C. F. Yong* and A. Deletic*
* Centre for Water Sensitive Cities, Department of Civil Engineering, Building 60, Monash
University, Wellington Rd, Clayton, Victoria, 3800, Australia (E-mail: fernyong@gmail.com; ana.deletic@monash.edu)
ABSTRACT
Porous pavements are one of many Water Sensitive Urban Design (WSUD) technologies that are widely used to manage stormwater runoff. They are easily retrofitted and are effective in improving water quality and hydrology, but are prone to clogging. Despite being a major determinant in the lifespan of porous pavements, there is a dearth of information on the physical clogging processes through these systems. The aim of this study was to develop a fundamental understanding of the main physical processes that govern clogging, and ultimately develop a simple model that predicts clogging through porous pavements over their lifespan. This paper presents findings from two laboratory experiments that were conducted over 3 years at Monash University on three commonly used porous pavements. The key variables that were hypothesised to influence clogging were pavement design, flow dynamics as well as drying conditions. The pavements investigated in this study were observed to clog at various times, with the exception of one, which showed no signs of clogging despite receiving 25 years of simulated Brisbane, Australia rainfall. Systems that were simulated under varied inflow conditions, along with drying sequences were found to double the lifespan of systems simulated under a constant continuous flow without drying.
KEYWORDS
Drying; hydraulic resistance; pavement design, physical clogging; porous pavements
INTRODUCTION
Compared to the level of porous pavement adoption in Europe, Japan and the USA, porous pavements have not been widely implemented in Australia despite being one of the most progressive countries with respect to the adoption of the WSUD philosophy. This is mainly due to a low level of confidence in its stormwater treatment performance under Australian conditions, and early perceptions relating to clogging due to numerous negative experiences in field installations
(Fletcher et al., 2005).
Although “new” porous pavements of any type can show impressive infiltration results in the first few months of service, it is the long-term infiltration performance of a pavement that determines their ultimate success or failure. Most studies tend to report the hydraulic performance of porous pavements immediately after field-installation (while the system is still “new”), but there are not many studies on installations after years of operation.
Laboratory studies of monolithic systems by Pratt et al.
(1995) have shown that clogging can result from fine particles accumulating in the void spaces of porous pavements. As smaller particles trap larger particles, the rate of clogging will therefore increase as more fines are trapped (Balades et al.,
1995) thus leading to lower infiltration rates, which eventually cause the ponding of surface waters over the monolithic pavements. Previous studies by Galli (1992) and Nozi et al.
(1999) have found clogging to be the main reason for the failure of stormwater infiltration systems, unless properly maintained. A small scale laboratory study on the clogging of stormwater infiltration systems by

Pokrajac and Deletic (2002) showed that clogging occurs at the interface between the filter media and the surrounding soil. Siriwardene (2007) studied physical clogging under both constant and variable water levels, and found that a clogging layer forms at the interface between the filter and underlying soil, irrespective of the inflow regime of both water and sediment. It was also found that clogging is much slower if the water level is kept at a constant level than if it varies within the column, due to a sediment plug that ‘shields’ the filter/soil interface. Most importantly, Siriwardene
(2007) found that sediment particles less than 6 µm are the main driver in the development of the clogging layer and hence causing system failure.
Most modular pavement systems are constructed with a geotextile layer to reduce the flow of pollutants beyond this barrier. Studies from the USA, Australia and Europe (Thompson 1995;
Legret and Colandini 1999; Shackel et al., 2003; Ferguson 2005; Newton et al., 2003) have shown evidence that effective pollutant removal can be achieved without the use of a geotextile layer.
However, there is also evidence that the geotextile layer is an important mechanism that may further enhance pollutant removal (Hogland et al.
, 1987, Niemczynowicz, 1990; Pratt, 1990). Although this effective removal of particles by the geotextile is rather desirable, it is unfortunately the same mechanism which leads to the reduction in infiltration capacity which is the principal operational concern for all forms of porous pavement (Newton et al., 2003). The decision to use a geotextile between the laying course and the sub-base is ultimately a balance between durability, structural performance of the pavement and possible improvements in water quality.
