You are on page 1of 7

Alternative Approach for Interpreting

Traffic Speed Deflectometer Results


Adam Zofka, Jacek Sudyka, Maciej Maliszewski, Przemysław Harasim,
and Dariusz Sybilski

This paper presents an alternative approach for interpreting data in this paper because the proposed approach cannot be applied to
collected with the traffic speed deflectometer (TSD). Although the idea data collected by the RWD in the current version.
of measuring pavement deflections at traffic speeds in a continuous
manner is not new, worldwide interest is recent, in part because of the
availability of a commercial TSD device. As with any new device, there is Traffic Speed Deflectometer
a need for robust, scientifically sound, yet practical interpretation pro-
cedures. This paper describes the details of such a procedure recently The first TSD prototype device was developed in Denmark in the
developed at the Road and Bridge Research Institute in Poland. In addi- late 1990s (17 ). The device has been constantly evolving with
tion to a detailed description of the proposed algorithm and its unique new features, but the concept remains the same: vertical deflection
features, this study demonstrates an example application that uses data velocities are measured by a set of Doppler sensors installed on a
collected by the TSD operated by the Road and Bridge Research Institute. semi truck with a rear axle load of 100 kN (22 kips). The deflec-
The proposed approach can be successfully applied to the TSD data and tion velocities divided by the instantaneous vehicle speed give the
allows for deep customization for a user’s and customer’s needs and deflection slopes (18, 19). Absolute deflections can be obtained
applications. by integrating the deflection slopes numerically (20) or by using
a closed-form solution of a mechanical model, such as an elastic
beam on a Winkler foundation (19). Once the deflection bowl is
Deflections allow for direct measurement of the bearing capacity of obtained from a set of Doppler sensors at one TSD location, it can
a pavement structure. Because of the growing number of pavements be compared with the FWD deflection bowl measured at the same or
in a deteriorated condition, information on structural capacity is an a nearby location. This comparison is important for many agencies
integral part of many pavement management systems. Deflections because it increases the credibility of the new method and allows for
can be implemented into the pavement management system through smooth adaptation. Trial comparisons were conducted in Denmark
standardized measured deflections, or they can be part of a stand-alone (21), the United Kingdom (22, 23), France (24), Australia (25), and
or combined condition index, for example, the structural adequacy Italy (26), and in most cases comparisons were satisfactory. Similar
index and the pavement quality index, respectively. Deflections also studies have been planned in Poland, the United States, and South
can be included in the maintenance and rehabilitation decision process, Africa. However, as recently pointed out in another study, a com-
for example, as a screening tool for homogeneous pavement segments, parison of TSD and FWD results should be done with caution, and
and they can be implemented into deterioration models to increase special statistical methods, such as limits of agreement, should be
their prediction accuracy (1–10). used rather than a simple linear regression (27).
Deflections traditionally have been measured with static devices,
such as the falling weight deflectometer (FWD) or the Benkelman
beam. However, interest has shifted in recent years toward con- Eulerian and LaGrangian Descriptions
tinuous deflectometers that can measure pavement deflections at
highway speeds. These devices provide a safe and efficient tool for In general, any combination of spatial and temporal data can be
relatively quickly evaluating even an extensive road network. A analyzed with two alternative (yet to some degree complementary)
recent SHRP 2 study identified two devices as the most promising approaches, Eulerian description and Lagrangian description. These
continuous deflectometers according to department of transportation descriptions are well known in continuum mechanics, and the key
needs: the rolling wheel deflectometer (RWD) and the traffic speed difference between them lies in the definition of the point of reference.
deflectometer (TSD) (1, 2). The RWD device operated by FHWA In Lagrangian description the focus is on an object (particle) mov-
uses four lasers to directly measure pavement deflections. The RWD ing through space, and specific equations are solved to determine
is well documented in the literature (11–16), but it is not discussed the location and status of individual objects solely as a function of
time t. Eulerian description focuses on a specific location in space
(defined by its position x) and observes the processes that occur
Road and Bridge Research Institute, 1 Instytutowa Str., PL 03-302 Warsaw, with time t at that location. With consideration of a set of points, the
Poland. Corresponding author: A. Zofka, azofka@ibdim.edu.pl. processes can be expressed for both position x and time t. The TSD
outputs are analyzed with a Lagrangian-type approach, that is, the
Transportation Research Record: Journal of the Transportation Research Board,
No. 2457, Transportation Research Board of the National Academies, Washington,
TSD device is followed as it travels on a pavement. This paper
D.C., 2014, pp. 12–18. investigates the Eulerian-type approach for interpreting data col-
DOI: 10.3141/2457-02 lected by the TSD, that is, the focus in on the specific location and

12
Zofka, Sudyka, Maliszewski, Harasim, and Sybilski 13

investigating the processes (i.e., deflections) that happen at the location


with time.

