This research article evaluates the performance of permeable pavement subjected to sediment loading over multiple years through laboratory experiments and modeling. The experiments showed the permeable pavement became completely clogged after seven hydrological years. Modeling revealed the ratio of horizontal to vertical hydraulic conductivity is 3.5, and hydraulic conductivity decreased 20% after three years. The permeable pavement was highly effective at removing total suspended solids (100% removal), moderately effective at removing ammonium (59% removal) and phosphate (23% removal), and resulted in a 12% increase in nitrate, likely due to nitrification processes. Proper annual cleaning of permeable pavements can maintain hydraulic performance and pollutant removal over their lifespan.

1. Research article
Evaluation of permeable pavement responses to urban surface runoff
Meysam Kamali a
, Madjid Delkash b
, Massoud Tajrishy a, *
a
Department of Civil and Environmental Engineering, Sharif University of Technology, Tehran, Iran
b
Department of Civil and Environmental Engineering, University of Delaware, Newark, DE, USA
a r t i c l e i n f o
Article history:
Received 25 July 2016
Received in revised form
9 November 2016
Accepted 13 November 2016
Available online 20 November 2016
Keywords:
Permeable pavement
Urban runoff
Clogging
Sediments removal
Nutrient removal
a b s t r a c t
The construction of permeable pavement (PP) in sidewalks of urban areas is an alternative low impact
development (LID) to control stormwater runoff volume and consequently decrease the discharge of
pollutants in receiving water bodies. In this paper, some laboratory experiments were performed to
evaluate the efficiency of a PP subjected to sediment loadings during its life span. Simple infiltration
models were validated by the laboratory experiments to evaluate the trend and extend of PP infiltration
capacity throughout the life of the pavement operation. In addition, performances of the PP in removing
total suspended solids (TSS) and selective nutrient pollutants such as NO
3 ; NHþ
4 and PO3
4 from the
surface runoff have been investigated. Experimental data showed that the PP was completely clogged
after seven hydrological years. The model revealed that the ratio of horizontal to vertical hydraulic
conductivity is 3.5 for this PP. Moreover, it was found that 20% reduction in hydraulic conductivity
occurred after three hydrological years. The PP showed 100%, 23% and 59% efficiencies in sediment
retention (TSS removal), ðPO3
4 Þ, and N NHþ
4 removal during the entire study, respectively. However,
the removal efficiency of ðN NO
3 Þ was 12% and we suspect the increase in effluent ðN NO
3 Þ is due
to the nitrification process in subsurface layers. This study demonstrated that when PPs are annually
cleaned, it is expected that PPs can function hydraulically and be able to remove particulate pollutants
during their life span by a proper maintenance.
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
Urban development increases impermeable surfaces such as
pavements and buildings that diminish water infiltration to the
ground and increase runoff volume (Finkenbine et al., 2000; Nie
et al., 2011). This runoff that was induced by impermeable sur-
faces, washes out pollutants from urban areas and carries them to
waterbodies (Davis et al., 2001). Urban runoff has been known as a
primary pollutant source. About 46% of surface water pollution is
attributed to urban runoff (Chai et al., 2012; USEPA, 1996). There-
fore, controlling urban runoff quantity and quality seems vital to
properly maintain watercourses. Among several practices devel-
oped to obviate the abovementioned issues, the PP is known as a
LID that can mitigate first flush impacts and decrease volume of
runoff as well as treatment costs (Sansalone and Teng, 2005;
Andersen et al., 1999).
Several researches have investigated infiltration rates of PPs in
sidewalks using double ring infiltrometer tests (e.g. Qin et al., 2013;
Valinski and Chandler, 2015); however as far as authors' knowledge
permits, literature lacks studies that cover the hydraulic perfor-
mances of a sidewalk PP under clogging. Therefore, the authors
present the literature that has studied the clogging process of PPs
with the most similarity to our experimental setup. Several in-
vestigations have been performed to evaluate the performance of
PPs for water quality and some representative example studies are
examined here. For instance, TSS contain attached heavy metals,
which remarkably prevent aquacultural growth (Brown et al.,
2009). PPs have revealed acceptable performances for TSS
removal from runoff. Morquecho et al. (2005) showed that a PP can
reduce TSS, turbidity and total phosphorus more than 50%. Tota-
Maharaj and Scholz (2010) reported that TSS, N NHþ
4 and
P PO3
4 removal efficiencies for a PP were 91%, 84.6% and 77.5%,
respectively. In another study, Collins (2007) asserted that a PP
noticeably decreased N NHþ
4 and increased N NO
3 effluent
concentrations compared to the asphalt pavement. Although PPs
have shown acceptable performances in TSS and some other
pollutant removal, the poor PP construction and maintenance lead
to sediments accumulation into the PP structure, which causes
* Corresponding author.
E-mail addresses: meisamkamali_63@yahoo.com (M. Kamali), delkash@udel.
edu (M. Delkash), tajrishy@sharif.edu (M. Tajrishy).
Contents lists available at ScienceDirect
Journal of Environmental Management
journal homepage: www.elsevier.com/locate/jenvman
http://dx.doi.org/10.1016/j.jenvman.2016.11.027
0301-4797/© 2016 Elsevier B.V. All rights reserved.
Journal of Environmental Management 187 (2017) 43e53

2. clogging. This clogging reduces PP popularity when compared to
the other practices (Bean et al., 2007). Several studies have been
investigated to assess clogging behaviors of PPs (Coleri et al., 2013;
Kayhanian et al., 2012a). Types of sediments as well as methods of
applying sediments to PPs vary in literature. Sediments introduced
to a PP might be real samples during urban runoff campaigns or can
be artificially provided (Siriwardene et al., 2007). Further, sedi-
ments can be applied to PPs either manually (Castro et al., 2007;
Coleri et al., 2013), or by precipitation (Pezzaniti et al., 2009);
however, the pavements in sidewalks are frequently cleaned and
majority of the sediments are entered through surface runoff;
therefore, applying the sediments by runoff is a more realistic
experimental setup.
From the literature review and some of the aforementioned
studies, generally the PP performances were evaluated under un-
realistic conditions. For instance, the simulated rainfall and applied
sediments were not in accordant with real sediment loads and
precipitation conditions. Therefore, experimental observations
might differ from field observations. Pezzaniti et al. (2009) assessed
the effective life of a PP in both laboratory and field works. Their
results showed that in the field, clogging happened at a faster rate
compared to the laboratory work. Our study was performed by
conducting experiments through applying clean simulated rainfall
(no suspended solid) from the top of the PP (representing rainfall)
and simulated runoff with sediments that was flown from a side. In
addition to this unique experimental setup, we developed a model
to understand the water flow in this PP structure. This approach
was used to evaluate the temporal and spatial clogging trends in
this PP and hence to find the vulnerability of the PP to sediment
loadings during rainfalls. We defined clogging in terms of the
accumulation of silt within the pavement structures that corre-
sponds to a decrease in the hydraulic conductivity of the PP (Castro
et al., 2007). Castro et al. (2007) related the clogging due to the
reduction in hydraulic conductivity to the entrapments of particles
in the upper surface layer as well as the accumulations of particles
on the geotextile fabric. Recognizing susceptible spots for clogging
in PPs would help to locate the failed locations and to properly
clean it through an optimum maintenance schedule. One other
important feature of our investigation was addressing the pattern
of water flow through the PP as a porous media to determine the
fraction of the total water that enters the PP and horizontally flows
for sizing the drainage system in PPs.