Clogging in porous pavements is inevitable eventually and the key mechanisms that govern clogging in porous pavements, specifically when exposed to typical Australian operational conditions require further investigation before any predictions can be made on the lifespan of these systems. The aim of this project is to understand and develop new insight into the long-term performance and clogging behaviour within three major porous pavement types used in Australia, with the ultimate aim of developing a simple physically-based model that predicts clogging. This paper presents the preliminary results from two extensive laboratory experiments, which were simulated under varied flows and drying sequences.
METHODS
To investigate the long-term clogging behaviour of porous pavements, two laboratory experiments were conducted in an accelerated time scale over 3 years. In the first experiment (E1), a constant continuous inflow regime without any drying sequence was simulated while varied inflow with drying sequences was simulated in the second experiment (E2). Three widely used porous pavements were chosen for this study; (i) monolithic porous asphalt (PA), (ii) a product by Boral
Clay and Concrete, modular Hydrapave (HP), and (iii) monolithic Permapave (PP). They were installed based on manufacturer guidelines in a 2.7 m x 0.45 m x 1.95 m rig, in three separate vertical compartments. The rig included a 550 L tank with aerator coils and a pneumatic stormwater distribution system, which consists of a peristaltic pump and a rotating sprinkler to ensure a random and equal distribution of stormwater and sediments over the pavements. The concentration of sediments was around 50-300 mg/L which is the typical range of stormwater sediment concentration as found in a number of studies (e.g. Francey et al.
, (2010) and Duncan (1999)).
Three tipping bucket-rain gauges, with a volume resolution of 0.2 mm/tip were also installed.
Further details can be found in Yong et al.
(2008, 2011).
Experimental Procedure
Experiment 1 (E1): Constant continuous inflow without drying.
A continuous delivery of stormwater, without drying regime was simulated. In order to cover a wide range of Australian

conditions, the chosen flow rate had to reflect both sub-tropical Brisbane and temperate Melbourne climate. This corresponded to a flow of 3.9 mm/hr through each pavement system (Table 1). As equivalent annual rainfall volumes treated by the porous pavement systems before it clogged were simulated, the actual lifespan of a real porous pavement system in practice could be determined, thus relating results to practice. In-depth details have previously been described in Yong et al.
(2008).
Table 1: The Flow Rate for Experiment 1, chosen from a 30 year Melbourne Rainfall Time Series from 1970 to 1999 (i.e. the 90 th
percentile rainfall intensity in Melbourne, which roughly approximates the mean rainfall intensity in Brisbane).
Frequency
90%
Mean
Flow rate per ha. m
3
/sec
0.0108
0.0088
Velocity m/sec
1.1E-06
8.8E-07 mm/hr
3.9
3.2
Pavement Area m
2
0.1564
0.1564 m
3
Flow rate
/sec ml/sec
1.7E-07
1.4E-08
0.169
0.137
A total of 13 and 26 years of operation were simulated under Brisbane and Melbourne climate respectively. The continuous delivery of stormwater in a week was equivalent to half the average annual Brisbane rainfall (1200 mm) or the average annual Melbourne rainfall (650 mm). To study the rate of clogging during typical floods, a 1 in 5 year Brisbane storm (of 5 min. duration – typical design flood for small catchments where porous pavements are likely to be deployed), that is also equivalent to a more than 1 in 100 year Melbourne storm (of 5 min. duration), was simulated in each of the 5 th
, 10 th
, 17 th
and 25 th
weeks of the experiment. The intensity of these storms was 191 mm/hr, which translates into a flow rate approximately 50 times higher than the average constant inflow into each system. Therefore, a possible storm in tropical Queensland and a major storm in
Melbourne were replicated, under which the systems may start to malfunction and cause flooding.
