Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 51
51
laboratory but was estimated to be similar to the US-2 HMA 2.3.1 NDT Devices for Unbound Layers
base with the PSPA (see Table 25). The reason for the large dif-
ference between the laboratory and field deviation from unity 2.3.1.1 Variability of Response Measurements
for this one mixture is unknown. Conversely, the FWD adjust- Figures 23 through 26 compare the COV to the average
ment factors are significantly different from unity. The FWD modulus measured by each device. All COV point com-
overestimated the SMA modulus for the overlay project and parisons were for the same test area. Thus, the material
underestimated the HMA base modulus for the reconstruction variance should be the same between the different NDT
projects suggesting that the calculated values from the deflec- devices.
tion basins are being influenced by the supporting materials. The GeoGauge consistently had the lower COV, and that
On the average, the PSPA can be used to estimate the value decreases with increasing material stiffness (Figure 26).
dynamic modulus measured in the laboratory HMA mixtures, The variations of the GeoGauge measurements were found to
while the FWD was found to be extremely variable. The PSPA be less dependent on type and size of aggregate, as well as less
ratios are variable, but that variability is less than the ratios dependent on the underlying materials for the thicker layers
for the unbound materials. These ratios were compared to tested. The reason for the higher COV values for the other
the binder type, gradation, and other volumetric properties devices is that the DCP penetration rate is dependent on the
but no relationship was found. It is suggested that dynamic amount and size of coarse aggregate particles, while the
modulus tests be performed to determine the target or design LWD modulus values are more dependent on the under-
value and that those results be used to calibrate the PSPA for lying materials. The DSPA is dependent on the water content
a specific mixture. The dynamic modulus test can be performed variations nearer the surface (water content-density gradi-
on bulk mixture compacted to the expected in-place density ents) and the amount of fines in coarse-gained materials.
during the mixture verification process or during construction The DSPA had higher variability when testing stiff mate-
of a control strip. rials that had water contents significantly below the opti-
mum value or where the surface had been primed. Some
2.3 Accuracy and Precision layers tested had a significant modulus gradient near the sur-
Important parameters in QA are the accuracy and precision face, which had a much larger effect on the DSPA responses.
of a test method. The higher the precision of a test method, Some sites had a positive gradient (modulus increases with
the fewer tests need to be completed at some confidence level depth), while other sites had a negative gradient. Those sites
for estimating properties of the population or lot and making with positive modulus gradients generally had higher adjust-
the "right" decision regarding the quality of the lot. This section ment ratios, while those with negative gradients had lower
evaluates and compares the variability measured within the ratios. These modulus gradients were confirmed with the
field evaluation projects with different NDT devices. The more DCP--the only device that could readily measure these gra-
precise result, however, does not automatically imply that dients in real time. Figure 27 shows some examples of the
the test method can identify physical differences or informa- change in modulus with depth, as calculated from the pene-
tion about the population related to performance. tration rate (see Equation 16).
Fine-Grained Coarse-Grained
70
Coefficient of Variation, %
60
50
40
30
20
10
0
0 10 20 30 40 50 60
Mean Elastic Modulus, DCP, ksi
Figure 23. Coefficient of variation versus the mean modulus
calculated from the DCP penetration rates.
OCR for page 52
52
Fine-Grained Coarse-Grained Log. (Coarse-Grained)
90
Coefficient of Variation, %
80
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60
Mean Elastic Modulus, LWD, ksi
Figure 24. Coefficient of variation versus the mean modulus
calculated from the LWD deflections.
Fine-Grained Coarse-Grained
90
Coefficient of Variation, %
80
70
60
50
40
30
20
10
0
0 50 100 150 200 250
Mean Elastic Modulus, DSPA, ksi
Figure 25. Coefficient of variation versus the mean modulus
determined from the DSPA responses.
Fine-Grained Coarse-Grained Log. (Coarse-Grained)
30
Coefficient of Variation, %
25
20
15
10
5
0
0 10 20 30 40 50 60
Mean Elastic Modulus, GeoGauge, ksi
Figure 26. Coefficient of variation versus the mean modulus
determined from the GeoGauge responses.
OCR for page 53
53
Ohio Crushed Stone North Dakota Crushed Stone
Florida Limerock Base
60
Calculated from DCP
Penetration Rate, ksi
Resilient Modulus
50
40
30
20
10
0 2 4 6 8 10 12 14
Depth Below Surface, inches
(a) Aggregate Base Materials/Layers.
