Runoff and infiltration characteristics of pavement structures--review of an extensive monitoring program.
ABSTRACT The stormwater runoff and infiltration performance of permeable pavements has been systematically evaluated within an intensive monitoring program. The primary objective of the investigation was to generate a broad database, which enables the development of an advanced simulation module for urban drainage modelling. Over 160 field and lab scale experiments have been completed and analyzed for surface runoff and infiltration characteristics. The test series include several pavement types under various boundary conditions such as diverse precipitation impacts, varying surface slope and layer construction as well as different stages of surface clogging and several base and subgrade layer characteristics. The results represent a reliable and comprehensive database that allows profound conclusions and substantial recommendations.
Chapter: Urban Soils in the Vadose Zone[Show abstract] [Hide abstract]
ABSTRACT: Between the soggy ceiling of the ground water aquifer and the uppermost interface of earth and air is the unsaturated space of soil particles and pores invisible to most surface dwellers – the vadose zone. In cities, this space can be frozen in time under buildings and sidewalks, and contaminated with various kinds and concentrations of polluting substances. With more than 50% of the world’s population living in cities as of 2007, research on the composition, function and dynamics of urban soils is of utmost importance for urban ecological questions as well as the for the wellbeing of city dwellers world wide. Even before the 50% demographic benchmark, interest in anthropogenic soils began stirring in Germany in the 1970s in Berlin and Essen (Burghardt 1995; Blume 1975). At that time, research concerns revolved around the proper classification of soils in urban areas and the dilemma of restoring and re-using former industrial sites. From the 1980s until the beginning of the 1990s, pollution of urban soils with organic and inorganic contaminants became the focus of many studies (Thornton 1991; Lux 1993; Radtke et al. 1997). Since then, research on urban soils has substantially broadened. The BMBF (Federal Ministry of Education and Research) project “Evaluation of Urban Soils” from 1993 to 1996, for example, included groundbreaking work on the chemical, physical and biological properties of anthropogenic soils, involving major soil science institutions from the universities of Kiel, Essen, Hohenheim, Halle, Rostock and Berlin. Results are presented in Blume and Schleuss (1997).Perspectives in Urban Ecology, Edited by Endlicher Wilfried, 01/2011: chapter Urban soils in the vadose zone: pages 89-133; Springer Berlin Heidelberg., ISBN: 978-3-642-17730-9
- [Show abstract] [Hide abstract]
ABSTRACT: This paper describes the infiltration performance of a new 18 m 2 PICP test pavement under a variety of different flow conditions and pavement slopes. The initial results show that there is a clear relationship between pavement slope and the infiltration capacity through the pavement surface. Generally this relationship has demonstrated that the infiltration capacity at the surface of PICP systems will reduce as the slope of the system increases. However, the results also showed that the PICP test pavement used in this study had high infiltration rates of , even at slopes up to 20%. This suggests that typical PICP design guideline recommendations of maximum pavement slopes of 5% are probably overly conservative. The results of this investigation also showed that a significant amount of lateral movement of water occurred within the pavement structure and this was thought to be mainly caused by a geofabric layer that was included between the bedding and basecourse layers.12th International Conference on Urban Drainage, Porto Alegre, Brazil; 09/2011
NOVATECH 2007 1023
permeable pavements – Review of an intensive
and Infiltration characteristics of
Analyse du fonctionnement hydraulique de pavés perméables –
compte rendu d'un programme intensif de suivi
Marc Illgen*, Kathrin Harting**, Theo G. Schmitt*, Antje Welker*
Ehrlich-Strasse 14, 67663 Kaiserslautern, Germany,
Institute of Urban Water Management, University of Kaiserslautern, Paul-
Email : email@example.com
** IKT - Institute for Underground Infrastructure, Exterbruch 1, 45886
Email : firstname.lastname@example.org
Le fonctionnement hydraulique de surfaces équipées a été systématiquement évalué
dans un programme intensif de suivi. Le premier objectif de la recherche est
d’engendrer une large base de données qui permet le développement d'un module
avancé pour modéliser le ruissellement superficiel des pavés poreux. Plus de 150
expériences de laboratoire et de terrain ont été accomplies et analysées pour des
caractéristiques d'infiltration et d’écoulement superficiel. Les séries d'expériences
incluent plusieurs types de pavés poreux sous diverses conditions aux limites telles
que différentes précipitations, plusieurs pentes, et différents degrés d’encrassement.
Les résultats représentent une base de données fiable et complète qui permet des
conclusions profondes et des recommandations substantielles.