Study Objectives

Two main objectives are considered in this study:

1. Develop an alternative procedure for interpreting data collected


by the TSD device.
2. Conduct a proof-of-concept analysis with the proposed pro­
cedure and data collected by the TSD operated by the Polish Road
and Bridge Research Institute.

The proposed algorithm has two unique features. First, the approach
focuses on discrete points (i) along the TSD route and determines
the deflection history at each point i as a function of time (t), that is,
δi = f (t). This approach parallels Eulerian description, which provides
a viable alternative to the Lagrangian description.
The second unique feature of the proposed approach is that it
employs a specific function for the slope versus sensor locations vector
at each TSD measurement point (j), that is, sj = g(dDopp). Function g
should fulfill three requirements: it should follow sj versus dDopp data,
it should have no more than three coefficients, and it should have a FIGURE 1   Eulerian approach concept.
smooth and explicit integral form over the entire range defined by
the dDopp vector. The proposed approach is independent from the
assumed function as long as the function meets these requirements. conceptually presented in Figure 2. Compilation of deflection data
The Greenwood TSD device uses the Doppler effect to determine for various Eulerian locations can be further analyzed depending on
the velocity of changes in vertical deflections in the front of the TSD user needs. In the example presented later in this paper, the deflec-
load axle (Greenwood). Once deflection velocities are measured for tion histories are used for constructing deflection isochrone maps of a
each Doppler sensor location, they are divided by the instantaneous pavement section. Other potential applications are pavement structural
vehicle speed, which produces the slopes of deflection basin at the condition clustering and comprehensive comparison with FWD data.
discrete points. In this study these points are defined as the dDopp Figure 3 shows the conceptual steps of the proposed algorithm.
vector. The TSD device owned and operated by the Road and Bridge Each step can be further described as follows:
Research Institute is equipped with 6+1 Doppler sensors at the
following locations: dDopp = [0.1 0.2 0.3 0.6 0.9 1.5] m. The sensor Step 1. For each TSD location defined by the vector xj (dj), extract
at 3.5 m is used to increase the accuracy of the measurements made measured vehicle speed vj and slope sj data for each sensor location
at other points (18). defined by vector dDopp. Using location and speed data, calculate the
precise time stamp tj for each TSD location.
Step 2. Fit specific function g to sj versus dDopp data and obtain
Proposed Data Processing corresponding parameters. Function g should be flexible but should
Algorithm for TSD have no more than three coefficients (for quick and good fit). Further­
more, function g should have an explicit integral form over the entire
Available methods for analyzing TSD data focus on the Lagrangian-
range of vector dDopp. This further simplifies the calculations because
type approach. Although such an approach is correct, there can be
deflections δij can be directly calculated by integrating the slopes sj
situations in which the proposed Eulerian-type approach is more
function according to the classic (Euler–Bernoulli) beam theory.
appropriate. The Eulerian approach is schematically presented in
Step 3. Assume the size of the buffer zone b in distance (or time)
Figure 1. The key element is to focus on a single location along the
units. The buffer zone contains all TSD locations that contribute to
TSD route and to record the change of deflections at that location as
the TSD wheel passes through. In Figure 1, the considered (Eulerian)
location is described by vector xi (di), and TSD wheel locations are
depicted by vectors xj (dj) at different times tij. Each TSD location
within the buffer zone b contributes to the deflection δij at the point
xi (di) at different time tij. The contribution of the TSD locations
situated after a given Eulerian point (i.e., locations j + 1 and j + 2 in
Figure 1) is determined assuming the symmetry of their deflection
basins (gray sensors in Figure 1). This is a reasonable assumption
that is necessary for determining a full deflection history at the point
xi (di). Once all pairs of deflection δij and corresponding times tij
are determined for a given location xi (di), the deflection history at FIGURE 2   Deflection history at Eulerian
that location can be plotted. An example of the deflection history is location related to several TSD locations.
14 Transportation Research Record 2457

locations (i) for a given buffer size is correct as long as the TSD speed
is approximately constant. Alternatively, the time on the deflection
isochrone map can be replaced by the distance dij. The demonstra-
tion example presented in the next section shows both cases, that is,
δij(di, tij) and δij(di, dij).