This study aims at examining performances of a PP, used in
sidewalks, during a flow of runoff that contains sediment loads. We
present the clogging steps as a runoff was exposed to the PP using
specific sediment samples that were collected from some streets in
Tehran, Iran. Using our experimental measurements, a simple
model was developed to find out the capacity of the PP, the chro-
nological clogging trend and its age (we defined the age of the PP as
the time before runoff overflows the PP). Although evaluating the
PP clogging by sediments was taken as the primary objective of this
study, the performances of the PP in removing some pollutants
such as TSS, PO3
4 ; NHþ
4 and NO
3 removal were also investigated.
2. Methodology
2.1. Experimental setup
Fig. 1 displays the schematic diagram of the experimental setup
used to evaluate the performance of our PP under different labo-
ratory conditions. Several important features of the experimental
setup are described below.
2.1.1. PP module
The schematic diagram of a PP module is depicted in Fig. 1a. The
experimental area of this PP was about 2 m2
(2 m by 1 m) and
pavement was constructed at 2% slope. The experimental PP
module comprised of three different layers: (1) top layer includes
concrete blocks with 4 cm height and 0.5 cm gaps between the two
adjacent blocks; the gaps were filled with granular gravel size
ranging from 2.36 mm to 4.75 mm, (2) middle layer that consists of
5 cm granular filter, which is comprised of granular gravel in the
range of 2.36e4.75 mm, and (3) the bottom subbase pebble gravel
layer with 12 cm height and granular particle size between 4.75 and
20 mm. In addition, a geotextile fabric was installed in the bottom
of the granular filter (second layer) and subbase layer (third layer).
The top geotextile layers were installed to prevent downward flow
of particles into the subbase layer. The bottom geotextile layer was
added to prevent particle upflow movements from subbase soil into
the subbase layer.
Using the rainfall records from the nearest weather station in
Tehran (Mehrabad weather station, about 2.5 km far from the
studied streets) as well as design method suggested by the Iowa
stormwater management manual for pavement systems (Iowa
Stormwater Management, 2009), we estimated the depth of
reservoir layer to be about 25 cm. According to this manual, curve
number, ratio of impermeable pavement to permeable pavement,
void ratio of aggregate base, and design rainfall event were
assigned 98, 3, 0.4, and 30 mm, respectively. However, due to
limitations in the PP modular dimension and weight, the depth of
12 cm was considered. The 13 cm reduction in depth was mostly
associated with the subbase layer that might be more significant for
PP design with higher traffic speeds as well as load applications and
will be insignificant for clogging investigations in sidewalks. For
example, it has been documented that pavement failure due to
clogging is mostly associated with particles trapped in surface or
upper subsurface of PP pavements (Teng and Sansalone, 2004;
Kayhanian et al., 2012a; Coleri et al., 2013), which is usually inde-
pendent of bottom layer depths. Therefore, the proposed depth in
our experimental setup is justified in accordance with the study
objectives.
2.1.2. Rainfall and sediment simulator setup
A schematic diagram of the rainfall and sediment loading
simulator is shown in Fig. 1b. Different setups were examined to
find the highest uniformity in the rainfall applied to the PP. In order
to achieve the highest uniformity, the number of nozzles and their
heights with respect to the surface of the PP were changed. The best
experimental setup, whose uniformity was high, had a 2 m height
from the PP surface and 3 nozzles in one row. With some limita-
tions in our experimental set up, we calibrated the intensity of the
rainfall simulator and found to be 36 mm/h, which is fairly close to
the average rainfall intensity in Tehran (Tehran's mean annual
rainfall ¼ 240 mm with majority of rainfall events last about 6 h and
therefore the average hourly rainfall intensity is equal to 40 [¼240/
6] mm/hr). In order to meet the annual rainfall, the system worked
for 6.67 h (¼240/36) in each hydrological year. The rainfall simu-
lator was operated with clean tap water and applied on top of the
PP and let the water flow downward through different pavement
layers.
In addition to applying clean rainfall from top, an artificial sur-
face runoff was also introduced to the PP from one side (see Fig.1b).
This type of PP is planned to be implemented in pedestrians of
Tehran. It will convert impermeable surfaces, which have three
times greater surface area than permeable surfaces, to permeable
pavements. Thus the flowrate of runoff was three times larger than
rainfall intensity to be consistent with real conditions. This applied
runoff containing sediments in some of the experiments allowed us
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53
44

3. to investigate the impact of particles in surface and subsurface
clogging.
2.1.3. Piezometers installation and hydraulic head measurements
As mentioned before, this study defines clogging in terms of a
decrease in hydraulic conductivity values of the PP due to the
accumulation of silt within the PP structure. On the other hand, a
decrease in hydraulic conductivity of the surface corresponds to a
drop in the piezometric head in that spot. Therefore, any unusual
drop in piezometric head in the PP indicates clogging in that spot.
These drops were utilized to confirm the results of our model about
clogging spots and times. In this study, piezometric heads were
measured by 20 piezometers that were installed in different loca-
tions of the PP. We assigned five piezometers to one segment of the
PP to have a good distribution of piezometric heads. Therefore, the
PP was divided into four segments that can contribute in the total
PP drainage. The numbers shown in the arrows in Fig.1b display the
areas that are swept by each segment. The discharge in these seg-
ments were measured by dividing the water level rise in a marked
bucket into the time intervals. These discharge values were
employed for model validation as well as prediction of the location
and time of clogging occurrences. The approach that these mea-
surements were used in the model is explained in the latter section.
Effluent water discharge from the first to fourth segments shows
the “drainage path” in the PP. This path is defined as the length
between the point that surface flow enters the PP and where it
leaves the PP. An increase in effluent water discharge in farther
segments is accompanied with an increase in drainage path and
clogged surfaces (Castro et al., 2007).
