1
5.82
10.78
8.14

3.78
3.90
3.27
10.4
4.64
6.57
10.7
3.57
7.85
5.92

2.52
2.01
1.17
8.14
3.96
2.92
5.94
3.47
4.48
3.27

11.2
15.0
18.0
99.0
223.1
15.5

29.22
38.24
40.84
39.81
26.71
35.42

13.99
15.24
18.70
14.10
14.43
16.99

24.67
17.06
13.93
15.84
21.52
18.82

15.86
13.50
13.74
13.21
18.43
14.97

7.24
6.24
5.73
6.83
8.28
6.60

4.83
6.21
4.77
6.35
6.46
4.77

4.19
3.50
2.29
3.86
4.18
2.43

12.6
113.3

42.80
28.22

15.84
15.58

15.17
22.49

10.27
19.03

5.81
8.22

6.69
5.24

3.42
1.22

Pervious concrete
PL1
0.39
PL2
6.15
PL3
1.20
PL4
51.25
PL5
22.22
PL6
68.16
PL7
104.66
PL8
3.33
PL9
9.23
PL10
80.56
PL11
Not measuredb
PL12
16.99
PL13
7.39
PL14
0.33
PL15
4.58
PL16
103.63
PL17
4.32
PL18
Not measuredb
PL19
9.35
PL20
12.63
a
b

Sediment larger than 1 mm was captured. These particles are less mobile and remain on surface and can easily be removed through sweeping or other cleaning methods.
Sediment was not measured due to vacuum equipment failure.

120

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

70
60

uL/L

50
40
30
20
10

.5
27
.1
37
.6
52
.4
72
.8
10
1
14
1
19
6

19

.1

14

24

10

7.

21
5.

72

3.7

97

2.

1.

1.

44

Particle Size (microns)

Fig. 4. Example particle size distribution for parking lot (PL8) sediments under 38 mm,
measured by LISST particle size analyzer.

handle stormwater runoff inltration and provide adequate void

space to limit surface clogging. Unfortunately, the original and
historical permeability measurement from these parking lots is not
available to perform such analysis, and differences in the original
concrete mix designs were not available.
The second important factor inuencing the eld permeability
measurement was the mass of particles less than 38 mm diameter.
The total amount of sediment mass and mass distribution with
respect to various particle fractions ranging from 1000 to 38 mm
from each parking lot is summarized in Table 6. As shown, the
majority of the sediment mass for all of the parking lots is related to
particles larger than 38 mm. This is due to the debris from
surrounding trees (i.e. twigs and leaves) and sometimes the
breakdown of the top pavement layer, causing broken pieces of
pavement to lie loose on the surface. However, particles smaller
than 38 mm can play a signicant role in surface pavement clogging
as is discussed further. As mentioned earlier in the Method section,
the distribution of particles smaller than 38 mm was determined
using the LISST portable particle size analyzer. To accomplish this,
a representative mass of particles below 38 mm was mixed with
milli-Q water and the solution was used to measure the particle size

distribution (PSD). One example particle size distribution for

particles smaller than 38 mm is shown in Fig. 4. As expected, the
peak measured particles were at or about 38 mm. Nevertheless,
larger particles have been measured that may be attributed to
natural coagulation during measurement. This occurs when dry
samples of ne sediment are placed in water and stirred, causing
the particles to increasingly clump together as they are stirred. In
general, however, the majority of particles was below 38 mm and
the particle concentration larger than 100 mm was insignicant.
These ne particles, particularly in clay and inorganic form,
collected from parking lots may play a major role in clogging. To
further investigate the nature of these particles, a representative
sample of solids <38 mm from each parking lot was used to
measure the organic content through the standard method
measurement of total volatile solids (TVS). The results of this
measurement are summarized in Table 7. As shown, the organic
content of particles below 38 mm for most parking lots was less
than 25%. This low organic content indicates that most solid
particles collected from parking lots are inorganic and hence there
is a low probability for these particles to degrade. Hence, the only
way to remove them is through cleaning.
The inuence of ne particles on surface pavement clogging can
be conrmed by two previous research studies. In one study,
Siriwardene et al. (2007) investigated physical clogging of permeable pavements and found that a clogging layer forms at the
interface between the lter course and underlying soil, and that the
main driver in the development of a clogging layer is the migration
of particles less than 6 microns in diameter. In a separate study,
Yong et al. (2008) used only the particles less than 6 microns
present in the simulated stormwater to clog the pavements. No
other materials were used to clog the pavements. The accumulation
of the particles less than 6 microns in diameter over the 17.5 weeks
resulted in signicant clogging of the permeable pavement when
the ood storms were simulated.