The stormwater runoff and infiltration performance of permeable pavements has been
systematically evaluated within an intensive monitoring program. The primary
objective of the investigation is to generate a broad database which enables the
development of an advanced simulation module for urban drainage modelling. Over
150 field and lab scale experiments have been completed and analyzed for surface
runoff and infiltration characteristics. The test series include several pavement types
under various boundary conditions such as diverse precipitation impacts, varying
surface slope and layer construction as well as different stages of surface clogging
and several base and subgrade layer characteristics. The results represent a reliable
and comprehensive database which allows profound conclusions and substantial
Minor permeable pavement, surface runoff, infiltration, clogging, HYDRUS.
1024 NOVATECH 2007
Permeable pavements are a common and effective technique to reduce stormwater
runoff from urban areas. Various types of more or less pervious structures are
available and are used for private as well as for public areas. Consequently, simplified
design methods as well as urban drainage models have to reproduce the specific
runoff and infiltration processes of these surface structures. However, validated
design parameters are rare and the common models are still using ancient methods
which have been adopted from the simulation of natural watersheds such as the
approaches of Horton or Green-Ampt. Even though permeable surfaces are receiving
more and more attention and the expectations regarding the accuracy of model
applications are continuously rising, these makeshift methods have not been
reconsidered or improved for years. Furthermore, the technical construction of a
pervious pavement including base layers significantly differs from natural soils.
Therefore, it is not surprising that simulation results show differences of up to 100%
compared to observed runoff volume or peak discharge (c.f. Hochstrate or Thorndahl
et al.). A lack of adequate methods for the computation of runoff and infiltration
processes on pavement structures is obvious; practical approaches which can be
easily implemented in common simulation models are needed.
The main objective of the overall investigation is to develop such an advanced
approach for modelling stormwater runoff from permeable pavements. Hereto, a
profound knowledge regarding the physical processes of stormwater infiltration on
and through the pavement construction is essential.
Valuable investigations have been realized in recent years (e.g., Timmermann , Field
et al., Pratt et al, Smith or Davies et al.) but are only focusing on particular aspects,
do only consider the infiltration capacity of new constructions or do not enable entire
water balances, mostly. Overall, more detailed investigations which include the
evaluation of the entire percolation processes through top and base layer into the soil
layer as well as clogging effects and other impacts are rare. Consequently, further
investigations which allow a detailed view into the construction were necessary to
obtain a reliable database for the development of a new computational method.
Due to this, an extensive monitoring program has been initiated including field
measurements on existing pavements as well as numerous lab scale tests on new
and gradually clogged pavements. The overall research project is divided into the
(1) Field measurements on existing pavements by infiltration tests
(2) Lab scale experiments of several pavement types under various conditions
(3) Data base extension by application of the detailed finite element model
HYDRUS-2D on further lab scale scenarios
(4) Evaluation of monitoring data supported by simulation results
(5) Development of an advanced approach on the basis of the monitoring data
(6) Calibration and validation of the new approach against the monitoring data
MATERIAL AND METHODS
NOVATECH 2007 1025
The infiltration capacity of existing pavements and the temporal distribution of the
infiltration rate have been evaluated by infiltrations tests. On various sites six different
types of permeable pavements have been tested several times using an infiltrometer
device (Illgen and Harting, 2006). The collected data indicate the infiltration capacities
of the pavement structures after several years of use and enable the evaluation of the
spatial and temporal variability of the infiltration rate depending on clogging effects,
mechanical impacts by car traffic and the particular weather condition, respectively. In
addition, the data are used as reference values for the imitation of clogging effects
within the lab scale test series.
The measured values show a high variability between the infiltration rates of a
particular pavement recorded at a certain site (different monitoring points on a
particular car park or even the same point at different times) as well as between the
rates observed at different locations (same pavement at different sites). For instance,
the infiltration rates recorded in the centre of a single parking space are significantly
higher than on the main traffic or tyre track, where mechanical impacts from cars and
the resulting wheel ruts increase clogging. However, surprisingly high infiltration rates
have been found at many locations, even on pavements which are generally
considered to be hardly permeable and even after long-lasting periods of operation
An overview of selected monitoring results recorded at a site in Lingen (Germany) is
given in Figure 1. At another car park with a pavement consisting of interlocked
concrete blocks with a ratio of 8% of gravel filled gaps considerable differences
between the infiltration capacity curves of three different parking spaces were
observed. Here, the initial infiltration rates amounted to 550, 750 and 1250 l/(s⋅ha),
the final rates varied between 370 and 770 l/(s⋅ha), all measured the same day.
