Home > CRWR Online Report 08-06

CRWR Online Report 08-06

CRWR Online Report 08-06


Porous Friction Course: A Laboratory Evaluation of Hydraulic
Properties


by


Remi M. Candaele, M.S.E.
Michael E. Barrett, Ph.D.
Randall J. Charbeneau, Ph.D.


May 2008



Center for Research in Water Resources
The University of Texas at Austin
J.J. Pickle Research Campus
Austin, TX 78712-4497
This document is available online via the World Wide Web at
http://www.crwr.utexas.edu/online.shtml




iii

ACKNOWLEDGMENTS


This research was funded by the Texas Department of Transportation under
project number 0-5220, ��Investigation of Stormwater Quality Improvements Utilizing
Permeable Pavement and/or the Porous Friction Course (PFC)��.


iv

TABLE OF CONTENTS
Contents 
List of Tables ................................................................................................................. vi 
List of Figures ............................................................................................................... vii 
1.  Overview of the Porous Friction Course Project ..................................................... 1 
1.1.  Objectives ......................................................................................................... 1 
1.2.  Locations of Study ............................................................................................ 2 
1.3.  Extraction of Cores ........................................................................................... 8 
1.4.  Outline of Thesis ............................................................................................ 11 
2.  POROUS ASPHALT AND HYDRAULIC PROPERTIES IN LITERATURE ...... 12 
2.1.  Measure of Porosity and Methods .................................................................. 13 
2.2.  Measure of Hydraulic Conductivity ............................................................... 14 
2.3.  Limitation of Darcy��s Law ............................................................................. 21 
2.4.  Maintenance of Porous Asphalt ...................................................................... 23 
3.  LABORATORY MEASUREMENTS OF EFFECTIVE POROSITY AND
RESULTS ......................................................................................................................... 35 
3.1.  Introduction .................................................................................................... 35 
3.2.  Water displacement method ........................................................................... 36 
3.3.  Image Analysis Method .................................................................................. 40 
3.4.  Effective Porosity Results and Interpretation ................................................. 43 
4.  LABORATORY MEASUREMENTS OF HYDRAULIC CONDUCTIVITY ........ 47 
4.1.  Analytical Description of the Testing Apparatus ........................................... 47 
4.2.  Description of the equipment ......................................................................... 52 
4.3.  Procedure of the constant head test ................................................................ 55 
4.4.  Verification Test ............................................................................................. 56 
4.5.  Cleaning of porous asphalt samples ............................................................... 58 
5.  ANALYSIS OF THE CONSTANT HEAD EXPERIMENT RESULTS ................. 61 
5.1.  Inertial Effects and Forchheimer Model ......................................................... 61 
5.2.  Linear Model: Darcy��s Law ........................................................................... 69 
5.3.  Characterizing the clogging phenomenon ...................................................... 72 
v

5.4.  Radial Flow Distribution ................................................................................ 81 
6.  CONCLUSIONS....................................................................................................... 83 
6.1.  General Trend ................................................................................................. 83 
6.2.  Specific Locations .......................................................................................... 84 
6.3.  Limitations ...................................................................................................... 85 
6.4.  Future Work .................................................................................................... 86 
APPENDICES .................................................................................................................. 88 
Appendix A: Permeameters & Measuring EAV Contents ....................................... 88 
Appendix B: Weight & Characteristics of Cores ..................................................... 93 
Appendix C: Experimental System – Blueprints ..................................................... 98 
Appendix D: Porous Friction Course Aggregate - Water Properties ..................... 105 
Appendix E: Hydraulic Conductivity Model Parameters ...................................... 107 
Appendix F: Head Gradient versus Flowrate Curves ............................................ 109 
References ................................................................................................................... 115 



vi

LIST OF TABLES

Table 1 - Cleaning guidelines from Van Leest et al. (1997) ............................................. 19 
Table 2 - Summary of conductivity data from Kandhal and Mallick (1999) ................... 20 
Table 3 - Literature Values for PFC Characterization ...................................................... 21 
Table 4 - Cleaning Machine Summary from Japan at PWRI (2005) ................................ 30 
Table 5 - Summary of Effective Porosity ......................................................................... 44 
Table 6 - Forchheimer Model - Hydraulic Conductivity and Standard Error ................... 68 
Table 7 - Summary of 95% Confidence Intervals of Hydraulic Conductivity ................. 70 
Table 8 - Extrapolated Air void content ........................................................................... 75 
Table 9 - Mass of Extracted Sediments ............................................................................ 76 
Table 10 - Hydraulic Conductivity of Initial and Non-Cleaned Drilled Cores ................. 80 
Table 11 - Weight & Geometric Characteristics of Cores – Sliced Cores ........................ 93 
Table 12 - Weight & Geometric Characteristics of Cores – Original Cores .................... 94 
Table 13 - Water Properties (FHA, 2007) ...................................................................... 105 
Table 14 - Aggregate Grain Size Distribution (Alvarez et al, 2006) .............................. 106 
Table 15 - Plain Cores - Darcy��s linear Model ............................................................... 107 
Table 16 - Forchheimer Model Parameters .................................................................... 107 
Table 17 - Drilled Cores - Darcy��s linear Model ............................................................ 108 
Table 18 - Drilled Cores - Forchheimer Model Parameters ........................................... 108 

vii

LIST OF FIGURES
Figure 1 - Overview of the PFC locations in the Austin area. ............................................ 3 
Figure 2 -Location of the three extraction sites – Photo Courtesy of Google Earth ........... 4 
Figure 3 - Northbound Loop 360 near Bull Creek Extraction Site – Photo Courtesy of
Google Earth ....................................................................................................................... 5 
Figure 4 - Eastbound RM1431 Extraction Site – Photo Courtesy of Google Earth ........... 6 
Figure 5 - Northbound RM 620 Extraction Site – Photo Courtesy of Google Earth .......... 7 
Figure 6 - Drill Press on RM 620 - March 14, 2007 ........................................................... 9 
Figure 7 - Extracted Cores on RM 620 - March 14, 2007 .................................................. 9 
Figure 8 - Quick pavement filling holes on RM 620 - March 14, 2007 ........................... 10 
Figure 9 - Separation of the overlay – April 09, 2007 ...................................................... 10 
Figure 10 - Drainage potential over time (Isenring et al., 1990) ...................................... 17 
Figure 11 - Deterioration of Permeability Coefficient k (Fwa et al., 1999) ..................... 26 
Figure 12 - The cleaning process of the Spec-keeper cleaning machine (PWRI, 2005) .. 28 
Figure 13 - The Spec-Keeper cleaning machine in Japan from PWRI (2005) ................. 28 
Figure 14 - Cleaning Machine with High Pressure Air only from PWRI (2005) ............. 29 
Figure 15 - Enclosing the Sample in the Plastic Bag – CoreLok device .......................... 38 
Figure 16 - Specific Gravity Bench .................................................................................. 38 
Figure 17 - Plane Section with Pores filled with Green Fluorescent Epoxy – RM 620 ... 41 
Figure 18 - Red Color Distribution with Image J – RM 620 ............................................ 42 
Figure 19 - Black and White Representation of the Plane Section – RM 620 .................. 42 
Figure 20 - Black and White Pixels Distribution – RM 620 ............................................. 43 
Figure 21 - Plane Section with Fluorescent Epoxy Filling Voids – RM1431 .................. 45 
Figure 22 - Setup of the Testing Apparatus ...................................................................... 48 
Figure 23 - Analytical View of the PFC Testing Apparatus ............................................. 49 
Figure 24 - Blueprint of the top plate................................................................................ 54 
Figure 25 - Verification of the hydraulic conductivity results .......................................... 57 
Figure 26 - Plunger unclogging a PFC core...................................................................... 59 
Figure 27 - Sediments in sink ........................................................................................... 59 
Figure 28 - Sediments in filter paper ................................................................................ 60 
Figure 29 - Core 1B – Relationship between Hydraulic Gradient and Flowrate .............. 64 
Figure 30 - Core 1B – Comparison of Hydraulic Conductivity and Head Gradient ........ 65 
Figure 31 - Core 3C – Limitations of the Quadratic Model at Low Flow Measurements 67 
Figure 32 - Core 2B Upside Down – Linear Relationship between Head Gradient and
Flux ................................................................................................................................... 69 
Figure 33 - Comparison of Quadratic and Linear model .................................................. 71 
Figure 34 - Actual Decrease of Hydraulic Conductivity .................................................. 73 
Figure 35 - Increase of Hydraulic Conductivity versus Normalized Extracted Mass ...... 77 
Figure 36 - Influence of Daily Traffic on Normalized Extracted Mass ............................ 78 
Figure 37 - Influence of Accumulated Sediments on K ................................................... 79 
Figure 38 - Isotropy of the Radial Flow – Test with KMnO4 ........................................... 81 
viii

Figure 39 - In-Situ Field Permeameter (Di Benedetto et al., 1996) .................................. 88 
Figure 40 - Constant Head – Automatic Permeameter (Di Benedetto et al., 1996) ......... 89 
Figure 41 - Determination of the EAV content according to Regimand et al. (2003) ...... 90 
Figure 43 - Core-lock air vacuum and the core sealed in plastic bag (Regimand et al.,
2004) ................................................................................................................................. 91 
Figure 42 - Sample and its plastic bag (Regimand et al., 2004) ...................................... 91 
Figure 44 - Measuring the weights of the core with and without the plastic bag
(Regimand et al., 2004) ..................................................................................................... 92 
Figure 45 - Loop 360 – Core 1C – September 07 - Face A .............................................. 96 
Figure 46 - Loop 360 – Core 1C – September 07 - Face B .............................................. 96 
Figure 47 - RM 1431 – Core 2C – September 07 - Face A .............................................. 96 
Figure 48 - RM 1431 – Core 2C – September 07 - Face B .............................................. 97 
Figure 49 - RM 620 – Core 3A – June 07 ........................................................................ 97 
Figure 50 - Top Plate & Top View ................................................................................... 98 
Figure 51 - Bottom Plate - Bottom View .......................................................................... 99 
Figure 52 - Side View of Top and Bottom Plates ........................................................... 100 
Figure 53 - Standpipe ...................................................................................................... 101 
Figure 54 - Plexiglas Box ............................................................................................... 102 
Figure 55 - Manometer Frame ........................................................................................ 103 
Figure 56 - Slanted Board ............................................................................................... 104 
Figure 57 - Head Gradient vs. Flowrate - 1A Top .......................................................... 109 
Figure 58 - Head Gradient vs. Flowrate - 2A Top .......................................................... 110 
Figure 59 - Head Gradient vs. Flowrate - 2B Top .......................................................... 110 
Figure 60 - Head Gradient vs. Flowrate - 3B Top .......................................................... 111 
Figure 61 - Head Gradient vs. Flowrate - 3C Top .......................................................... 111 
Figure 62 - Head Gradient vs. Flowrate - 1A Upside Down .......................................... 112 
Figure 63 - Head Gradient vs. Flowrate - 2A Upside Down .......................................... 112 
Figure 64 - Head Gradient vs. Flowrate - 2B Upside Down .......................................... 113 
Figure 65 - Head Gradient vs. Flowrate - 3B Upside Down .......................................... 113 
Figure 66 - Head Gradient vs. Flowrate - 3C Upside Down .......................................... 114 

1

1. OVERVIEW OF THE POROUS FRICTION COURSE
PROJECT
1.1. OBJECTIVES
Porous Friction Course or open-graded friction course (OGFC) is commonly known as
porous asphalt. The porous pavement is commonly used in Europe and the United States.
The pavement consists in a porous overlay allowing rainwater to flow down to the bottom
the overlay and then to drain on the edges of the pavement. The safety benefits from
Permeable Friction Course (PFC) are well known, but the porous structure of the asphalt
also may result into a significant reduction of pollutants associated with stormwater. Two
preliminary aspects compose this project. On one hand, a study has been conducted by
Christina Stanard on the in-situ monitoring of stormwater in terms of water quality and
runoff quantifications. For this purpose, two sites located on Loop 360 near Bull Creek in
Austin are being monitored. Another aspect of the project is the characterization of the
hydraulic properties of the porous asphalt. Determining the porosity and the hydraulic
conductivity of the pavement is motivated by two facts:
• The clogging phenomenon is persistent within PFC. Once the pores of the PFC
become plugged, then runoff will begin flowing on the surface of the pavement and
all the benefits, both in terms of safety and water quality, will be lost.
• The actual methods utilized to determine the hydraulic parameters are strongly
dependent on the vertical hydraulic conductivity and the assumed isotropy of the
flow. The hydraulic conductivity of samples of PFC is usually tested by constant or
falling head permeameter (Appendix A: Permeameters & Measuring EAV Contents).
EAV defines the effective air void content a porous media.
Facing these two facts, the laboratory experimentations are focused on determining
porosity and hydraulic conductivity of a series of cylindrical PFC samples extracted at
2

three different locations around Austin on March 14, 2007. The results of the experiments
will help us to determine:
• The rate at which pavement gets clogged. Another series of samples were extracted at
the same locations on February 27th, 2008 and will be tested using the same
procedure. A quantification of the clogging phenomenon will be established.
• The characteristics of the vertical and radial flow within the samples.

The purpose of determining the vertical and radial components of the flow within the
samples is to develop in a later study a hydraulic model predicting the movement of
water within the porous asphalt. At turn, this information will be useful in the evaluation
of the effects of cross slope, highway width, curves, and superelevations on the flow
within the pavement. This information will be valuable for supporting the development of
guidelines for the appropriate use of PFC.


1.2. LOCATIONS OF STUDY
The Texas Department of Transportation utilizes the benefits of porous asphalt in terms
of safety and water quality at sites around the Austin area. Six different sections of PFC
overlay have been installed on I-35 Southbound, 183 North, Loop 360 near Bull Creek,
RM 1431 and RM 620. The length of the sections varies approximately from a half-mile
to a couple of miles. Most sections were installed during the years 2004 and 2005 on the
highways. Figure 1 gives an overview of the location of the different porous asphalt
sections in the Austin area. The PFC sections are represented by the blue color while the
conventional pavement ones are represented in red and rose colors.

3


Figure 1 - Overview of the PFC locations in the Austin area.

From all these sections, three locations were chosen to be studied according to safety,
traffic factors and the compatibility with in-situ measurements. On March 14th, 2007 a
TxDOT crew drilled cylindrical samples at these three sites:
• Loop 360 near Bull Creek,
• RM 1431,
• RM 620.
At each site, four core samples were extracted. All samples were taken from between the
tire tracks on the roadway. Figure 2 shows each coring location.


