See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259310808

Development of Mechanistic-Empirical Design Procedure for Fully Permeable

Pavement Under Heavy Traffic

Article  in  Transportation Research Record Journal of the Transportation Research Board · December 2012
DOI: 10.3141/2305-09

CITATIONS READS

8 205

3 authors, including:

Hui Li David Jones

University of California, Davis University of California, Davis
40 PUBLICATIONS   335 CITATIONS    84 PUBLICATIONS   419 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Full depth reclaimation View project

All content following this page was uploaded by Hui Li on 11 February 2014.

The user has requested enhancement of the downloaded file.

Development of Mechanistic–Empirical
Design Procedure for Fully Permeable
Pavement Under Heavy Traffic
Hui Li, David Jones, and John Harvey
Fully permeable pavements are those in which all layers are intended to
be permeable and the pavement structure serves as a reservoir to store
water during storm periods to minimize the adverse effects of storm-
water runoff. Fully permeable pavements are currently used primar-
ily for light traffic (e.g., parking lots that only allow cars, not trucks).
The development of fully permeable pavement designs as a potential
best management practice for stormwater management in areas that
carry heavy truck traffic is of significant interest. Results are presented
from one phase of a project to develop mechanistic–empirical design
procedures for fully permeable pavement designs that will carry heavy
vehicles operating primarily at slow speeds (e.g., in maintenance yards,
parking lots, streets, and highway shoulders). Mechanistic–empirical
computer modeling of the structural designs was performed for three
types of surface material (open-graded asphalt, open-graded concrete,
and concrete slabs with cast drainage), considering different traffic
loading based on typical California highway truck traffic, different
traffic speeds, and several typical California climate regions, as well as
materials properties developed from laboratory testing. A preliminary
design method and preliminary design charts that considered both
hydraulic and structural performance were developed; some examples of
their use are provided. The mechanistic–empirical approach used in this
project for the development of fully permeable pavement designs could
increase the speed of development and implementation of this technology.
Fully permeable pavements are defined for the purposes of this
study as those in which all layers are intended to be permeable and
the pavement structure serves as a reservoir to store water during
storm periods to minimize the adverse effects of stormwater runoff.
The California Department of Transportation (Caltrans) and local
governments are interested in developing fully permeable pavement
designs for use in areas that carry heavy truck traffic as a potential
best management practice for stormwater management.
Since the late 1970s, fully permeable pavement projects have
been constructed in many states for low traffic areas and light vehi-
cles. Most of the information in the literature about these applica-
tions touts their success, but a few failures have been reported. The
authors observed several projects that exhibited localized failures
University of California Pavement Research Center, Department of Civil and Envi-
ronmental Engineering, University of California, Davis, 2001 Ghausi Hall, 1 Shields
Avenue, Davis, CA 95616. Corresponding author: H. Li, hili@ucdavis.edu.
Transportation Research Record: Journal of the Transportation Research Board,
No. 2305, Transportation Research Board of the National Academies, Washington,
D.C., 2012, pp. 83–94.
DOI: 10.3141/2305-09
83
as a result of clogging on the permeable surface or severe surface
raveling or cracking caused by construction processes.
Reflecting road owner concerns about durability, most applications
in North America have been for pavements with no high-speed traf-
fic or truck traffic (e.g., automobile parking lots). Structural design
methods have been empirical in nature, with little or no long-term
monitoring data to support the empiricism. Purely empirical design
methods require good, comprehensive long-term performance data for
all of the expected design conditions, and the high cost of learning
from inevitable failures has limited the speed of technology develop-
ment for fully permeable pavements. The need for a large factorial
set of performance data that considers different materials, climates,
subgrades, and structural cross sections makes it difficult for purely
empirical design methods to consider all the design variable permuta-
tions. A review of design practice across the United States shows the
limited scope of current applications for fully permeable pavements,
even by the leading design firms specializing in this type of design
(1). The limited scope of current applications also is reflected in man-
uals recently produced by the National Asphalt Pavement Associa-
tion (NAPA) (2), American Concrete Pavement Association (3), and
Interlocking Concrete Pavement Institute (4) for the design of porous
asphalt, pervious concrete pavements, and permeable ­interlocking
concrete pavements, respectively.
The mechanistic–empirical approach used in this project to
develop new fully permeable pavement designs will increase the
speed of development. The mechanistic–empirical design develop-
ment process consists of determining the relevant material prop-
erties in the laboratory, using them in inexpensive and risk-free
computer ­models to evaluate pavement performance, then empiri-
cally validating and calibrating the failure mechanisms and perfor-
mance of the most promising designs through accelerated pavement
testing and field test sections.
This paper summarizes the development of a preliminary design
method and preliminary design charts that consider both hydraulic
and structural performance for fully permeable pavements. Hydrau-
lic performance information was sourced from a companion project
also conducted by the University of California Pavement Research
Center (UCPRC) (5).
Performance Modeling
Pavement Structures
A mechanistic–empirical approach is used to develop detailed pave-
ment designs in this study. Caltrans is in the process of implement-
ing this approach as a replacement for the empirical R-value design
84
method for flexible (asphalt-surfaced) pavement designs and has
replaced the previous design charts for rigid (concrete-surfaced) pave-
ments with a new catalog of designs based on mechanistic–empirical
analysis. The assumptions of the R-value design method for flexible
pavements, including standard compaction and pavement structural
layering, are not appropriate for fully permeable pavements.
In this project, the mechanistic–empirical approach was used for
both flexible porous pavements [surfaced with open-graded hot-mix
asphalt (HMA-O)] and two types of rigid, fully permeable pavement
(often referred to as pervious pavement) to produce a set of designs for
different traffic index (TI), climate, and soil conditions. Parking lots,
low-traffic-volume roadways, and highway shoulder retrofits were
considered. The different pavement types are summarized in Figure 1.
The two types of rigid pavement were surfaced with either open-
graded portland cement concrete (PCC-O), in which the surface is
permeable because of the aggregate gradation, or ordinary dense-
graded portland cement concrete (PCC), in which the surface has
drainage holes cast into it during construction (referred to as cast
PCC). The results of the analyses were used to produce a catalog of
designs similar to those prepared by UCPRC for the Caltrans Rigid
Pavement Design Catalog currently used in the Caltrans Highway
Design Manual. All calculations considered two subbase options:
• No subbase and
• A PCC-O subbase 150 mm thick to provide support and confine-
ment to the granular layer and help protect the saturated subgrade.
Analytical Processes
PCC Surface
A comprehensive factorial was followed to model the performance
of fully permeable pavement surfaced with PCC-O or cast PCC;
1,536 cases were run. Variables for the PCC layer included sur-
face material type (PCC-O or cast PCC), slab thickness, slab length,
material properties, climate zone, season, diurnal peak temperature
gradient, axle type, axle load, load location, and traffic ­volume. This
analytical process is summarized in Figure 2a.
Hot-Mix Asphalt Surface
A similar comprehensive factorial was followed to model the perfor-
mance of permeable hot-mix asphalt (HMA) wearing courses; a total of
20,736 cases were run. Variables for the HMA layer included ­material
type, layer thickness, material properties, climate zone, season, diurnal
peak temperature, axle type, axle load, traffic speed, and truck traffic
volume. This analytical process is summarized in Figure 2b.
Factorial for Performance Modeling
The factorial for performance modeling included climate, traffic,
material properties, mechanical response, and pavement performance,
which are described as follows.
Climate
Two California climate zones were selected: Sacramento and Los
Angeles. Climate conditions determine thermal gradient and tem-
Transportation Research Record 2305
perature in the pavement, which affect the thermal stress of the
PCC, HMA stiffness, and, consequently, the mechanical responses
of the pavement structures.
PCC Surface   The thermal gradients of PCC slabs were calculated
from 30 years (1961 to 1990) of California climate data by using the
Enhanced Integrated Climatic Model (EICM). The maximum, mini-
mum, and average of the 30-year thermal gradient were calculated
at each hour in each day for January and July (winter and summer
seasons, respectively). The maximum and minimum of the average
day for those 2 months were chosen as the thermal gradients during
the day and night, respectively, for the finite element calculations
of mechanical responses.
HMA Surface   The temperatures at one-third of the depth of the
HMA layer were calculated from the same California climate data
with the EICM. Maximum, minimum, and average temperatures
were calculated at each hour in each day for January, April, and July
(winter, spring, and summer, respectively). These three seasons were
used to better capture the temperature-related changes in stiffness
that occur in HMA and the changes in subgrade stiffness and shear
strength that control subgrade rutting in a fully permeable pavement.
The maximum and minimum of the average day for each of those
3 months were chosen as the day and night temperatures, ­respectively,
for layer elastic theory calculations of mechanical responses.
Traffic
Axle loads (including components and distributions across seasons
and time) were obtained from a database of California weigh-in-
motion (WIM) stations (6). Because thousands of calculations are
required to determine critical stresses and strains with layer elastic
theory for HMA-surfaced pavements and finite element analysis for
open-graded and cast PCC, the traffic loads included in the ­calculations
were reduced to two each for single and tandem axles:
• The traffic repetitions between the 50th percentile load and the
legal maximum load, with the representative load taken at approxi-
mately halfway between the 80th percentile load and the legal maxi-
mum load (this representative load imparts some conservatism to
the designs), and
• The traffic repetitions between the legal maximum load and the
maximum load—excluding a few outliers that are heavier—with
the representative load taken at approximately halfway between the
legal maximum load and the maximum load (this representative
load imparts additional conservatism to the designs).
Steering-, single-, and tridem-axle loads were not considered
because they contribute little to damage, according to calculations
from the Caltrans and UCPRC WIM data (6). These loads were
selected on the basis of (a) the understanding that most pavement
damage is caused by the heaviest loads, particularly for concrete,
and (b) examination of typical axle-load spectra from the WIM data-
base, which indicated that the first load group was representative of
a significant percentage of the loaded axles on California highways
and the second load group was representative of the few loads that
cause the most damage.
The allowable truck traffic [equivalent single-axle load (ESAL) or
TI] during the design life was calculated from a set of factors, includ-
ing season (winter, spring, or summer), time (day or night), axle type
 Permeable surface Permeable surface