Oklahoma High PI Clay North Dakota Embankment
25
Calculated from DCP
Penetration Rate, ksi
Resilient Modulus
20
15
10
5
0
0 2 4 6 8 10 12 14
Depth Below Surface, inches
(b) Subgrade and Embankment Materials/Layers.
Figure 27. Modulus gradients in unbound layers, as
determined with the DCP.
0.64 The LWD had higher variability in test results and
292
E R = 17.6
(DPI )1.12
(16) lower success rates. The higher COV value is related to the
variability in the underlying layers and their influence on
Where: the measured response with the deflection measuring
ER = Resilient modulus, MPa. devices, as well as thickness variations of the layer being
DPI = Penetration rate or index, mm/blow. evaluated. A constant layer thickness and subsurface con-
dition were used.
The DSPA was also placed in different directions relative to The variability of the GPR and EDG for measuring the vol-
the roller direction for measuring modulus; the other NDT umetric properties (density and fluids content) was found to
devices do not have this capability--only an equivalent or be significantly different from each other, as well as from the
average modulus value is reported for all directions. Figure 28 agencies' QA data, when available. Both of these devices had
compares the difference between the modulus values parallel very poor success rates in identifying physical differences
and perpendicular to the roller's direction to the modulus between different sections. The EDG resulted in very low
measured parallel to roller direction. For less stiff materials variability in its estimates of dry density and water content
(especially fine-grained materials), there is no difference within a specific area or test section. Most of the COV values
between the two readings. For stiffer, coarse-grained materials, for both properties were less than 2 percent (see Tables 27 and
however, there is a slight bias. The moduli measured parallel 28). Thus, the average values determined at a test point and
to roller direction were slightly higher, on the average. This within a test section did not deviate significantly from the
difference and bias resulted in a higher COV for the clustered project average that was determined from nuclear density
measurements. gauges and/or sand-cone tests.
OCR for page 54
54
Coarse-Grained Fine-Grained Zero Residual Line
40
Residual (E, Parallel - E,
30
Perpendicular), ksi
20
10
0
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80
Adjusted Seismic Modulus Parallel to Roller Direction,
DSPA, ksi
Figure 28. DSPA modulus values measured parallel to roller
direction versus the difference between modulus values
parallel and perpendicular to roller direction.
Conversely, the GPR resulted in high variability of the tion used to convert the dielectric values to dry densities--a
dielectric values (see Table 29), as well as for the dry densities. constant water content for all areas within a lot was assumed.
The dry densities determined in some areas exceeded 160 pcf As a result, the GPR data interpretation technique needs to be
(see Figure 29)--an unlikely value. The reason for the improb- improved to determine the dry density and water content
ably high as well as low values within a project was the assump- along the project prior to day-to-day use in QA programs.
Table 27. Dry densities measured with the EDG, pcf.
Project Identification Area A B C D
I-85 Embankment, Silty Clay; Mean, pcf 107.92 108.9 108.6 107.7
Section 1, Before IC Rolling COV, % 1.3 0.5 1.1 1.7
I-85 Embankment, Silty Clay; Mean, pcf 107.2 107.5 108.9 107.2
Section 2, Before IC Rolling COV, % 0.8 0.8 1.1 1.9
I-85 Embankment, Silty Clay; Mean, pcf 108.1 108.2 108.5 108.4
Section 1, After IC Rolling COV, % 1.0 0.5 0.7 0.3
I-85 Embankment, Silty Clay; Mean, pcf 107.4 107.7 108.0 107.6
Section 2, After IC Rolling COV, % 0.5 0.5 0.8 1.3
TH-23 Embankment, Silt-Sand- Mean, pcf 123.9 123.7 124.4 ---
Gravel Mix; North Section COV, % 0.4 0.1 1.0 ---
TH-23 Embankment, Silt-Sand- Mean, pcf 122.5 122.9 122.9 ---
Gravel Mix; South Section COV, % 1.8 1.8 0.8 ---
SH-130 Improved Mean, pcf 123.7 123.7 124.9 ---
Embankment; Section 1 COV, % 0.3 0.1 0.6 ---
SH-130 Improved Mean, pcf 122.6 123.1 122.7 ---
Embankment; Section 2 COV, % 2.0 2.0 0.8 ---
SH-130 Improved Mean, pcf 123.3 122.3 123.7
Embankment; Section 3 COV, % 1.4 0.1 0.2
TH-23 Crushed Aggregate; Mean, pcf 129.9 129.8 129.8 ---
North Section COV, % 0 0 0 ---
TH-23 Crushed Aggregate; Mean, pcf 129.8 129.8 129.8 ---
Middle Section COV, % 0 0 0 ---
TH-23 Crushed Aggregate; Mean, pcf 129.8 129.9 129.8 ---
South Section COV, % 0.1 0.1 0 ---
US-280 Crushed Stone; Section Mean, pcf 147.4
1 COV, % 0.7
US-280 Crushed Stone; Section Mean, pcf 148.8
2 COV, % 0.3
US-280 Crushed Stone; Section Mean, pcf 145.9
3 COV, % 0.5
US-280 Crushed Stone; Section Mean, pcf 148.2
4 COV, % 0.3
Note: The shaded cells designate those areas with anomalies (refer to Table 14); the black cells denote the
weaker areas, while the gray cells denote the stronger areas tested within a specific project.