Table 7
Organic content of sediment mass less than 38 microns.
Site

Total sediment
mass for
particles <1000 mm (g)

Pervious concrete
PL1
0.39
PL2
6.15
PL3
1.20
PL4
51.25
PL5
22.22
PL6
68.16
PL7
104.66
PL8
3.33
PL9
9.23
PL10
80.56
PL11
Not measured
PL12
16.99
PL13
7.39
PL14
0.33
PL15
4.58
PL16
103.63
PL17
4.32
PL18
Not measured
PL19
9.35
PL20
12.63

Sediment mass for p

articles <38 mm

Organic
content

(g)

(%)

(% VS)

0.14
1.32
0.15
2.67
0.64
2.12
33.14
0.37
3.31
3.37

2.52
2.01
1.17
8.14
3.96
2.92
5.94
3.47
4.48
3.27

NES
22.2
NES
25.3
NES
27.9
9.5
NES
19.7
23.6

1.39
1.17
0.12
1.13
13.85
1.12

4.19
3.50
2.29
3.86
4.18
2.43

28.8
16.7
NES
15.9
20.5
25.9

1.28
0.77

3.42
1.22

17.0
10.4

NES not enough sediment to perform VS test.

Sediments from PL11 and PL18 were not collected due to vacuum equipment failure.

Fig. 5. Examples of original and segmented scanned images for a high permeable (top)
and moderately permeable (bottom) pervious concrete pavement (while solid,
black air void).

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

3.3. Evidence of clogging using CT scanning image evaluation

To assess the evidence of clogging, CT scan image of seven core
samples obtained from four different parking lots were evaluated.
Numerous slice images produced during CT scanning were processed and underwent segmentation in which the void space and
solid fraction of each slice was determined. Fig. 5 shows an example
of an original and segmented image for two core samples with
relative permeability of high (top) and low (bottom). In the
segmented image the white color represents solids and the black
color represents the air voids.
Fig. 6(i) show the porosity proles with depth of four core
samples from different parking lots. To aid visual inspection, typical
horizontal (inset images) and vertical (presented as (ii)) slices are
presented along with the porosity proles. The average porosity is
also shown on each prole. It can be seen that the porosities for
parking lots PL6 and PL12 are lower compared to porosities of
samples from parking lots PL2 and PL4. Visual evidence of these is
found on the inset slice images shown. These results can be
conrmed with permeability results shown in Tables 2 and 3. From
these results it can be concluded that core samples with higher

121

average porosity also had a higher permeability. The high correlation between these two parameters was conrmed by measuring
total air voids (porosity) and permeability of different open graded
concrete pavement specimens in the lab. The relationship developed is shown in Fig. 7. The only difference is that the permeability
measurement can not distinguish the difference in porosity at
different depths of the pavement. For instance the average porosity
for the following samples may result the same permeability
measurement:
A. Low porosity at the top 5 cm layer and much higher porosity in
the rest of the core,
B. Low porosity in mid depth and higher porosity in the rest of the
core,
C. Low porosity in bottom section of the core and higher porosity
in the rest of the core,
Scanning, on the other hand, provides an opportunity to
investigate the porosity prole throughout the core depth and that
possibly be used as a tool to investigate clogging. Preliminary CT
scanning images evaluated in this study generally showed a lower

Fig. 6. CT scanning porosity proles of core samples from four different parking lots.

122

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

and generally found to be similar. Gravimetric measurements were

performed with specimens vacuum sealed in plastic bags using the
CoreLok device. In a number of samples, the porosity obtained by
the gravimetric method is greater than the average value computed
over the depth from CT scanning. This may be attributed to differences in the boundary denitions at the specimen surfaces for the
two methods and the inability of the CT method to measure pores
that are smaller than the resolution of the scanning equipment. The
gravimetric method is useful for ensuring the CT scanning values
are reasonable. However, the CT scanning results are accurate
enough to detect the clogs and their location, which makes it
a superior method for the purposes of this investigation.