4


Figure 2 -Location of the three extraction sites – Photo Courtesy of Google Earth


Some more information is given below for each site in order to visualize the situation and
the type of the road. Some traffic data is given and corresponds to a study headed by
TxDOT. The Annual Average Daily Traffic Counts (AADT) are comprised of a
mathematical computation where the counts are made over a given period-of-time and the
high and low counts are removed. ��The counts are done during the non-summer weeks;
excluding Fridays, Saturdays, Sundays, and Holidays. The AADT counts are what could
be expected during a normal workday of a given week.�� (Texas Department of
Transportation, 2005)

Site 1 – Northbound Loop 360 near Bull Creek
The first site is positioned at the following geographic coordinates:
5

⎩⎨

��
��
W
N
"03'4797
"22'2230
.
Figure 3 shows the exact location of the extracted samples on a map; cores were
extracted northbound just of the bridge over Bull Creek.


Figure 3 - Northbound Loop 360 near Bull Creek Extraction Site – Photo Courtesy
of Google Earth


The pavement was installed in October 2004. The statistics show an average daily traffic
of 45,000 vehicles per day for the last seven years between Spicewood Springs Rd. and
RM 2222. In 2005, the daily traffic reached 48,000 vehicles per day.
Loop 360 at this location consists of a two lane highway in each direction. The two roads
are separated by a large V- shape median.
6

The proximity of the in-situ monitoring site of the location to the extraction site is an
advantage because hydraulic properties determined from the extracted cores can be
compared to the hydrographs obtained from the runoff analysis.

Site 2 – Eastbound RM 1431

The second site is positioned with the following geographic coordinates:
⎩⎨

��
��
W
N
"20'5297
"00'3130
.
Figure 4 shows the exact location of the extracted samples on a map; cores were indeed
extracted eastbound near the Industrial park Hur of Cedar Park.


Figure 4 - Eastbound RM1431 Extraction Site – Photo Courtesy of Google Earth

The PFC was installed in February 2004. The statistics show an average daily traffic that
has been slightly increasing since 2001 from 14,000 to 18,200 in 2005 between Nameless
7

Road and Lime Creek Road on RM1431. Two lanes in each direction are separated by a
double yellow line at this site on RM1431.

Site 3 – Northbound RM 620

The third extraction site is positioned with the following geographic coordinates:
⎩⎨

��
��
W
N
"12'4397
"06'3030
.
Figure 5 shows the exact location of the extracted samples on a map; cores were
extracted northbound on RM 620 at the intersection with Connor Road.


Figure 5 - Northbound RM 620 Extraction Site – Photo Courtesy of Google Earth

The PFC overlay was installed in June 2004. The statistics show that the average daily
traffic has been fluctuating over the last 5 years of data (from 2001 to 2005). In 2001, the
traffic reached a count of 50,000 vehicles per day but was 38,000 vehicles per day in
8

2005. The section of RM620 corresponding to our site is the one between FM 734
(Parmer Lane) and Lake Creek Pkwy.Two lanes in each direction are separated by a
double yellow line at this site on RM620.

1.3. EXTRACTION OF CORES
Four cores were drilled and extracted for each location. The extraction was organized by
Gary Lantrip of TxDOT. The operation was realized under extreme precaution since
cores were drilled from between the tire track on the right lane of double lanes roads.
Two signalization trucks closed the right lane about 400 yards before the extraction site
and most locations were chosen after an intersection in order to avoid traffic accidents.

An asphalt engineering company was contracted by TxDOT for the day. Cylindrical
samples composed of different layers of asphalt were extracted using a drill press
attached to the truck��s bed. A core pin of inner diameter 6-inches was connected to the
drill press. Figure 6 was taken on RM 620, the large cylinder rotating is the core pin. On
the left to the pin, a 6 inch diameter hole stands open on the road after a first core was
drilled.

9


Figure 6 - Drill Press on RM 620 - March 14, 2007

The four extracted cylindrical cores at each location are principally composed of two
layers. The lower one is a 10-15 cm thick impervious asphalt layer which acts as a rigid
and impermeable base. The upper layer is a 5 cm thick porous asphalt layer. Figure
7shows two extracted cores on RM 620.


Figure 7 - Extracted Cores on RM 620 - March 14, 2007
The holes on the road were refilled with a quick pavement mixture, which consists in
standard impervious repair asphalt (Figure 8).
10


Figure 8 - Quick pavement filling holes on RM 620 - March 14, 2007
In order to conduct experiments on the layer of porous asphalt itself, the last operation
required separating the PFC overlay from the impervious layer. This separation was done
at the TxDOT Asphalt Laboratory in Cedar Park by Byron Kneipfel. Figure 9 shows the
operation of separation of the two layers using a circular saw.


Figure 9 - Separation of the overlay – April 09, 2007

11

All cores were stored in a cold room in order to preserve the properties of the binders. For
each site, the four cores were classified by alphabetical order:
• A and B would be tested as are for their porosity by the water displacement method
and for their hydraulic conductivity,
• C would be sliced in two halves and its porosity would be determined by the method
of image analysis,
• D would be preserved in the cold room for possible later experiments.



1.4. OUTLINE OF THESIS
In order to answer the defined objectives, the present thesis will develop and analyze the
results of porosity and hydraulic conductivity measurements conducted on the extracted
cores. In the first part, the present methods for determining the hydraulic properties of
porous asphalt but also the preponderant need of maintenance will be developed as a
summary of the scientific literature. The thesis will then focus on the porosity
experiments either by the water displacement or image analysis methods. An
interpretation of the results on porosity will be given at this point. In a third part, the
thesis will precisely look at the analytical and experimental characteristics of the
methodology for determining the hydraulic conductivity of the sample. The results on
hydraulic conductivities will be analyzed and verified: an interpretation of the results on
the general patterns but also for every site will conclude the analysis. A discussion on
potential uncertainties and recommendations will conclude this thesis.

12

2. POROUS ASPHALT AND HYDRAULIC PROPERTIES IN
LITERATURE



The widely-accepted advantages of PFC are the road safety improvements in wet
conditions as well as the noise reduction from the roadway. The safety benefits during
rain events include reduced hydroplaning, greater skid resistance at high speeds, less
spray and light reflection from the roadway due to the improved drainage, and therefore,
better visibility. In addition to these benefits, PFC has been found to reduce pollutant
concentrations in stormwater runoff (Pagotto et al., 2000).

These advantages come with greater initial and maintenance costs and sometimes shorter
service lives than conventional pavement. Cost-benefit analyses (van der Zwan, 1990)
have shown that even with a greater yearly maintenance cost and shorter service life, the
benefits outweigh the increased costs. Also, recent improvements in mix designs have
increased the expected service life of these types of pavements. Other commonly noted
disadvantages are reduced performance over time due to clogging of the pavement and
the winter maintenance requirements. The literature review focuses on the hydraulic
properties, water quality benefits and common maintenance practices for porous
pavements. The majority of the cited articles are from online databases or journals. The
search for articles focused on recent publications with relevant experiments and results or
discussion. Part of this chapter utilizes the ��State of Practice of Porous Friction Course
(2007)��, written in collaboration with Christina Stanard.

13

2.1. MEASURE OF POROSITY AND METHODS
As stated, the drainage capacity of PFC depends greatly on the porosity, or void content.
Studies on different mix designs are used to improve the durability and strength of new
mix designs, and also to decrease issues associated with clogging. Based on experiences
with mixes with different void contents, a study in Spain (Ruiz et al., 1990) found that
pavements with greater than 20% void content are more durable than ones with less than
20% void content.

Regimand et al. (2004) developed a method for evaluating air void contents for
compacted materials. This method determines the effective air void (EAV) content of a
porous sample which is a subset of the total void content. The EAV content includes the
voids that are accessible by water and other environmental fluids and excludes the
portion of voids that will not be reached by liquid during the compacted material��s use.
The EAV parameter is beneficial since it strongly correlates to mixture permeability. To
determine the EAV, a compacted material sample is encased in a sealant material of
known weight and air is evacuated from the encasement. The vacuum-sealed material is
weighed both in air and then under water. After weighing the sealed sample in water, the
seal is opened to allow water to contact the compacted sample. Since the volume of the
bag is non-negligible, the dimensions of both the sample and the sample in the bag should
also be recorded. With these recorded values, Equation 1 can be used to calculate the
EAV parameter.

Equation 1 100
2
12 ⋅⎟⎟⎠

⎜⎜⎝
⎛ −= ��
�Ѧ�EAV
where ��1 = the density of the vacuum sealed compacted material sample
14

��2 = the density of the vacuum sealed sample after opening the seal under water
Illustrations of the techniques and methods of measuring the EAV content of porous
asphalt can be seen in Appendix A: Permeameters & Measuring EAV Contents.

In 1999, researchers in Denmark used samples from test sections of porous asphalt
surfacing to conduct porosity measurements with image analysis (Bendsten et al., 2002).
The Danish Road Institute has been using the image analysis method since the early
1980s. This method of analyzing thin sections was developed to detect clogged pores
sooner than would otherwise be possible. Other benefits of image analysis besides
porosity and clogging of pores include the observation of the size of voids and their
distribution. The surfaces of the plane sections are impregnated with a florescent epoxy
that fills all of the void space in the section. Porous stones exposed in the thin sections
are marked with black ink to avoid any confusion in the analysis. The void content
number is determined by looking at the plane section under ultraviolet light. Bendsten et
al. (2002) obtained air void contents of 0.182 to 0.224 depending on the size of the
aggregates.


2.2. MEASURE OF HYDRAULIC CONDUCTIVITY
Permeability is a measure of a materials ability to transmit fluids. Measured values
for permeability, hydraulic conductivity, and infiltration rate vary widely in the literature.
While these terms are often used interchangeably, they do not represent the same
quantity. In this thesis, hydraulic conductivity is used as the permeability coefficient
which relates the volumetric flux and the hydraulic gradient. Its units are length per time.
15

Nearly every reported value is derived from a different test, thus preventing a direct
comparison. However, the test methods do share some similarity in that many utilize a
falling head apparatus to achieve a sense of the flow capacity of the pavement. The test
documents the time required for a certain amount of water to drain into the pavement,
which can be converted to a flow rate measurement. One problem with this test is that
the water surfaces around the apparatus rather than flowing through the pavement. Thus,
the data from such tests do not accurately represent a theoretical quantity.

Bear (1972) describes the methods commonly used in determining the hydraulic
conductivity of a porous media. The experiments consist of characterizing the unsteady
or steady flow in the vertical or horizontal direction through a cylindrical specimen with
an instrument called a permeameter. The two types of permeameters can be
distinguished.

The constant head permeameter applies a constant head loss ��H over a porous media of
height L, and cross-sectional area A. The measured discharge Q flows through the
sample, and the hydraulic conductivity is computed according to Equation 2:


HA
QLK ��=
Equation 2

Several runs are necessary in order to reduce the uncertainty. In France, the standardized
test consists in a constant head of 1.50 meters of water (Di Benedetto et al., 1996).

16

The falling head permeameter measures a percolation velocity (vp) through a specimen of
porous media of height L and cross section A. Initially, a standpipe of section a
constrains the sample to a fixed volume of water V. The time T for the water level to
drop in the standpipe from heights Hi to Hf is recorded and the hydraulic conductivity is
computed with Equation 3 (Terzaghi et al., 1996):

⎟⎟⎠

⎜⎜⎝
⎛=
f
i
H
H
AT
aLK ln
Equation 3


Experimental data is needed to calibrate both permeameters. Standardized diagrams for
these two types of permeameters can be seen in Appendix A.

Isenring et al. (1990) discuss the testing apparatus used to measure the drainage potential
of porous asphalt in Switzerland. The Institute for Transportation, Traffic, Highway and
Railway Engineering (IVT) of the Swiss Federal Institute of Technology developed the
��IVT permeameter��. A cylindrical tube with putty around the bottom is used to time how
long it takes for 2.27 L of water to flow into the porous pavement, or ��drainage
potential��. It is extremely similar to TxDOT��s cylindrical field permeameter, specified in
Tex-246 (TxDOT, 2004). The tested pavements had void contents ranging from 11 to
22% and layer thicknesses of 28 to 50 mm. The average measured discharge through this
new porous asphalt was 3.4 L/min.

Isenring et al. point out that a single point measurement does not represent the true flow
rate since the porous mixture is not homogeneous. Nonetheless, it does allow for
17

comparison of the porous asphalt over time. As expected, the project in Switzerland
found that the drainage potential decreased with time at all of the testing locations. The
greatest reduction in the results was found in the years directly following construction of
the overlay. The measurements from multiple sites of two types of porous asphalt
(maximum aggregate sizes of 10 mm and 16 mm) from this project are shown in Figure
10.

Figure 10 - Drainage potential over time (Isenring et al., 1990)

Ruiz et al. (1990) also perform similar testing on porous asphalt mixtures in Spain.
Using an ��LCS Drainometer��, the time to drain 1.735 L of water is recorded. This test
18

includes a large plate around the bottom of the tube in an effort to prevent the water from
surfacing. The drainage time (T) is related to the hydraulic head gradient (H) by
Equation 4, which was developed with laboratory tests.

Equation 4 305.0
6.58
T
H =

The drainometer test was conducted on porous mixes with void contents less than 20%
and mixes with void contents greater than 20%. The mixes with the lower void content
experienced a larger decrease in drainage capacity over time. Overlays with initial air
voids of 16% and heavy traffic had clogged up after only two years according to the drain
time. Other mixes totally clogged at different times up to nine years. Clogged overlays
are defined as having a drainage time from the LCS drainometer greater than 600
seconds. The mixes with the void content above 20% had initial drainage times in the
range of 15 to 25 seconds. These mixes had longer drainage times after many years of
service and did not totally clog or deteriorate. After 9 years, drainage times were still
under 300 seconds.

In Denmark, porous asphalt is characterized with a similar instrument called ��Becker��s
Tube�� as described by Bendtsen et al. (2002). Measurements taken immediately after
construction and over time, before and after cleaning, were compared to cleaning
guidelines for Dutch porous asphalt surfaces from Van Leest et al. (1997), which are
shown in Table 1. Once the outflow time gets too high, the porous asphalt is considered
clogged and cannot be flushed clean.


19



Table 1 - Cleaning guidelines from Van Leest et al. (1997)
Degree of Clogging of
Porous Asphalt
Outflow time
(seconds) Permeability Class
New 30 High
Partly Clogged 50 Medium
Clogged 75 Low

A study on pervious pavement in Germany by Stotz and Krauth (1994) evaluated the
drainage capacity of a highway section based on the percentage of rainfall that ran off the
highway. The porous asphalt was 40 mm thick with a porosity of 19%. Comparison of
values in the summer and winter revealed that the infiltration rate was approximately
50% larger in the winter than in the summer. Since Germany has cold and damp winters,
it is assumed that the greater conductivity is due to fewer losses from high temperatures
and dry pavements.