2.2. Formulation of the runoff transport in the PP
Schematic diagram for experimental setup related to the
formulation of runoff transport over and through PP module is
shown in Fig. 2. As shown, the PP module was divided into four
segments. Water discharge ðqijÞ was measured for each of these
segments (j index) at different times (i index). Water mass balance
was taken into account for each segment. The applied runoff can
infiltrate into each segment of the PP or keep flowing as surface
runoff and infiltrate into the next segment (based on infiltration
rate and hydraulic conductivity values). Water inside the segment
can discharge from that segment (and the system) downward or
drain in the adjacent segment horizontally according to the hy-
draulic conductivity ratio
Kh
Kv
. This horizontal flow is expressed by
(eq. (1)). Mass balance considers a steady precipitation (P)
distributed uniformly on the PP as well as applied influent runoff,
horizontal water drainage ðdijÞ from each segment, and effluent
discharge ðqijÞ from that. The mass balance relationship for each
segment is shown in (eq. (2)). A quasi-steady state condition was
assumed here for infiltration meaning that infiltration rate is con-
stant between two consecutive measurements.
dij ¼
Kh
Kv
*qij (1)
Iij þ diðj1Þ ¼ dij þ qij (2)
where Iij infiltration rate (m/s) at time (i) into segment (j), diðj1Þ
Joints
Permeable block paved layer
Filter layer (Ds=2.36-4.75 mm)
Geotextile
Subbase layer (Ds= 4.75 - 20 mm)
Geotextile
40 mm
50 mm
120 mm
(a)
Artificial runoff (Tap
water with sediments)
(diameter150 μm))
First
segment
Second
segment
Third
segment
Fourth
segment
600 cm 320cm 400 cm 600 cm
Bypass
Rainfall simulator
Permeable pavement
Artificial runoff
P
P
Rainfall tank
F
F
Tap water
Tap water
(b)
Fig. 1. (a) PP module with specified layers, and (b) schematic diagram of experimental setup.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53 45

4. horizontal water drainage (m/s) at time (i) from segment (j-1) into
segment (j), qij water exiting discharge (m/s) from segment (j) at
time (i).
Porous media infiltration is dependent of hydraulic conductivity
at a certain soil saturation ðKqÞ and a hydraulic head gradient
DH
L
as expressed by (eq. (3)):
I ¼ Kq
DH
L
(3)
where L is the total height of a porous media (m), DH is the dif-
ference of total head between up and downstream of fluid into the
porous media (m). Total head in the porous media is made of matric
potential (h) and elevation height (z): (H¼h þ z).
To study the PP responses to urban runoff and to evaluate the
clogging trends, a modeling approach was used to estimate hy-
draulic conductivity. In this paper, it was assumed that PP rapidly
meets the saturated condition. Main reason for this assumption was
that the major portion of total depth (21 cm) of the PP was domi-
nated with granular and pebble gravels, so matric potential is
neglected for these large particles ðDðhÞ DðzÞÞ. Hence, it is ex-
pected that the hydraulic conductivity immediately reaches its
saturated values ðKqzKsatÞ. In addition, it was assumed that
ponding height ðHpondingÞ on the PP is much smaller than the PP
column (L) ðz ¼ Hponding þ L; Hponding≪LÞ. Thus, the infiltration
rate (I) value was estimated by the saturated hydraulic conductivity
ðKsatÞ that is shown in (eq. (4)).
I ¼ Kq
DH
L
¼ Kq
Dðh þ zÞ
L
z Ksat
Hponding
L
þ
L
L
z Ksat
(4)
Whenever overflow (runoff and precipitation) is greater
than Ksat , infiltration rate shown in (eq. (5)) equals Ksat and under
the condition that overflow is lower than Ksat the infiltration rate is
equal to the overflow rate.
Iij ¼
Ksat Ksat Rij þ P
Rij þ P Ksat Rij þ P
(5)
where Iij is the infiltration rate (m/s) at time i and segment j, Rij
runoff overflowing on the PP (m/s) and P the precipitation applied
on the PP (m/s).
2.3. Laboratory experiments
Three different experimental sets were performed under labo-
ratory conditions. The first experiment was implemented to
calculate the hydraulic features of the PP as a basis to evaluate
clogging. Then the obtained hydraulic conductivity ratio will be
validated by another experiment. The second experimental set was
executed under sediment loadings and particles entrapments in
void spaces, which cause clogging. The third experiment was car-
ried out to evaluate the performance of the PP in removal of some
selective pollutants, which are concern in urban road surface
runoff.
2.3.1. Experiment set 1: performance evaluation of PP based on
hydraulic response (No sediment)
Since there is no water storage and clogging in the first exper-
iment, this experiment was performed to obtain the ratio of the
horizontal to vertical hydraulic conductivities
Kh
Kv
of the PP. This
ratio plays an important role in the horizontal water transport in
porous mediums. It is estimated to be around 10 for natural soils
that have clay and other cohesive materials. However, the PP used
in this study was primarily made of granular and pebble gravels and
hence, it is expected that this ratio would be smaller than normal
soils (less horizontal movements due to less matric suctions). To
find the value of
Kh
Kv
in this PP, rainfall with the intensity of
36 mm/h and runoff (without sediment) with the runoff rate of
108 mm/h were applied to the PP. Piezometric heads in the storage
layer were measured by piezometers to monitor the level of PP
saturation. Total volume of water fed into the PP was 298 L during
three-hour rainfall and runoff application.
An additional experiment was run during this experimental set
to assess the accuracy of predicted overflow under higher flow
intensity. Water would recess from the saturated segments during
this experiment. This overflow would occur because of the larger
runoff and rainfall rates than infiltration rate. Thus, it was assumed
that water storage was equal to the difference between inflow and
outflow in each segment. The mass balance for this condition is
expressed by (eq. (6)). Water would recess from each segment to
the surface when accumulation volume, VACij
(m), was greater than
segment capacity. Water accumulation is the multiplication of
accumulation rate and time intervals (eq. (7)). Segment capacity is
Fig. 2. Scheme of PP module with variables used to develop a model to formulate runoff transport in the PP (Note: P ¼ simulated rainfall precipitation, Rj ¼ runoff rate over jth
segment, Ij ¼ infiltration rate into the jth segment, dj ¼ horizontal drainage rate into jth segment, qj ¼ effluent discharge rate from jth segment).
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53
46

5. the total void volume in each segment wherein water can occupy. It
is worthy to note that all the calculations were done in unit of area.
Sij ¼ Iij þ diðj1Þ dij qij ¼ Iij þ diðj1Þ
Kh
Kv
þ 1
qij (6)
VACij
¼ Sij*DðtÞ (7)
During this experiment, 62 mm/h uniform and steady precipi-
tation was applied to the PP. This applied precipitation is about 70%
times greater than the first experiment to study the PP capacity and
overflow. As compared with the first experiment, the runoff rate
applied to the first segment during this experiment was three times
greater than precipitation (3*62¼192 mm/h). In addition, the
application of source water (precipitation and runoff) on the PP
lasted one hour and then the PP was left to drain the applied runoff
and precipitation.