Experiment 2 (E2): Varied inflows with drying/wetting sequences.
Five different flows were simulated along with a drying sequence (Table 2). Similar to E1, the 5 flow rates in E2 were reflective of both sub-tropical Brisbane and temperate Melbourne climate, the latter of which was approximately half of Brisbane’s rainfall intensity. Similar storm conditions as per E1 were simulated on the 6 th
, 8 th
, 12 th
, 16 th
, 20 th
and 24 th
weeks of the experiment. A total of 26 years of operation of Brisbane climate was simulated. Further details including drying simulations have previously been described in Yong et al. (2011).
Table 2. The Flow Rates for Experiment 2, chosen from a 10 year Brisbane Rainfall Time Series from 1988 to 1997.
Flow Frequency
(percentile range)
Duration hours
Flow rate/ha. m
3
/sec
Velocity m/sec mm/hr
Pavement
Area m
2 m
3
Flow rate
/sec ml/sec
A
B
C
D
0-39
40-59
60-79
80-100
96
48
48
48
0.0006 5.8E-08
0.0029 2.9E-07
0.0071 7.1E-07
0.2
1.0
2.6
0.0609 6.1E-06 21.9
0.1564 9.0E-09 0.009
0.1564 4.5E-08 0.045
0.1564 1.1E-07 0.111
0.1564 9.5E-07 0.953
Data Collection
Sampling Procedure and Analytical Methodology.
Prior to the commencement of both experiments, a series of hydraulic conductivity (k) measurements was performed to determine the porosity of the individual pavement systems when the systems were in their “new” condition. Water quality samples were collected from both the inflow and outflow points and analysed for Total Suspended

Solids (TSS). Particle size distributions (PSD) were also measured using a Beckham Coulter
LS100Q Laser Diffraction Particle Size Analyser.
In E1, an intensive daily sampling regime for TSS was conducted during the initial 2 weeks to determine the required inflow and outflow sampling frequency. This was subsequently reduced to 2 separate inflow and outflow composites per week. The 1 st
composite consisted of 3 samples taken on Mondays, Wednesdays and Fridays, while the 2 nd
composite consisted of 2 samples, taken on
Fridays and Mondays. Each inflow and outflow composite was of a volume of 500 ml and 1000 ml respectively. The outflow samples required a higher volume as they were lower in concentration.
Similar inflow composites were also taken for PSD every alternate week.
In E2, TSS analyses were performed as per E1, with similar inflow and outflow volumes collected for each of the 5 simulated flow rates. The analyses were initially performed on a regular basis but were reduced to fortnightly intervals as the concentration of samples started to stabilise and occur at a more predictable manner. PSD analyses were also performed fortnightly for the inflow samples throughout the duration of the experiment.
In both experiments, the duration taken per sampling session to achieve the required volume was dependent on the type of flow rate (low flow or storm) as well as the age of the system. Clogging observations were made by regularly measuring the level of ponding above each pavement surface as they occurred, with an accuracy of 0.5 cm. Each experiment was conducted, until the systems were clogged, defined as when ponding above the pavement surfaces were observed to overflow, or when the outflow was 10 % of the initial outflow.
Data Cleaning and Processing. To improve the efficiency and integrity of the collected data to be used in statistical analyses, local and global checks were first performed manually, and then automatically on the collected tippings. The clean tippings were then analysed in hourly time steps before being converted to flow rates. From here onwards, four predictor variables that were of interest were then obtained: flow rates, cumulative volume, cumulative mass and cumulative mass <
6 microns (µm). Flow rates were pre-determined parameters, while cumulative volume and mass were straightforward calculations obtained by using total volume and total mass. Cumulative mass
< 6 µm were calculated by multiplying the passing percentage of sediments with mass < 6 µm with the total mass. To measure the evolution of clogging over time, hydraulic resistance for each sampling point was calculated in both experiments. Hydraulic resistance (R) is defined as the ratio between the length of filtration media and saturated hydraulic conductivity, as per Bouwer’s work
(1969, 2000). In the case of surface clogging, whereby the clogging layer is of negligible thickness,
R is equal to a ratio between the ponding depth (measured from the clogging layer) and the infiltration velocity that in fact is equal to the saturated hydraulic conductivity of the clogging layer
R=L/K=h/v (Eqn. 1) where, while L is thickness of a layer, K is saturated hydraulic conductivity of a layer, h is ponding depth above a thin clogging layer, and v is infiltration (Darcy’s) velocity.