centercenter centertyre trackcentertyre track center tyre trackcentertyre track
block pavement with narrow slots of 4 mmblock pavement with slots
of 10 mm
pavement with special
porous concrete pavement
with narrow slots
infiltration rate [l/(s⋅ ⋅ha)]
initial infiltration rate (after 10 minutes)
final infiltration rate (after 60 minutes)
1853 ?? 1710
Figure 1. Selected infiltration rates of four different types of pavement
measured at a car park in Lingen (Germany)
As expected, clogging effects due to fine material accumulating into the slots or voids
are greatly influencing the infiltration capacity and can cause a pointwise decrease of
the infiltration rate by factor 10 or even 100 compared to new built pavements. As the
process of clogging shows stochastic attributes, probabilistic aspects should be taken
into account for any further assessment and detailed recommendations. Admittedly,
the recorded data are not sufficient for a probabilistic evaluation yet.
1026 NOVATECH 2007
The major data base for the further evaluation of the runoff and infiltration
performance has been generated within an extensive lab scale test series. Several
types of pervious pavements have been evaluated under various conditions. Included
were experiments of various constructions for diverse precipitation impacts, varying
surface slope and layer construction as well as for different stages of surface clogging
and several base und subgrade layer characteristics. Over 120 tests have been
completed and analyzed for surface runoff, infiltration and percolation rates as well as
for changes of volumetric water content at several depths of the base layer.
The test facility is composed of a major steel framework with an integrated hopper
and a removable sprinkling unit on top. A steel vat of 1m x 1m x 0.5m contains the
entire pavement construction including base layer, bedding layer and pavers and is
put on the hopper into the major framework (Figure 2). At the bottom of the steel vat a
fine sieve of stainless steel enables an almost free exfiltration of percolating water but
prevents erosion of the base layer material. The sprinkling unit provides variable rain
intensities between 30 l/(s⋅ha) and 1000 l/(s⋅ha) (= 10.8-360 mm/h) and is controlled
by an inductive flow meter and a flexible tube pump. Surface runoff is measured
continuously by a mechanical tipping counter with a volume of 2.4 litres on a precision
balance; drainage flow is collected in a second tipping counter below the hopper
outlet and with a volume of 1.8 litres. The volumetric water content in the base layer is
observed by time domain reflectometry (TDR) measures. Six TDR probes can be
placed in the base layer at various depths and give a profound insight into the main
processes inside the pavement construction during sprinkling and during drying
periods, also. The monitoring data as well as the actual rain intensity are continuously
recorded and processed in time steps of 10 seconds.
Lab scale experiments
CONTROL UNITCONTROL UNIT
150,0 l/(s x ha)150,0 l/(s x ha)
78,3 l/(s x ha) 78,3 l/(s x ha)
44,5 l/(s x ha)44,5 l/(s x ha)
0% -10% 0% -10%
1.269 kg1.269 kg
tipping counter on
(runoff measurement)(runoff measurement)
(drainage flow measurement)(drainage flow measurement)
including pavers, bedding
and base layerand base layer
tipping counter on
including pavers, bedding
Figure 2. Scheme of the lab scale test facility.
The major advantage of the lab scale test series is the opportunity to evaluate
systematically the impact of each single key influence under well defined boundary
conditions whereas all other parameters are left unmodified. The impact of the
surface slope of a particular pavement, for instance, has been analyzed by
experiments with a gradually varied slope of 2.5%, 5.0% and 7.5%, each for several
NOVATECH 2007 1027
As old pavements after several years of use cannot be removed, afterwards
reconstructed and examined in the test facility without damaging the jointing structure,
clogging effects are imitated by the application of silica flour on new pavements. The
fine silica powder has proved to be the best alternative to reproduce clogging effects
within laboratory experiments (e.g. Davies et al.). Various states of clogging have
been considered by dispersion of several amounts of silica flour on the pavement
surface. Depending on the type of pavement, amounts between 200 g/m² and 4000
g/m² have been applied and washed into the joints or voids.
To link these amounts to the infiltration capacity of pavements in situ, infiltrometer
tests have been carried out in the test facility as well. The resulting infiltration curves
have then be compared with our own field measurements and with the results of 49
former infiltrometer tests (c.f. Nolting et al.) as well as with recorded infiltration rates
reported in the literature. Thus, it was possible to combine the advantage of clearly
defined boundary conditions of the lab scale experiments on the one hand side and
the necessity to consider clogging phenomena as they can be found after some years
of operation on the other hand.