2.3.2. Experimental set 2: performance evaluation of PP for clogging
under sediment loading
The aim of the second experimental set was to get the age of the
PP before having any overflow bypass. Infiltration of runoff that
contains sediments leads to PP clogging. This clogging depends on
PP hydraulic conductivity, sediment and runoff loadings.
An artificial runoff was prepared by addition of dry sediments
collected from the streets to the tap water in a mixing tank depicted
in Fig. 1b. This mixing tank was used to keep the TSS concentration
constant during the experiments. Based on a study carried out by
Kamali et al. (2012), the median concentration of TSS was about
400 mg/L in urban runoff of Tehran during nine field measurements
from different surfaces including urban stream, rusted iron roof,
galvanized iron roofs, asphalt street and storm channels. Therefore,
runoff with average TSS concentration of 400 mg/L was exposed to
the PP. The runoff was transported to the surface of the PP by a
pump. A float ball used to fix the water elevation in tank. During
6.67 h, steady-state precipitation rate of 36 mm/h and runoff
108 mm/h were applied to the PP during each hydrological year.
Generally, there was a day gap between each hydrological simula-
tion to make the needed TSS for the next hydrological year.
Dry sediments were swept up from the edges of streets with
low, medium and high traffic volumes. Average of annual daily
traffic of these streets is about 1200. After sampling, they were
carried to the laboratory to be washed and granulated, respectively,
and finally were dried in an oven. The sieving results of these
sediments revealed that sediment sizes in streets with medium and
low traffic volumes are somehow similar. One representative
example of a sample sediment size distribution is displayed in
Fig. S1 (supplementary document). As shown, about 50% of sedi-
ment size was smaller than 150 mm. For this reason and the fact that
particle less than 150 mm is more responsible for clogging in PPs
(Coleri et al., 2013), these particle size ranges were used during our
experiments.
2.3.3. Experiment set 3: performance evaluation of PP for pollutants
removal
This part of the experiment aims to determine pollution removal
efficiencies of the PP and detect influences of drainage path length
in nutrient removal efficiencies. Four different water quality pa-
rameters were taken into account to assess PP performances
including: TSS, PO3
4 , NHþ
4 and NO
3 . The pollutants were selected
because an elevated concentration of these constituents were
measured in storm drain channels; especially when mixed with
wastewater streams (Kamali et al., 2012). Discharge of storm drains
containing these pollutants into surface water bodies causes
eutrophication and reduces the overall quality of water.
To carry out these experiments, the 150 lit tank was filled with
an urban stream with base flow around the studied streets. The TSS,
PO3
4 , N NHþ
4 and N NO
3 concentrations in the tank were about
720, 3, 3.5 and 2.2 mg/L. For a better control of influent and effluent
rates and path of pollutants, the direct rainfall is removed from the
setup. It should be noted that the PP surface had been fully cleaned
before this stage was begun. Pollutants had a higher influent con-
centration at the beginning of this experiment to consider the first
flush phenomena. To achieve this goal, tap water was added to the
tank as the polluted water was conveyed to the PP. This substitution
makes the pollutants diluted and keeps the water at a constant
flowrate (108 mm/h). Runoff flew for 240 min and the measure-
ments went on 30 min after the runoff stoppage. Seven samples
were harvested from influent runoff and effluent discharge of each
segment. A spectrophotometer device was utilized to measure the
concentrations of the pollutants. The concentration differences
between influent and effluent from the segments indicated the role
of PP in the nutrients removal efficiency.
3. Results and discussion
3.1. Performance results based on hydraulic behavior
The results of this experiments are categorized into two
different parts: (1) finding the ratio of hydraulic conductivities and
(2) validating the model by predicting the runoff overflow under
more intense rainfall and runoff rates. This validated model is used
to better understand the governing mechanisms and evaluate the
long-term hydraulic performance of the pavement structure.
3.1.1. Ratio of hydraulic conductivities
The results of exiting discharge for four different segments
during the experiment set 1 are tabulated in Table S1 (supple-
mentary document). As shown, higher drainages occurred in the
farther segments to the influent and the first segment underwent
the least water drainage, which can be explained by slope of the
experimental setup (2%) that enforces the water to move down-
stream. This observation is consistent with literature (e.g. Castro
et al., 2007). Castro et al. (2007) investigated the role of PP's
slope on the water discharge and found that the slope affects the
drainage path. They reported that although over 70% of total water
was drained from the first half of the PP under zero slope condition,
the water drainage path shifted to the middle and end of the PP
when the slope gently increased to 2%.
The water influent into the PP in 180 min was 898.2 lit and the
amount of effluent was 821 lit. Afterward, 7.15 lit water exited from
the PP in 24 h. Totally, 8% difference was found between water fed
into the PP and the outlet volume. This difference between the
volumes of water in the influent and effluent has a descending
trend. We associated this observation with 1) grain affinity in
adsorbing water, 2) errors in measurements, 3) incomplete infil-
tration of the rainfall into the PP (e.g. forming ponds on the PP), and
4) water retention in the PP due to the slope, which brings about
incomplete discharge.
Insignificant storage in all the segments was assumed during
this experiment, so water would only be discharged out or drained
to the adjacent segment in
Kh
Kv
ratio over vertical discharge. A
global
Kh
Kv
value was searched for the PP, which means that this
ratio was considered to be for the whole of the PP during the
experimental steps. The least square error between the observed
discharge ðqiÞ and predicted values was used to get the most
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53 47

6. representative value
Kh
Kv
. The representative ratio for
Kh
Kv
ob-
tained was 3.5, which is reasonable for a porous media with large
void ratio, less matric suction, and lower horizontal water
movement.
3.1.2. Evaluating the PP capacity
As discussed in the methodology section, the applied rainfall
and runoff intensities were 62 and 186 mm/h, respectively. The PP
was saturated 26 min after experimental initiation and a pond was
formed in the outlet. A higher discharge rate was seen for the first
hour when we had both runoff and rainfall. A decreasing trend in
water drainage after the first hour is related to stopping water flow
entrance the PP. The third and fourth segments made the highest
contributions in discharge, which was related to the horizontal
water movement in the porous structure. In this experiment, the
relative difference between the total water that came into the PP
module and exited was about 1.4%, which infers that water was not
noticeably stored in the PP.
Knowing the
Kh
Kv
, which is porous medium specific, helps us
model more experimental setups. PP performance under high
rainfall intensity and drainage prediction would be another inter-
esting point about the PP. Applying higher rainfall and runoff rates
than PP capacity would lead to a runoff overflow. Observation of
runoff that is overflowing from the fourth segment infers that the
PP failed at controlling runoff quantity. The data of this experi-
mental setup are displayed in Table S2.