Permeable Permeable
base/reservoir layer base/reservoir layer

PCC-O subbase

Subgrade Subgrade

No subbase With subbase

(a)

Travelled lane (Shoulder Paint Line) Permeable Shoulder Embankment Travelled lane (Shoulder Paint Line) Permeable Shoulder Embankment

Plan View Plan View

Existing shoulder structure

New surfacing New surfacing

Permeable shoulder Permeable shoulder
Existing surfacing (Design dependent on Existing surfacing (Design dependent on
site and traffic) site and traffic)

Existing base & subbase Existing base & subbase

Subgrade Subgrade

Filter fabric/geogrid Filter fabric/geogrid

Side View New edge drain Side View New edge drain
Impermeable composite liner Impermeable composite liner Impermeable composite liner Impermeable composite liner
(optional ) (optional)
Not to scale Not to scale

(b) (c)

FIGURE 1   Analyses of fully permeable pavement structural designs: (a) pavement cross sections with and without subbase, (b) example of shoulder retrofit in which permeable shoulder
is placed against pavement layers with fabric separator, and (c) example of shoulder retrofit in which portion of existing shoulder is retained to prevent damage to pavement layers and
to maintain separation from reservoir.
 Fully Permeable PCC Pavement Design Fully Permeable HMA Pavement Design

Laboratory Testing Phase Laboratory Testing Phase

(Test Results: Mechanical (Test Results: Mechanical
Properties)
Properties)