OCR for page 55
55
Table 28. Water content measured with the EDG, percent.
Project Identification Area A B C D
I-85 Embankment, Silty Clay; Mean, % 16.9 16.8 16.9 16.9
Section 1, Before IC Rolling COV, % 0.8 0.3 0.3 1.0
I-85 Embankment, Silty Clay; Mean, % 16.9 16.9 16.8 17.0
Section 2; Before IC Rolling COV, % 0.7 0.3 0.3 1.5
I-85 Embankment, Silty Clay; Mean, % 16.9 16.9 16.9 16.9
Section 1, After IC Rolling COV, % 0.5 0.3 0.4 0
I-85 Embankment, Silty Clay; Mean, % 17.0 16.9 16.9 16.9
Section 2, After IC Rolling COV, % 0.5 0.3 0 0.7
TH-23 Embankment, Silt-Sand- Mean, % 8.0 8.0 7.6
Gravel Mix; North Section COV, % 5.1 1.1 11.9
TH-23 Embankment, Silt-Sand- Mean, % 9.8 8.7 7.6
Gravel Mix; South Section COV, % 7.5 7.3 15.8
SH-130 Improved Mean, % 8.1 8.05 7.23
Embankment; Section 1 COV, % 4.4 1.2 6.8
SH-130 Improved Mean, % 8.85 8.43 8.7
Embankment; Section 2 COV, % 19.8 21.6 8.4
SH-130 Improved Mean, % 8.35 9.1 8.05
Embankment; Section 3 COV, % 14.4 1.6 0.9
TH-23 Crushed Aggregate; Mean, % 4.26 4.28 4.34
North Section COV, % 1.3 1.0 2.1
TH-23 Crushed Aggregate; Mean, % 4.24 4.28 4.30
Middle Section COV, % 1.3 2.0 1.6
TH-23 Crushed Aggregate; Mean, % 4.18 4.18 4.38
South Section COV, % 3.9 3.9 1.0
US-280 Crushed Stone; Mean, % 3.92
Section 1 COV, % 3.1
US-280 Crushed Stone; Mean, % 4.18
Section 2 COV, % 2.9
US-280 Crushed Stone; Mean, % 3.77
Section 3 COV, % 2.9
US-280 Crushed Stone; Mean, % 4.06
Section 4 COV, % 2.6
Note: The shaded cells designate those areas with anomalies (refer to Table 14); the black cells denote weaker
areas, while the gray cells denote the stronger areas tested within a specific project.
2.3.1.2 Standard Error 2.3.2 NDT Devices for HMA Mixtures
Another reason for using the adjustment ratios in evaluat- 2.3.2.1 Variability of Response Measurements
ing each NDT device is to eliminate or reduce bias by assum-
ing that the target value is the laboratory resilient modulus Figure 32 compares the COV between different tech-
measured at a specific stress state. Figure 30 compares the lab- nologies and devices (PSPA, FWD, PQI, and GPR). The
oratory measured resilient modulus values to those estimated PQI consistently had a low COV relative to the other
with different NDT devices but adjusted to laboratory con- devices, while the FWD had the largest value. It should be
ditions, while Figure 31 presents the residuals (laboratory noted that a low COV does not necessarily mean that the
resilient modulus minus the NDT modulus), assuming that device is providing an accurate measure of the HMA mix-
the laboratory value is the target value. On the average, the ture property and variability. One reason for the lower
adjusted elastic modulus from all devices compare reasonably COV values for the PQI relative to the other devices is that
well with the laboratory measured resilient modulus. Table 30 five tests were performed at each test point. In other words,
contains the tabulation of the mean of the residuals and stan- the testing and sampling error or differences get averaged
dard error for the NDT devices that provide a direct measure out through the testing sequence.
of material stiffness. Two versions of the GPR air-coupled antennas were used.