8.0

Permeability (cm/sec)

7.0
-6

y = 1.4x10 * x^(4.5) R = 0.91

6.0

Pavement type = PCC-O

5.0
4.0
3.0

4. Conclusions

2.0

The conclusions drawn from this study are summarized below:

1.0

20

22

24

26

28

30

32

Air voids (%)

Fig. 7. Relationship between permeability and air voids (porosity) of concrete
specimens.

porosity on the upper surface for most core samples. However,

a decrease in porosity in the mid-section for few of the core
samples (not shown here) was also noticed and could not be
conrmed for clogging. Nevertheless, it is important to note that
the decrease in porosity did not signicantly change the permeability to create surface ponding or surface overow. To pose
a serious reduction in stormwater inltration, nearly 90% of the
void ratio must be plugged (Al-Omari and Masad, 2004).
In a separate study, Ongel et al. (2008) used both eld permeability and scanning methods to evaluate the clogging of open
graded asphalt porous friction course (PFC) pavements. Results
revealed that the top of the cored samples typically had a lower air
void content than the bottom of the cored samples. From this
preliminary nding, they concluded that generally clogging takes
place at the top zone of the surface of open graded mixes, especially
surface layer mixes with thickness of about 25 mm. The top surface
clogging noticed in the investigation presented in this paper and
other previous studies can be viewed as good news for two reasons:
(1) there is a possibility of self cleaning by tire pressure during rain
event, and (2) it is easier to remove these particles by vacuum or
other cleaning mechanisms.
The average porosity computed by scanning was veried by
measuring the overall porosity of the samples in the laboratory
using a gravimetric method (Candaele et al., 2008). The results
obtained from both methods were compared as shown in Table 8,

Table 8
Average porosity of core samples based on scanning and gravimetric methods.
Sample

PL2-1
PL2-2
PL4-1
PL6-2
PL6-3
PL12-1
PL12-3

Average porosity (%)

Scanning method

Gravimetric method

34.51
17.39
25.87
9.60
8.26
14.1
22.03

26.54
19.68
14.87
14.36
9.19
23.55
22.27

 The permeability results indicate that there is very large variability in permeability within each parking lot and when
compared with other parking lots. The variability in K value
between no-trafc and trafc areas can range up to a factor of
1000. In general, the permeability of older parking lots was
lower than newer parking lots; indicating that age is a factor
affecting permeability.
 The statistical analysis revealed that the age of pavement is the
main factor that explains variations in measured permeability.
Another important factor affecting pavement permeability is
the mass of ne particles less than 38 mm. Other factors that
explain the observed variations include number of days with
temperature greater than 30
C, thought to inuence tire
rubber deposition, and the amount of vegetation surrounding
the parking lots.
 The combined image analysis and porosity prole showed that
the porosity of the top layer surface (25 mm) of core samples
was generally lower and there may be an indication of clogging.
In some core samples the evidence of lower porosity has also
been noticed in the mid layer and the clogging effect could not
be conrmed. Nevertheless, if the lower porosity in the mid
layer is due to the penetration of ne particles from surface of
the pavement then a regular cleaning maintenance program
will improve the average porosity prole.
Acknowledgments
This research study was funded by the California Department of
Transportation, Division of Environmental Analysis through
Agreement 43A0247. The content of this paper reect the views of
the authors and do not necessarily reect the ofcial views or
policies of the state of California or the Regents of the University of
California. The authors are thankful: to the staff of the University of
California Pavement Research Center for their assistance on sample
coring and to Mr. Neil Willits from UC Davis Statistical Laboratory
for his advice on statistical analysis. We appreciate the assistance of
research staff especially Kalehiwot Nega from the Washington State
University WAX-CT Laboratory who performed the CT scanning and
image analysis. We are especially thankful to the ACPA for
providing the list of pervious concrete parking lots in California and
the related pavement information that was used for our test site
selection for clogging investigation.
References
Al-Omari, A., Masad, E., 2004. Three dimensional simulation of uid ow in X-ray
CT images of porous media. International Journal for Numerical and Analytical
Methods in Geomechanics 28 (13), 1327e1360.