Get Mechanical Properties

Get Mechanical Properties (Materials: HMA,
(Materials: PCC-O/Cast Slab, Gravel, PCC-O Subbase, Hydraulic
Hydraulic Subgrade Soils. Properties
Gravel, PCC-O Subbase,
Subgrade Soils. Properties Properties: Fatigue,
Properties: Fatigue, Strength, Stiffness, etc.)
Strength, Stiffness, etc.)
Identify Critical Factors
(Region, HMA type, Structure
Identify Critical Factors Identify Critical Factors
Type, Thickness, Load, Speed,
(Region, PCC type, Structure etc.)
Identify Critical Factors
Type, Thickness, k-values, Load,
etc.)
Calculate Mechanical Responses
(Tensile Strain at bottom of HMA Simulate Hydraulic
and vertical stress at top of SG.) Behavior
Calculate Mechanical Responses
[Tool: LEAP & GT-Pave]
(Tensile Stress in PCC-O/Cast Simulate Hydraulic
Slab.) Behavior
[Tool: EverFE]
Calculate Fatigue Life (TI) of HMA Calculate Hydraulic
(Tool: Miner’s Law) Responses

Calculate Fatigue Life (TI) in

Calculate Hydraulic
PCC-O/Cast Slab
Responses
(Tool: Miner’s Law)
Calculate Stress/Strength Ratio
Determine Hydraulic
(Ratio of shear stress to shear
Performance
strength of SG soil)

Draw Design Plots of Determine Hydraulic

TI vs. Slab thickness/slab length Performance
Draw Design Plots of
TI vs. HMA thickness x base
thickness

Interpolation
(TI interpolation for different Slab
thickness & length using Bilinear Interpolation
Interpolation Function) (TI/SR Interpolation for different
HMA and base thickness using
Bilinear Interpolation Function)

Structural Design
Hydraulic Design
(Thickness & Length of
(Thickness of Base + PCC-O)
PCC-O/Cast Slab) Structural Design
Hydraulic Design
(HMA + base + subbase
(Thickness of base + subbase)
thickness)

Design Fully Permeable Pavement Design Fully Permeable Pavement

(a) (b)

FIGURE 2   Analytical processes for developing structural designs for fully permeable pavements: (a) portland cement concrete and (b) hot-mix asphalt.
Li, Jones, and Harvey
(single or tandem), ESAL (average ESALs per axle), and load bin (i.e.,
percentage of total axle repetitions for each axle in each load range) as
described in the following list. Each factor value was determined from
the statistical analysis of statewide traffic information obtained from
the WIM database (6).
Season  Because the UCPRC WIM study indicates that axle loads
are evenly distributed across months, the seasonal factor is calculated
according to the number of months in the season (6):
• Winter: PCC = 0.5, HMA = 0.33;
• Spring: PCC = 0.0, HMA = 0.25; and
• Summer: PCC = 0.5, HMA = 0.42.
Time  The UCPRC WIM study indicates that more loaded trucks
travel by night than by day:
• By day = 0.45 and
• By night = 0.55.
Axle  The UCPRC WIM data study indicates that Truck Type 5
(typically with one steering single and one loaded single axle) and
Truck Type 9 (typically with one steering axle and two loaded tan-
dem axles) dominate the truck composition on California highways
and that, on average, there are twice as many Type 9 as Type 5 trucks
across the state. Therefore, one-third of the trucks have one single
axle and two-thirds of the trucks have two tandem axles, thus result-
ing in 20% single axles and 80% tandem axles in the total population
of axles shown in the table [additional information is available from
UCPRC (6, Figure 12)]:
• Single axle = 0.2 and
• Tandem axle = 0.8.
ESAL  The number of ESALs per axle repetition (based on an
80-kN single axle) is calculated for each axle type on a statewide
average of all Caltrans WIM stations between 1993 and 2001 (6).
The ESAL calculations used an exponent of 3.8, recommended by
FHWA, rather than the 4.2 exponent Caltrans normally uses (6).
This factor converts ESALs in the TI into total axle repetitions
[additional information is available from UCPRC (6, Figure 35); or
the equation TI = 9(ESAL/106)0.119 may be used]:
• Single axle = 0.17 and
• Tandem axle = 0.3.
Load Bin   The percentage of axle repetitions for each axle type out
of the total repetitions of that axle type is determined for the two
load ranges used in the calculations (half the legal load to the legal
load, and the legal load to the maximum load):
• Single axle: 0.8 × legal–legal (75 kN)= 0.492, legal–max
(93 kN) = 0.008, and
• Tandem axle: 0.8 × legal–legal (135 kN) = 0.46, legal–max
(155 kN) = 0.04.
Traffic Speed for HMA Surface
Because traffic speed affects load frequency and thus HMA stiff-
ness in HMA structures, two truck traffic speeds (7 and 40 km/h) are
87
used in the calculations. The slower speed represents truck opera-
tions during traffic congestion on highways (in this case, a detour
onto the shoulder) and in maintenance yards or parking areas; the
faster speed represents truck operations on a street or a shoulder
where traffic has been diverted but is not severely congested. These
speeds are somewhat slower than the average speeds for these con-
ditions. The speeds are conservative because under slow speeds,
HMA is less stiff, which increases the strains that cause fatigue
cracking and the stresses in the granular base and subgrade that can
lead to rutting.
Material Properties
The key material properties for each layer, including PCC and
HMA stiffness and the resilient modulus of the base and subgrade,
were determined from a laboratory study (7).
PCC Surface   The rigid PCC pavements were modeled as two-layer
systems (the slab and the supporting layers) with a composite k-value
(modulus of subgrade reaction) simulating all layers below the slab
acting together. A separate factorial for the supporting layers was
used to derive two input k-factors for the supporting layers (7).
HMA Surface   Three types of HMA-O were considered in the cal-
culations: 9.5 mm with conventional binders; 9.5 mm with rubberized
binders; and 12.5 mm with polymer-modified asphalt, fibers, and lime.
HMA stiffness was calculated from the master curves for each com-
bination of temperature and load frequency corresponding to loading
time from flexural beam frequency sweep testing during the labora-
tory study (7). The loading frequency at one-third the thickness of the
HMA layer was approximately ­estimated with Equation 1:
1 1 v v
freq = = = = (1)
t L L Th
D+2×
v 3
where
t = loading time,
L = loading distance,
v = loading speed,
D = loading tire-to-pavement contact diameter, and
Th = thickness of HMA layer.
The stiffness of each type of HMA material was averaged for each
layer thickness to reduce the number of calculation combinations.
Consequently, the HMA stiffness used in the calculations was inde-
pendent of layer thickness. The master curves and time–temperature
relationships used are summarized in Table 1.
Granular Base and Subgrade   The resilient moduli of the granu-
lar base and subgrade were determined from the estimated stress
status in the corresponding layer and the relationship between resil-
ient modulus and stress status obtained from the laboratory study
(7). The resilient moduli of the subgrade differ by season because
of changes in moisture content.
Before the layer elastic analysis was conducted for HMA structures,
the stiffness of the granular base was evaluated with nonlinear elastic
models in the GT-Pave software package (8). A range of values for
different structural factors were selected for the structural response
values of the granular base stiffness (9). The Uzan model was used to
88 Transportation Research Record 2305