The results presented in Table S2 reveal that water was more
discharged from the third and fourth segments compared to the
first and second one. The authors believe that this distribution of
water discharge highly depends on PP slope, PP particle size dis-
tributions and water flow rates. As water influent stops (after one
hour), discharged volume was decreased in all the segments except
the first one. We related this observation to water backward
movement from the farther segments to the first one due to un-
balance water distribution in segments after stoppage of water
influent.
Predicted discharges (qijÞ for each segment at different times are
plotted vs. observed discharges in Fig. 3. Predicted water discharge
from each segment was compared to the measured value to assess
the accuracy of
Kh
Kv
. The results unveil an acceptable prediction of
this ratio for the studied PP. Total measured and predicted overflow
are 7.7*105 and 3.6*105
m/s, respectively. It can be seen that they
are in order of magnitude; however, this difference might be
associated with the quasi-steady state assumption, complicated
behaviors of water during recession and averaging long intervals of
measurements.
3.2. PP performance for clogging under sediment loading
Evaluating the PP clogging trends under sediment loading was
the primary objective of this stage of the experiments. The water
discharge of these four segments and runoff coefficient during all
seven hydrological years are displayed in Fig. 4. The PP had a free
drainage condition that was governed in the first two years. The
drainages prevented releasing of water in the first hour of each
other hydrological year. As shown in Fig. 4, runoff overflow was not
observed in the first five years. The runoff coefficient in the sixth
year was 15%. During the seventh period, the runoff coefficient
increased from 15% to 35%. During all these hydrological years,
coarse sediments infilled the gaps between the blocks, which in-
crease the runoff coefficient (Pezzaniti et al., 2009). Castro et al.
(2007) compared infiltration rate and runoff percent between
two conditions (no clogging, fully clogging) in 2% slope. They found
that the filter water volume decreased from 98.45% to 70.98% and
runoff percentage increased from 1.6% to 29%, which is consistent
with our observation. During all seven years of our experiment,
water discharge from segments 1 and 2 decreased from 48.4% to
10.5%, and water discharge from segments 3, 4 and the weir
increased from 51.6% to 89.5%.
The hydraulic heads recorded by piezometers for the experi-
ments since the fourth until seventh hydrological years are shown
Fig. 3. Predicted and observed discharge for each segment at each measurement time.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53
48

7. in Fig. 5. As seen, there is a noticeable clogging in the PP in a dis-
tance about 80 cm from the zone where runoff was applied during
the fourth and fifth hydrological years. However, the observations
are different in last two hydrological years. During these years, a
sharp drop occurred in the third segment. The clogging in these two
years occurred in a distance about 130 cm from the beginning of the
PP, which is located in the third segment. It can be seen that the
drop was shifted from the second to the third segment during the
last three years. According to Fig. 5, water discharge from the first
and second segments decreased from 22% to 10.5%, and from the
third, fourth segments and weir increased from 78% to 89.5%.
Pezzaniti et al. (2009) reported that these changes are associated
with 1) accumulations of fine sediments on the geotextile layers,
and 2) coarse sediments on PP surface that bring about an increase
in drainage passage from the fourth to the seventh year.
Eqn. (3) was taken into account as the governing equation to
obtain the temporal and spatial clogging trends. Although the
maximum infiltration rate of the PP might not be affected by par-
ticles trapped in the initial years (because of higher hydraulic
conductivity compared to the runoff and precipitation rates),
infiltration rate should be decreased as sediments are introduced to
the PP. A cake was formed on top of the pavement as the
experiments went on. It was related to sediments trapped in the
particles of the PP at the surface, which is concurrent with former
studies (James, 2004; Teng and Sansalone, 2004; Kayhanian et al.,
2012a). James (2004) and Pratt et al. (1995) reported that the ma-
jority of TSS usually accumulate on a very thin layer on top of the PP.
Teng and Sansalone (2004) called this thin layer “schmutzdecke”
cake and found that this layer forms when the ratio of sediment
diameter to the medium diameter is less than 10. Here, since the
ratio of the medium diameter was about 2.7 mm and the sediment
diameter was about 150e300 mm, sediments were strained in this
cake on top of our PP. In another study, Kayhanian et al. (2012a)
utilized scanned image analysis and found that top 2.5 cm of a PP
had noticeable smaller porosities compared to the bottom of the PP.
Further, no water receded from inside the PP into the surface during
this experimental stage. It means that the reason for runoff effluent
after a few years is surface clogging and capacity of the PP is still
higher than water that enters the PP. Thus, a thin cake layer (1 cm)
was assumed to be formed on top of the PP. Further, since water can
be drained easily, no water storage inside the PP was assumed
within each two consecutive measurements. It means that as water
comes into each segment, it discharges out of the system or travels
to the next segment. This quasi-steady state assumption eases the
0
10
20
30
40
50
60
1 2 3 4 5 6 7
Outflow
percentage
(%)
Hydrological year
segment 1 segment 2 segment 3 segment 4 runoff
Fig. 4. Comparison between the percent output water volume water effluent percentage from the drainages and weir in seven hydrological periods during the experiments with
clean runoff.
0
20
40
60
80
100
120
140
160
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Pizeometeric
head
(mm)
distance from upsteam of the PP (mm)
Forth Simulation Period Fifth Simulation Period
Sixth Simulation Period seventh Simulation Period
Fourth Simulation Period
First segment Second segment Third segment Fourth segment
Fig. 5. Temporal and spatial variations in piezometric heads.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53 49

8. modeling of water flow in the PP.
The infiltration model was used to simulate the temporal and
spatial clogging trends. Hydraulic conductivity variations during
seven hydrological years are shown in Fig. S2. It can be seen that the
first and fourth segments did experience insignificant variations,
while the second and the third segments got clogged (about 12%),
which complies with literature. During an experiment with
200 mg/L sediment, a formpave showed 10% decline in hydruclic
conductivity after 5 hydrological years (Valkman, 1999). Based on
the results presented in Fig. S2, the second segment was clogged
since the third hydrological year, which is somehow consistent
with piezometric observations (Fig. 4). Furthermore, it was found
that the third segment had clogging since the sixth hydrological
year, which is concurrent with the piezometric measurements.
Comparing our observations with other studies (Castro et al.,
2007; McKenzie et al., 2008; Pezzaniti et al., 2009) reveals that
the clogging of the PP highly depends on sediments loading, the
way of applying sediments to the PP, PP hydraulic conductivity, and
particle size distribution. We learnt that sediment loadings have a
big influence on age of the PP. Pezzaniti et al. (2009) reported that
their PP age was about 35 years in experimental setup; however,
our PP clogged after 7 years. They imported sediments with a rate
32.62 gr/yr to the PP, while we applied 583.2 gr/yr of sediments to
the PP. Therefore, our sediment load was about 18 times greater
than the load in Pezzaniti et al. (2009).
Particle size distribution plays an important role in PP clogging.
