Академический Документы
Профессиональный Документы
Культура Документы
Current mechanistic procedures for structural design of flexible pave- control. Field procedures for measuring the modulus and Poisson’s
ments consider the modulus and Poisson’s ratio of each layer. Unfortu- ratio of each pavement layer shortly after placement are presented.
nately, the construction specifications are not based on these engineering The devices and procedures introduced here have one thing in com-
properties. The acceptance criteria typically are based on adequate den-
mon: they measure seismic wave velocities. Seismic wave velocities
sity of the placed and compacted materials. To successfully implement
any mechanistic pavement design procedure, and to move toward per- can be transformed easily to moduli by using fundamentally cor-
formance-based specifications, it is essential to develop tools that can rect relationships. Modulus of each layer is one of the main para-
measure the modulus and Poisson’s ratio of each layer. Presented is an meters input to any mechanistic pavement design for flexible
approach to such a program based on seismic testing. Field protocols and pavements. A discussion on the nature of the moduli measured with
test equipment, which in a rational manner combine the results from labo- the seismic methods is included. A brief description of each method
ratory and field tests with those used for quality control during construc- is also included because some of them may be new to the pavement
tion, are discussed. A series of simplified laboratory tests that are com-
patible with the field tests also can be used; these methods are discussed. industry.
Several case studies are included to present some results that can be Several case studies are included to show some results that can be
obtained with the methodology. Several issues that remain to be addressed obtained with some of the devices. That section also contains the
are included. issues that have yet to be resolved.
develop correlations that can be used readily in the field. Simplified mined with different methods are shown in Table 1. The existing
laboratory tests will be discussed later. problems with each method are also included in the table. In the
For field testing, one should consider the differences and similar- MR test, secant moduli at different confining pressures and devia-
ities between “material characterization” and “design simulation.” toric stresses are measured (see Figure 1). After a resilient modu-
In material characterization one attempts, in a way that is the most lus test is performed on a specimen, a mathematical model is fitted
theoretically correct, to determine the engineering properties of a to the data. The recommended model at this time, based on a recent
material (such as modulus or strength). The material properties mea- NCHRP project (3), is in the form of
sured in this way are fundamental material properties that are not
related to a specific modeling scenario. To use these material prop- E = MR = k1 σ ck2 σ kd3 (3)
erties in a certain design methodology, they should be combined
with an appropriate analytical or numerical model to obtain the design where σd and σc are the deviatoric stress and confining pressure,
output. In the design simulation, one tries to the best of one’s abil- respectively. Parameters k1 through k3 are coefficients statistically
ity to experimentally simulate the design condition, and then back- determined from the results of the laboratory test.
figure some effective material parameters that are relevant only to The relation of the seismic moduli with the other two moduli is
that condition. The seismic methods can be considered as methods also of concern. Nazarian et al. (4) present an extensive derivation
that provide material characterization, whereas the deflection- of these relationships. In general
based methods are geared more toward the design simulation. Both
of these approaches have advantages and disadvantages. The design σ
k2
FIGURE 1 Schematic demonstration of moduli as function In the time-domain analysis, one relies on identifying the time at
of stress.
which different types of energy arrive at each sensor. The velocity
of propagation, V, typically is determined by dividing the distance
where between two receivers, ∆X, by the difference in the arrival time of a
specific wave, ∆t. In general, the relationship can be written in the
α = a number approximately 0.05 to 0.10, following form:
σgeo = geostatic stress due to the weight of the overlying materials
and is equal to σv (1 + 2ko) /3, ∆X
ko = coefficient of earth pressure at rest, and V= (5)
∆t
σv = vertical stress due to the weight of the soil or pavement
layers. In the equation, V can be the propagation velocity of any of the three
waves [i.e., compression wave, VP; shear wave, VS; or surface
The conceptual way of representing the seismic moduli is simple
because the states of stress before and after the seismic loading are (Rayleigh) wave, VR]. If wave velocity is known, modulus can be
very similar. Referring to Equation 4, this method yields a funda- determined in several ways. Young’s modulus, E, can be determined
mental material property that can be used in quality control and can from shear modulus, G, through Poisson’s ratio (ν) by using
be used in performance-based specification, provided values for k2
E = 2 (1 + ν) G (6)
and k3 are known. Laboratory tests are necessary to determine the
values for k2 and k3.
Shear modulus can be determined from shear wave velocity, VS,
by using
OVERVIEW OF METHODS γ
G = VS2 ( 7)
g
On the basis of the background information provided, the goal is to
develop modulus-based tests that can be readily used in the field for To obtain modulus from surface wave velocity, VR is first converted
quality control of any layer in the flexible pavement system. In the to shear wave velocity by using
next section, the procedures necessary for this task are introduced.