TABLE 1   Summary of Master Curves and Time–Temperature Relationships

Time–Temperature
Master Curve Relationship

Mix Type n A B C D E F

9.5-mm rubber (AR95) 3 21,478.93 15.77917 −9.57447  94.4856 −17.8532 53.2251

12.5-mm polymer (G125) 3 22,927.80 10.79402 −10.01293 147.9806 12.0135 −35.4865
9.5-mm conventional (RW95) 3  8,420.84  3.976235 −5.32720 143.6314 7.62143 −24.3910

Note: Reference temperature = 20°C. Flexural controlled-deformation frequency sweep tests were conducted at following testing conditions:
  Frequencies = 15, 10, 5, 2, 1, 0.5, 0.2, 0.1, 0.05, 0.02, and 0.01 Hz;
  Temperatures = 10°C, 20°C, and 30°C; and
  Strain level = 100 or 200 microstrain.
  Master curve gamma fitting equations:

If n = 3,

  (x − C) i  x − C (x − C)  
2
E * = D + A i  1 − exp  −   1 + + 
 B  B 2B  
2
 
 where x = ln freq + ln aT.

  Time–temperature relationship:

  T − Tref  
ln ( aT ) = E i  1 − exp  −  
  F

consider the nonlinear behavior of the granular base by using GT-Pave were calculated with LEAP to evaluate the permanent d­ eformation
(10). The procedure proposed by Tutumluer and Thompson (11) was ­potential of the subgrade.
used to obtain cross-anisotropic parameters of the granular base for
GT-Pave (9). From the results of these calculations, three represen-
tative values of granular base stiffness (60, 90, and 120 MPa) were Pavement Performance
chosen for the final structural calculations.
Linear Cumulative Damage   The Miner’s law equation, also referred
to as the linear cumulative damage (LCD) equation, was used to calcu-
Mechanical Responses late the fatigue damage under specific conditions (pavement structure,
traffic loading, and climate conditions) (15):
PCC Surface  The distress type modeled was cracking due to
tensile stresses in the PCC slab. Four locations of loading and ni
responses were considered: D=∑ (2)
i Ni

• Midslab edge at the top of the slab,

where
• Midslab edge at the bottom of the slab,
• Near the corner of the slab at the top of the slab, and D = damage from fatigue;
• Near the corner of the slab at the bottom of the slab. ni = actual repetitions under ith condition of axle type, climate
condition, and pavement structure, calculated from the load
Mechanical responses in terms of tensile stress in the slab from spectrum; and
different load configurations were determined with the EverFE finite Ni = allowable repetition under ith condition, calculated from
element software package for concrete pavement (12). The stiffness the fatigue equation.
of cast slabs was estimated from conventional PCC by a factor of 0.92
(i.e., 30 GPa × 0.92 = 27.6 GPa) (13). Stresses from EverFE in the cast In addition,
slabs were increased by a factor of 3.0 to reflect the stress concentra-
1
tion around the holes in the cast slabs based on separate finite element ni = ESAL × Fseason × Ftime × Faxle × × FLB (3)
analyses completed before the EverFE calculations. FESAL

HMA Surface   The distresses analyzed included (a) fatigue crack-
ing of the HMA layer associated with the tensile strain at the bot-
tom of the HMA layer and (b) unbound layer rutting associated with
the vertical stresses at the top of the base, subbase (where included),
and subgrade. Mechanical responses in terms of tensile strains at the
bottom of the HMA layer from different load configurations were
determined with the layer elastic model in LEAP layered elastic
ana­l­ysis software (14). Vertical stresses at the top of the subgrade also
where ESAL is the ESALs for the TI and Fx are the factors for
season (winter, spring, or summer), time (day or night), axle type
(single or tandem), ESALs per axle repetition per ESAL coefficient
(6), and load bin (LB). The allowable repetitions to failure (n) were
calculated with Miner’s law to determine the number of ESALs
(later converted to TI; D = 1.0 in Equation 1) for each combination
of pavement type, slab dimensions, thickness, and climate region.
The allowable repetitions were converted back to ESALs, then to TI.
Li, Jones, and Harvey 89

1.E+10
B1
1.E+09 B2
9.5s
1.E+08 Zero-Maintenance

1.E+07

Repetitions
1.E+06

1.E+05

1.E+04

1.E+03

1.E+02

1.E+01

1.E+00
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Stress Ratio
(a)

1E+8

G125 AR95
1E+7
y = 2E-19x-7.5307 y = 3E-12x-5.3569
R2 = 0.9101 R2 = 0.6778
1E+6
Fatigue Life (Nf)

1E+5

RW95
1E+4
y = 1E-07x-3.5268
R2 = 0.694
1E+3

AR95 G125 RW95

1E+2
0.0001 0.001
Test Strain
(b)

FIGURE 3   Fatigue life of surface materials: (a) open-graded PCC (PCC-O)
and (b) three types of open-graded HMA (HMA-O).