The Georgia DOT uses open-graded friction course as the overlay on all interstate
projects. Watson et al. (1998) compare the three different types of OGFC mix designs
that have been developed by GDOT: conventional, modified and European. The
conventional mix has a maximum aggregate size of 9.5 mm, while the modified and
European mixtures have a max aggregate size of 12.5 mm. The air voids percentage of
the mixes ranges from 10-20 %. A falling head permeameter is used by GDOT to
determine the hydraulic conductivity of the mixes. The conventional mix had the
smallest hydraulic conductivity of all of the mixes with an average value of 39 m/day
20

(0.045 cm/s). The European mix, with coarser gradation and largest thickness of 32 mm,
had the greatest drainage capacity of approximately 100 m/day (0.116 cm/s).

Birgisson et al. (2006) evaluated the use of OGFC in Florida. Field tests were performed
to analyze the latest PFC design. A falling head permeameter was used to measure the
hydraulic conductivity of test sections of porous pavement. Hydraulic conductivity tests
in and between the wheel paths gave values of 0.81 cm/s and 0.74 cm/s respectively. The
hydraulic conductivity was greater in the wheel path even only 2 months after
construction. Part of the test section was repaved with lower design asphalt content and
similar field tests were again performed two months later. The permeability of the PFC
increased overall, and the permeability between the wheel paths was greater with a value
of 1.27 cm/s.

Kandhal and Mallick (1999) compared four OGFC mixes with different gradations using
Florida DOT��s laboratory falling head permeameter. The results show that the mixes
with the smallest percentage of fine aggregates (4.75 mm) have the largest hydraulic
conductivity. The reported values are shown in Table 2. For comparison, the values
found in the literature are presented in Table 3.

Table 2 - Summary of conductivity data from Kandhal and Mallick (1999)

Gradation
(percent passing 4.75 mm
sieve)
Hydraulic Conductivity
(m/day)
Hydraulic
Conductivity
(cm/s)
15 117 0.135
25 88 0.102
30 28 0.032
40 21 0.024
21



Table 3 - Literature Values for PFC Characterization



2.3. LIMITATION OF DARCY��S LAW
In a porous media, the hydraulic conductivity K represents the specific discharge per unit
hydraulic gradient, which means that the coefficient depends on both matrix and fluid
Location Age of Pavement Flowrate
Hydraulic
Conductivity
(cm/s)
Void
Content
Layer
Thickness
Max
Agg.
Size
Source
Switzerland Initial 3.4 L/s - 11-22% 28-50 mm 10 mm
Isenring
et al.
1990
Spain Initial - - >20% 40 mm 10 mm
Ruiz et
al. 1990
Belgium Design Spec < 1.4 L/s - 19-25% 40 mm -
Van
Heystraet
en et al.
1990
Germany 3 years - 0.047 – 0.11 19% 40 mm
Stotz and
Krauth
1994
Netherland
s
Design
Spec - - > 20% 50 mm
11
mm
Van der
Zwan et
al. 1990
Georgia Design Spec - 0.116 10-20% 30 mm
12.5
mm
Watson
et al.
1998
Florida 2 months 1.2 1.4��
Birgisson
et al.
2006
Oregon Design Spec - - 50 mm
19
mm
Moore et
al. 2001
Florida Design Spec - 0.78 18-22% 32 mm
10
mm
Bjorn et
al. 2006
22

properties (Bear, 1972). From a dimensional analysis, the hydraulic conductivity can be
derived as (Nutting, 1930):

Equation 5 ��
gkK ⋅=

Where k is the intrinsic permeability, v the kinematic viscosity and g the gravity
acceleration. The intrinsic permeability is only a function of the matrix composing the
porous media and its characteristics such as grain size distribution, tortuosity and
porosity. For porous media, the Reynolds number (Re) can be defined as (Charbeneau,
2000):

Equation 6 ��
dq ⋅=Re
Where q is Darcy��s velocity and d is the average grain diameter or d10 of the size
distribution profile of the porous media. Experiments have shown that Darcy��s law
remains valid as long as the Reynolds number doesn��t exceed a value between 1 to 10
(Venkataraman, 1999). As the Reynolds number doesn��t exceed this value, the flow
remains laminar and is governed by viscous forces. However the inertial forces start to
govern the flow through the porous media in transitional flow when Re becomes higher
than the transition value. As the Reynolds number continues to increase, the flow
becomes turbulent at some point (Bear, 1972).

A quadratic model was suggested by Forchheimer (1901) in order to describe the inertial
effects in the relationship between the hydraulic gradient and the flowrate:


Equation 7
2~ QQH ⋅+⋅=��−=�� �¦�
23


Where �� and �� are experimental coefficients depending respectively on the properties of
the fluid and the matrix. Using a dimensional analysis, Ward (1964) derived analytical
expressions for both coefficients:


Equation 8 Kgk
1=⋅=
�ͦ�
And

Equation 9 gk
C
⋅=��
Where C is a characteristic constant of the porous media. The first term is attributed to
the viscous forces, or Darcy��s linear relationship. The second term quantifies the inertial
effects within the porous media. The characteristic length was reported by Ward as the
square root of the intrinsic permeability.

Another approach was based on drag forces (Rumer, 1966). The drag coefficient is a
function of both the Reynolds number and the shape of the grains. When the flow
becomes non-laminar, the drag resistance is not proportional to Darcy��s velocity
anymore.


2.4. MAINTENANCE OF POROUS ASPHALT
The service life of porous asphalt depends not only on the deterioration of the overlay,
but also the drainage capacity (Fwa et al., 1999). Only a few studies have been
24

conducted on the long-term behavior of porous pavements which are necessary to help
determine the functional life of PFC. Since the pores in the overlay collect particles and
are susceptible to clogging, some transportation authorities perform regular maintenance
on the porous overlays (FHWA, 2005). This section discusses the causes of deterioration
and commonly used maintenance procedures.
Causes of Deterioration
Graff (2006) discusses different aspects of asphalt pavement preservation, in particular
the most common causes of deterioration. Asphalt is made from by-products of refining
crude oil. Over the past few decades, overall asphalt quality has declined due to better
refining of crude oil. To offset the decreasing quality of asphalt, it is now common to
add materials to asphalt mixes in order to improve the properties of the asphalt.

According to Graff (2006), two main physical factors lead to the aging of asphalt:
ultraviolet (UV) light and heat. The surface or the chip seal of overlay ages first because
of their exposure to UV light and heat. Porous asphalt allows more light penetration,
which results in faster aging than regular asphalt. The high temperature that porous
asphalt is manufactured and placed under also contributes to a shorter service life.

All pavements are also subject to stresses that cause failure in the form of cracking or
raveling. Thermal expansion causes internal stresses in porous asphalt since the
coefficients of expansion are different for the aggregates and the asphalt. Traffic loading
and expansive soils also cause internal stresses. As the asphalt ages, it cannot handle
these stresses as well as new asphalt. Water in the pores of the pavement, combined with
25

traffic loading and temperature changes, also creates extreme internal pressures on the
pavement. (Graff, 2006)

Clogging
Since porous asphalt allows surface runoff to flow within it, particles in the runoff are
often collected by the pores of the pavement. The particles are generally sand particles or
debris released from tires (Fwa et al., 1999). The drainage capacity of the pavement can
be drastically affected by the clogging of the pores (Van Heystraeten and Moraux, 1990).
Therefore, clogging of the pores will decrease the benefits of wet weather traction and
noise reduction. Regular maintenance is often required to ensure adequate drainage
capabilities over time.

In many locations, porous asphalt is only used on high-speed roadways to help avoid
clogging problems. The tires push water into the voids and suck it back out as they drive
over the surface. At high speeds, this helps to clean the pores at the surface. For this
reason, it is not recommended to use porous asphalt on low-volume or slow-traffic
roadways (Van Heystraeten and Moraux, 1990). Also, less traffic results in more debris
on the roadway since there is not enough wind created by the cars to keep the roadway
clean (NCHRP, 2000).

Fwa et al. (1999) performed laboratory testing on porous asphalt samples to evaluate the
clogging potential. The testing involved manually clogging the porous samples with soil
and measuring the permeability of the sample throughout the clogging process. The
permeability coefficient (k) in this experiment was calculated with an equation based on
26

Darcy��s law. The results consistently showed that the permeability coefficient decreased
quickly in the beginning of the test and then asymptotically approached a terminal value,
as shown in Figure 11.

Figure 11 - Deterioration of Permeability Coefficient k (Fwa et al., 1999)

The curves for the deterioration of k are comparable with an average of 33 mm/s overall
reduction of k from the different initial values. This information could be used in design
to establish a required initial permeability depending on the expected terminal value.

According to NCHRP (2000), larger aggregate sizes are being specified since they create
larger air voids. Larger voids are less likely to clog because they are cleaned by the
pressure from traffic during rain events.

27

Cleaning Machines
Different techniques are used to clean porous pavement around the world. One major and
recurrent technique is the use of cleaning machines. The most common type of these
cleaning machines spray water at high-pressures into the overlay and then vacuum out the
resulting sludge. This process is referred to as ��captive hydrology��. According to
Newcomb and Scofield (2004), pressure cleaning is recommended on fine-graded open-
graded overlays in Europe once or twice a year. Bäckström and Bergstrom (2000)
recommend cleaning the porous asphalt every 2 to 4 years with this high-pressure water
cleaning.

The stormwater quality study in Israel (Pacific Water Resources, 2004) tested the
performance of a cleaning machine which used the ��captive hydrology�� technique. The
cleaning machine used in Israel was supplied by Netivey Hamifratz Ltd. (NH). It was
tested on its pick-up performance by cleaning a test area with a known weight of soil
applied over it. The NH machine was found to have very high pick-up performance of
99.7%.

A report from the Public Work Research Institute (PWRI, 2005) in Japan gives some
insight on porous asphalt cleaning machines. The first types of machines developed in
Japan, similar to ones in Europe, had to clean at very slow speeds (1 km/hr) and
attempted to fully recover the pavement. The newer machines can run at greater speeds
(10-20 km/hr) and are designed to be used more frequently to maintain the function of the
pavement.

28

One machine, the ��Spec-keeper��, sprays water into the pavement and creates high
pressure air around the cleaning area to push the water back out of the pavement with the
collected particles. The water and particles are then separated so the water can be reused.
A schematic of the Spec-keepers cleaning process is shown below in Figure 12 and a
picture of the machine is shown in Figure 13.


Figure 12 - The cleaning process of the Spec-keeper cleaning machine (PWRI, 2005)


Figure 13 - The Spec-Keeper cleaning machine in Japan from PWRI (2005)
29

Another machine uses high pressure air only to loosen the clogging particles. The
particles are then vacuumed up. The machine blows air from both sides of the pavement
and a vacuum in the middle collects the dirt and dust from the pavement. This machine
runs at an average speed of 20 km/hr. Experiments on the frequency of use of this
machine found that cleaning should be conducted as often as possible to maintain the
most effective function of the pavement. The machine is shown in Figure 14.


Figure 14 - Cleaning Machine with High Pressure Air only from PWRI (2005)
Upon comparison of the different types of machines, the original slow-speed machines
collected the most mass of particles per area and had to be used with the lowest
frequency. However, the cleaning costs were much greater. The high-pressure water and
air machines had lower efficiencies but were much more cost effective. An overview in
effectiveness and costs of the three types of cleaning machines is given in Table 4.




30

Table 4 - Cleaning Machine Summary from Japan at PWRI (2005)

Cleaning Machine
Collected
mass/cycle
(g/m2)
Frequency
(times/year)
Cleaning
Costs
($/m2)
Cleaning
Costs
($/m2/year)
Conventional slow speed 100 3 8.55 25.65
High speed with high-
pressure water 10 30 0.27 8.18
High-pressure air blow 6 50 0.10 4.96

A study was conducted by the Federal Highway Administration (FHWA, 2005) to learn
more about common practices with noise-reducing pavements in European countries that
have more experience with porous pavements. Different general and winter maintenance
methods are used according to the policy priorities and environment conditions. In most
countries, porous pavements are cleaned in order to maintain a certain acoustic ability
(noise reduction) affected by clogging. A summary of each country��s policy is given
below.

In Denmark, porous asphalt is cleaned with high pressure water spray (125 psi) followed
by vacuuming. Cleaning is performed three months after construction and then
semiannually. Experience at the Danish Road Institute (DRI) has found that the
pavement can become too clogged to clean effectively after two years if regular cleaning
is not performed.

In the Netherlands, porous pavements are also cleaned semiannually with a captive
hydrology cleaning machine. The Dutch have found that the noise reduction and
drainage benefits are reduced immediately after cleaning since the clogging material is
brought to the surface. However, after a short time these properties improve to an
31

unknown extent. Porous pavements are not used in urban areas because of the problem
with clogging and questioned effectiveness of cleaning.

Porous pavements in France are not cleaned as the French have found cleaning to be
ineffective. The mixes are designed to avoid clogging and the expected service life is
greater than ten years.

Fog Sealing
Fog sealing is a process where asphalt is sprayed onto an existing pavement surface. Fog
seals are used to replace asphalt that has deteriorated at the surface due to weathering
(NCHRP, 2000). They can also be used to stop the pavement from raveling or reduce
aggregate loss. Fog seals are applied by spraying dilute asphalt emulsion over the surface
of the pavement. The fresh application of asphalt can lengthen the service life of the
pavement and even seal small cracks in the pavement. However, applying too much
asphalt can result in slick pavement surfaces and tracking of excess asphalt (Caltrans,
2003). Fog seals reduce the air void content of open-graded pavements. Some
transportation authorities apply fog seals every 3-5 years as a part of surface
maintenance. (NCHRP, 2000)

Winter Conditions
Winter maintenance is a commonly noted disadvantage of porous asphalt. A few studies
on the performance of PFCs under winter conditions have been conducted. Bäckström
and Bergstrom (2000) studied drainage through porous asphalt in freezing temperatures
and snowmelt conditions. Laboratory experiments were performed in a climate room to
32

determine the infiltration rate through porous asphalt samples at cold temperatures. The
results indicated that the infiltration rate decreased significantly at freezing temperatures
and was nearly zero at -5ºC. To simulate snowmelt conditions, periods of freezing
temperatures were combined with rainfall. After a few days of these conditions, the
infiltration rate decreased to about 90% of the initial infiltration rate.