RESULTS AND DISCUSSION
Influence of Design (Rate of Clogging). Figure 1 shows the ponding depth against the cumulative volume of inflow received for Porous Asphalt in both E1 and E2. In Porous Asphalt, ponding was observed to occur above the pavement surface, indicating the formation of a clogging layer at 12 m
(Table 3). Ponding in E2 did not occur until almost twice the volume had infiltrated the pavement at
18 m. A closer inspection of the pavement showed accumulated sediment on the pavement surface,

and at the interface between the pavement and the crushed aggregate bedding layer, the former of which was completely covered in sediment by the end of the experiment. Over the course of both experiments, E1 did not overflow, while E2 overflowed at approximately 31 m, which is equivalent to receiving approximately 19 years of Brisbane rainfall.
40
20
0
-20
-40
-60
-80
-100
-120
-140
0 asphalt
5 10 15 20
Cumulative Inflow (m)
25 30 crushed aggregate
35 reservoir bed
40
E1 Flow E1 Storm E1 E2 Flow A E2 Flow B E2 Flow C E2 Flow D E2 Storm E2
Figure 1: Ponding Observations (mm) of Porous Asphalt versus Cumulative Volume (m) for both
Experiment 1 and Experiment 2.
Table 3: Recordings of the First Ponding Observation along with the Corresponding Explanatory
Variable Values in Porous Asphalt and Hydrapave, for Experiments 1 and 2.
PA E1 Flow Rate
(ml/min)
10.14
500
PA E2 0.54
2.72
6.67
57.18
500
HP E1 10.14
500
HP E2 0.54
2.72
6.67
57.18
500
Ponding
(mm)
5
2
5
17.5
27
35
5
120
160
40
40
20
20
50
Cum. Vol.
(m)
14
12
26
26
24
24
18
13
7
28
28
27
24
18
Cum. Mass
(g)
2353
2133
3610
3611
3345
3362
2304
2350
2792
3702
3702
3612
3245
2305
Cum. Mass <6
µm (g)
425
384
835
835
772
777
539
397
478
824
824
812
750
559
R
(hour)
1.41
0.01
51
51
18
19
0.03
35.5
0.96
223.4
47
86
0.9
0.3
In Hydrapave (Figure 2), similar trends of ponding was observed to occur, not on the pavement surface as per Porous Asphalt but above the geotextile layer, In E1, the first ponding observation was recorded at 7 m, while in E2, ponding occurred after more than double the amount of volume had been infiltrated (18 m). In both Hydrapave experiments, the maximum height of ponding above the pavement surface was observed at 15.5 m and 30 m respectively, with no overflow occurrences.
Throughout the duration of both experiments, ponding in Permapave was not observed. This does not come as a surprise as Permapave pavement and sub-base are constructed with similar sized aggregates, thus facilitating a smooth flow of water through the system. As such, Porous Asphalt, with its transition from regular pavement aggregates to fine-sized bedding layer had the shortest lifespan, followed by Hydrapave and Permapave. Regardless of pavement design however, the impact of clogging appears to be consistently delayed in E2, compared to E1, suggesting the influence of drying and multiple runoff conditions.

40
20
0
-20
-40
-60
-80
-100
-120
-140
0 bricks
5
5 mm bedding stone
10 15 20
Cumulative Inflow (m)
25 30 35 geotextile
40
E1 Flow E1 Storm E1 E2 Flow A E2 Flow B E2 Flow C E2 Flow D E2 Storm E2
Figure 2: Ponding Observations (mm) of Hydrapave versus Cumulative Volume (m) for both
Experiment 1 and Experiment 2.