Proof-of-Concept Example

Example slope data from the Polish TSD device were used to ver-
ify the proposed algorithm. The considered pavement section was
approximately 200 m long; although this may appear to be short,
the primary purpose of the example is to demonstrate the proposed
algorithm itself and determine deflection histories δij. Once the deflec-
tions are known, interpretation depends on the application. In this
example, deflections were presented in terms of the aforementioned
maps δij(di, tij) and δij(di, dij).
For Step 1, the TSD locations were selected every 0.1 m and
appropriate data (speed and slope sj) were extracted from the TSD
readings. The actual measurements were made at a frequency of
0.02 m by the high-speed collection and acquisition systems installed
in the TSD device (18). Combining the distance and precise speed
data (reported to 1/10,000th in meters per second units) allowed for
calculating the exact time stamps tj at each TSD location.
In the next step, function gj was fit to the slope sj versus dDopp data
FIGURE 3   Flowchart of proposed algorithm (vs. 5 versus). at each TSD location. In this example, the Weibull function was used
in the following general form:

γ
( γ −1)  x −µ 
γ  x − µ −
 α 
the deflections at a given Eulerian point. The size of the buffer zone f (x) = e x ≥ µ; α , γ > 0 (1)
α  α 
shown in Figure 1 should be at least 7 m (2 × 3.5 m) to reach beyond
the furthermost sensor installed in the TSD device.
Step 4. Assume the Eulerian locations, that is, points where where
history of deflections will be determined. Eulerian locations are γ = shape parameter,
defined by distance vector xi (di): xi is the distance from the admin- α = scale parameter, and
istrative beginning of the route (e.g., milepost), and di is the distance µ = location parameter.
from the arbitrary point (e.g., beginning of a pavement section).
Spacing of the Eulerian locations is theoretically arbitrary. In practice, If the location parameter µ is assumed to be equal to 0, the Weibull
the spacing should be determined iteratively because it depends on function has the following integral:
the quality of the collected TSD data, the desired accuracy of pre-
dicted deflections δij, and the application in general. For example,  γ 
∫ f ( x ) = −γ
1−γ
i Γ  γ, x (2)
Eulerian locations can be assumed at the exact locations of the FWD  α 
measurements for the direct comparison.
Step 5. For a given Eulerian location, identify all TSD locations where Γ implies incomplete gamma function given by
that are situated within buffer zone b. Calculate the distances dij
between given Eulerian point and selected TSD locations. ∞
Step 6. Use distance dij and calculate deflections δij by integrat- Γ ( γ , x ′ ) = ∫ t γ −1e − t dt (3)
ing corresponding function fits gj from Step 2. Because functions x′

gj were determined at the specific time stamps tj, the deflections δij
occur at times tij = tj for a given Eulerian point (i). Practically, this step determines a set of two parameters (γj and αj)
Step 7. For data analysis and visualization, normalize the time at each TSD location using the nonlinear least squares trust region
tij vector and the distance dij vector by their first values, that is, approach (28). An example of the raw slope data together with the
tij = tij − ti1 and dij = dij − di1, respectively. The ti1 and di1 values cor- Weibull fit and its integral form is presented in Figure 4.
respond to the first TSD location within the buffer zone for a given The TSD operated by the Road and Bridge Research Institute
Eulerian point. Depending on data quality, deflections δij can be has 6+1 sensors, which give six slope values at distances dDopp =
smooth if a robust numerical procedure is used, for example, the [0.1 0.2 0.3 0.6 0.9 1.5] m. If the slope at 0 m and the slope at
Savitzky–Golay filter. 3.5 m are equal to zero, a two-parameter Weibull function can be
Step 8. Analyze the deflection data δij as a function of time tij or fit to the set of eight points. However at some locations the slope
distance dij for a specific Eulerian point or for all Eulerian locations values were not reliable and could not be used for fitting. To address
simultaneously. This step depends on the specific user or customer this issue, a filter was set to omit the TSD locations with fewer than
needs and the particular application. Linking ti1 = 0 among Eulerian five reliable data points (of eight possible). This filtering disrupts
Zofka, Sudyka, Maliszewski, Harasim, and Sybilski 15