As in any other quality management program, acceptance criteria VS = VR (1.13 − 0.16 ν) (8)
should be developed. The proposed acceptance criteria are based on
seismic testing of specimens prepared in the laboratory. The speci- The shear modulus then is determined by using Equation 7.
mens used for this purpose are similar to those used in determining Typical records from two sensors are shown in Figure 2(b) for a
the job mix formula for asphalt concrete pavement (ACP) and for granular material and Figure 2(c) for an ACP. As an example, the
moisture-density tests for base and subgrade. The laboratory test arrivals of compression, shear, and surface waves are marked on
setup and procedures are included herein. Figure 2(c). The compression wave (or P-wave) energy is reason-
Another aspect of the quality management is providing feedback ably easy to identify because it is the earliest source of energy to
to the designer about the design moduli of different layers during appear in the time record. Because only less than 10 percent of the
construction. The designer, in cooperation with the construction seismic energy propagates in this form, the peak compression wave
engineer, ideally would use that information to adjust the thick- energy in the signal sometimes is only several times above the inher-
nesses of different pavement layers to achieve the envisioned ent background noise. This limitation may make it difficult always
remaining life. This task, which is being developed, is not discussed to estimate reliably the arrivals of these waves.
FIGURE 2 Seismic testing setup and typical time records.
54 Paper No. 99-1561 TRANSPORTATION RESEARCH RECORD 1654
The shear wave (or S-wave) energy is about one-fourth of the mining the modulus. The method is called spectral analysis of surface
seismic energy and as such is better pronounced in the record. The waves (SASW) (7).
practical problem with identifying this type of wave is that it prop- The goal with the SASW method is to generate and detect surface
agates at a speed that is close to that of the surface waves. Thus, the waves over a wide range of wavelengths. The time records collected
separation of the two energies, at least for short distances from the with the setup described earlier are transformed to a so-called dis-
source, may be difficult. persion curve—a plot of velocity of propagation of surface waves
Surface (Rayleigh) waves contain about two-thirds of the seismic with wavelength. If the goal is only to determine the modulus of the
energy. As marked in Figure 2, the most dominant arrivals are top layer, the method becomes straightforward.
related to the surface waves, and thus it should be easy to measure Consider the time records shown in Figures 2(b) and 2(c). By per-
them. If a layer does not have surface imperfections, and if the forming a fast Fourier transform on the two signals, and by dividing
impact is “sharp” enough to generate only waves that contain energy the two transformed signals by one another, one obtains a phase
for wavelengths shorter than the thickness of the top layer, this spectrum (i.e., variation in phase with frequency). Such phases are
method can be used readily to determine the modulus. However, it shown in Figures 3(a) and 3(b) for a granular material and ACP,
may be difficult to observe these two restrictions. The frequency- respectively. The phase shown in Figure 3(a) or 3(b) can be
domain analysis, although more complex to implement, is by far “unwrapped” and fitted by a straight line. The slope of such a line,
more robust than the time-domain analysis. m, is directly related to Young’s modulus, E, by using (6)
γ 360 ∆ X 2
E = 2 (1 + ν) (1.13 − 0.16 ν) ( 9)
Frequency-Domain Data Reduction g m
Because most of the energy in a seismic wave train is carried by sur- where ν is Poisson’s ratio and ∆X is the sensor spacing. As before,
face waves, one can take advantage of the signal processing and and γ and g are the unit weight and the acceleration of gravity,
spectral analysis to develop a more robust methodology for deter- respectively. Alternatively, one can construct a dispersion curve, as
FIGURE 3 Typical phase spectra and dispersion curves for granular materials and ACP.
Nazarian et al. Paper No. 99-1561 55
γ
EUSW = 2 (1 + ν) [(1.13 − 0.16 ν)Vph ]
2
(10)
g
where Vph is the average phase velocity of the top layer. Baker et al.
(6) have developed a device that can perform this test in the field in
less than 1 min per point [see Figure 2(a)].