Shao et al. (1994) developed a model to predict the state of porous asphalt pavement
surfaces. The thermal properties of the porous overlay depend on the porosity of the
mixture. A higher porosity, or more air voids, creates a faster thermal response of the
asphalt to the ambient air temperatures since it is an open structure. The air in the pores
insulates the mixture from heat from the road sub-layer or ground. Water is retained in
the pores of the pavement after wet conditions and incoming heat from the sub-layer is
first consumed by evaporation of this water. Therefore, the surface of the pavement
reaches freezing temperatures faster and stays below freezing longer than conventional
pavements. Porous asphalt studies discussed by van der Zwan et al. (1990) also found
similar results. Experience in France (FHWA, 2005) found that porous pavements
reached freezing temperatures about 30 minutes before conventional pavements.

The ��Icebreak Model�� developed by Shao et al. (1994) was validated against actual
temperature measurements at a site in England. The model successfully predicted 90% of
the occurrences of below freezing temperatures at the pavement surface. Predicting
freezing temperatures of the porous road surface would allow highway authorities to react
in time and take appropriate actions. Therefore, road safety would not be compromised
in winter conditions.

33

The pores in the pavement also affect the application of salt for de-icing in the wintertime
since a portion of the salt will simply run into the pavement. This problem can be dealt
with by increasing the frequency of salt applications on the roadway and also using wet
salt instead of dry salt. (Van der Zwan et al., 1990)

A survey of DOT districts in Texas found that sand is the most commonly used material
even though it causes clogging of the pavement. Anti-icing chemical agents, instead of
de-icing agents, are the most effective winter maintenance procedure in these districts.
Pre-wetted salts and chemicals are effective if they are applied at the right time.
(Yildirim et al., 2006)

Pre-wetted salt is the most effective form of salt application for winter maintenance of
porous asphalt in Europe. This type of salt application will stick to the porous surface
instead of draining into the pavement and clogging the internal pores. This allows it to be
useful for longer periods of time. Many users expect that the winter maintenance on the
porous asphalt will require twice as much salt as the original dense-graded asphalt
(Newcomb, 2004).

Camomilla et al. (1990) advise using almost three times the amount of salt used for
conventional pavements to prevent icy conditions on roadways in Italy. Snowplow
efforts are most effective shortly after snowfall before the snow penetrates the pores.
Since the snowplows push some snow into the pores which can easily freeze, salt
application is especially important following the snowplows. Experiences in Oregon also
reveal that PFC should not be used in areas where snowplows are frequently used since it
often damages the surface (Moore et al. 2001).
34


In Denmark, the DRI uses a wetted-salt solution with calcium-chloride to allow an even
distribution of the solution. Experience has found that porous asphalt winter maintenance
increases the salt use by 50%. The DRI also recommends avoiding short-sections of
porous asphalt since the change in pavement type can startle drivers and make
maintenance changes difficult. (FHWA, 2005)

In Italy, highway runoff of salt brine is an environmental concern, and therefore, a
combination of magnesium and calcium is used for winter maintenance on the porous
pavements (FHWA, 2005).

The Switzerland summary of porous asphalt performance in winter conditions by
Isenring et al. (1990) found that porous asphalt did not behave worse than conventional
pavements. They concluded that any disadvantages of porous asphalt in the winter can
be avoided by intensive winter maintenance. The same conclusions were draw from
porous asphalt use in Belgium (Van Heystraeten and Moraux, 1990).

The hydraulic properties as well as the maintenance of Porous Friction Course in
literature were summarized in this chapter. The methods and fundamental equations of
hydraulic conductivity and porosity were reported. Limitations of Darcy��s law were also
detailed. The next chapter deals with the air void content characterization of the extracted
samples of porous asphalt.

35

3. LABORATORY MEASUREMENTS OF EFFECTIVE
POROSITY AND RESULTS
3.1. INTRODUCTION
Evaluating the effective porosity of the samples is an important objective of the project
because it quantifies the clogging phenomenon and verifies the homogeneous distribution
of effective pores between the samples. Surface runoff can appear in the case of major
clogging and both the safety for drivers and the water quality are impacted. Verifying the
good effective porosity of the porous pavement was indeed necessary.

The first hydraulic parameter which was measured from the cylindrical samples of PFC is
the porosity or air void content. The porosity can be seen as the water-holding potential
of a porous media and is defined as the volume of voids per bulk volume (Charbeneau,
2000):

Equation 10
t
v
V
V
volumebulk
voidsvolumen ==

Where Vv is the volume of voids and Vt is the total volume.
In our case, we were interested in characterizing the runoff within the pavement or the
flow through the pores. The samples of porous asphalt are indeed susceptible to contain
isolated air voids or clogged pores that do not contribute to the advection of water. The
effective porosity describes this phenomenon and was determined in the experiments.
Equation 11 defines the effective porosity:

Equation 11
samplePFCofvolumetotal
circulatetoablewaterofvolumene =
36

The effective porosity was measured using two different methods:
• the water displacement method
• the image analysis method as verification
For each location, two samples A and B were analyzed using the water displacement
method and one sample C was evaluated by the image analysis method. This chapter
deals with the explanation of both methods in terms of procedure, equipment and
mathematical solution. An interpretation of the results is given at the end.

3.2. WATER DISPLACEMENT METHOD
The method is commonly known and used in different asphalt laboratories as the
Method of Test for determining the specific gravity and unit weight of compacted
bituminous mixtures (Section 2.1). The water displacement method is accurate for open-
graded pavements containing more than 2.0% of open or interconnected air voids. Two
samples A & B for each location were investigated using the water displacement method
on April 11, 2007.

Equipment required
The cylindrical cores were tested at the TxDOT asphalt laboratory in Cedar Park. The
laboratory owns the required equipment:
• A specific gravity bench measures the volume of water displaced by the core plunged
in water. More detail on this device is given in the procedure explanation (Appendix
A: Permeameters & Measuring EAV Contents).
• A CoreLok apparatus, or automatic vacuum sealing apparatus for sealing the PFC
sample in plastic bags,
37

• Plastic bag in conformity with the vacuum sealing device in order to enclose the PFC
into them,
• An electronic ruler with a 1/100 millimeter precision,
• An electronic scale with a 1/10 gram precision.

Procedure and Mathematical Derivation
In a first step, the geometric characteristics of the cores were recorded using the
electronic ruler:
• The minimum and maximum values of the diameter were measured. The average
diameter was defined as Dcore.
• The minimum and maximum values of the thickness were measured. The average
thickness was defined as bcore.
The weight of the bag and core were recorded separately using the electronic scale as
Mbag and Mcore.

The plastic bags were of unique standardized dimensions: Length 30 cm and Width 22.5
cm. The assumption of a specific plastic density of ��plastic = 1 g/cm3 was taken and the
volume of the plastic bag was determined as in Equation 12:

Equation 12 ( )
bag
bag
bag
M
V ��=
In a second step, the CoreLok device and the specific gravity bench were successively
used. The principle of the CoreLok is to vacuum up the air from inside the asphalt
sample and seal this sample into a plastic bag. The plastic bag contours the shape of the
sample after a 2 minute run of the device. A picture of the CoreLok device is presented in
38

Figure 15. The user is enclosing the asphalt sample into a yellow plastic bag, specific to
the vacuum sealing device.

Figure 15 - Enclosing the Sample in the Plastic Bag – CoreLok device
The core sealed in the bag was then scaled on the specific gravity bench (Figure 16).


Figure 16 - Specific Gravity Bench
39

The specific gravity bench is composed of a tank filled with water at a standardized
temperature of 68��F or 20��C, a basket immersed in the water which will contain the
sample and a calibrated scale attached to the basket that weights the volume of water
displaced by the specimen. In this step, the core in the bag was weighed and the displaced
volume was recorded as M1, measured.

Finally, the third step consisted in weighing the volume of water displaced by the core
itself using the specific gravity bench. The displaced volume was recorded as: M2, measured.

The mathematical derivation is based on Archimedes Principle, which states that a body
immersed in a fluid experiences a buoyant force equal to the weight of the displaced
fluid. In this case, the measures carried out with the specific gravity bench conform to the
principle since samples were submerged.

The results of the second step (core sealed in plastic bag) are derived in the force balance
(Equation 13):

Equation 13 ( ) ( ) gMFwhereFgVVgMM measuredbagcorewatercorebag ⋅=+⋅+⋅=⋅+ ,111��

We define ��water = 0.9981 g/cm3 as the specific density of water at 20��C, F1 as the force
indirectly measured by the scale and g = 9.81 m/s2 as the gravity acceleration. The right
hand side is composed of the buoyant force and the force applied on the scale.

The results of the third step (core itself) are similarly derived in the following force
balance (Equation 14):

Equation 14 gMFwhereFgVgM measuredgrainswatercore ⋅=+⋅⋅=⋅ ,222��
40


F2 is the force indirectly measured by the scale and Vgrains is the volume of the grains and
binders composing the core. Taking into account the two force balances, the effective
porosity is derived as in Equation 15:

Equation 15
core
grains
e V
V
n −= 1

The same methodology was applied to the six investigated cores. This method was simple
to use since the required devices were available at the asphalt laboratory.



3.3. IMAGE ANALYSIS METHOD
This method is specifically used by the Danish Road Institute in Europe (DRI, 2005) and
allows looking at the air voids, grains and binders of a sample of porous asphalt at a
microscopic level. The principle of the method is to examine a plane section of the core
and analyze the void content of the surfacing, the size of the voids, their form and
distribution. The cylindrical core was previously sliced in half and pores and other
porosities in the matrix were filled by a fluorescent epoxy. A more detailed description of
the device is available in the literature review (Section 2.1).
In order to illustrate the procedure, the example of RM 620 is detailed. Looking at the
following pictures, it is interesting to note that binders have been removed from the plane
section.

Procedure and Analysis
The cores were sent to the National Petrographic Service laboratory located in Houston,
specialized in sample analysis and preparation of thin and polished sections for
41

petrographic studies. A first core from RM 620 was sent in May 2007 and the two others
from Loop 360 and RM 1431 in August 2007.

A core was sliced and pores and other porosities in the matrix were filled by a fluorescent
epoxy. A picture was taken with a camera in a dark room; a black light illuminated the
cores.

Using an image editor such as Image J or Adobe Photoshop, the picture was cropped to
the effective pores boundaries. In this example, Image J was used since it is freeware
intended for research purposes. The cropped Figure 17 shows how the plane section
appears: grains and binder connections are in dark colors and air voids come out in light
green color.


Figure 17 - Plane Section with Pores filled with Green Fluorescent Epoxy – RM 620

Further analysis was done with Image J such as the red color distribution. Image editors
provide the histogram of basic colors for a RGB (Red-Green-Blue) Standard picture. On
the red color histogram (Figure 18), the first peak corresponds to the grains and binders
of the pavement. The second peak mostly corresponds to the effective voids and other
porosities.

42


Figure 18 - Red Color Distribution with Image J – RM 620

Looking at the color distribution of the picture and at the contours of the pores on a
grayscale basis, a threshold was calibrated to delineate grains, binders and pores.
Considering that grains and binders are represented by the first 43 colors of the 256
composing the RGB system, the image was transformed into a black/white representation
using a threshold function. The result is shown in Figure 19; white areas correspond to
pores while dark areas represent grains and binders.


Figure 19 - Black and White Representation of the Plane Section – RM 620
The black and white picture was analyzed by its color histogram or distribution. The
porosity was defined as the ratio of the total number of white pixels to the total number of
pixels in the image. In the example, the figure has:
• 503504 black pixels,
• A total number of 631760 pixels.
43


The distribution of pixels is shown in Figure 20, the mode gives the number of black
pixels and the count gives the total number of pixels:


Figure 20 - Black and White Pixels Distribution – RM 620

The ratio of pixels implies an air void content of 20.30%. The same method was carried
out on Adobe Photoshop for similar results; we obtained a porosity number of 20.34%
One core for each location was tested using the image analysis method. A summary of
the results and their interpretation is in Section 3.4.



3.4. EFFECTIVE POROSITY RESULTS AND INTERPRETATION
When the PFC overlay was installed on the roads in 2004, the aggregate was designed to
maintain a porosity specification of about 20%. The results obtained by both the water
displacement and image analysis methods confirm the relative homogeneity of the
cylindrical PFC samples in terms of air void contents. Table 5 summarizes the different
calculated effective porosities.

44

Table 5 - Summary of Effective Porosity

Basic statistical numbers are given at every location:
• On Loop 360, the average effective porosity equals 21.59% with a standard deviation
of 1.21%.
• On RM 1431, the average effective porosity equals 21.55% with a standard deviation
of 1.41%.
• On RM 620, the average effective porosity equals 19.76% with a standard deviation
of 0.47%.

The weight and geometric characteristics of the cores are summarized in Appendix B:
Weight & Characteristics of Cores. The method of water displacement determines
accurately the effective porosity and is easy to realize as long as the required equipment
is available.

The method of image analysis carries advantages but also weaknesses with it. The main
advantage is the analysis of the plane section at the scale of a porous media. The structure
of the aggregate is revealed such as the grain size, the binders and the voids and a
description of the kinematical flow through the interconnected voids can be viewed.

Specific Gravity Image analysis
1A 22.78
1B 21.64
1C 20.36
2A 23.17
2B 20.51
2C 20.98
3A 20.3
3B 19.44
3C 19.55
Effective Porosity (%)
Loop 360
RM 1431
RM 620
Location Core
45


Figure 21 - Plane Section with Fluorescent Epoxy Filling Voids – RM1431
Observing Figure 21, the distributions of pores and grains are not homogeneous and the
radial flow through pores might not be isotropic. The analysis of the other images leads to
the same conclusions. In this image, the bottom layer corresponds to the impervious
pavement and the size of the pores is heterogeneous being larger than the grains for some
of them.

One main disadvantage is the technique of impregnating fluorescent epoxy through pores.
The hot epoxy is vacuumed through the pavement for several hours before cooling down.
Most of the binders are dissolved due to their hydrocarbon composition and the particles,
which were supposed to clog the media, are flushed out. This method wasn��t accurate to
calculate the defined effective porosity.

Another disadvantage is the requirement of one core, which will be sliced in half sections
just for the image analysis. The displacement method keeps the cores integral. Finally,
the defined threshold of 43 colors remains a subjective value, even if experiments were
coherent between the 3 cores examined.

46

The plane sections of the other pavement samples can be found in Appendix B: Weight
& Characteristics of Cores.

For later studies on porosity calculations, it would be recommended to only carry out the
water displacement method.

In this chapter, we have reviewed two different methods used to determine the effective
porosity of the cores. The first method, or water displacement method, is based on
Archimedes Principle and offers accurate results. The second one, or image analysis
method, interprets the grains and voids composing a plane section. Results were
mitigated by the drawbacks of the methods. Concerning the results, the aggregate
specifications of porosity were respected with a porosity evaluated at 20-21% according
to the site. The next chapter presents the laboratory experiments realized in order to
determine the hydraulic conductivity of the PFC cylindrical samples.