Generally, the range of urban runoff particle size has been reported
in great variety. For instance, just 3% of the dry solids in runoff were
smaller than 50 mm (Li et al., 2005). In another study, Sansalone and
Tribouillard (1999) reported that the fraction of particles less than
50 mm in a PP was 10%, while in another research, majority of
particles on a PP were less than 50 mm (McKenzie et al., 2008).
According to Stoke's law, this variety in particle size distribution
would lead to different settling velocities. Sphere shapes were
assumed for the particles to calculate the horizontal distance that
the particles pass to settle down; however, small particles might
have non-sphere shapes, which makes them more suspended than
sphere shapes. The results of Stoke's law reveal that if the second
segment in the PP (50e100 cm) was noticeably clogged, the ma-
jority of sediments should have particle sizes less than 7 mm. In an
investigation about road dust and stormwater sampling, McKenzie
et al. (2008) found that the majority of particles are smaller than
7 mm, which confirms the results found here.
The way of importing sediments to the PP is one of important
elements in evaluating PP performances. In this study, sediments
were imported to the PP from one side and clogging occurred from
upstream of the water flow, which is called “clogging front”. In this
condition, upstream of the PP is initially clogged and then clogging
extends to downstream regions. Although what Pezzaniti et al.
(2009) found in their field work accords well with our observa-
tion, they did not observe this process in their experimental effort.
Moreover, their results showed that clogging occurs in a faster rate
in the field compared to the laboratory experiment. Therefore, this
comparison infers that our experimental setup is closer to what
happens in reality (fields).
Another key element in the PP performance is hydraulic con-
ductivity, which causes easier flow in PPs. The PP utilized in
Pezzaniti et al. (2009) had a hydraulic conductivity more than two
orders (4.7*102 m/s) of magnitude greater than our PP
(2.1*104
m/s). This difference can somehow explain the higher
duration of their PP. However, a hydrapave constructed in Australia
with a hydraulic conductivity 6.5*103
m/s (one order of magnitude
greater than ours) had about 10 year as effective life (Yong et al.,
2008), which is closer to the age of our PP.
3.3. Performance evaluation of PP in pollutants removal
As mentioned before, if PPs are properly designed, constructed,
and maintained, they can be a good alternative method to manage
stormwater runoff in urban areas. Beside the runoff volume
reduction, PPs are capable to remove most of the particulate and
some dissolved pollutants in the urban runoff. The performance of
the PP for the removal of selected pollutants investigated in our
study (namely, TSS, NHþ
4 , NO
3 and PO3
4 ) are discussed below.
3.3.1. Total suspended solids (TSS)
Total suspended solids are an important water quality param-
eter that is commonly used to design and evaluate the performance
of best management practices or sustainable urban drainage sys-
tems. Although TSS intuitively are not a pollutant of concern, their
measurement could be used as a surrogate parameter to evaluate
particle bound organic and inorganic pollutants in runoff. TSS could
also impact the turbidity of runoff and influence the quality of
receiving waters. The concentration of TSS in urban runoff may vary
between 100 and 3000 mg/lit. Generally, the size of particles found
in runoff influences the measurement of TSS in laboratory and
settling or their removal in BMPs (Kayhanian et al., 2012b).
Permeable pavements are generally considered a good LID to
remove TSS or particle bound pollutants with an efficiency varying
between 72% and 100% (Pezzaniti et al., 2009; Yong et al., 2008;
Rowe et al., 2009; Brown et al., 2009; Tota-Maharaj and Scholz,
2010). A complete removal (100%) of TSS with our PP module was
accomplished during 180 min. Although TSS are captured
throughout the PP depth, the geotextile layer improved the effi-
ciency of removal TSS in this experiment. In a study, a geotextile
layer between gravel layers increased efficiency of TSS removal by
about 30% (Rowe et al., 2009). Although PPs have great perfor-
mances in TSS removal from urban runoff, this TSS capturing
hardens water flow in PPs. These TSS trapped in structures of PPs
occupy pore spaces that leads to a decrease in porosity of PPs and as
a result, a decrease in hydraulic conductivity of PPs.
3.3.2. Orthophosphate ðPO3
4 Þ
Fig. 6 displays effluent PO4
3
concentrations from the first and
fourth segments. The PP removed 32% and 35% of PO3
4 from the
first and fourth segments in the first 30 min, respectively. The total
removal efficiency was 9% and 38% for the first and fourth seg-
ments, respectively. This better PO3
4 removal seen at initial time of
the experiment can be explained by chemical adsorption. Lower
effluent PO3
4 concentrations in the fourth (last) segment than the
first segment implies time dependency of PO3
4 removal from this
porous medium. It infers that as the travel time lengthens, PO3
4
removal increases. The results obtained from our study are com-
parable with other studies. For instance, the PO3
4 removal effi-
ciencies reported by Tota-Maharaj and Scholz (2010) and Collins
(2007), were about 78% and 58%, respectively. The lower removal
efficiencies of PO3
4 in our study may be due to minimum adsorp-
tion capacity by granular gravel (and aggregates), and lack of
insufficient nutrient and time for biogeochemical processes
including biological activities within the short depth of PP depth.
3.3.3. Ammonia nitrogen (N NHþ
4 Þ
Fig. 7 compares N NHþ
4 effluent concentrations from the first
and fourth segments. As shown, the concentration of N NHþ
4
increased in the effluent of the fourth segment is greater than the
first segment between 70 and 170 min. This event was coincident
with the runoff overflow of the PP. However, the concentration in
this segment dropped compared to the first one after this period.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53
50

9. We relate less N NHþ
4 concentrations in the fourth segment
compared to the first segment after 170 min to thriving nitrification
conditions that convert N NHþ
4 to N NO
3 . The removal effi-
ciencies of N NHþ
4 during early part of the experiment (e.g.,
30 min) were 68% and 79% in the first and the fourth segments,
respectively. However, removal efficiencies decreased to 57% and
60% in the first and fourth segments in total of the experiment time
period. We believe that oxygen is being consumed as runoff moves
through the PP (increasing degree of saturation) and this low level
of oxygen could prevent nitrification process that is required for the
oxidation of available N NHþ
4 . Comparing our results with liter-
ature [e.g. Tota-Maharaj and Scholz (2010)], the lower N NHþ
4
removal in our PP could be related to lower retention times and less
available nutrients to support effective biological conversion in our
experiment.