One of the tools that typically are used in pavement engineering
is the dynamic cone penetrometer (DCP). This test can quantify
the layers and qualify the type of material used. A three-dimensional
accelerometer package has been retrofitted in the cone of a DCP
to measure the modulus and Poisson’s ratio. The test proposed is
the so-called downhole seismic test (8). The receivers are placed
at the depth at which tests have to be performed. The pavement
surface then is impacted with a small hand-held hammer. The
records from the receivers are retrieved and saved for future analy-
sis. The reduction of data consists of determining the arrivals of
different waves very similar to those carried out for the time-
domain analysis described above. If the shear and compression
wave velocities are known, Poisson’s ratio, v, can be determined
readily by using:
where α = VP /VS. (VS and VP are shear and compression wave veloci-
ties, respectively.) For the sake of brevity, this equipment is not dis- sophisticated ones during the design process. Two procedures are
cussed any further. For more information, refer to Nazarian et al. (5). proposed in this study.
A schematic of the test setup for the free-free resonant column
tests is shown in Figure 4. The specimen is either suspended by two
Seismic Laboratory Tests wires or placed on a material that is substantially less stiff than the
specimen (e.g., Styrofoam). A accelerometer is securely placed on
One of the major goals of the project is to develop field tests that are one end of the specimen, and the other end is impacted with a ham-
compatible with laboratory results. As indicated, the existing tests mer instrumented with a load cell. The signals from the accelero-
used to determine the modulus of asphalt concrete (AC), base, and meter and load cell are used to determine the resonant frequency, f,
subgrade in the laboratory are cumbersome and time consuming. as shown in Figure 5. After the frequency; mass density, ρ; and length
Simplified laboratory tests can be used in conjunction with the more of the specimen, L, are known, Young’s modulus can be found from
where ρ and ν are the bulk density and Poisson’s ratio of the speci-
men, respectively.
CASE STUDIES
modulus. A significant drop in modulus is associated with a small moduli basically are bulk moduli that are seriously affected by the
variation in dry density. This may partially explain the significant surface imperfections. Moduli obtained from the frequency domain
differences between the laboratory models and the field. analysis are the average modulus over a range of thickness explic-
Similar tests were performed on the base material about 1000 m itly defined during the data reduction. Use of the frequency domain
away from the subgrade site. Tests again were carried out at 2-m analysis is recommended because although more complex to imple-
intervals. The variation in modulus with test location is shown in ment, it is by far more robust. The average modulus from the time
Figure 8(a). Unlike with the subgrade, some differences are evident domain is less than the frequency domain. This usually happens
between the results from the time and frequency domain analyses. because the bulk wavelength of the signal may be longer than the
These differences typically show up when the surface is coarse and thickness of the base. Poisson’s ratio along the test section, as shown
microcracks are present at a site. As indicated, the time domain in Figure 8(b), varies between 0.38 and 0.43. The reason for one
outlier is not known at this time.
A comparison of different moduli related to this base is shown in
Figure 8(c). The average seismic moduli is about 840 MPa with a
coefficient of variation of about 23 percent. The resilient modulus
of the base measured as about 470 MPa when the specimen was pre-
pared at the average moisture content and density of the field, which
is about 80 percent less than the seismic modulus. As presented in
Table 1, these levels of differences are anticipated. The average
modulus measured by using free-free resonant column test on labo-
ratory specimens prepared to the average density and moisture of the
base is about 745 MPa. However, when the specimens were pre-
pared to maximum density as per the Proctor method, the seismic
modulus was about 387 MPa. These experiments indicate that the
relationship between the field and the laboratory compaction should
be considered and addressed more carefully.
From this case study we learned that trends exist between field
moisture content, density, CBR, and seismic modulus. These rela-
tionships should be explored and described in more detail in the
future. We also learned that specimens prepared as per the Proctor
method may yield moduli that are less than those measured in the
field. However, if the laboratory specimens are prepared at the den-
sity and moisture level measured in the field, closer relationships
between laboratory and seismic moduli can be developed. A means
of balancing these two parameters should be considered.
AC Layer
A 45-m section of the access road of I-20 near Odessa was tested with
the seismic methods. The pavement section at this site consisted of
about 50 mm of AC over about 250 mm of base over subgrade. The
portable seismic pavement analyzer (6) was used in field tests.
As shown in Figure 9, the modulus of the AC layer adjusted to a
temperature of 25°C is relatively constant except for an area be-
tween 25 m and 35 m. This area coincides with the entrance to a For AC layers, moduli measured in situ and in the laboratory with
business entity where a new drainage pipe was installed. On the different seismic testing devices are very close. They also are close
average the modulus of the AC layer is about 13.8 GPa when all to those measured with deflection-based methods such as falling
points are included and 14.9 GPa when the results from the area weight deflectometer (FWD) tests.