47

4. LABORATORY MEASUREMENTS OF HYDRAULIC
CONDUCTIVITY

In-situ rainwater drains down to the impervious surface and flows towards the edge of the
pavement: the flow through porous media has both vertical and radial components. The
design of the testing apparatus has attempted to comply with the in-situ flow
characteristics. The objective of this chapter is to introduce the analytical and
experimental methods carried out to determine the hydraulic conductivity from the
cylindrical samples of PFC. A method for assessing the clogging phenomenon is also
presented.

4.1. ANALYTICAL DESCRIPTION OF THE TESTING APPARATUS
The testing apparatus consists in an immersed sample of Porous Friction Course in a
large tank of water. Two metallic plates as well as rubber gaskets ensure the
impermeability of the top and bottom surfaces of the core. A peristaltic pump imposes a
constant head gradient between the levels in the standpipe and in the tank. Heads are
measured from measurement ports and lines by a manometer board and a bubbler
respectively. The configuration of the testing system is observable in Figure 22.

48


Figure 22 - Setup of the Testing Apparatus

The Porous Asphalt sample can be analytically interpreted as a cylindrical core with
radius R and thickness b. The flow during the test has vertical flow from a standpipe of
radius = a centered on the top of the sample and radial flow along the edges at radius R.
The established head = hA is uniform over the disk 0 < r < a, z = 0. The established head
= hB is also uniform over the radial periphery of the sample, r = R, 0 < z < b.

49


Figure 23 - Analytical View of the PFC Testing Apparatus

The goal is to derive an expression of the steady head in this porous media due to a
constant head gradient applied over the circular area. The steady-state continuity
equations along with associated boundary conditions are expressed as followed:

Equation 16 01 2
2
=∂
∂+⎟⎟⎠
⎞⎜⎜⎝
⎛ ⎟⎠
⎞⎜⎝





z
hK
r
hr
rr
K zr


Equation 17 ( ) [ ]0;0, =�ܡ�= zarhzrh A

Equation 18 ( ) [ ]0;0, =��<=∂
∂ zRra
z
zrh

Equation 19 ( ) [ ]bzRrhzrh B �ܡ�== 0;,

Equation 20 ( ) [ ]bzRr
z
zrh =�ܡ�=∂
∂ ;00,

50

Equation 16 – 20 present a well-posed boundary value problem whose solution h(r,z)
exists. The solution to this problem is analogous to heat transfer through a finite medium
due to heat supply over a circular area (Carslaw and Jaeger, 1959). The discharge through
the porous media is derived from the known solution of head distribution (Equation 21).

Equation 21
( ) ( )dz
r
zRhRKdr
z
rhrKQ
b
r
a
z �ҡ� ∂∂=∂∂= 00
,20,2 �Ц�

Due to the heterogeneous distribution of effective pores and the potential clogging of the
pavement, the flow rate is not expected to be uniform over either the inflow or outflow
surface.
We consider that the radial and vertical hydraulic conductivities are the same (for K = Kr
= Kz) since one controls the other. If both R and b are considered infinite compared to a,
the solution becomes (Equation 22):

Equation 22 ( ) ( ) ( ) ( ) ⎟



⎜⎜⎝

++++−
−= −
2222
1 2sin
2
,
zarzar
ahhzrh BA��
The discharge calculated using the first of Equation 21 is

Equation 23 ( )ahhKQ BA −= 4

Since R and b are not infinite, the effects of the finite vertical dimension are addressed
using the method of images in Equation 24. This gives:

Equation 24
( ) ( ) ( ) ( ) ( ) ( ) ⎟



⎜⎜⎝

++++++−
−= −
��
−��=
�� 22221 22
2sin
2
,
jbzarjbzar
ahhzrh
j
BA
��

51

In Equation 24, the image j = 0 corresponds to the basic solution given by Equation 22.
However the core has a thickness of b and the image j = -1 corresponds to a source disk
located a distance z = 2b below the surface (z = 0) making the surface z = b a no flow
boundary. The drawback of this image is the apparition of an upward gradient across the
z = 0 surface because of the imaginary part. The conjugate j=-1 attempts to cancel this
gradient. The same derivation happens infinitely between conjugate components at z =
4b, 6b��

The image solution satisfies Equation 16 and the boundary conditions from Equation 18
to Equation 20. The analogy with heat transfer doesn��t take into account the non-uniform
head on the disks 0 < r < a; z = 0 and r=R, 0<z<b disks because of the presence of grains
and open pores. The boundary conditions (17) and (19) are not intrinsically verified nor
mathematically satisfied.

The veracity of the images equation was tested and confirmed by quantitative measures
(Charbeneau, 2006). Charbeneau found that the finite effects of Equation 24
( ) ( ) ( ) ( ) ( ) ( ) ⎟



⎜⎜⎝

++++++−
−= −
��
−��=
�� 22221 22
2sin
2
,
jbzarjbzar
ahhzrh
j
BA
�� can be
represented using a shape factor. The shape factor corresponds to the geometric
characteristics a, b and R of the sample (Equation 25).


Equation 25
a
Q
R
b
R
aF
4
ˆ
, =⎟⎠
⎞⎜⎝


52

Q is the normalized discharge through the porous media. The magnitude of the shape
factor is close to unity. The head gradient can be related to the flowrate through the shape
factor (Equation 26).


Equation 26 ⎟⎠
⎞⎜⎝
⎛��=
R
b
R
aFhaKQ ,4
In the subsequent experiments, the shape factor will be determined by measuring the
geometric characteristics of the sample. The hydraulic conductivity will be evaluated by
monitoring the head gradient as well as the flowrate.



4.2. DESCRIPTION OF THE EQUIPMENT
In order to put in practice the analytical principle detailed in the last section, a measuring
system was designed and assembled together. The equipment is listed according to the
associated functionality and the purpose of the equipment is explained.

Tests
Different types of tests were carried out on the measuring device such as the importance
of clogging and the isotropy of the radial flow. The radial flow was tested using a
solution of potassium permanganate (KMnO4), coloring the water in the standpipe.

The clogging phenomenon was quantified by the mass extracted from the cores:
• Filter paper circles of characteristics: Ashless, Grade 40 and diameter 12.5cm. They
were used to collect and dry the extracted mass.
• A scale ��Scout Pro SP202�� of maximum capacity 200g and precision 0.01g. The
scale allowed weighing the extracted mass.
53

• A regular plunger of 4 inches diameter.

Measure of hydraulic heads
The hydraulic head was evaluated in the standpipe, at different radial positions and in the
box. In order to address this purpose, two devices were used:
• A Bubbler Flow Meter (Isco Reference 4230). The flowmeter line was connected to
the standpipe, converting the pressure necessary to force a bubble in the standpipe
into a level reading.
• An inclined manometer board with measuring tubes connected to the different
pressure tabs on the plates, the box and the standpipe. The manometer was built on a
similar basis as the manometer board designed for the Rain Simulator at the Center
for Research in Water Resources. Dimensions were changed to the purpose of our
project: a slanted board of dimensions 34.5 x 16 inches and a frame offering different
incline angles down to 5��. Blueprints and correspondence are available in Appendix
C: Experimental System – Blueprints.

Measure of flowrate
The flowrate was measured from the discharge hose of the peristaltic pump in use with:
• Standard laboratory measuring glassware
• A standard stopwatch.

Testing system
Following the testing system described analytically (Section 4.1), several components
were designed, built or ordered:
54

• Two boxes made with ½ inch thick Plexiglas plates. Both dimensions are of 23 inches
by side and 18 inches of height. There are two threaded holes where a faucet and a
pressure barb are attached. One box was used.
• Two metallic squared plates of side 12 inches side and thickness ½ inches. The plates
were designed to analyze 12 inches diameter cylindrical cores and ¼ inch holes were
threaded at 0, 1.375, 2, 2.625, 3.75 and 5.25 inches radius positions for pressure
measurements. On the top plate, a 2�� diameter open hole was drilled in order to insert
the acrylic standpipe within the plate. Figure 24 depicts the design of the top plates.
The locking system of the cylindrical PFC core was constructed with 3/8 inch holes
on the sides.

55


Figure 24 - Blueprint of the top plate.

• Two pieces of wood of dimensions L12x H 6x W1.5 inches constituting the support
of the system. The two pieces were waterproof treated. 3/8 inch holes were drilled at
the extremities and threaded rods were mounted in these holes. The locking system
would use bolts on the four corners, squeezing the material between the 2 plates.
• An acrylic standpipe of characteristics 2.50 OD x 2.00 ID inches.
• Several pieces of rubber gaskets maintaining the impermeability of the top and
bottom surfaces of the cylindrical cores. The rubber gaskets were ½ inch thick, and
supple. Holes were drilled at the pressure tabs radial positions and also for the
standpipe.
56

• Two peristaltic pumps working at different flowrates. The first peristaltic pump,
Heidolph PD5106, was acquired with a Single Channel pump Head and silicone
tubing (ID.250xOD.438 inches) according to technical criteria: low pressure, no
particle matter, self-priming, high-accuracy dispensing, flowrates fluctuating between
100 mL to 2 L/min. This pump was primarily used for higher flowrates. Another
pump, VWR peristaltic pump, was used for more accuracy within the low flow
measurements.

All the blueprints of the testing apparatus are presented in Appendix C: Experimental
System – Blueprints.



4.3. PROCEDURE OF THE CONSTANT HEAD TEST
This section summarizes the methodology followed when experiments were carried out.

Preparation of the measuring system
Tested cores were immersed in water overnight and shaken before being placed between
two pieces of rubber gaskets. The whole assembling was locked between the two metallic
plates using the locking bolts. The tank was completely filled with water and the
peristaltic pump was turned on for 30 minutes to evacuate the last air voids from the core.
The air in the connection between the manometer tubes and the devices was flushed out
using a hose. When ready, the tank was drained until the water level was about 5cm
above the top plate. Note: the torque applied on the system was as high as possible,
consistent with the structure of the porous asphalt core.

57

Run of the experiment
The hydraulic conductivity was measured by decreasing the flowrate. Between two
measures, the level on the bubbler was calibrated and set to zero. A hydraulic gradient
was imposed on top of the open core surface by turning on one of the pumps. The levels
were read on the manometer or bubbler once the levels were stabilized. The flowrate
delivered by the pump was then measured using the stopwatch and the laboratory
glassware. High flowrate measures were carried out with the Heidolph pump whereas
low flowrates were delivered by the VWR peristaltic pump. 12 to 18 different head
difference measures were reported for each core.



4.4. VERIFICATION TEST
Porous asphalt is subject to clogging. The main measure for hydraulic conductivity
considers that the vertical and radial hydraulic conductivities are equivalent since the
slowest velocity drives the behavior of the other velocity inside the pavement. Since the
cylindrical samples were extracted from pavement installed in 2004, the pavement might
be clogged on surface. An experiment was carried out in order to verify the veracity of
the results. Experimented cores were drilled on their center by a core pin of 1.25 inches
(1.59 cm) outer diameter at the TxDOT asphalt laboratory.

Comparing the situation to the one in (Section 4.1), we have a uniform established head
H1 imposed on the inner area of the cylinder at R= R1. The established head H2 is uniform
over the radial periphery of the sample at R= R2.

58


At steady state, we consider that the continuity equation is independent of the vertical
position and constrained to the following potentials (Equation 27):

Equation 27
⎪⎪


⎪⎪



==
==
=⎟⎠
⎞⎜⎝


∂⋅∂

22
11
)(
)(
0
HRrh
HRrh
r
hr
r

From the velocity of the flow and the cross-section of the cylindrical
sample brA ⋅⋅⋅= ��2 , we can derive the flow rate, which is considered
uniform over the inflow and outflow surfaces this time:

Equation 28
dr
dhKbrQ r ⋅⋅⋅⋅⋅=− ��2
Or

Equation 29
r
dr
Kb
Qdh
r
⋅⋅⋅⋅=− ��2

R1 R2
b
Figure 25 - Verification of the hydraulic conductivity results
r
hKq r ∂
∂⋅=
59

Integrating between the two boundaries of the cylindrical core, the hydraulic conductivity
is derived as a function of the flowrate and the geometric characteristics of the core:

Equation 30 ⎟⎟⎠

⎜⎜⎝
⎛⋅−⋅⋅⋅= 1
2
21
ln
)(2 R
R
HHb
QKr ��

As described in Section 4.3, the same experimental procedure was followed during the
verification tests.



4.5. CLEANING OF POROUS ASPHALT SAMPLES

Experiments were carried out in order to quantify the effects of clogging on the
hydraulic conductivity. The comparison was done on the drilled cores using the same
methodology.

The immersed cores were unclogged using a plunger on both top and bottom sides
(Figure 26).

60


Figure 26 - Plunger unclogging a PFC core
Letting the particulate matter to decant for a few minutes, the sink was drained out with a
high size cap. Accumulated sediments were gathered from the bottom of the tank using a
filter paper (Figure 27).


Figure 27 - Sediments in sink

61

The sediments were then dried in the filter paper contained in glassware. The weight of
collected sediments was obtained using the electronic scale after calibration.


Figure 28 - Sediments in filter paper
The hydraulic conductivity of the unclogged core was determined by running the PFC
experimental apparatus.

In this chapter, the experimental apparatus employed for the determination of the
hydraulic conductivity was presented. In both testing cases, simple equations of the
hydraulic conductivity were derived. Results are summarized and interpreted in the next
chapter, introducing the limitation of Darcy��s law.
62

5. ANALYSIS OF THE CONSTANT HEAD EXPERIMENT
RESULTS

Using the experimental system described in the previous chapter, the PFC samples
were evaluated under different conditions and positions. Due to the high velocity of the
radial flow within the pavement, the system operating region deals with the limitation of
Darcy��s Law: inertial effects appear to be consequent and are characterized within this
chapter. An estimate of the clogging phenomenon is also carried out.
It is also important to summarize the differences between run experiments in order to
understand the overall analysis:
• An initial test was run on the core 1B from Loop 360
• The five other cores were sliced on the bottom and characteristics were evaluated
from the standpipe sitting on top and upside down.
• The six cores were finally drilled on their center. Hydraulic conductivity was
evaluated before and after being cleaned up.


5.1. INERTIAL EFFECTS AND FORCHHEIMER MODEL
Empirical studies have shown the limitations of Darcy��s Law in characterizing the flow
within porous media at large velocities (Ward, 1964). In literature, the transition between
the laminar and turbulent flow through porous media is seen at low Reynolds numbers
(1<Re<10), (Charbeneau, 2000). Due to the geometric characteristics of the experimental
apparatus, the formulas derived in the literature for Re are not applicable. The Reynolds
number is then calculated from a non-unique method, allowing the author to have an idea
63

of Darcy��s velocity. An equivalent radius is defined as the radius at which the flow would
have covered half the volume of the core (Equation 31).