PCC Surface   Results of the EverFE stress calculations were input
into a Miner’s law equation to calculate the fatigue performance of
the PCC slabs. From the laboratory fatigue testing results (7) shown in
Figure 3a, the zero-maintenance fatigue equation (16) was used to cal-
culate the allowable repetitions under the ith condition (­combination
of factors shown in Equation 3):
 1− σ i 
17.61
 MR 
N i = 10
where σi is the maximum tensile stress in the slab under the ith
condition and MR is the modulus of rupture (flexural strength)
[MR = 2.3 MPa for PCC-O based on testing results (7) and MR =
2.6 MPa for cast slabs considering a stress concentration based on
finite element analysis].
HMA Surface   The mechanical responses from LEAP were input-
ted in the Miner’s law equation to calculate the fatigue performance
of HMA in terms of an allowable TI. The fatigue equations for the
three types of HMA-O are shown in Figure 3b. The allowable rep-
etitions to failure (n) were calculated to determine the number of
ESALs for each combination of HMA-O type, thickness, and climate
region. The allowable repetitions to failure were converted back to
ESALs, then to TI.
The ratio of shear stress to shear strength at the top of the subgrade
was calculated to evaluate the permanent deformation potential of the
subgrade. According to the FAA subgrade soil evaluation report,
the shear stress was estimated as one-half the vertical stress at the top of
(4) the subgrade (17). The saturated shear strength for clay was estimated
as 51.7 kPa—a worst-case scenario that takes into consideration the
natural confinement in the pavement. Continued permanent deforma-
tion of the subgrade after initial densification under traffic is unlikely
when the stress-to-strength ratio is less than 0.3, which was the design
criterion selected for this project (17). Continual rutting at a steady rate
after initial embedment is expected when the stress-to-strength ratio is
less than 0.7 but greater than 0.3 times the shear strength.
Results from these calculations were plotted to assess the influ-
ence of pavement layer combinations on the stress-to-strength ratio of
the subgrade. Stress-to-strength ratios for different HMA-O and base
thicknesses then were interpolated to identify a range of a­ ppropriate
90 Transportation Research Record 2305

layer thicknesses for the heaviest traffic loads and each traffic speed
and temperature condition. The process of structural design selec-
tion involves using the results of both fatigue life and subgrade
stress-to-strength ratio.
Example Results
Example predictions of design life (TI) for PCC surfaces for vari-
ous variable combinations in the experimental design (PCC-O,
cast slabs, climatic zone, and base stiffness in terms of k-values)
are shown in Figure 4a (9). For HMA surfaces, example predic-
tions of the shear stress-to-strength ratio at the top of the subgrade,
which is the main contributing factor to permanent deformation
(rutting) of the granular base and subgrade, are provided in Fig-
ure 4b for various combinations of variables in the experimental
design (mix type, subbase inclusion, traffic speed, and climatic
zone) along with example plots of the fatigue design life of the
same pavements (9).

(a) (b)

(c) (d)

FIGURE 4   Example fatigue design life in Sacramento County: traffic index (TI) for PCC fatigue damage 5 1 in (a) Riverside PCC-O
(k 5 0.08 MPa/mm) and (b) Sacramento cast slab (k 5 0.05 MPa/mm); for HMA-O, (c) TI for fatigue damage 5 1, and (d) stress-to-
strength ratio at the top of subgrade (Sacramento AR95 HMA mixes, PCC-O subbase stiffness of granular base 5 60 MPa, and speed 5 40
km/h for both HMA examples) (Th_PCC 5 thickness of PCC slab; Th_GB 5 thickness of granular base; Th_AC 5 thickness of AC layer;
k 5 modulus of subgrade reaction).
Li, Jones, and Harvey 91