Influence of Flow Dynamics and Drying. Figures 3 and 4 each show the hydraulic resistance from
E1 and E2, plotted against cumulative volume of inflow that passed through Porous Asphalt and
Hydrapave respectively. Hydraulic resistance was also plotted against cumulative mass and cumulative mass < 6 µm (showing similar trends to cumulative volume), but is not presented in this paper. It is clear that clogging was slowed dramatically (twice the volume required) when multiple runoff conditions along with drying events are present, as simulated in E2. This was true for both pavement types despite their differences in design. The degree of clogging which is expressed as hydraulic resistance (R) increases with a decrease in flow. PA, which had the shortest lifespan, consistently showed a smaller range in resistance values than Hydrapave, which did not overflow. A comparison of results from E1 and E2 suggest that drying and wetting has a direct influence on the longer lifespan of these systems and indicates that previous laboratory studies on porous pavements, which have only been based on wetting conditions, may have been underestimating the actual lifespan of porous pavement systems.
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 10 20
Cumulative Inflow (m)
E1 Single Runoff No Drying
30
E2 Multiple Runoff with Drying
40
800
700
600
500
400
300
200
100
0
Figure 3: Runoff Rate of Porous Asphalt for both Experiment 1 and Experiment 2.

50
40
30
20
10
0
0 40
0
400
200
1200
1000
800
600
5 10 15 20 25
Cumulative Inflow (m)
30 35
E1 Single Runoff no Drying E2 Multiple Runoff with Drying
Figure 4: Runoff rate of Hydrapave for both Experiment 1 and Experiment 2.
Influence of inflow variables on clogging. Whilst cumulative volume is an important factor in determining the rate of clogging, cumulative mass, cumulative mass <6 µm and flow rate were 3 other investigated inflow variables that were useful predictors in the development of clogging.
Using these results, a 4-parameter regression model as a function of hydraulic resistance that could predict clogging is currently being developed with the hope of allowing more accurate predictions to be made on the performance of systems under normal drying conditions, as well as varying inflow rates, which would be experienced in reality.
CONCLUSIONS
Modelling of clogging behaviour in porous pavements over their lifespan is important to provide a reliable prediction method in future watershed behaviour. However, such a model has yet to be developed, particularly under various Australian rain and drought conditions. The results from this study help to predict the average life expectancy of porous pavement systems under two different climates in Australia as well as their treatment performance. The results also showed that the clogging behaviour and lifespan of porous pavements varied by their design, and thus the position at which the clogging layer is formed. Porous Asphalt for example, had its clogging layer on the surface, causing water to pond relatively quicker compared to Hydrapave, which had its clogging layer above the geotextile layer, positioned 130 mm below the pavement. As clogging starts considerably deeper in the Hydrapave system, where it is not visible to pedestrians, failure of the system will not be noticed until significantly later, when ponding is observed above the pavement surface, thus masking the actual efficiency of the system unless regular testing is performed. On the other hand, Permapave did not show signs of clogging at all. While the use of geotextile has its advantages and disadvantages as seen from the Hydrapave system, the decision to incorporate a geotextile layer in future installations should be based on a careful consideration of balancing the life expectancy and pollutant removal performance of the system, without compromising one or another.
The rate of clogging was also dependent on the conditions on which the pavements were exposed to. Systems that received variable flow magnitudes, along with long drought conditions were found to have doubled the lifespan of systems receiving low flow magnitudes with the absence of drought.
These observations help inform designers and asset managers on the average life expectancy of porous pavement systems under drying and various wetting conditions in Australia. The decision to choose a particular system should not only be based on the climate data and environmental

conditions of a potential site but also on the life expectancy and pollutant removal performance of the system, without compromising one or the other.