0.3 150
Slope data @ TSD locations
0.25 Weibull fit
125
Slope (mm/m)

0.2

Number of Points (count)


0.15
100
0.1

0.05 75

0
0 1 2 3 4 50
(a)

25
0

−0.05 0
Deflection (mm)

20 30 40 50 60 70
−0.1
Number of TSD Locations per Eulerian Point
−0.15 Calculated deflection
FIGURE 5   Histogram of TSD locations assigned to Eulerian points.
−0.2

−0.25
deflection δij was assigned an exact time stamp tij determined in
0 0.5 1 1.5 2 2.5 3 3.5 Step 1. Figure 6 shows the two examples of the deflection basins
Distance from the TSD Load Axle (m) expressed in terms of the normalized time tij. The smoothing in
Step 7 incorporated the Savitzky–Golay filter, which is a general-
(b)
ized moving average based on the two-degree polynomial fitted on
FIGURE 4   Example plot: (a) slope data s j and with Weibull g j (d Dopp) 19 adjacent points. Although data in Figure 6 are scattered, they fol-
and (b) corresponding deflections d ij (integrated Weibull function). low a logical and expected trend discussed in Figure 2. It is believed
that the scatter is driven primarily by the quality of the collected
slope data sj as well as the quality of the fit gj in Step 2.
In the final step of the proposed procedure (Step 8), interpretation
the uniform spacing of the TSD locations but it does not influence of the deflections δij data can be customized to meet the application.
the entire algorithm as long as the filtered locations are randomly In this demonstration example, deflections δij were combined with
distributed. the time tij and distance dij data to construct isochrone maps: tempo-
In Step 3, the buffer zone was set to 2 × 5 m. The extra space ral δ(di, tij) and purely spatial δij(di, dij). Examples of such maps are
beyond the range of the TSD sensors (2 × 3.5 m) was reserved for shown in Figures 7, 8, 9, and 10. All these figures were constructed
smoothing and other numerical operations on the deflection data for the same 200-m pavement section. Besides considering use of
that are highlighted in the following steps. time tij versus distance dij data, a factor studied in these figures is the
The Eulerian locations were chosen as between the TSD locations number of Eulerian points (two scenarios introduced in Step 4). The
(Step 4). In the first scenario, the number of Eulerian points was set following observations can be made:
to 1,000, and one Eulerian point followed each TSD location. In the
second scenario, there were 10 times fewer Eulerian points, that is, • Time tij versus distance dij data (Figure 7 versus Figure 9;
one Eulerian point was located at every 10 TSD locations. Figure 5 Figure 8 versus Figure 10) produce very similar maps. This should
shows the histogram of the number of TSD locations assigned to be the case as long as the TSD speed is reasonably constant through-
the Eulerian points in the first scenario (Step 5). The median value out the considered pavement section. This example used a 200-m
is equal to 46, which is less than the theoretical value of 100 points segment, and it appears that the speed was consistent.
[the total buffer size (10 m) divided by the spacing of the TSD loca- • Increasing the number of Eulerian points (Figure 7 versus
tions (0.1 m)]. The result was expected because some TSD locations Figure 8; Figure 9 versus Figure 10) introduces undesirable noise in
were filtered out and the number of points within the buffer zone the maps. Although the noise depends on the quality of the data sj and
must decrease. Forty-six TSD locations within the buffer zone pro- fit gj, the spacing of Eulerian points should be carefully selected with
duced on average deflection value every 0.22 m, which is sufficient consideration for pavement section length and analysis objective.
for plotting the entire deflection basin at each Eulerian point.
Once the TSD locations ( j ) were established for a particular The maps in this example were constructed with a straightforward
Eulerian point (i), Equation 2 was used to calculate deflections at interpolation method, in which the value at an interpolated point
that point due to every TSD location in the set ( j ) (Step 6). The matched the value of the nearest data point. Although other methods
deflections were calculated with the distances dij and appropriate could be implemented, the nearest neighbor method introduces
coefficients γj and αj derived in Step 2 for every TSD location. Each marginal data alterations and appears to be computationally effective.
0 0

−0.1 −0.1

−0.2 −0.2
Deflections at Eulerian Point

Deflections at Eulerian Point


−0.3 −0.3

−0.4 −0.4

−0.5 Raw data −0.5


Smooth data
−0.6 −0.6

−0.7 −0.7

−0.8 −0.8 Raw data


Smooth data
−0.9 −0.9
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Time (s) Time (s)
(a) (b)

FIGURE 6   Examples of deflection history at two Eulerian points.