The lab scale test series comprises the examination of several types of minor
permeable pavements with narrow slots (various block and flag pavements), as well
as medium permeable pavements with wider joints of 8-15 mm and porous concrete
pavement. Extraordinary permeable pavements have not been considered as they
are of low relevance for urban drainage management due to their high infiltration
capacities. All tested pavement constructions are similar to the pavements
investigated within the on-site test series. Moreover, additional experiments have
been carried out to examine the impact of a subgrade with a low hydraulic
conductivity on the runoff and infiltration performance. Also, the infiltration capacities
of various pavements and base layers have been tested in single arrangement
without base layer and pavement layer, respectively.
The test matrix of the extensive series of an interlocked concrete pavement with
narrow slots of 4 mm is exemplarily shown in Figure 3 and includes 45 single tests.
The ratio of joints amounts approx. 4% for this pavement.
slope stage of clogging
2 hours with
3 hours with
3 hours with
3(4) x 30 min
each 150 l/(s×ha)
30 mm with
200 g silica powder
400 g silica powder
500 g silica powder
200 g silica powder
400 g silica powder
200 g silica powder
400 g silica powder
PRECIPITATION IMPACT (IRRIGATION CHARCTERISTICS)
Figure 3. Test matrix for an interlocked pavement with narrow slots of 4 mm (45 single tests; grey
coloured cells indicate one or more tests for the particular boundary condition).
1028 NOVATECH 2007
To extend the data base, the very detailed finite element model HYDRUS-2D is
currently applied on further lab scale scenarios. HYDRUS-2D is a modelling
environment for simulating the two-dimensional movement of water, heat and multiple
solutes in porous media. The program numerically solves the Richards' equation for
saturated-unsaturated water flow. The unsaturated soil hydraulic properties are
described using van Genuchten, Brooks and Corey as well as modified van
Genuchten type analytical functions. The particular lab scale test assembly described
above is an ideal application case for the sophisticated HYDRUS-2D model and
enables reliable results.
The model is calibrated against the monitoring data and afterwards applied for
additional scenarios which have not been considered within the test series. Thus,
simulations for additional rain intensities, varying base layer characteristics
(thickness, compression, hydraulic conductivity) or different conditions at the
beginning of a rain event, for instance, are carried out instead of executing additional
time and cost intensive lab experiments. The application of the detailed soil hydraulic
model gives the opportunity to densify the database and supports a better
understanding of the hydraulic processes within the pavement construction as well.
The simulation results of the volumetric water content θ in a pavement construction
over several time steps are visualized in Figure 4. In this example, HYDRUS-2D has
been applied on a pervious pavement with slots of 10 mm on a 40 cm base layer of
gravel (0/45 mm) for a rain intensity of 100 l/(s⋅ha).
Data base extension by model application
Figure 4. Volumetric water content θ in a HYDRUS-2D simulation on a pervious pavement with
10 mm wide slots after (a) 5 min., (b) 10 min., (c) 20 min., (d) 25 min..
The monitoring data of the lab scale experiments together with the first simulation
results of the HYDRUS-2D application have been analyzed regarding major runoff
and infiltration characteristics and their interrelations with the several key parameters
such as rain intensity, rain duration, grade of clogging and base layer characteristics
(diameter, hydraulic conductivity, density, pore volume).
Mean infiltration and runoff rates from selected tests and their correlation to rain
intensity and grade of clogging are exemplarily shown in Figure 5. The results
originate from a test series on a pavement with slots of 4 mm on a 40 cm base layer
with a slope of 2.5% and are related to constant rain intensities over a period of 20
DATA BASE ANALYSIS
water content θ θ
NOVATECH 2007 1029
The figure indicates that the infiltration rate of a particular pavement depends on the
particular rain intensity even in cases where runoff occurs.
Figure 5. Mean infiltration and runoff rates depending on rain intensity and grade of clogging for a
pavement with joints of 4 mm and slope of 2,5% over 20 min. of irrigation.
In comparison to clogging effects and rain intensities, the surface slope of a
pavement construction is of lower relevance regarding runoff formation, especially for
pavements with higher infiltration capacities or lower degrees of clogging (Figure 6).
Also, the particular duration of a rain event has only for short-term events a significant
impact. The infiltration rate of a pavement construction decreases during the first 30
minutes mostly and remains more or less constant afterwards.
Figure 6. Mean infiltration rates depending on rain intensity, surface slope and grade of clogging
for a pavement with joints of 4 mm and slope of 2,5% over 15 min. of irrigation.
The investigated block pavements with slots of 7 and 10 mm, both, showed very high
infiltration capacities. Even for tests with large amounts of silica powder only very low
runoff rates of 50 and 25 l/(s⋅ha) were observed for rain intensities of 1000 l/(s⋅ha).