M. Kayhanian et al. / Journal of Environmental Management 95 (2012) 114e123

APHA Standard 2540E, 1989. Total Volatile Solids Ignition at 550
C, Standard
Methods for the Examination of Water and Waste Water, 17th ed. American
Public Health Association (APHA), New York.
Ballestero, T., Roseen, R., Briggs, J., Puls, T., August 2009. Reduction in Inltration
Capacity of Porous Asphalt and a Survey of Cleaning Methods and Their
Effectiveness. StormCon, Anaheim, CA, pp. 8e18.
Candaele, R., Barrett, M., Charbeneau, R., 2008. Porous Friction Course: A Laboratory
Evaluation of Hydraulic Properties. Center for Research in Water Resources
Online Report 08-06. University of Texas at Austin.
Cooley Jr., Allen, 1999. Permeability of Superpave Mixtures: Evaluation of Field
Permeameters NCAT Report No. 99-1.
Dierkes, C., Kuhlmann, L., Kandasamy, J., Angelis, G., 2002. Pollution retention
capability and maintenance of permeable pavements. In: 9th International
Conference on Urban Drainage, Portland, OR, USA.
Ferguson, B.K., 2005. Porous Pavements-porous Pavement Hydrology. CRC Press,
Taylor and Francis Group, Boca Raton, Florida.
Gogula, A., Hossain, M., Romanoschi, S., Fager, G.A., 2003. Correlation between the
laboratory and eld permeability values for the superpave pavements. In:
Proceedings of the 2003 Mid-continent Transportation Research Symposium,
August, Ames, Iowa.
Kayhanian, M., Vichare, A., Rasa, E., Green, P., Alaimo, C., Hwang, H.-M., Signore, J.,
Troxler, M.P., Guada, I., Farshidi, F., Harvey, J., 2008. Water Quality and Toxicity
Evaluation of Discharge Generated from Asphalt Surfacing Materials. Caltrans
Division of Environmental Analysis, Sacramento, CA. Report No. CTSW-RT-08168-20.3.
Maupin Jr., G.W., 2000. Asphalt permeability testing in Virginia. Transportation
Research Record: Journal of the Transportation Research Board 1723, 83e91.
Montes, F., Haselbach, L., 2006. Measuring hydraulic conductivity in pervious
concrete. Environmental Engineering Science 23 (6), 960e969.

123

Nielsen, C.B., 2007. Clogging of Porous Pavements e The Cleaning Experiment. Road
Directorate, Danish Road Institute. www.roaddirectorate.dk.
Ongel, A., Harvey, J., Partl, M.N., 2008. Assessment of clogging in open graded
asphalt mixes in California. In: International Society for Asphalt Pavements
(ISAP) Symposium on Asphalt Pavements and the Environment, August 18e20,
Zurich, Switzerland.
Prowell, B.D., Dudley, M.C., 2002. Evaluation of Measurement Techniques for
Asphalt Pavement Density and Permeability. Transportation Research Record
1789 Paper No. 02-3907.
Schlotzhauer, S.D., Littell, R.C., 1997. SAS System for Elementary Statistical Analysis,
second ed. SAS Institute Inc, Cary, NC.
Siriwardene, N.R., Deletic, A., Fletcher, T.D., 2007. Clogging of stormwater gravel
inltration systems and lters: insights from a laboratory study. Water
Research 41 (7), 1433e1440.
St. John, S.T., Matthias, S., Horner, R.R., 1997. Effect on Road Shoulder Treatments on
Highway Runoff Quality and Quantity. WA-RD-4291. Washington State
Department of Transportation, Olympia, WA.
USEPA, 1999. Stormwater Technology Factsheet on Porous Pavement. Ofce of
Water. http://www.epa.gov/owm/mtb/intdrn.pdf.
Wei, Irvine W, 1986. Installation and Evaluation of Permeable Pavement at Walden
Pond State Reservation, 77e12 and 80e22. Northeastern University Department of Civil Engineering.
Williams, S.G., 2008. Field permeability measurements of coarse-graded asphalt
pavements. In: International Society for Asphalt Pavements (ISAP) Symposium
on Asphalt Pavements and the Environment, August 18e20, Zurich,
Switzerland.
Yong, C.F., Deletic, A., Fletcher T.D., Grace, M.R., 2008. The clogging behavior and
treatment efciency of a range of porous pavements. In: 11th International
Conference on Urban Drainage, Edinburgh, Scotland, UK.