0 0.1
0.05 0
0.1 −0.1
Time (s) at Eulerian Point

0.15 −0.2
0.2
−0.3
0.25
−0.4
0.3
−0.5
0.35
−0.6
0.4
−0.7
0.45
0.5 −0.8

20 40 60 80 100 120 140 160 180 200


Eulerian Points i (0.1 m) (−−−−−−−> TSD Direction)

FIGURE 7   Deflection isochrone map d ij(d i, t ij) for time: 1,000 Eulerian points, approximately 200 m, buffer size
b 5 2 3 5 m (color-coded deflections in millimeters).

0 0.1

0.05 0

0.1 −0.1
Time (s) at Eulerian Point

0.15 −0.2
0.2 −0.3
0.25 −0.4
0.3
−0.5
0.35
−0.6
0.4
−0.7
0.45
−0.8
20 40 60 80 100 120 140 160 180 200
Eulerian Points i (0.1 m) (−−−−−−−> TSD Direction)

FIGURE 8   Deflection isochrone map d ij(d i , t ij) for time: 100 Eulerian points, approximately 200 m, buffer size
b 5 2 3 5 m (color-coded deflections in millimeters).
Zofka, Sudyka, Maliszewski, Harasim, and Sybilski 17

−5 0.1

−4 0

Distance dij (m) at Eulerian Point i


−3 −0.1
−2 −0.2
−1 −0.3
0
−0.4
1
−0.5
2
−0.6
3
−0.7
4
−0.8
5
20 40 60 80 100 120 140 160 180 200
Eulerian Points i (0.1 m) (−−−−−−−> TSD Direction)

FIGURE 9   Deflection isochrone map d ij (d i, d ij) for distance: 1,000 Eulerian points, approximately 200 m,
buffer size b 5 2 3 5 m (color-coded deflections in millimeters).

−5 0.1

−4 0
Distance dij (m) at Eulerian Point i

−3 −0.1
−2
−0.2
−1
−0.3
0
−0.4
1
−0.5
2
−0.6
3
−0.7
4

5 −0.8
20 40 60 80 100 120 140 160 180 200
Eulerian Points i (0.1 m) (−−−−−−−> TSD Direction)

FIGURE 10   Deflection isochrone map d ij(d i, d ij) for distance: 100 Eulerian points, approximately 200 m, buffer
size b 5 2 3 5 m (color-coded deflections in millimeters).

Summary and Conclusions The deflection isochrone maps can be used, for example, to
identify uniform pavement sections that have a similar struc-
This paper presented an alternative approach for processing data tural capacity by using appropriate clustering procedure. In other
collected by the TSD device. Such data provide valuable informa- applications, deflection values at the Eulerian points (or inter­
tion on the structural capacity of pavements at the network level polated between) can be compared with the FWD measurements
that is particularly important to agencies facing aging infrastructure as a function of time or distance, easing transition to new equipment
and limited budgetary resources. A growing number of agencies and increasing data reliability. In another application, deflection
are shifting toward incorporating structural information into their histories at the discrete (Eulerian) points can be used to determine
comprehensive pavement management systems. the necessary pavement structural index for pavement rehabilitation
This paper discussed in detail a proposed algorithm and presented design.
a successful demonstration example. Two unique features of the The proposed approach is expected to be advanced in the near
proposed algorithm are emphasized: future, and its precision and accuracy will increase because of at
least three factors:
• The Eulerian approach allows for analyzing deflection history
at a particular location in a pavement. • Better quality of slope sj = f (dDopp) data once the technology
• Flexibility and customization permit adaptation to a variety of develops, more TSD devices are in routine use, and advanced
applications. smoothing procedures are available (29);
18 Transportation Research Record 2457