Proposed Structural Design Procedure
Overview
A preliminary catalog-type design procedure was developed for
the preliminary design of fully permeable pavement test sections in
­California on the basis of region (rainfall), storm event design period,
design traffic, design truck speed, surface type (HMA or PCC), sub-
base type, and the ratio of shear stress to shear strength at the top of the
subgrade. The example hydraulic design chart presented in Figure 5
includes 2-, 50-, and 100-year storm design events (which consider
infiltration and drawdown for full storm duration and repeat storm
events) for three California regions: Sacramento, Los Angeles, and
Eureka, described in the companion hydrological modeling study (5).
These three storm design events were selected to test the sensitiv-
ity of the design to a wide range of events (e.g., effect of storm inten-
sity, storm duration, geometry, drawdown, and degree of clogging
on infiltration) and were not necessarily intended to be representa-
tive of typical Caltrans storm event design procedures. The design
charts for PCC-O and cast PCC include two regions (­Sacramento
and Los Angeles) and two k-values (18). Design charts for asphalt
include three mix designs [9.5-mm HMA-O with conventional
binder, 9.5-mm HMA-O with rubberized binder (RHMA-O), and
12.5-mm HMA-O with polymer-modified binder plus lime and
fibers (G125)], the same two regions as the PCC options, two
subbase options (no subbase and PCC-O subbase), and two truck
speeds (7 km/h and 40 km/h) (18). All example charts assume a
shoulder width of 3.0 m and cover designs up to a TI of 18 (about
340 million ESAL—clearly more than would ever use such a pave-
ment, but included to show the range of thicknesses). The shoulder
is considered to be a lane for drainage design purposes. The design
charts have not been validated in full-scale experiments and are
intended to be used only for the design of experimental test sections.
The proposed procedure generally entails the following:
1. Select the permeability of the subgrade, region, storm design
event period, and number of lanes drained. For HMA pavements,
a subbase option also must be selected. This information is used to
determine the thickness (in terms of hydraulic performance) of the
gravel base or reservoir layer. Whether occasional overflows are per-
mitted (e.g., during a series of heavy storms on consecutive days,
prolonged rainfall, and so on) should be specified because the answer
will influence the choice of storm design period and dictate the thick-
ness of the base or reservoir layer. Permeabilities should be measured
for each project at a range of depths around the expected depth of the
top of the subgrade after the excavation of material for the reservoir
layer(s). The lowest permeability should be used in the design.
2. Select the surface type. The variables used to identify the
thickness of the surface layer differ by pavement type. For PCC pave-
ments, the variables are slab length, base thickness, design traffic
(design life depends on project circumstances), and design speed.
For HMA pavements, the variables are base thickness, design traf-
fic, and design speed. After the HMA layer thickness has been deter-
mined, this thickness and base thickness are used to determine whether
the ratio of shear stress to shear strength at the top of the s­ ubgrade is
adequate to prevent permanent deformation in the subgrade.
To improve the durability of the surface material, the design method
assumes that water should reach only the top of the granular base
layer for the design storm and not be stored in the surface layer during
the infiltration period.
These preliminary charts can be used for test section designs
of highway shoulder retrofits as well as parking lots, maintenance
yards, and similar facilities. Two typical cross sections for shoul-
der retrofits were developed (Figure 1), and two large highway
construction contractors reviewed them for constructability. Both
contractors indicated that construction appeared to be feasible and
that they would be comfortable bidding on projects with similar
designs. The selection of the most appropriate structure will depend
on whether the existing pavement structure can maintain a verti-
cal cut face equal to the height of the fully permeable pavement
shoulder structure. If the structure can maintain such a cut, then the
first cross section shown in Figure 1 would be the preferred option,

Rainfall Region
Sacramento (Sac) Riverside (LA) Eureka
Storm design
Subgrade soil
(years) Thickness of Granular Base + Thickness of Granular Base + Thickness of Granular Base +
permeability
(Full storm PCC-O Subbase (mm) PCC-O Subbase (mm) PCC-O Subbase (mm)
(cm/s) 1
duration)
Number of highway lanes2 Number of highway lanes2 Number of highway lanes2
2 3 4 5 2 3 4 5 2 3 4 5
1.00E-05 270 450 600 700 270 400 480 680 600 900 1270 1570
1.00E-04 2 130 180 250 420 130 150 320 400 350 650 850 1200
1.00E-03 130 130 130 130 130 130 130 130 130 130 130 150
1.00E-05 480 700 1050 1250 580 860 1180 1600 800 1270 1720 2150
1.00E-04 50 190 420 680 950 360 700 950 1350 500 850 1300 1770
1.00E-03 130 130 130 130 130 130 130 230 130 130 220 500
1.00E-05 600 800 1150 1430 680 1050 1300 1800 1150 1720 2300 2900
1.00E-04 100 210 500 750 1070 400 850 1200 1450 830 1300 1890 2500
1.00E-03 130 130 130 150 130 130 150 320 130 220 650 950
1
Note that draw down times will vary significantly and are dependent primarily on subgrade soil permeabilities but also on other factors such as number of lanes drained, storm
recurrence interval, etc as well. Draw down times could vary between one hour for subgrades with a permeability of 1.00E-03 to several months for subgrades with a
permeability of 1.00E-05 and higher. Refer to Reference 5 for discussion on the calculation of draw down times.

2
The number of highway lanes must include the shoulder. Shoulder width is 10 ft. (3.0 m).

FIGURE 5   Example design chart for hydraulic performance: preliminary granular base thickness based on hydraulic performance simulations.
(Source: 18, Appendix B, Table B.1.)
92 Transportation Research Record 2305

providing a greater reservoir area for the permeable pavement. If
the cut face cannot be maintained, then the second cross section
should be used. A drain between the existing traveled way and the
new fully permeable shoulder will need to be installed to allow any
water in the traveled way to drain away from the road while not
allowing any water from the permeable area to flow into the pave-
ment structure. An impermeable composite liner is included in the
diagrams to prevent water from flowing sideways from the reservoir
layer and causing a slip failure in the embankment. Including this
liner is project dependent and not always required.
Example Design Procedure
The design of an RHMA-O shoulder retrofit of three lanes of high-
way in the Sacramento area illustrates the combined use of the
structural and hydraulic design results that were developed.
Project Design
1. Choose base thickness on the basis of hydraulic performance.
From Figure 5, select the minimum thickness of granular base for
a subgrade soil permeability of 10−4 cm/s, 50-year design storm,
and Sacramento region. These variables require a minimum base
or ­reservoir layer thickness of 680 mm for a shoulder retrofit of
a highway draining three lanes plus the shoulder (i.e., select four
lanes in the chart).
2. Choose the RHMA-O layer thickness on the basis of RHMA-O
fatigue damage for the given TI. With Figure 6, identify the minimum
RHMA-O layer thickness for a base thickness of 700 mm (rounded
up from 680 mm from Step 1) and a TI of 13. The minimum required
thickness of RHMA-O is 395 mm.
3. Check the stress-to-strength ratio at the top of the subgrade.
With Figure 7, check the ratio of shear stress to shear strength at the
top of the subgrade based on the minimum required granular base
thickness of 700 mm and the minimum required RHMA-O thick-
ness of 395 mm. The stress-to-strength ratio is G, which implies that
the shear stress is less than 0.3 of the shear strength. Consequently,
permanent deformation in the subgrade should not be a problem for
this pavement design.