Even though relatively high rain intensities have been evaluated, the water content
below the pavement does usually not reach the state of saturation within a period of
1-2 hours in cases of a free drainage of the base layer. For the scenario of a
subgrade with significantly lower hydraulic conductivity, the base layer is functioning
as a storage tank with a throttled outflow. In such a case, the water content inside the
(200 g silica flour) (400 g silica flour) (500 g silica flour)
infiltration rate [l/(s x ha)]
100 l/(s×ha) 150 l/(s×ha)200 l/(s×ha)
(200 g silica flour) (400 g silica flour) (500 g silica flour)
runoff rate [l/(s x ha)]
100 l/(s×ha)150 l/(s×ha)200 l/(s×ha)
100 150200250 300
rain intensity [l/(s×ha)]
infiltration rate [l/(s⋅ ⋅ha)]
new pavement: slope 2,5%
new pavement: slope 5,0%
new pavement: slope 7,5%
slightly clogged pavement: slope 2,5%
slightly clogged pavement: slope 5,0%
slightly clogged pavement: slope 7,5%
medium clogged pavement: slope 2,5%
medium clogged pavement: slope 5,0%
medium clogged pavement: slope 7,5%
1030 NOVATECH 2007
base layer may rise quickly and can restrict the infiltration into the construction during
long-lasting rain events.
The soil water content at the beginning of a rain event is also influencing the runoff
and infiltration processes of the pavement construction but on a much lower level
than for natural soils. A relatively high soil water content represses the higher
infiltration capacities commonly observed at the beginning of a rain event and causes
lower and more equal infiltration rates instead. After a dry period of 8 to 24 hours
depending on the particular soil and weather conditions the soil water content has
reached its initial level and the infiltration capacity has recovered correspondingly.
Consequently, the length of a dry period as well as the process of drying itself are
important factors and thus not to neglect.
On the basis of the widespread data set and the therefrom derived empirical
relationships, an advanced approach for modelling runoff and infiltration processes of
permeable pavements is going to be developed. As the approach should be easy to
implement into common urban drainage models, it is planned to work only with a
limited number of parameters which are easy to determine. For this reason, a detailed
approach as it is implemented in a soil hydraulic model like HYDRUS is not useful.
The new approach will be probably based on a simplified bi-directional layer model. In
comparison to the common methods used in urban drainage models, as the approach
of Horton for instance, the approach will include rain dependent infiltration capacities,
the main processes taking place within base and subgrade layers and their interaction
with the pavement layer. Clogging will be taken into account by adequate parameter
values. Moreover, detailed recommendations for reasonable infiltration rates, runoff
coefficients and model parameters are going to worked out which may help planners
to estimate the runoff contribution from pavements.
LIST OF REFERENCES
Davies J.W., Pratt C.J. and Scott M.A. (2002). Laboratory study of permeable pavement systems
to support hydraulic modeling. Proc. 9th Int. Conf. on Urban Drainage, Portland, USA.
Field R., Masters H. and Singer, M. (1982). An Overview of Porous Pavement Research. Water
Resources Bulletin, Vol. 18, No. 2, pp. 265-270.
Hochstrate S. (2003). Assessment of common approaches to model stormwater runoff and
infiltration processes of permeable pavements (in German). Master Thesis at the University
of Kaiserslautern (unpublished).
Illgen M. and Harting K. (2006). Modelling stormwater runoff and infiltration processes of
permeable pavements – a new approach for an old problem. Proc. 7th International
Conference on Hydroinformatics, HIC 2006, Nice, France.
Nolting B., Schönberger O., Harting K. and Gabryl P. (2005). Investigation on permeable
pavements after several years of operation (in German). Technical report free for download
Pratt C.J., Wilson S. and Cooper P. (2002). Source controle using constructed pervious surfaces
– hydraulic, structural and water quality performance issues. CIRIA C582.
Smith D.R. (1984). Evaluation of concrete grid pavements in the United States, Proc. 2nd Int.
Conf. on Concrete Block Paving, Delft, pp. 330-336.
Thorndahl S., Johansen C. and Schaarup-Jensen K. (2005). Assessment of runoff contributing
catchment areas in rainfall runoff modelling. Proc. 10th Int. Conf. on Urban Draiange,
Copenhagen, Denmark, 21-26 August 2005.
Timmermann U. (2000). Entsiegelung von Verkehrsflächen (in German). Neue Landschaft, No.
7/2000, pp. 445-448.