• Incorporation of the dynamic force effect that would have sig- 16. Rada, G. R., and S. Nazarian. The State-of-the-Technology of Moving
nificant influence on the resultant deflections; incorporation of the Pavement Deflection Testing. Final report. FHWA-DTFH61-08-D-00025.
FHWA, U.S. Department of Transportation, 2011.
force effect in the TSD interpretation procedure would require a more 17. Hildebrand, G., and S. Rasmussen. Development of a High Speed Deflec-
advanced mechanical model, which is already available (30–33); and tograph. Report 117. Danish Road Institute, Copenhagen, Denmark,
• Incorporation of the pavement temperature effect that also has 2002.
an impact on deflection, especially in a case of the flexible pavements. 18. Rasmussen, S., J. Krarup, and G. Hildebrand. Non-Contact Deflection
Measurement at High Speed. Presented at 6th International Conference
on the Bearing Capacity of Roads, Railways, and Airfields, Lisbon,
­Portugal, 2002.
References 19. Krarup, J., S. Rasmussen, L. Aagaard, and P. Hjorth. Output from the
Greenwood Traffic Speed Deflectometer. Presented at 22nd ARRB
1. Flintsch, G., S. Katicha, J. Bryce, B. Ferne, S. Nell, and B. Diefenderfer. Group Conference, Canberra, Victoria, Australia, 2006.
Assessment of Continuous Pavement Deflection Measuring Technologies. 20. Muller, W., and J. Roberts. Revised Approach to Assessing Traffic
SHRP Report S2-R06F-RW-1. Transportation Research Board of the Speed Deflectometer Data and Field Validation of Deflection Bowl
National Academies, Washington, D.C., 2013. Predictions. International Journal of Pavement Engineering, Vol. 13,
2. Flintsch, G. W., B. Ferne, B. Diefenderfer, S. Katicha, J. Bryce, and S. Nell. No. 4, 2013, pp. 388–402.
Evaluation of Traffic-Speed Deflectometers. In Transportation Research 21. Rasmussen, S., L. Aagaard, S. Baltzer, and J. Krarup. A Comparison
Record: Journal of the Transportation Research Board, No. 2304, Trans- of Two Years of Network Level Measurements with the Traffic Speed
portation Research Board of the National Academies, Washington, D.C., Deflectometer. Presented at Transport Research Arena Conference,
2012, pp. 37–46. ­Ljubljana, Slovenia, 2008.
3. Bryce, J., G. Flintsch, S. Katicha, and B. Diefenderfer. Developing a 22. Ferne, B., P. Langdale, and N. Round. Development of the UK High-
Network-Level Structural Capacity Index for Structural Evaluation of ways Agency Traffic Speed Deflectometer. Presented at 8th International
Pavements. Final Report VCTIR 13-R9. Virginia Department of Trans- Conference on the Bearing Capacity of Roads, Railways, and Airfields,
portation, Richmond, 2013. Champaign, Ill., 2009.
4. Martin, T. Benefits and Risks of Investing in Network Level Deflection 23. Ferne, B. W., P. Langdale, N. Round, and R. Fairclough. Development
Data Collection. AP-T217-12. Austroads, Sydney, New South Wales, of a Calibration Procedure for the U.K. Highways Agency Traffic-
Australia, 2012. Speed Deflectometer. In Transportation Research Record: Journal of
5. Stubstad, R., R. Carvalho, R. Briggs, and O. Selezneva. Simplified Tech- the Transportation Research Board, No. 2093, Transportation Research
niques for Evaluation and Interpretation of Pavement Deflections for Board of the National Academies, Washington, D.C., 2009, pp. 111–117.
Network-Level Analysis: Guide for Assessment of Pavement Structure 24. Simonin, J.-M., D. Lievre, G. Hildebrand, and S. Rasmussen. Assess-
Performance for PMS. FHWA-HRT-12-025. FHWA, U.S. Department ment of the Danish High Speed Deflectograph in France. Presented at
of Transportation, 2012. 7th International Conference on the Bearing Capacity of Roads, Railways,
6. Kelley, J., and M. Moffat. Review of the Traffic Speed Deflectograph. and Airfields, Trondheim, Norway, 2005.