• Compacted subgrade permeability: 10−4 cm/s, In this example, the minimum required thickness of granular base
• Storm design: 50 years, is 700 mm and the minimum RHMA-O thickness is 395 mm for the
• Design TI: 13 (design life dependent on project circumstances), design requirements and site conditions.
• Design truck speed: 7 km/h due to congestion,
• Surface layer: RHMA-O with 9.5-mm nominal maximum
aggregate size, and Summary
• Subbase: no cemented subbase.
This paper summarizes the testing, calculations, and analyses per-
formed during the development of mechanistic–empirical design
Design Procedure procedures for fully permeable pavements. The deliverable from
this research is a preliminary design procedure and an example
The three-step design procedure discussed here refers to a series of set of catalog-type design charts that can be used to design pilot
tables published in a UCPRC report that are reproduced in this paper and experimental test sections of fully permeable pavements in
as the example design charts in Figures 5 through 7 (18). Cells that ­California. Recommendations toward the implementation of fully
are referred to in the steps are circled and highlighted in the charts. permeable pavements as a best management practice for storm­water

Granular Base (GB) Layer Thickness (mm)

500 550 600 650 700 750 800 850 900 950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
200 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0 7.0
215 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5
230 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0
245 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5
260 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0 9.0
275 9.5 9.5 9.5 9.0 9.5 9.5 9.5 9.0 9.5 9.5 9.5 9.5 9.5 9.0 9.5 9.5 9.5 9.0 9.5 9.5 9.5
HMA Layer Thickness (mm)

290 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5
305 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
320 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5
335 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0 11.0
350 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5
365 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0
380 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5
395 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0 13.0
410 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5
425 14.0 14.0 14.0 13.5 14.0 14.0 14.0 13.5 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.5 14.5 14.5 14.5
440 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.0 14.5 14.5 14.5 14.5 15.0 15.0 15.0 15.0 15.5 15.5
455 14.5 14.5 14.5 14.5 14.5 14.5 14.5 14.5 14.5 14.5 14.5 15.0 15.0 15.0 15.5 15.5 15.5 16.0 16.0 16.0 16.5
470 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.0 15.5 15.5 15.5 16.0 16.0 16.5 16.5 17.0 17.0 17.0
485 15.5 15.5 15.5 15.5 15.5 15.5 15.5 15.5 15.5 15.5 15.5 16.0 16.0 16.5 16.5 17.0 17.0 17.5 17.5 18.0 18.0
500 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.5 16.5 17.0 17.0 17.5 18.0 18.0 18.0 18.0 18.0

FIGURE 6   Example design chart for selecting RHMA-O thickness: Preliminary TI for HMA-O fatigue damage 5 1, Sacramento County
(RHMA-O, no subbase) (E_GB 5 60 MPa; speed 5 7 km/h). (Source: 18, Appendix E, Table E.5.)
Li, Jones, and Harvey 93

Granular Base (GB) Layer Thickness (mm)

500 550 600 650 700 750 800 850 900 950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
200 Y Y Y Y Y G G G G G G G G G G G G G G G G
215 Y Y Y Y G G G G G G G G G G G G G G G G G
230 Y Y Y Y G G G G G G G G G G G G G G G G G
245 Y Y Y G G G G G G G G G G G G G G G G G G
260 Y Y Y G G G G G G G G G G G G G G G G G G
275 Y Y G G G G G G G G G G G G G G G G G G G
HMA Layer Thickness (mm)

290 Y G G G G G G G G G G G G G G G G G G G G
305 G G G G G G G G G G G G G G G G G G G G G
320 G G G G G G G G G G G G G G G G G G G G G
335 G G G G G G G G G G G G G G G G G G G G G
350 G G G G G G G G G G G G G G G G G G G G G
365 G G G G G G G G G G G G G G G G G G G G G
380 G G G G G G G G G G G G G G G G G G G G G
395 G G G G G G G G G G G G G G G G G G G G G
410 G G G G G G G G G G G G G G G G G G G G G
425 G G G G G G G G G G G G G G G G G G G G G
440 G G G G G G G G G G G G G G G G G G G G G
455 G G G G G G G G G G G G G G G G G G G G G
470 G G G G G G G G G G G G G G G G G G G G G
485 G G G G G G G G G G G G G G G G G G G G G
500 G G G G G G G G G G G G G G G G G G G G G

FIGURE 7   Example design chart for checking stress-to-strength ratio at top of subgrade: preliminary stress-to-strength ratio at top of
subgrade for HMA-O structure, Sacramento County (RHMA-O, no subbase) (E_GB 5 60 MPa; speed 5 7 km/h) (G 5 stress-to-strength
ratio < 0.3; Y 5 0.3 Ä stress-to-strength ratio Ä 0.7; R 5 stress-to-strength ratio > 0.7). (Source: 18, Appendix E, Table E.6.)

management, and the preparation of guidelines for the design, con-
struction, and maintenance of these pavements, will be possible only
after a performance evaluation in full-scale field experiments and
pilot studies.
Conclusions
Several key findings resulted from the computer modeling of structural
capacity and the development of structural designs:
• The use of mechanistic–empirical pavement design equa-
tions developed in this project appears to be effective in estimating
required structural thicknesses for fully permeable pavements to
carry heavy truck traffic.
• The pavement structures developed were considered to be rea-
sonable, with all pavement structures less than 1.5 m thick (includ-
ing reservoir and subbase layers) and most concrete slabs less than
0.46 m thick for the heaviest traffic. The use of a PCC-O subbase will
provide considerably greater protection against the risk of subgrade
rutting for asphalt pavements.
• Construction and maintenance experts reviewed design cross
sections developed for the shoulder retrofit of highways as well as
low-speed trafficked areas such as parking lots and maintenance
yards and considered them to be feasible for construction and
maintenance.
pavement testing and pilot study field validation for hydraulic and
structural design considerations. The actual designs for those sections
(following the procedure discussed above) will depend on the sub-
grade soil permeability, climate, and design truck traffic levels. The
pavements should be instrumented to monitor stresses, strains, deflec-
tions, temperatures, thermal gradients, moisture content, and water
quality. In PCC-O test sections, optimal mix designs and the use of
proprietary additives should be discussed with ready-mix suppliers.
Life-Cycle Considerations
Detailed life-cycle cost analysis and environmental life-cycle analy-
sis should be performed after the construction, evaluation, and per-
formance validation of accelerated and field test sections. These
analyses will provide more realistic initial cost information and
improved cost estimates for maintenance and rehabilitation.
Other Pavement Types
It is recommended that this study be extended to consider inter-
locking concrete blocks as an alternative to asphalt- and concrete-
surfaced pavement. It is recommended that a life-cycle cost analysis
be performed as part of that study to provide example information
for comparison with the pavement types considered in this project.