REFERENCES
Balades, J. D., Legret, M. and Madiec, H. (1995). Permeable pavements: Pollution Management Tools. Water Science and Technology, 32 (1):49-56.
Bouwer, H. (1969(Theory of seepage from open channels” Adv.Hydrosci.
5 :121-170.
Bouwer, H. (2002). "Artificial recharge of groundwater: hydrogeology and engineering." Hydrogeology 10 (1): 121-142.
Ferguson, B. K. (2005). Porous Pavements. Boca Raton, Florida, CRC Press LLC.
Fletcher, T. D., Duncan, H. P., Poelsma, P. and Lloyd, S. D. (2005). Storm water flow and quality, and the effectiveness of non-proprietary storm water treatment measures-a review and gap analysis (Technical Report 04/8). Melbourne:
Cooperative Research Centre for Catchment Hydrology.
Galli, J. (1992). Analysis of urban BMP performance and longevity in Prince George's County, Maryland, 92711,
Prince George's Country Department of Environmental Resources.
Hogland, W., Niemczynowicz, J. (1987). Test of porous pavements performed in Lund, Sweden, in Topics in Drainage
Hydraulics and Hydrology. B.C.Yen (Ed.), pub. Int. Assoc. for Hydraulic Research, pp 19-80.
Legret, M. and Colandini, V. (1999). Effects of a porous pavement with reservoir structure on runoff water: Water quality and fate of heavy metals. Water Science and Technology 39 (2): 111-117.
Newton, D. B., Jenkins, G. A. and Phillips, I. (2003). The potential of porous pavement for urban storm water management. Paper presented at the 28 th
International Hydrology and Water Resources Symposium, NSW.
Niemczynowicz, J. (1990). Swedish way to stormwater enhancement by source control, in Torno, HC: Urban stormwater quality enhancement: source control, retrofitting and combined sewer technology. Proceedings of an
Engineering Foundation conference, American Society of Civil Engineers, New York.
Nozi, T., Mase, T. and Murata, K. (1999). Maintenance and management aspect of stormwater infiltration system. Paper presented at the Proceedings of the 8th International Conference on Urban Drainage, Sydney, Australia.
Pokrajac, D. and Deletic, A. (2002). Clogging of infiltration drainage systems. In International Conference on Sewer,
Operation and Maintenance, Bradford, UK.
Pratt, C. J. (1990). Permeable pavements for stormwater quality enhancement, in Urban Stormwater Quality
Enhancement: Source Control, Retrofitting and Combined Sewer Technology. H.C.Torno (ed.), pub. American Soc. of
Civil Engineers, New York, U.S.A., ISBN 0-87262-759-4, pp. 131-155.
Pratt, C. J., Mantle, J. D. G. and Schofield, P. A. (1995). UK Research into the performance of permeable pavement, reservoir structures in controlling stormwater discharge quantity and quality. Water Science Technology, 32 (1): 63-69
Shackel, B., Ball, J. E. and Mearing, M. (2003). Using permeable eco-paving to achieve improved water quality for urban pavements. Proceedings of the 7th International Conference on Concrete Block Paving, Africa.
Siriwardene, N., Deletic, A. and Fletcher, T. D. (2007). Clogging of stormwater gravel infiltration systems and filters:
Insights from a laboratory study. Water Research, 41 (7): 1433-1440.
Thompson, M. K. (1995). Design and installation of test sections of porous pavements for improved quality of parking lot runoff, Guelph University, Canada.
Yong, C.F., Deletic, A., Fletcher, T.D. & Grace, M.R. (2008). The Clogging Behaviour and Treatment Efficiency of a
Range of Porous Pavements. In Proceedings of the 11th International Conference on Urban Drainage, September 2008,
Edinburgh, Scotland.
Yong, C.F., Deletic, A., Fletcher, T.D. and Grace, M.R. (2011). Hydraulic and treatment performance of pervious pavements under variable drying and wetting regimes. Water Science and Technology, 64 (8): 1692-1699.