In summary, the GeoGauge, DSPA, and DCP all provide The first version was a single-antenna method, which was
good estimates with negligible bias (effect of adjustment only used in Part A of the field evaluation. The second version
ratios) of the laboratory measured resilient modulus val- included the use of multiple antennas and the EPIC Hyper
ues. The GeoGauge has the lower standard error. The LWD OpticsTM proprietary data interpretation system. The EPIC
has a higher bias and over two times the standard error, in GPR system was supposed to be used along the NCAT, Mis-
comparison to the GeoGauge. souri (US-47), and Texas (I-20) sections; however, weather
OCR for page 56
56
Table 29. Dielectric values measured with the GPR on the unbound layers.
Project Identification Area A B C D
I-85 Embankment, Silty Clay; Section Mean 15.38 15.79 14.29 15.19
1, Before Rolling COV, % 17.8 23.3 53.6 25.7
I-85 Embankment, Silty Clay; Section Mean 13.91 17.47 16.82 16.38
2, Before IC Rolling COV, % 29.0 20.5 30.7 24.1
I-85 Embankment, Silty Clay; Section Mean 20.37 21.23 21.61 23.23
1, After IC Rolling COV, % 15.8 10.6 15.0 12.6
I-85 Embankment, Silty Clay; Section Mean 19.13 23.75 23.77 25.36
2; After IC Rolling COV 10.2 10.7 17.6 8.4
TH-23 Embankment, Silt-Sand- Mean 23.004 13.468 19.334 ---
Gravel Mix; South Section COV, % 11.3 7.0 14.4 ---
TH-23 Embankment, Silt-Sand- Mean 20.324 34.438 23.882 ---
Gravel Mix; North Section COV, % 22.2 32.7 22.7 ---
SH-130 Improved Embankment; Mean 9.225 10.00 7.65 ---
Section 1 COV 33.1 42.3 42.9 ---
SH-130 Improved Embankment; Mean 12.875 8.875 9.825 ---
Section 2 COV 90.3 47.4 20.1
SH-130 Improved Embankment; Mean 8.775 9.025 11.85
Section 3 COV, % 51.5 50.8 48.7 ---
TH-23 Crushed Aggregate; North Mean --- 8.796 10.042 ---
Section COV, % --- 1.6 5.4 ---
TH-23 Crushed Aggregate; Middle Mean --- 8.950 10.87 ---
Section COV, % --- 6.1 10.9 ---
TH-23 Crushed Aggregate; South Mean --- 9.792 10.378 ---
Section COV, % --- 8.2 4.3 ---
US-280 Crushed Stone; Section 1 Mean 11.723
COV, % 8.3
Mean 12.222
US-280 Crushed Stone; Section 2
COV, % 11.4
Mean 11.919
US-280 Crushed Stone; Section 3
COV, % 7.3
Mean 11.569
US-280 Crushed Stone; Section 4
COV, % 7.0
Notes:
The shaded cells designate those areas with anomalies (refer to Table 14); the black cells denote the weaker areas, while the
gray cells denote the stronger areas tested within a specific project.
Due to construction sequencing, lane A of the TH-23 crushed aggregate base sections could not be tested with the GPR after it
arrived on site.
delays and equipment/plant problems resulted in changes to by assuming that the target value is the laboratory dynamic
the testing schedule. These schedule changes resulted in con- modulus measured at a specific load frequency and an aver-
flicts with other projects, so ultimately, this system was used age in-place mix temperature. Figure 33 compares the
only on the NCAT test sections. PSPA and FWD modulus values that have been adjusted to
Data were made available for use from other projects in laboratory conditions using the factors or ratios listed in
Florida, which were not included in the original field evalu- Table 26. On the average, the adjusted modulus values
ation (Greene 2007; Greene and Hammons 2006). The EPIC compare reasonably well to one another. Table 31 contains
system is reported to have much more accurate and repeat- the mean of the residuals (laboratory dynamic modulus
able estimates of HMA volumetric properties. One reason minus the NDT modulus) and standard error from the
for this increased accuracy and precision is that it does not expected laboratory value--excluding all measurements
rely on the assumptions that were included in the single taken in areas with anomalies, segregation, and along lon-
antenna method used along the Part A projects. The preci- gitudinal joints.
sion and bias for both devices and systems are provided in While the difference between the two NDT devices is small,
the next section. the PSPA had the lower residual and standard error.
2.3.2.2 Standard Error 2.3.3 Summary
As for the unbound materials, the adjustment ratios Tables 32, 33, and 34 contain the statistical analyses of the
were used in evaluating the PSPA and FWD to reduce bias NDT devices included in the field evaluation projects. This