Recommendations Acknowledgments

Accelerated Pavement Testing and Pilot Studies
It is recommended that the design method and cross sections to meet
hydraulic and structural performance requirements developed in this
project be used to design appropriate test sections for accelerated
This paper describes research activities that were requested and spon-
sored by the California Department of Transportation (­Caltrans),
Division of Environmental Analysis Stormwater Program. The
authors gratefully acknowledge Caltrans sponsorship, particularly
the technical review of Bhaskar Joshi.
 94
References
  1. Hansen, B. Porous Pavement Increases Storage Area at Portland Marine
Terminal. Civil Engineering, Vol. 77, No. 3, 2007, American Society of
Civil Engineers, Reston, Va., pp. 24–27.
 2. Hansen, K. Porous Asphalt Pavements for Stormwater Management.
Information Series 131. National Asphalt Pavement Association, ­Lanham,
Md., 2008.
 3. Stormwater Management with Pervious Concrete Pavement. American
Concrete Pavement Association Skokie, Ill., 2009.
 4. Smith, D. R. Permeable Interlocking Concrete Pavements. Selection,
Design, Construction, and Maintenance. Interlocking Concrete Pavement
Institute, Herndon, Va., 2006.
  5. Kayhanian, M., L. Chai, and B. Givens. Hydraulic Performance Evalu-
ation of Permeable Pavement Under Heavy Load and Heavy Traffic.
CTSW-RT-10-247.03D. University of California, Davis, 2010.
 6. Lu, Q. Truck Traffic Analysis Using Weigh-In-Motion (WIM) Data
in California. University of California Pavement Research Center,
­Richmond, 2002.
  7. Jones, D., J. Harvey, H. Li, and B. Campbell. Summary of Laboratory
Tests to Assess Mechanical Properties of Permeable Pavement Materials.
CTSW-TM-09-249.01/UCPRC-TM-2009-05. University of C ­ alifornia
Pavement Research Center, Richmond, 2009.
  8. Tutumluer, E. Predicting Behaviour of Flexible Pavements with Granular
Bases. PhD dissertation. Georgia Institute of Technology, Atlanta, 1995.
  9. Li, H., J. Harvey, and D. Jones. Summary of a Computer Modeling
Study to Understand the Performance Properties of Fully Permeable
Pavements. CTSW-TM-09-249.02/UCPRC-TM-2010-04. University
of California Pavement Research Center, Richmond, 2010.
10. Uzan, J. Characterization of Granular Material. In Transportation
Research Record 1022, TRB, National Research Council, Washington,
D.C., 1985, pp. 52–59.
11. Tutumluer, E., and M. R. Thompson. Anisotropic Modeling of Granu-
lar Bases. Final Report to Federal Aviation Administration Center of
View publication stats
Transportation Research Record 2305
Excellence for Airport Pavements. COE Report No. 2. University of
Illinois at Urbana–Champaign, Champaign, 1998.
12. Davids, W. G., G. M. Turkiyyah, and J. P. Mahoney. EverFE: Rigid
Pavement Three-Dimensional Finite Element Analysis Tool. In Trans-
portation Research Record 1629, TRB, National Research Council,
­Washington, D.C., 1998, pp. 41–49.
13. Signore, J., J. Harvey, M. Kayhanian, N. Santero, R. Wu, M. Troxler,
and F. Farshidi. Performance Evaluation of Permeable Pavements for
Water Quality Under Heavy Traffic. UCPRC-RR-2007-02. University
of California Pavement Research Center, Richmond, 2007.
14. LEAP 2.0: A Layered Elastic Analysis Program (Release 2.0, March
2004). Symplectic Engineering Corporation, Berkeley, Calif., 2004.
15. Miner, M. A. Cumulative Damage in Fatigue. Journal of Applied
Mechanics, Vol. 12, 1945, pp. A159–A164.
16. Darter, M. I., and E. J. Barenberg. Design of Zero-Maintenance Plain
Jointed Concrete Pavement. Volume 1: Development of Design Proce-
dures. Report FHWA-RD-77-III. FHWA, U.S. Department of Trans-
portation, 1977.
17. Bejarano, M. O., and M. R. Thompson. Subgrade Soil Evaluation for the
Design of Airport Flexible Pavements. Report to FAA Center of Excel-
lence for Airport Pavements. COE Report No. 8. University of Illinois
at Urbana–Champaign, Champaign, 1999.
18. Jones, D., J. Harvey, H. Li, T. Wang, R. Wu, and B. Campbell. Labo-
ratory Testing and Modeling for Structural Performance of Fully
Permeable Pavements Under Heavy Traffic: Final Report. CTSW-
RT-10-249.04/UCPRC-RR-2010-01. University of California Pavement
Research Center, Richmond, 2010.
The contents of this paper reflect the views of the authors and do not necessarily
reflect the official views or policies of the State of California or the Federal Highway
Administration.
The Flexible Pavement Design Committee peer-reviewed this paper.