3.3.4. Nitrate nitrogen (NO
3 Þ
Fig. 8 compares effluent N NO
3 concentrations of the first and
fourth segments. Comparing Fig. 8 and 11 reveals that N NO
3
concentrations are lower in the fourth segment compared to the
first segment between 70 and 220 min, which is coincident with
higher N NHþ
4 concentrations in the fourth segment than the first
one. This negative correlation between N NHþ
4 and N NO
3
clearly represents the importance of the nitrification process. After
this period, we believe that nitrification process was evolved and
N NO
3 concentration is higher in the fourth segment. During our
measurements, N NO
3 removal efficiencies were 3% and 15% for
the first and fourth segments at first 30 min, respectively. Con-
centrations of N NO
3 in the effluent were usually lower in the
fourth segment than the first segment. In total, this removal effi-
ciency decreased to 17% and 6% for the first and fourth seg-
ments, respectively. It can be associated with the higher N NO
3
production rate in the nitrification than its removal in the denitri-
fication process. This low N NO
3 removal efficiency is consistent
with the findings in literature. Tota-Maharaj and Scholz (2010) and
Collins (2007) reported that removal efficiencies for N NO
3 in PPs
are low.
3.4. Limitations and uncertainty
Although this study reveals some interesting points about the
behaviors of PPs, there are some uncertainties and limitations in
this investigation that should be pointed out to be considered in
future studies. These limitations originated from our experimental
setup, measurements, and model assumptions include:
Limitation in the size of the PP because of module weight and
laboratory limitations.
0
0.5
1
1.5
2
2.5
3
3.5
0 30 60 90 120 150 180 210 240 270
Concentration
(mg/L)
Time (min)
influent concentration
effluent concentration of the first segment
effluent concentration of the fourth segment
Fig. 6. Performance of PP in Orthophosphate removal. Time in X-axis begins as the water discharge was seen in each segment.
0
1.5
3
4.5
0 30 60 90 120 150 180 210 240 270
concentration
(mg/l)
Time (min)
influent concentration
effluent concentration of the first segment
effluent concentration of the fourth segment
Fig. 7. Performance of PP in ammonia removal.[Note: Time in X-axis begins as the water discharge was seen in each segment].
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53 51

10. Systematic errors in the measurements and human errors that
are inevitable during our measurements. For instance, the water
might not be completely discharged from the PP due to the
slope.
Some assumptions were made during model development that
might be violated under some conditions. For instance, although
water was not significantly accumulated in the PP, a small
fraction might be stored in the PP. Other assumption is quasi
steady-state condition. Since one measurement was carried out
during a long period (e.g. 30 min for the first experiment), this
assumption was inevitable. However, water discharge might
alter within each measurement time interval. Utilizing a data
logger, which obtains data in a shorter time interval, can
improve the accuracy of the model.
Although our experimental setup has high similarity to actual
conditions, we believe that PPs should have better performances
compared to our experiment for some reasons. We loaded one-
year rainfall to the PP during only 6.7 h, which puts the PP under
more stress compared to the actual conditions. Some exterior
parameters such as sunlight can dry the sediments on the PP
and makes a hard cluster on it, which improves the PP perfor-
mance (Brown et al., 2009).
Although we evaluated the performance of the PP for one
greater rainfall, the sensitivity of the PP performance to water
and sediment loadings are of interest in this regard. The authors
suggest more investigations about the sensitivity of the PP to
these loadings.
4. Conclusions
This study investigated the performance of a PP under sediment
loadings during its life span. Main purpose of this study was to
evaluate the temporal and spatial clogging trends of this PP and
find the vulnerability of the PP to sediment loadings during rain-
falls. Afterward, simple infiltration models were validated by the
laboratory experiments to evaluate the trend and extend of the PP
infiltration capacity throughout the life of the operation. At the end,
performances of the PP in removing TSS and selective nutrient
pollutants such as NO
3 ; NHþ
4 and PO3
4 from the surface runoff were
investigated. The conclusions drawn from this study are:
The Kh
Kv
ratio was determined to be about 3.5, which is much
smaller than the Kh
Kv
ratio for normal soils. This low value for Kh
Kv
ratio
for the PP system was related to low matric suction that mostly due
the larger aggregate and sand size.
The experimental results with applied sediment loads showed
that the PP failure occurred in the seventh hydrological year. Runoff
overflow was not observed during the beginning five hydrological
years with zero runoff coefficient. The runoff coefficient in the sixth
year increased to 15%. During the seventh period, the runoff coef-
ficient increased from 15% to 35%. Coarse sediments infilled the
gaps between pavements blocks during all these years, which
increased the runoff coefficient.
The experimental results were further used in a simple infil-
tration model to simulate the clogging effect. Both experimental
observations and simulation performance revealed that the mid-
section (e.g., 9 cm from the pavement surface) of the PP were
clogged during the hydrological years.
The performance of the PP highly depends on the local condi-
tions including particle size distribution, sediments loadings, and
the way of importing sediments to the PP. It means that each region
should have specific experiments to evaluate the performances of
the PP instead of comparing the results of different PP under
different conditions.
The experimental results also showed that PPs could completely
remove TSS; however, medium removal efficiencies were measured
for N NHþ
4 and PO3
4 , while the removal efficiency for N NO
3
was negative.
We suggest further researches about influences of the rainfall
intensity, slope, particle size distribution of sediments, block di-
mensions of pavements, and space between blocks on hydraulic
conductivity as well as lifespan of PPs. These investigations can
evaluate these influences on performance of PPs in pollutants
removal. Moreover, they can help develop more accurate infiltra-
tion models.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jenvman.2016.11.027.
References
Andersen, C., Foster, I., Pratt, C., 1999. The role of urban surfaces (permeable
pavements) in regulating drainage and evaporation: development of a labora-
tory simulation experiment. Hydrol. Process 13 (4), 597e609. http://dx.doi.org/
10.1002/(SICI)1099-1085(199903)13:4597::AID-HYP7563.0.CO;2-Q.
Bean, E.Z., Hunt, W.F., Bidelspach, D.A., 2007. Field survey of permeable pavement
surface infiltration rates. J. Irrig. Drain. Eng. 133 (3), 249e255. http://dx.doi.org/
10.1061/(ASCE)0733-9437(2007)133:3(249).
0
1
2
3
0 30 60 90 120 150 180 210 240 270
Concentration
(mg/l)
Time (min)
influent concentration
effluent concentration of the first segment
effluent concentration of the fourth segment
Fig. 8. PP performance inN NO
3 removal. Time in X-axis begins as the water discharge was seen in each segment.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53
52

11. Brown, C., Chu, A., Van Duin, B., Valeo, C., 2009. Characteristics of sediment removal
in two types of permeable pavement. Water Qual. Res. J. Can. 44 (1), 59e70.
https://www.cawq.ca/journal/temp/article/413.pdf.
Castro, D., Gonz
alez-Angullo, N., Rodríguez, J., Calzada, M., 2007. The influence of
paving-block shape on the infiltration capacity of permeable paving. Land
Contam. Reclam 15 (3), 335e344. http://dx.doi.org/10.2462/09670513.855.
Chai, L., Kayhanian, M., Givens, B., Harvey, J.T., Jones, D., 2012. Hydraulic perfor-
mance of fully permeable highway shoulder for storm water runoff manage-
ment. J. Environ. Eng. 138 (7), 711e722. http://dx.doi.org/10.1061/(ASCE)
EE.1943-7870.0000523.