Equation 31
222 2)( erbarb ⋅⋅⋅=−⋅⋅ �Ц�
This leads to (Equation 32):

Equation 32 2
22 arre
−=

Where b is the height of the cylinder, r and a are the radiuses of the core and the
standpipe respectively.
Darcy��s velocity was therefore calculated at the re radial position (Equation 33):


Equation 33
br
Qq
e ⋅⋅⋅
= ��2

Where Q is the flowrate in cm3/s flowing through the PFC sample. Knowing Darcy��s
velocity, the associated Reynolds number is derived (Equation 34):


Equation 34 ��
10Re
dq ⋅=

Where d10 is the effective grain size of the aggregate, defined as the size at 10% of the
size distribution and �� is the kinematic viscosity of water at the recorded temperature.
For every core, the value of the Reynolds number was determined at each flow rate
condition. The equivalent radius was calculated according to (Equation 35):

64


Equation 35 cm
arre 15.52
52.153.7
2
2222
=−=−=

The effective grain size was obtained from the aggregate grain size distribution and
equals d1-20% = 4.75 mm (Appendix D: Porous Friction Course Aggregate - Water
Properties). Temperatures changed between experiments on a range of 17��C – 25��C and
the kinematic viscosity was determined from the Water properties table (Appendix D:
Porous Friction Course Aggregate - Water Properties). The geometric characteristics of
the cores have slightly changed between experiments but the range of associated Darcy
velocities and Reynolds number remain approximately the same:
• Darcy��s velocity, computed from Equation 33, fluctuates from .017 cm/s to 0.35
cm/s.
• The Reynolds number, derived from Equation 34, varies from 0.7 to 16.9.

Before analyzing any results on hydraulic conductivity, the Reynolds number shows that
the range of flows within the porous asphalt samples might be characterized as
transitional to turbulent.

At higher Reynolds numbers, the inertial effects are not negligible compared to the
viscous forces (Rumer, 1969). These patterns were observed on the following Gradient
(Head Difference) – Flowrate graph. Examples are focused on the sample 1B from Loop
360, representative of the observed trends.
65


Figure 29 - Core 1B – Relationship between Hydraulic Gradient and Flowrate

As the discharge of water through the PFC sample increases, the hydraulic gradient
increases more than proportionally to Darcy�� law. A quadratic model was applied in order
to describe the observed inertial effects (Forchheimer, 1901):


Equation 36 2QQH ⋅+⋅=�� �¦�

Where �� and �� are parameters defined by the quadratic model. In this case, the following
parameter values were determined: �� = 0.0477 and �� = 0.00281.
The hydraulic conductivity was computed from Darcy��s law:

Equation 26 ⎟⎠
⎞⎜⎝
⎛��=
R
b
R
aFhaKQ ,4

0.00
1.00
2.00
3.00
4.00
5.00
6.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
Q (cm3/s)
D
el
ta
H
(c
m
)
66

The effects of the convective terms in Navier-Stokes equation are also viewable on the
hydraulic conductivity versus head difference graphs (Figure 30). The higher the head
difference is, the lower the hydraulic conductivity is since the inertial effects increase the
rate of head loss along the characteristic flow line.


Figure 30 - Core 1B – Comparison of Hydraulic Conductivity and Head Gradient
The standard error of the model was determined from Equation 37 and appeared to be
very small in this case: 0.125 cm.


Equation 37 ( )��
=
��−��−=
N
j
quadraticmeaured HHN
SE
1
2
2
1
Where ��Hmeasured and ��Hquadratic are respectively the measured head gradient and the head
derived by the quadratic model. According to Ward��s approach (2.3), the first term of our
quadratic model corresponds to the Darcy��s linear law and one would anticipate:
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Head Difference (cm)
H
yd
ra
ul
ic
c
on
du
ct
iv
ity
(c
m
/s
)
67


Equation 38
⎟⎠
⎞⎜⎝
⎛=
R
b
R
aaKF ,4
1��
This gives:

Equation 39 scm
R
b
R
aaF
K /63.2
,4
1 =
⎟⎠
⎞⎜⎝
⎛= ��


The value K = 2.63 cm/s seems reasonable, especially given the scatter for small head
difference values seen on figure. One benefit from using a Forchheimer model remains in
the characterization of the deviation from Darcy��s linear law.

Equation 40 ( )QQQQH 0589.010477.01 +=⎟⎠
⎞⎜⎝
⎛ +⋅=�� ��
�¦�

This suggests that the model equation departs from the linear equation (Darcy��s law) by a
factor of 11 percent when the minimal discharge is Q = 1.89 cm3/s. Deviations are of
course much higher when the flowrate increases so that measurements within the linear
range would be extremely difficult to make.

The other five cores presented lower hydraulic conductivity and their bottoms were sliced
up in order to determine their degree of clogging. The same procedure was carried out
with the hydraulic gradient imposed either on the top or bottom surfaces.

There were fewer experimental points which induced discrepancies into the fit of a
Forchheimer model at low flow values. This trend can be seen in Figure 31, the plot
corresponding to the relationship between the hydraulic head gradient and the flowrate.
68

One can observe in blue the experimental points and the quadratic model is represented
by a red curve. The applied model doesn��t fit the experimental points as well for flow
rates lower than 4 cm3/s.


Figure 31 - Core 3C – Limitations of the Quadratic Model at Low Flow
Measurements

The hydraulic conductivity was calculated from the coefficient �� and the geometric
characteristics of the different cores. Table 6 summarizes the results. ��Top�� corresponds
to the tests where the standpipe was sitting on the top surface of the pavement. ��Bottom��
corresponds to the case where the sample was tested upside down.

0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
0.00 5.00 10.00 15.00 20.00 25.00
Flowrate (cm3/s)
H
yd
ra
ul
ic
G
ra
di
en
t (
cm
)
69

Table 6 - Forchheimer Model - Hydraulic Conductivity and Standard Error


Results are globally homogeneous within the samples extracted from RM 1431 (Index 2)
and RM 620 (Index 3). Hydraulic conductivities on RM 1431 range approximately from
0.5 to 0.6 cm/s; on RM 620 a typical value is 0.49 cm/s. Results between the two cores
from Loop 360 are disparate: one averaging 0.19 cm/s of hydraulic conductivity, the
other 2.63 cm/s. The difference can be interpreted as the importance of clogging in some
samples.

Disparities between the top and bottom are also observed but are examined in the
clogging phenomenon section. Parameters of the Forchheimer model are available in
Appendix E: Hydraulic Conductivity Model Parameters. Due to the discrepancies of the
quadratic model, the standard errors are higher for the five sliced cores, ranging from
0.19 cm to 0.63 cm. The same comment is made on the minimal deviation, ranging from
2% to 15.5%.

The advantage of a Forchheimer equation is the distinction of the linear terms (Darcy��s
law) and the inertial effects. In the next section, a linear model is applied to the low flux
experimental points and a comparison is carried out.
K (cm/s) SE (cm) K (cm/s) SE (cm)
1A 0.19 0.63 0.18 0.49
1B 2.63 0.13
2A 0.7 0.22 0.58 0.19
2B 0.32 0.37 0.64 0.29
3B 0.49 0.51 0.1 0.36
3C 0.49 0.36 0.48 0.26
Forchheimer Model
Top BottomCore
70

5.2. LINEAR MODEL: DARCY��S LAW
Scatter was consequent at low flow measurements but a linear relationship between the
hydraulic gradient and the flux of water seemed to be graphically consistent to these
points. The linear regression was computed on the subjectively chosen low flow points.
Figure 32 shows the linear relationship for the core 2B, tested upside down.


Figure 32 - Core 2B Upside Down – Linear Relationship between Head Gradient
and Flux
The linear correlation is derived with a correlation factor ��:

Equation 41 QH ⋅=�� ��

In this example, the coefficient �� = 0.2143 leads to a hydraulic conductivity of 0.61 cm/s.

Equation 42 scm
R
b
R
aaF
K /606.0
,4
1 =
⎟⎠
⎞⎜⎝
⎛= ��

0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
Q (cm3/s)
H
(c
m
)
71


The associated regression coefficient (R2) was 0.981 and the associated standard error
0.61cm appeared to be reasonable. The scatter of points was taken into account by
computing 95% confidence intervals of hydraulic conductivity. Results are summarized
in Table 7.

Table 7 - Summary of 95% Confidence Intervals of Hydraulic Conductivity


The 95% confidence intervals are still precise; the bandwidth of the confidence intervals
doesn��t go over 29% of the average hydraulic conductivity. The comparison of the
hydraulic conductivities obtained either by application of quadratic model or from
Darcy��s linear law is graphed in Figure 33.

Min Max Min Max Min Max
1A 0.21 0.27 0.2 0.24
1B 1.88 2.22
2A 0.57 0.7 0.49 0.6
2B 0.33 0.47 0.51 0.68
3B 0.3 0.37 0.08 0.09
3C 0.45 0.58 0.41 0.52
Darcy's Law -Hydraulic Conductivity - 95% Confidence Interval
Core
Initial Top Upside Down
72


Figure 33 - Comparison of Quadratic and Linear model

The hydraulic conductivity computed from Forchheimer��s model is represented by bright
yellow bars bordered by the low and high values of the intervals. Results of the quadratic
model globally fall within the range of values from the linear regression. However two
exceptions subsist in cores 1B and 3B where the quadratic model goes over the upper
value of the linear model respectively by 18% and 28%.

Until now, experiments were run on original cores, providing us the actual hydraulic
potential of the porous asphalt of the three different roadways. In the following section,
the PFC samples were manipulated in order to characterize the clogging phenomenon.



0 0.5 1 1.5 2 2.5 3
1A
1B
2A
2B
3B
3C
C
or
e
Hydraulic Conductivity (cm/s)
DarcyMax
Forchheimer
Darcy Min
73

5.3. CHARACTERIZING THE CLOGGING PHENOMENON
Clogging remains one of the main issues with porous asphalt, since rubber particles and
sediments change the hydraulic characteristics of the pavement (Section 2.4). The goal of
this section is to understand the effects of clogging in terms of hydraulics. Note that the
hydraulic conductivity obtained from the quadratic model is used in the following cases.
The variability of results between the top and bottom experiments can be associated with
the distribution of the grain size and no consistent trend appears.

Actual Rate of Clogging
The six cores were drilled and cleaned using the technique described in the methods
section (Section 4.5). The hydraulic conductivity of the drilled cores was determined
using Equation 39 from Forchheimer��s model. The computed value of K may be
corresponding to the original hydraulic conductivity of the porous asphalt when the
pavement was installed in 2004. This assumption is limited by the fact that porous asphalt
can remain permanently clogged (Section 4.5).

A comparison of the original and actual values of K provides explanation for the rate of
clogging that porous asphalt is subject to at the different studied locations.
74


Figure 34 - Actual Decrease of Hydraulic Conductivity

Looking at Figure 34, the general trend is a higher hydraulic conductivity of the original
cores. Only core 1B doesn��t follow the tendency and observes a decrease from 2.6 cm/s
to 1.7 cm/s. In the actual state, we have compared to the original hydraulic conductivities:
• A decrease of 50% in hydraulic conductivity on Loop 360; the original K value could
be set as 1.5 cm/s, note that in this case one sample (1B) was not clogged.
• A decrease of 80% in hydraulic conductivity on RM 1431; the original K value could
be set as 2.5 cm/s.
• A decrease of 50% in hydraulic conductivity on RM 620; the original K value could
be set as 1.0 cm/s.

-90.0%
-60.0%
-30.0%
0.0%
30.0%
60.0%
90.0%
0
0.5
1
1.5
2
2.5
3
1A 1B 2A 2B 3B 3C
Va
ri
at
oi
n
(%
)
Hy
dr
au
lic
C
on
du
ct
iv
ity
(
cm
/s
)
Actual
Original
Variation (%)
360 1431 620
75

In general, the decrease in hydraulic conductivity reaches approximately 50% after 3
years of utilization besides on RM 1431 where particular conditions, such as the
proximity to a large construction site, have contributed to a decrease estimated in average
at 80%.

The pattern of 1B could also be explained by the fact that the core has been tested more
than twelve times since it was the pilot core. The structure of the core could have been
stressed by the torque applied by the measuring system: this would have led to the
deformation of the matrix of grains and de facto to a change in hydraulic properties.

The actual decrease in hydraulic conductivity is induced by the presence of material
clogging the voids of the samples. The volume occupied by the material leads to a
decrease in porosity of the sample. Fair and Hatch (1933) derived the hydraulic
conductivity as a function of the fluid properties and the size and packing of the porous
medium (Equation 43).

Equation 43
( )
12
2
3
1

⎥⎥⎦

⎢⎢⎣

⎟⎟⎠

⎜⎜⎝
⎛⋅⋅−⋅= ��i mid
f
n
n
ngK ����


Where �� is a shape factor describing the geometry of the grains, fi the fraction of
sediments held between adjacent sieves, and dm is the geometric mean of the rate size of
adjacent sieves. Since the fluid properties and the grain size of the samples have not
changed between the two experiments, the change in hydraulic conductivity is derived as
a function of the original 1 and actual 2 porosity numbers (Equation 44).

76

Equation 44
( )
2
1
2
2
3
2
3
1
2
1
)1(
1
n
n
n
n
K
K

−⋅=


The original porosity number was computed from Equation 44 and results are
summarized in Table 8. The original air void content was taken as the one calculated
using the water displacement method (Table 5).

Table 8 - Extrapolated Air void content

Porosity has sensibly increased for all the samples except 1B. The pattern of 1B was
explained above and corresponds to the deformation of the grain matrix. The porosity has
decreased by 0.07 in average in comparison to the 50% average decrease in hydraulic
conductivity. The removed grains have a major effect on the increase of hydraulic
conductivity since sand clogs critical throats of the porous medium. In the actual state,
the porosity has decreased by:
• 0.04 on Loop 360 corresponding to a 15% clogging of the air void content
• 0.08 on RM 1431 corresponding to a 37% clogging of the air void content
• 0.04 on RM 620 corresponding to a 19% clogging of the air void content
The decrease in porosity is also in correlation with the amount of sediments extracted
from the cores during the cleaning operation (Table 9). The decrease of hydraulic
conductivity can also be associated with the effect of drilling as well as the potential
deformation of the samples.
K (cm/s) n K (cm/s) n
1A 0.19 0.131 1.26 0.228
1B 2.63 0.243 1.72 0.216
2A 0.7 0.169 2.11 0.232
2B 0.32 0.108 2.73 0.205
3B 0.49 0.154 1.08 0.194
3C 0.49 0.160 0.97 0.196
Core
Extrapolated Actual Original
77

Factors & Consequences of clogging
During the cleaning operation, the extracted mass was conserved and weighed. The
residual was principally composed of sediments such as sand grains. This section focuses
on the extracted sediments and their impact on hydraulic conductivity, as well as the
effect of traffic on the volume of sediments.