between 25 m and 35 m are ignored. Seven cores retrieved from this For the base and subgrade, there is good agreement between the
site were tested in the laboratory using the ultrasonic device. The seismic moduli measured in the field and that measured in the labo-
average modulus from the laboratory tests is about 14.3 GPa. The ratory as long as the laboratory specimens are prepared at the den-
point-by-point comparison of moduli from the laboratory and the sity and moisture content of the field materials. Moduli measured
field is provided in Figure 9. The results from the two tests are quite with seismic methods are higher than those obtained from other test-
close with an average difference of about 4 percent and a maximum ing methods such as the resilient modulus and FWD tests. Large
difference of about 12 percent. variabilities in the base and subgrade moduli were observed and
In addition to the cores, loose AC material was saved for further found to be related to the test location in most cases.
laboratory study. About a dozen briquettes were made by using a SHRP
gyratory compactor. The specimens were subjected to different num-
ber of gyrations so that specimens with different voids in total mix ACKNOWLEDGMENTS
(VTM) could be built. The variation in modulus with the VTM on
these specimens is shown in Figure 10. A linear relationship between This work was supported by the Texas Department of Transporta-
the modulus and the VTM can be observed. The least-squares best- tion. The authors would like to express their sincere appreciation
fit line to the data yield an R2 of about 0.88. Also shown in the fig- to Stephen Smith, Roger Cisneros, and Mark McDaniel for their
ure are the results from the actual field measurement. The slopes ever-present support and valuable advice. Thanks also to the lab-
of the best-fit lines through the field data and the laboratory- oratory personnel of the El Paso district for their assistance and
prepared briquettes are more or less similar. However, the two support.
curves are shifted. The shift perhaps can be attributed to the dif-
ferences in the compaction method. If this trend is observed at
other sites, one can use either a different method of compaction or REFERENCES
perhaps one or two field cores to calibrate the modulus values
while maintaining the slope. 1. Finn, F., S. Saraf, R. Kulkarini, K. Nair, W. Smith, and A. Abdullah. The
Use of Distress Prediction Subsystems for the Design of Pavement Struc-
tures. Proc., 4th International Conference on the Structural Design of
SUMMARY AND CONCLUSIONS Asphalt Pavements, 1977, pp. 3–37.
2. Shook, J. F., F. N. Finn, M. W. Witczak, and C. L. Monismith. Thickness
Design of Asphalt Pavements—The Asphalt Institute Method. Proc., 5th
The nature, role, and feasibility of measuring the properties of flex- International Conference on the Structural Design of Asphalt Pavements,
ible pavement material in the laboratory and in the field with differ- 1982.
ent types of nondestructive methods and devices are discussed and 3. Barksdale, R. D., J. Alba, P. N. Khosla, R. Kim, P. C. Lambe, and M. S.
Rahman. Laboratory Determination of Resilient Modulus for Flexible
summarized. In general, results from field tests should be combined
Pavement Design. Interim Report Project 1-28. FHWA, U. S. Department
with those from laboratory tests for their use in quality control or of Transportation, Washington, D.C., 1994.
realistic design. 4. Nazarian, S., D. Yuan, and V. Tandon. Specifications for Tools Used in
Seismic methods measure the engineering properties of pave- Structural Field Testing of Flexible Pavement Layers. Research Report
ment materials. Seismic testing both in the laboratory and in the 1735-1. Center for Highway Materials Research, The University of Texas
at El Paso, 1998.
field is rapid and quite repeatable. Seismic modulus is sensitive to 5. Nazarian, S., I. Abdallah, D. Yuan, and L. Ke. Design Modulus Values
variations in moisture content and dry density of the base and Using Seismic Data Collection. Technical Memorandum 1780. Center for
prepared subgrade. Highway Materials Research, The University of Texas at El Paso, 1998.
60 Paper No. 99-1561 TRANSPORTATION RESEARCH RECORD 1654
6. Baker, M. R., K. Crain, and S. Nazarian. Determination of Pavement 8. Woods, R. D. Measurements of Dynamic Soil Properties. Proc., Geo-
Thickness with a New Ultrasonic Device. Research Report 1966-1. Cen- technical. Engineering Division Specialty Conference on Earthquake
ter for Highway Materials Research, The University of Texas at El Paso, Engineering and Soil Dynamics. Pasadena, Calif., Vol. 1, ASCE, New
1995. York, 1978, pp. 91–180.
7. Nazarian, S., D. Yuan, and M. R. Baker. Rapid Determination of Pave-
ment Moduli with Spectral-Analysis-of-Surface-Waves Method. Research Publication of this paper sponsored by Committee on Flexible Pavement
Report 1243-1. The University of Texas at El Paso, 1995. Construction and Rehabilitation.