Coleri, E., Kayhanian, M., Harvey, J.T., Yang, K., Boone, J.M., 2013. Clogging evaluation
of open graded friction course pavements tested under rainfall and heavy
vehicle simulators. J. Environ. Manag. 129, 164e172. http://dx.doi.org/10.1016/
j.jenvman.2013.07.005.
Collins, K.A., 2007. A Field Evaluation of Four Types of Permeable Pavement with
Respect to Water Quality Improvement and Flood Control. Master thesis. Uni-
versity of North Carolina. http://www.bae.ncsu.edu/info/permeable-pavement/
CollinsThesis.pdf.
Davis, A.P., Shokouhian, M., Ni, S., 2001. Loading estimates of lead, copper, cadmium,
and zinc in urban runoff from specific sources. Chemosphere 44 (5), 997e1009.
http://dx.doi.org/10.1016/S0045-6535(00)00561-0.
Finkenbine, J., Atwater, J., Mavinic, D., 2000. Stream health after urbanization. Am.
Water Resour. Assoc. 36 (5), 1149e1160. http://dx.doi.org/10.1111/j.1752-
1688.2000.tb05717.x.
Iowa Stormwater Management Manual, 2009. Version 3.
James, W., 2004. Clogging of permeable concrete block pavement by street par-
ticulates and rain. J. Water Manag. Model. http://dx.doi.org/10.14796/
JWMM.R220-29. R220-29.
Kamali, M., ghazvinizadeh, S., Tajrishy, M., Kayhanian, M., 2012. Urban runoff
characteristics in Tehran, Iran. In: 9th International Conference on Urban
Drainage Modeling. Belgrade.
Kayhanian, M., Anderson, D., Harvey, J.T., Jones, D., Muhunthan, B., 2012a. Perme-
ability measurement and scan imaging to assess clogging of pervious concrete
pavements in parking lots. J.Environ.Manage. 95 (1), 114e123. http://dx.doi.org/
10.1016/j.jenvman.2011.09.021.
Kayhanian, M., Fruchtman, B.D., Gulliver, J.S., Montanaro, C., Ranieri, E., Wuertz, S.,
2012b. Review of highway runoff characteristics: comparative analysis and
universal implications. Water Res. 46 (20), 6609e6624. http://dx.doi.org/
10.1016/j.watres.2012.07.026.
Li, Y., Lau, S., Kayhanian, M., Stenstrom, M.K., 2005. Particle size distribution in
highway runoff. J. Environ. Eng. 131 (9), 1267e1276. http://dx.doi.org/10.1061/
(ASCE)0733-9372(2005)131:9(1267).
McKenzie, E.R., Wong, C.M., Green, P.G., Kayhanian, M., Young, T.M., 2008. Size
dependent elemental composition of road-associated particles. Sci. Total Envi-
ron. 398 (1e3), 145e153. http://dx.doi.org/10.1016/j.scitotenv.2008.02.052.
Morquecho, R.E., Pitt, R., Clark, S., 2005. Pollutant associations with particulates in
stormwater. Impacts Glob. Clim. Change 173 (216), 1e12. http://dx.doi.org/
10.1061/40792(173)216.
Nie, W., Yuan, Y., Kepner, W., Nash, M.S., Jackson, M., Erickson, C., 2011. Assessing
impacts of Landuse and Landcover changes on hydrology for the upper San
Pedro watershed. J. Hydrol. 407 (1e4), 105e114. http://dx.doi.org/10.1016/
j.jhydrol.2011.07.012.
Pezzaniti, D., Beecham, S., Kandasamy, J., 2009. Influence of clogging on the effective
life of permeable pavements. Proc. ICE-Water Manag. 162 (3), 211e220. http://
dx.doi.org/10.1680/wama.2009.00034.
Pratt, C., Mantle, J., Schofield, P., 1995. UK research into the performance of
permeable pavement, reservoir structures in controlling stormwater discharge
quantity and quality. Water Sci. Technol. 32 (1), 63e69. http://dx.doi.org/
10.1016/0273-1223(95)00539-Y.
Qin, H., Li, Z., Fu, G., 2013. The effects of low impact development on urban flooding
under different rainfall characteristics. J. Environ. Manage 129, 577e585.
Rowe, A., Borst, M., O’Connor, T., Stander, E., 2009. Pervious pavement system
evaluation. World Environmental and Water Resources Congress, pp. 1e8.
http://dx.doi.org/10.1061/41036(342)141.
Sansalone, J., Teng, Z., 2005. Transient rainfall-runoff loadings to a partial exfiltra-
tion system: implications for urban water quantity and quality. J. Environ. Eng.
131 (8), 1155e1167. http://dx.doi.org/10.1061/(ASCE)0733-9372(2005)131:
8(1155).
Sansalone, J., Tribouillard, T., 1999. Variation in characteristics of abraded roadway
particles as a function of particle size: implications for water quality and
drainage. Transp. Res. Rec. J. Transp. Res. Board 1690, 153e163. http://
dx.doi.org/10.3141/1690-18.
Siriwardene, N.R., Deletic, A., Fletcher, T.D., 2007. Modeling of sediment transport
through stormwater gravel filters over their lifespan. Environ. Sci. Technol. 41
(23), 8099e8103. http://dx.doi.org/10.1021/es062821v.
Teng, Z., Sansalone, J., 2004. In situ partial exfiltration of rainfall runoff. II: particle
separation. J. Environ. Eng. 130 (9), 1008e1020. http://dx.doi.org/10.1061/
(ASCE)0733-9372(2004)130:9(1008).
Tota-Maharaj, K., Scholz, M., 2010. Efficiency of permeable pavement systems for
the removal of urban runoff pollutants under varying environmental condi-
tions. Environ. Prog. Sustain. Energy 29 (3), 358e369. http://dx.doi.org/10.1002/
ep.10418.
U.S. Environmental Protection Agency (USEPA), 1996. “Environmental Indicators of
Water Quality in the United States.” Rep. No. EPA-841R9602, Washington, DC.
Valinski, N.A., Chandler, D., 2015. Infiltration performance of engineered surfaces
commonly used for distributed stormwater management. J. Environ. Manage
160, 297e305.
Valkman, R., 1999. “Lifespan” of Permeable/Permeable Paving Systems, Internal
Report. Urban Water Resources Centre, University of South Australia.
Yong, C., Deletic, A., Fletcher, T., Grace, M., 2008. The Clogging Behaviour and
Treatment Efficiency of a Range of Porous Pavements. 11th International Con-
ference on Urban Drainage. Edinburg, Scotland, UK.
M. Kamali et al. / Journal of Environmental Management 187 (2017) 43e53 53