The extracted mass was recorded for all cores besides one from Loop 360 (1B). Results
are reported in Table 9:

Table 9 - Mass of Extracted Sediments

The mass of sediments ranges from 2.9 g to 13.8 g except for core 2B whose extracted
mass reached 27.2 g.

In the following development, the extracted mass was normalized by the volume of the
core in question. The first focus was on quantifying the increase in hydraulic conductivity
induced by extracted mass. The K value was determined for each drilled core before and
after the cleaning operation. The increase of K value was reported as a percentage of K
value associated to the non-cleaned drilled core (Figure 35).

Core Mass extracted (g)
1A 13.79
1B
2A 9.45
2B 27.82
3B 4.31
3C 2.86
78


Figure 35 - Increase of Hydraulic Conductivity versus Normalized Extracted Mass

The scatter of points is important and no specific relationship can be established from the
experimental data. One could at least note that the more sediment removed, the higher the
increase in hydraulic conductivity.

The second interest is the impact of traffic density on the mass recovered from the PFC
samples. Once again in Figure 36, the extracted mass was normalized by volume and the
daily traffic was reported in thousand of cars.
y = 6306.4x + 267.14
R2 = 0.3938
0
100
200
300
400
500
600
700
0.000 0.010 0.020 0.030 0.040 0.050 0.060
g/cm3
(%
)
79


Figure 36 - Influence of Daily Traffic on Normalized Extracted Mass

The highest normalized mass was removed from RM 1431 even if the daily traffic
(17,000 cars/day) is lighter than on the two other highways. This pattern was explained
earlier and could be attributed to the proximity of a large construction site. The porous
asphalt texture is also damaged at some locations.

There is no general trend appearing from the traffic density. The discrepancy of results
between Loop 360 and RM 620 might be explained by the type of traffic and not the load.
The porous asphalt samples on RM 620 were extracted in proximity to a large shopping
mall and a large intersection which may induce a permanent clogging of the pavement.

Normalized Extracted mass = F(Daily Traffic)
0.000
0.010
0.020
0.030
0.040
0.050
0.060
0 10 20 30 40 50
(Kcars/day)
(g
/c
m
3)
Loop 360
RM 1431
RM 620
80

A first evaluation of influence of sediments on permeability was given in Section 2.2
(Fwa et al., 1999). Figure 11 described the deterioration of permeability versus the mass
of clogging soil added. In a similar way, Figure 37 describes the decrease of hydraulic
conductivity versus the normalized accumulated mass for the six experimented cores.


Figure 37 - Influence of Accumulated Sediments on K

At each location, the slopes of K reduction are subjects to variations but the trends
between both cores of the same roadway concord. In analogy to Fwa (1999) and looking
at the original (cleaned) hydraulic conductivity, the aggregate has a different matrix
structure at the three locations. Results of Figure 37 will be implemented with a new set
of experiments in the future. The implemented results will lead to the identification of a
trend at the different locations.

0
0.5
1
1.5
2
2.5
3
0.000 0.010 0.020 0.030 0.040 0.050 0.060
H
yd
ra
ul
ic
C
on
du
ct
iv
ity
(c
m
/s
)
Normalized accumulated mass (g/cm3)
1A
1B
2A
2B
3B
3C
81

Influence of Drilling
Experimental cores were successively drilled from the roadway, sliced up on their bottom
and finally drilled on their center. This part studies the contribution of drilling to the
change of hydraulic conductivity. Table 10 summarizes the hydraulic conductivities
calculated initially from the plain cores and from the drilled cores before cleaning.

Table 10 - Hydraulic Conductivity of Initial and Non-Cleaned Drilled Cores


Thinner cores of volume lower than 470 cm3, such as the ones from RM 620, were
subject to a 30% decrease in hydraulic conductivity. However thicker cores with a higher
volume (higher than 520 cm3) were subject to a 10% increase in hydraulic conductivity.
This pattern could be explained by the principle of testing apparatus: the flow is only
radial in the drilled core experiment and doesn��t have any vertical component. The fact
that thinner cores get clogged could be related to the low number of effective pores
participating to the radial flow.
Note that the trend of core 1B was not taken into account since the core could have
endured a deformation of its structure. Since no particular mathematical trend has been
discovered, it is unrealistic to differentiate the real contribution of clogging to the drilling
effect.


K (cm/s) Top Initial Drilled Initial
1A 0.19 0.21
1B 2.63 1.13
2A 0.7 0.86
2B 0.32 0.38
3B 0.49 0.19
3C 0.49 0.23
82

5.4. RADIAL FLOW DISTRIBUTION
Characterizing the radial flow distribution was carried out at different levels. The first
level was the use of a solution of Potassium Permanganate in order to verify the isotropy
of the radial flow. Using a visual criterion, the dye seemed to flow out equally around the
circumference of the core, as observed in Figure 38.

Figure 38 - Isotropy of the Radial Flow – Test with KMnO4

The last level of study was the determination of head contours. The head was measured at
different radial positions of the core by an inclined manometer whose inclination was set
to 17�� or 23��. Experimental runs were performed at different angular positions (0��, 20��
and 180��) on core 1B, whose results are summarized in Figure 29. There is no significant
discrepancy in head gradient between the experiments and proof is that the quadratic
model has a small associated standard error (Equation 37).
83

Results of head contours were limited by the head gradients measured by the manometer:
at some radial positions, the head gradient would be negative. The relatively low velocity
of the flow through the pavement dismisses the theory of a Venturi effect. In order to
keep the accuracy of the study, head contours are not reported in this thesis and possible
improvements are discussed in Section 6.3.

The geometry of the experimental system induces a transitional to laminar flow through
the tested asphalt core. Two models were used to describe the evolution of the head
gradient as a function of the flux: a quadratic model taking into account the inertial
effects and a Darcy��s linear model on the low flow points. Results of hydraulic
conductivity have revealed the importance of clogging and the influence of the highway
environment. Hydraulic properties of the pavement at the three locations are summarized
in the next chapter.


84

6. CONCLUSIONS

Different experiments and methods were used to determine the hydraulic properties of
samples of porous asphalt. PFC cores were extracted on March 14, 2007. The porosity
was evaluated either by analyzing the image of a half core filled with fluorescent epoxy
or by the water displacement method. In addition, a testing apparatus was designed and
built to determine the hydraulic conductivity of the samples.


6.1. GENERAL TREND
In terms of porosity, the aggregate was designed with an effective porosity of about 20%.
The measured air void contents confirm the specifications after 3 years of utilization.
Results were relatively homogeneous between samples. The image analysis method has
to advantage of revealing the structure of the aggregate: grain and void sizes,
interpretation of the kinematical flow through the interconnected voids can be observed.
However it is recommended in the future to use the water displacement method for the
following reasons:
• the water displacement method remains relatively accurate,
• the image analysis method monopolizes a core, unusable after operation,
• the preparation of the cores during the image analysis method requires the vacuum of
hot epoxy through the pores: binders are dissolved and sediments flushed out.

In terms of hydraulic conductivity, results have shown an actual potential of about 0.5 –
0.6 cm/s besides on Loop 360 where results were disparate. Since the Reynolds numbers
were high during the experiments, inertial effects were interfering with the viscous forces
85

and two models were fit to the experimental points. A quadratic model or Forchheimer
model takes both effects into account; the standard errors associated to the models were
reasonable fluctuating from 0.19 to 0.63 cm. A linear model or Darcy��s model was fitted
to the low flow points. The standard errors are acceptable (about 0.7cm) and the
bandwidth of the confidence intervals didn��t go over 29% of the hydraulic conductivity.

The clogging phenomenon is a reality and was observed and quantified during the
experiments. The original hydraulic conductivity was not estimated since no data was
available at this point but values are reported compared to the definitive clogging of the
pavement. After three years of utilization, the decrease in hydraulic conductivity averages
50% and the decrease of air void content averages 0.04. These trends are of course
subjects to variations according to the locations. The more sediment removed, the higher
the hydraulic conductivity but no specific correlation of clogging versus traffic density
was observed. Finally, the drilling operation influences the hydraulic conductivity
according to the size of the cylindrical sample.


6.2. SPECIFIC LOCATIONS
On Loop 360, the air void content was reported to be 21.6% and the actual hydraulic
conductivity fluctuating between 0.2 cm/s and 2.6 cm/s. There is indeed an important
discrepancy between the two samples. Estimating the clogging, the highway seemed to
have a decrease of 50% in hydraulic conductivity and 15% of the effective pores are
clogged. Note that results from the pilot core have changed after the drilling operation,
changes attributed to the deformation of the sample matrix.

86

On RM 1431, the porosity was evaluated to be 21.6% and the actual hydraulic
conductivity fluctuating between 0.5 cm/s and 0.6 cm/s. The pavement has actually a
reduction estimated at 80% of its hydraulic conductivity and at 37% of its porosity. The
large percentage is tentatively explained by the presence of a large construction site at
proximity of the extraction site: the sediments removed were much higher.

On RM 620, the effective porosity was reported to be 19.8% and the actual hydraulic
conductivity 0.49 cm/s. The actual reduction in hydraulic conductivity was estimated at
50% and the clogged air void content was estimated at 19%. The highway seemed more
clogged than the other roads since the maximum definitive hydraulic conductivity is
estimated at 1.0 cm/s. The higher clogging of this road could be correlated with the
proximity of a large intersection and a shopping mall inducing more debris in the
pavement.


6.3. LIMITATIONS
The measure of air void content didn��t encounter any major difficulties but the one of
hydraulic conductivity did. The interpretation of hydraulic conductivity was limited at
two levels.

The first limitation was the domain of operation of the testing apparatus. The dimensions
of the standpipe and the cylindrical samples were respectively too large and too small
inducing a high Darcy��s velocity within the pavement. The inertial effects were
consequent, associating variations to the determination of the hydraulic permeability. At
this level, two actions are possible in the future in order to minimize the collateral effects
of a turbulent flow through the samples: the inner radius of the standpipe a has to
87

decrease and the radius of the pavement R has to increase; a second set of cores should
help us to determine accurately the intrinsic permeability of the pavement.

The second limitation was the measure of the head of water with the inclined manometer
board. At some radial positions, the head gradient was reported to be negative. The radial
measures would have helped to quantify and characterize the inertial effects within the
pavement as well as the contours would have confirmed the developed theory. This
pattern was observed with every core at low flow measurements. The manometer board
was changed twice and rebuilt one time, the connectors were tested and the air was
always flushed out. The problem seemed to occur at the interface between the core and
the rubber gasket.


6.4. FUTURE WORK
A new set of 8 inches cores was extracted on February 27th, 2008. The same procedure
will be carried out on these cores for determining their porosity and hydraulic
conductivity. The larger radius will help to dampen the inertial effects within the cores.
The comparison of results of this set with the present set will help to determine the
evolution of clogging in time as well as status on the hydraulic properties of the
pavement.

Another axis of the project on PFC is the correlation between laboratory results and in-
situ measurements. An in-situ falling head permeameter is used on site and the hydraulic
conductivity will be determined by comparison with the hydraulic characteristics
evaluated in laboratory. This constitutes an easy and efficient way to determine directly
the hydraulic properties of Porous Friction Course in Austin.
88


The final axis of the project is the development of a numerical model of flow within
porous asphalt whose parameters will be based on the hydraulic properties evaluated
through these experiments.


89

APPENDICES

APPENDIX A: PERMEAMETERS & MEASURING EAV CONTENTS


Figure 39 - In-Situ Field Permeameter (Di Benedetto et al., 1996)


90



Figure 40 - Constant Head – Automatic Permeameter (Di Benedetto et al., 1996)



91


Figure 41 - Determination of the EAV content according to Regimand et al. (2003)

92



Figure 43 - Core-lock air vacuum and the core sealed in plastic bag (Regimand et
al., 2004)
Figure 42 - Sample and its plastic bag (Regimand et al., 2004)
93







Figure 44 - Measuring the weights of the core with and without the plastic bag
(Regimand et al., 2004)






94

APPENDIX B: WEIGHT & CHARACTERISTICS OF CORES


Table 11 - Weight & Geometric Characteristics of Cores – Sliced Cores



Volume (cm3) Shape Factor
Sample min max avg min max avg
1A 22.58 35.24 28.91 150.34 151.01 150.675 515.5 0.992
2A 24.92 33.17 29.045 150.57 150.6 150.585 517.3 0.994
2B 28.46 31.95 30.205 150.35 150.41 150.38 536.5 1.01
3B 21.46 25.49 23.475 150.89 150.94 150.915 419.9 0.904
3C 27.66 24.29 25.975 151.69 151.69 151.69 469.4 0.946
Thickness (mm) Diameter (mm)
95

Table 12 - Weight & Geometric Characteristics of Cores – Original Cores
96

B
ag
W
t.
D
ry
W
t.
W
at
er
W
t.
w
ith
B
ag
G
a
W
at
er
W
t.
w
ith
ou
t b
ag
To
ta
l V
ol
um
e
of

th
e
co
re
V
ol
um
e
of

th
e
gr
ai
ns
P
or
os
ity
S
am
pl
e
m
in
m
ax
av
g
m
in
m
ax
av
g
(g
ra
m
s)
(g
ra
m
s)
(g
ra
m
s)
??
??
(g
ra
m
s)
(c
m
3)
(c
m
3)
43
.4
45
.6
44
.5
15
0.
21
15
0.
31
15
0.
26
30
.1
13
56
.7
60
9.
3
1.
83
5
77
9.
5
74
8.
79
57
8.
24
0.
22
8
35
.2
5
39
.5
2
37
.3
85
15
0.
72
15
0.
85
15
0.
78
5
30
.4
11
62
.1
53
1.
5
1.
86
7
66
7.
9
63
1.
79
49
5.
09
0.
21
6
31
.1
40
.0
2
35
.5
6
15
0.
14
15
0.
52
15
0.
33
30
.1
10
97
.9
48
7.
8
1.
82
3
62
9.
1
61
1.
25
46
9.
64
0.
23
2
39
.6
6
41
.9
3
40
.7
95
15
0.
15
15
0.
71
15
0.
43
30
.4
12
38
.8
56
1.
8
1.
85
2
70
0.
6
67
8.
27
53
9.
17
0.
20
5
37
.8
8
42
.4
4
40
.1
6
15
0.
27
15
1.
17
15
0.
72
30
.2
13
02
.9
61
2.
7
1.
91
74
6.
8
69
1.
49
55
7.
10
0.
19
4
38
.2
6
40
.7
9
39
.5
25
15
0.
69
15
1.
08
15
0.
88
5
30
.9
12
73
.5
59
9.
5
1.
91
3
73
1.
2
67
5.
26
54
3.
27
0.
19
5
��

C
)
20
te
r (
g/
cm
3)

0.
99
82
1
m
en
si
on
s
of

e
ba
g
B
ag

W
t.(
gr
)
Le
ng
th
(c
m
)
W
id
th
(c
m
)
Th
ic
kn
es
s
of

on
e
la
ye
r
(m
m
)
V
ol
um
e
of

th
e
ba
g
(c
m
3)
30
.1
30
22
.5
0.
22
3
30
.1
30
.4
30
22
.5
0.
22
5
30
.4
30
.1
30
22
.5
0.
22
3
30
.1
30
.4
30
22
.5
0.
22
5
30
.4
30
.2
30
22
.5
0.
22
4
30
.2
30
.9
30
22
.5
0.
22
9
30
.9
te
:
w
e
co
ns
id
er
th
at
th
e
pl
as
tic
b
ag
h
as
a
d
en
si
ty
o
f 1
g/
cm
3.
Th
ic
kn
es
s
(m
m
)
D
ia
m
et
er
(m
m
)
97

Image Analysis Method – Cross Section of Cores Filled with Fluorescent Epoxy


Figure 45 - Loop 360 – Core 1C – September 07 - Face A


Figure 46 - Loop 360 – Core 1C – September 07 - Face B


Figure 47 - RM 1431 – Core 2C – September 07 - Face A




Figure 48 - R
Figur
M 1431 – C
e 49 - RM 6
98
ore 2C – Se
20 – Core
ptember 07
3A – June 0
- Face B
7


99

APPENDIX C: EXPERIMENTAL SYSTEM – BLUEPRINTS

Figure 50 - Top Plate & Top View
100


Figure 51 - Bottom Plate - Bottom View
101


Figure 52 - Side View of Top and Bottom Plates

102


Figure 53 - Standpipe

103


Figure 54 - Plexiglas Box
104


Figure 55 - Manometer Frame

105


Figure 56 - Slanted Board


106

APPENDIX D: POROUS FRICTION COURSE AGGREGATE - WATER PROPERTIES




Table 13 - Water Properties (FHA, 2007)


107

Table 14 - Aggregate Grain Size Distribution (Alvarez et al, 2006)

108

APPENDIX E: HYDRAULIC CONDUCTIVITY MODEL PARAMETERS

Plain Cores

Table 15 - Plain Cores - Darcy��s linear Model


Table 16 - Forchheimer Model Parameters


Min Max Min Max Min Max
1A 0.21 0.27 0.2 0.24
1B 1.88 2.22
2A 0.57 0.7 0.49 0.6
2B 0.33 0.47 0.51 0.68
3B 0.3 0.37 0.08 0.09
3C 0.45 0.58 0.41 0.52
Darcy's Law -Hydraulic Conductivity - 95% Confidence Interval
Core
Initial Top Upside Down
K (cm/s) alpha beta Deviation (cm) Min. Deviation (%)
1A
1B 2.63 0.477 0.0028 0.1253 11.1
2A
2B
3B
3C
K (cm/s) alpha beta Deviation (cm) Min. Deviation (%)
1A 0.19 0.7 0.0143 0.663 2.83
1B
2A 0.7 0.19 0.0093 3.22 44
2B 0.32 0.403 0.015 6.26 5.2
3B 0.49 0.296 0.039 3.66 15.1
3C 0.49 0.286 0.014 4.99 6.9
K (cm/s) alpha beta Deviation (cm) Min. Deviation (%)
1A 0.18 0.733 0.012 7.83 2
1B
2A 0.58 0.229 0.009 3.46 11
2B 0.64 0.204 0.015 4.42 15.5
3B 0.1 1.437 0.023 7.128 2
3C 0.48 0.29 0.021 4.37 10
Upside Down
Top
Initial
Core
Core
Core
109

Drilled Cores

Table 17 - Drilled Cores - Darcy��s linear Model



Table 18 - Drilled Cores - Forchheimer Model Parameters



Min Max Min Max
1A 0.16 0.18 0.89 1.17
1B 0.94 1.28 1.52 1.88
2A 0.83 0.93 1.76 2.21
2B 0.34 0.39 2.68 3.53
3B 0.16 0.18 0.67 0.83
3C 0.19 0.21 0.81 0.88
Darcy's Law -Hydraulic Conductivity - 95% Confidence Interval
Core
Initial Cleaned
K (cm/s) alpha beta Deviation (cm) Min. Deviation (%)
1A 0.21 0.409 0.016 2.61 9
1B 1.13 0.058 0.005 2.88 19
2A 0.86 0.099 0.005 2.39 12
2B 0.38 0.214 0.002 2.73 2
3B 0.19 0.559 0.012 1.92 4
3C 0.23 0.415 0.011 2.12 5
K (cm/s) alpha beta Deviation (cm) Min. Deviation (%)
1A 1.26 0.068 0.006 2.41 18
1B 1.72 0.038 0.0015 1.05 9
2A 2.11 0.041 0.002 1.18 9
2B 2.73 0.03 0.0011 0.82 8
3B 1.08 0.098 0.009 0.95 25
3C 0.97 0.099 0.0035 1.46 10
Core
Cleaned
Core
Initial
110

APPENDIX F: HEAD GRADIENT VERSUS FLOWRATE CURVES

Figure 57 - Head Gradient vs. Flowrate - 1A Top
0.00
5.00
10.00
15.00
20.00
25.00
30.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
111


Figure 58 - Head Gradient vs. Flowrate - 2A Top

Figure 59 - Head Gradient vs. Flowrate - 2B Top

0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
0.00
5.00
10.00
15.00
20.00
25.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
112


Figure 60 - Head Gradient vs. Flowrate - 3B Top


Figure 61 - Head Gradient vs. Flowrate - 3C Top

0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
H
(c
m
)
Q (cm3/s)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
113


Figure 62 - Head Gradient vs. Flowrate - 1A Upside Down

Figure 63 - Head Gradient vs. Flowrate - 2A Upside Down

0.00
5.00
10.00
15.00
20.00
25.00
30.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
114


Figure 64 - Head Gradient vs. Flowrate - 2B Upside Down


Figure 65 - Head Gradient vs. Flowrate - 3B Upside Down

0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
H
(c
m
)
Q (cm3/s)
0.00
5.00
10.00
15.00
20.00
25.00
30.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
H
(c
m
)
Q (cm3/s)
115


Figure 66 - Head Gradient vs. Flowrate - 3C Upside Down




0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
0.00 5.00 10.00 15.00 20.00 25.00
H
(c
m
)
Q (cm3/s)
116

REFERENCES

Alvarez, A., and Button, W. (2006). Synthesis of current practice on the design,
construction, and maintenance of porous friction course, Texas Transportation Institute,
pp. 28.

Bäckström, M., and Bergström, A. (2000) Draining function of porous asphalt during
snowmelt and temporary freezing. Canadian Journal of Civil Engineering, Vol. 27, pp.
594-598.

Bear, Jacob (1972). Dynamics of Fluids in Porous Media, Dover editions, pp 125-171.

Bendtsen, H., Nielsen, C., Raaberg, J and Macdonald, R. (2002) Report 120: Clogging of
porous bituminous surfacing – an investigation in Copenhagen. Road Directorate,
Danish Road Institute.

Birgisson, B., Roque, R., Varadhan, A., Thai, T. and Lokendra, J. (2006) Evaluation of
thick open graded and bonded friction courses for Florida. University of Florida.

California Department of Transportation (Caltrans) (2003) Fog Seal Guidelines.

Carslaw, H.S. and Jaeger, J.C. (1959). Conduction of Heat in Solids, Oxford University
Ed., pp 214-231.

117

Charbeneau, R. (2000). Groundwater Hydraulics and Pollutant Transport, Waveland
Press Editions, pp. 12-104.

Charbeneau, R. (2006). Analysis of CRWR PFC Testing Apparatus.

Di Benedetto, H., Jouni, M., Boillon, P. and Salhi, J. (1996) In situ and laboratory
measurements of flows in porous bituminous concrete: modeling and validation, Bulletin
des Laboratoires des Ponts et Chaussees, Vol 204.

Fair, G.H. and Hatch, L.P. (1933) Fundamentals Factors Governing the Streamline Flow
of Water through Sand, Journal of American Water Works Association, Vol 25, pp.
1551-1565

Federal Highway Administration (FHWA) (2005) Quiet Pavement Systems in Europe,
United States Department of Transportation, Federal Highway Administration,
http://international.fhwa.dot.gov/quiet_pav/chapter_two_f.htm

Federal Highway Administration (FHWA) (2004) Tidal Hydrology, Hydraulics, and
Scour at Bridges, Hydraulic Engineering Circular N��25, Appendix A,
http://www.fhwa.dot.gov/engineering/hydraulics/hydrology/hec25appa.cfm

Forchheimer, P. (1901) Wasserbewegung dur Bodem, Z. Ver. Deutsch, Ing., Vol. 45, pp.
1782-1788.

118

Fwa, T., S. Tan, and Guwe, Y. (1999) Laboratory Evaluation of Clogging Potential of
Porous Asphalt Mixtures. Transportation Research Record: Journal of the Transportation
Research Board, No. 1681, TRB, National Research Council, Washington, D.C., pp. 43-
49.

Graff, J. (2006) Asphalt Pavement Preservation, Lone Star Roads Magazine, Issue 1.

Isenring, T., Koster, H. and Scazziga, I. (1990) Experiences with Porous Asphalt in
Switzerland. Transportation Research Record: Journal of the Transportation Research
Board, No. 1265, TRB, National Research Council, Washington, D.C., pp. 41-53.

Kandhal, P. and Mallick, R. (1999) Design of New-Generation Open-Graded Friction
Courses. NCAT Report No. 99-3.

Moore, L., Hicks, R. and Rogge, D. (2001) Design, Construction, and Maintenance
Guidelines for Porous Asphalt Pavements. Transportation Research Record: Journal of
the Transportation Research Board, No. 1778, TRB, National Research Council,
Washington, D.C., pp. 91-99.

National Cooperative Highway Research program (NCHRP) (2000) Synthesis of
Highway Practice 284: Performance Survey on Open-Graded Friction Course Mixes.
Transportation Research Board, National Research Council, Washington, D.C.

Newcomb, D., and Scofield, L. (2004) Quiet Pavements Raise the Roof in Europe. Hot
Mix Asphalt Technology, September-October, pp. 22-28.
119

http://www.quietpavement.com/hotmixasphalt/docs/Quiet_Pavements_Raise_the_Roof_i
n_Europe.pdf

Nutting, P.G. (1930). Physical analysis of oil sands, Bulletin of American Association in
Petroleum and Geology, Vol. 14, pp. 1337-1349.

Pacific Water Resources, Inc. (2004) Cross Israel Highway Stormwater Quality Study
Phase II Monitoring Report.

Pagotto, C., M. Legret and Le Cloirec, P. (2000) Comparison of the Hydraulic Behaviour
and the Quality of Highway Runoff Water According to the Type of Pavement. Water
Resources, Vol. 34, No. 18, pp. 4446-4454.

Public Works Reseach Institute (PWRI) (2005) Noise Reducing Pavement in Japan,
Technical Note 31, Presentation on Function recovery and maintenance of porous
pavements in Japan.
http://www.vejdirektoratet.dk/publikationer/VInot031/html/chapter06.htm

Regimand A. and Lawrence, J. (2004) Systems and methods for determining the porosity
and/or effective air void content of compacted material, United States Patent 6684684.

Ruiz, A., Alberola, R., P��rez, F., and S��nchez, B. (1990) Porous Asphalt Mixtures in
Spain. Transportation Research Record: Journal of the Transportation Research Board,
No. 1265, TRB, National Research Council, Washington, D.C., pp. 87-94.

120

Rumer R.R. and Drinker P.A. (1966) Resistance to laminar flow through porous media.
Journal of Hydraulic Division, American Society of Civil Engineers, V92, HY5, pp. 155-
163.

Shao, J., Lister, P. and McDonald, A. (1994) A surface-temperature prediction model for
porous asphalt pavement and its validation. Meteorological Applications, pp. 129-134.

Stotz, G. and Krauth, K. (1994) The pollution of effluents from pervious pavements of an
experimental highway section: first results. The Science of the Total Environment, Vol.
146/147, pp. 465-470.

Terzaghi Karl, Peck Ralph B. and Mesri Gholanreza (1996). Soil Mechanics in
Engineering Practice, Wiley-Interscience editions, pp. 78-80.

Texas Department of Transportation (2005), Capital Area Metropolitan Planning, Traffic
Count. http://www.campotexas.org/programs_rd_traffic_counts.php

Van der Zwan, J., Goeman, T., Gruis, H., Swart, J. and Oldenburger, R. (1990) Porous
Asphalt Wearing Courses in the Netherlands: State of the Art Review, Transportation
Research Record: Journal of the Transportation Research Board, No. 1265, TRB,
National Research Council, Washington D. C., pp.95-110.

Van Leest et al. 1997. Guidelines for determining the cleaning effect on porous
asphalt. In the European Conference on Porous Asphalt; Proceedings, Madrid,
1997.
121


Van Heystraten, G., and Moraux, C. (1990) Ten Years�� Experience of Porous Asphalt in
Belgium. Transportation Research Record: Journal of the Transportation Research
Board, No. 1265, TRB, National Research Council, Washington, D.C., pp. 34-40.

Venkataraman, P. and Rama, P. (1998). Darcian, Transitional, and Turbulent Flow
Through Porous Media, American Society of Civil Engineers, Journal of Hydraulic
Engineering, Vol. 125, pp. 63-71.

Ward, J. (1964) Turbulent flow in porous media. Journal of Hydraulic Division, V90,
HY5, pp. 1-12.

Watson, D., Johnson, A., and Jared, D. (1998) Georgia Department of Transportation��s
Progress in Open-Graded Friction Course Development. Transportation Research
Record: Journal of the Transportation Research Board, No. 1616, TRB, National
Research Council, Washington, D.C., pp. 30-33.

Yildirim, Y., Dossey, T, Fults, K, and Trevino, M. (2006) Winter Maintenance Issues
Associated with New Generation Open-Graded Friction Courses. Center for
Transportation Research Technical Report 0-4834-1, The University of Texas at Austin.



Search more related documents:CRWR Online Report 08-06
Download Document:CRWR Online Report 08-06

Set Home | Add to Favorites

All Rights Reserved Powered by Free Document Search and Download

Copyright © 2011
This site does not host pdf,doc,ppt,xls,rtf,txt files all document are the property of their respective owners. complaint#nuokui.com
TOP