AP-R395-12. Austroads, Sydney, New South Wales, Australia, 2012. 25. Baltzer, S., D. Pratt, J. Weligamage, J. Adamsen, and G. Hildebrand.
7. Choummanivong, L., and T. Martin. Predicting Structural Deteriora- Continuous Bearing Capacity Profile of 18,000 km Australian Road
tion of Pavements at a Network Level: Interim Models. AP-T159/10. Network in 5 Months. Presented at 24th ARRB Group Conference,
Austroads, Sydney, New South Wales, Australia, 2010. Melbourne, Victoria, Australia, 2010.
8. Martin T., and M. Moffatt. Assessment of Remaining Service Life 26. Drusin, S. ANAS TSD: Acceptance Test and Calibration Comparison
of Pavements. AP-R332/08. Austroads, Sydney, New South Wales, Between TSD and FWD. Presented at DaRTS Meeting, 2012.
Australia, 2008. 27. Katicha, S., G. Flintsch, B. Ferne, and J. Bryce. Limits of Agreement
9. Tepper, S. Comparison of Project Level and Network Level Pavement Method for Comparing TSD and FWD Measurements. International
Strength. AP-T21/03. Austroads, Sydney, New South Wales, Australia, Journal of Pavement Engineering, Vol. 15, No. 6, 2014, pp. 532–541.
2003. 28. Moré, J. J., and D. C. Sorensen. Computing a Trust Region Step.
10. Kim, Y. R. NCHRP Project 10-48: Assessing Pavement Layer Condition SIAM Journal on Scientific and Statistical Computing, Vol. 3, 1983,
Using Deflection Data. Transportation Research Board of the National pp. 553–572.
Academies, Washington, D.C., 2001. 29. Katicha, S., G. Flintsch, and B. Ferne. Optimal Averaging and Localized
11. Arora, J., V. Tandon, and S. Nazarian. Continuous Deflection Testing of Weak Spots Identification for TSD Deflection Slope Measurements.
Highways at Traffic Speeds. Research Report 0-4380-1. Texas Department Presented at 92nd Annual Meeting of the Transportation Research Board,
of Transportation, Austin, 2006. Washington, D.C., 2013.
12. Elseifi, M., A. Abdel-Khalek, K. Gaspard, Z. Zhang, and S. Ismail. 30. Kim, J. General Viscoelastic Solutions for Multilayered Systems
Evaluation of Continuous Deflection Testing Using the Rolling Wheel Subjected to Static and Moving Loads. Journal of Materials in Civil
Deflectometer in Louisiana. Journal of Transportation Engineering, Engineering, Vol. 23, No. 7, 2011.
Vol. 138, No. 4, 2012. 31. Basu, D., and K. Rao. Analytical Solutions for Euler–Bernoulli Beam
13. Abdel-Khalek, A. M., M. A. Elseifi, K. Gaspard, Z. Zhang, and on Visco-Elastic Foundation Subjected to Moving Load. International
K. Dasari. Model to Estimate Pavement Structural Number at Net- Journal for Numerical and Analytical Methods in Geomechanics, Vol. 37,
work Level with Rolling Wheel Deflectometer Data. In Transporta- No. 9, 2012, pp. 945–960.
tion Research Record: Journal of the Transportation Research Board, 32. Yu, H., and Y. Yuan. Analytical Solution for an Infinite Euler–Bernoulli
No. 2304, Transportation Research Board of the National Academies, Beam on a Visco-Elastic Foundation Subjected to Arbitrary Dynamic
Washington, D.C., 2012, pp. 142–149. Loads. Journal of Engineering Mechanics, Vol. 140, No. 3, 2014,
14. Elseifi, M., A. Abdel-Khalek, and K. Dasari. Implementation of Roll- pp. 542–551.
ing Wheel Deflectometer (RWD) in PMS and Pavement Preservation. 33. Sun, L. A Closed-Form Solution of a Bernoulli–Euler Beam on a Visco-
FHWA/11.492. Louisiana Department of Transportation and Develop- elastic Foundation Under Harmonic Line Loads. Journal of Sound and
ment, Baton Rouge, 2012. Vibration, Vol. 242, No. 4, 2001.
15. Diefenderfer, B. K. Investigation of the Rolling Wheel Deflectometer as
a Network-Level Pavement Structural Evaluation Tool. VTRC 10-R5.
Virginia Transportation Research Council, Charlottesville, Va., 2010. The Pavement Monitoring and Evaluation Committee peer-reviewed this paper.

You might also like