Light Weight Deflectometer Compaction Test TERRATEST 4000 ...
CHARACTERIZATION OF A TWO-LAYER SOIL SYSTEM …docs.trb.org/prp/10-1653.pdf · 2 LIGHTWEIGHT...
Transcript of CHARACTERIZATION OF A TWO-LAYER SOIL SYSTEM …docs.trb.org/prp/10-1653.pdf · 2 LIGHTWEIGHT...
CHARACTERIZATION OF A TWO-LAYER SOIL SYSTEM USING A 1
LIGHTWEIGHT DEFLECTOMETER WITH RADIAL SENSORS 2 Paper # 10-1653 3
4
By 5
6 Christopher T. Senseney, P.E., Maj, USAF (Corresponding Author) 7
Ph.D. Student 8
Division of Engineering 9
Colorado School of Mines 10
1610 Illinois Street 11
Golden, CO 80401 12
Ph: (303)384-2153 13
F: (303)273-3602 14
email: [email protected] 15
16
Michael A. Mooney, Ph.D., P.E. 17
Associate Professor 18
Division of Engineering 19
Colorado School of Mines 20
1610 Illinois Street 21
Golden, CO 80401 22
Ph: (303)384-2498 23
F: (303)273-3602 24
email: [email protected] 25
26
27
Word Count: 28
Abstract = 0218 29
Text = 4710 30
Tables (4 x 250) = 1000 31
Figures (6 x 250) = 1500 32
33
Total = 7428 34
35
36
37
38
39
40
41
42
43
Re-submitted on November 10, 2009 for presentation and publication by Transportation 44
Research Board, 88th
Annual Meeting, January 2010, Washington, D.C. 45
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 2
ABSTRACT 46 47
Non-destructive tests to estimate stiffness modulus, such as the light weight 48
deflectometer (LWD), have experienced increased popularity, but very little research has been 49
performed to evaluate the LWD with radial sensors. Results are presented from LWD testing 50
with radial sensors that measured the deflection bowl on one- and two-layer field test beds 51
consisting of unbound materials. LWD testing produced a measureable deflection bowl on 52
medium stiffness granular materials to a radial sensor spacing of 750 mm (30 in). When limited 53
to a stiff over soft layered system, the LWD with radial sensors demonstrated the ability to 54
accurately backcalculate layered moduli. Backcalculated moduli closely matched laboratory 55
determined moduli from triaxial testing at a similar stress state as in the field. The measurement 56
depth for the LWD with radial sensors was found to be 1.8 times plate diameter versus the 57
measurement depth of conventional LWD testing of 1.0 to 1.5 times plate diameter. The LWD 58
with radial sensors was able to measure deeper than conventional LWD testing because the radial 59
geophones measure vertical surface deflections caused almost entirely by deeper material. As 60
compared to other configurations, the 300 and 600 mm (12 and 24 in) radial sensor configuration 61
is recommended for unbound materials because it produced the most accurate moduli 62
backcalculation results and captures deflections critical to the backcalculation process. 63
64
INTRODUCTION 65 66
The light weight deflectometer (LWD) was developed to estimate the in-situ stiffness 67
modulus of soils. The device can be used for quality control/quality assurance and structural 68
evaluation of compacted earthwork and pavement construction. Over the past decade or more, 69
resilient modulus, analogous to stiffness modulus, was established as the primary input 70
parameter for characterizing subgrade, subbase and base materials for highway pavement design 71
in the United States. As a result, non-destructive tests to estimate stiffness modulus, such as the 72
LWD, have experienced increased popularity [1-5]. Conventional LWD testing uses deflection 73
measured from a center geophone or accelerometer coupled with static, linear-elastic half space 74
theory to calculate one stiffness modulus for the composite soil system. 75
LWD manufacturers have recently begun offering an LWD with radial geophones. This 76
evolution suggests that layered soil systems may be characterized in a similar method as the 77
falling weight deflectometer (FWD). If so, a significant advancement would be achieved because 78
earthwork is often layered. The modulus obtained from conventional LWD testing represents the 79
composite modulus of the layers within the influence depth rather than the true modulus of the 80
tested layer [6]. Very little research has been performed to evaluate the LWD with radial sensors. 81
To the authors’ knowledge there is just one published study presenting actual LWD radial 82
deflection results [7] and no published studies of backcalculated moduli from radial sensors. 83
This paper presents results from LWD testing that employed radial sensors to measure 84
the deflection bowl of one- and two-layer soil systems. Layer moduli were then backcalculated 85
based on measured deflections. The objectives of this paper are to: (1) verify the LWD produces 86
a deflection bowl on unbound materials and describe the nature of the deflection bowl on one- 87
and two-layer systems, (2) assess the capability of LWD with two radial sensors to accurately 88
characterize moduli of known two-layer stratigraphies (limited to stiff over soft), (3) evaluate 89
measurement depth of LWD with two radial sensors and show how and why it is different than 90
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 3
conventional LWD testing, and (4) evaluate the influence of radial sensor spacing on 91
backcalculation. 92
93
BACKGROUND 94
95
LWD Characteristics 96
97 The LWD is a portable device that applies an impulse load from a drop weight impacting 98
a circular plate resting on the ground. The LWD consists of a circular load plate (150 mm, 200 99
mm or 300mm diameter), housing, urethane dampers, guide rod, drop weight and geophone 100
sensors (Figure 1a). During testing the drop weight releases from a variable height, slides down 101
the guide rod and applies a dynamic force impulse to the load plate lasting 15-25 ms. Surface 102
deflections (through integration of velocity) are measured from a geophone mounted in the 103
center of the load plate and from up to two radial geophones mounted on a support bar resting on 104
the soil. Applied force (P) is measured with a force transducer mounted within the housing. In 105
conventional LWD testing, the maximum deflection measured by the center geophone (d0) and 106
maximum measured P are coupled with homogeneous, isotropic, static linear-elastic half space 107
theory to calculate a conventional LWD modulus (Evd or ELWD) that follows as: 108
109
( )π
ν
0
21
ad
PAELWD
−= (1) 110
111
where ν is the Poisson’s ratio of the soil, A is the contact stress distribution parameter (A = 2 for 112
a uniform stress distribution, π/2 for an inverse parabolic distribution, 8/3 for a parabolic 113
distribution) and a is the plate radius. The user has the option to input values for ν and A. The 114
LWD is designed for an impact force and load plate diameter to deliver an average contact stress 115
of 100 to 200 kPa (14.5 to 29 psi) on the soil surface. This stress range mimics the approximate 116
stress level on a typical subgrade, subbase or base course due to vehicle loading on the pavement 117
surface [8]. Two commercial LWD devices that employ radial geophone sensors are the Prima 118
100 and Dynatest LWD produced by Grontmij - Carl Bro Pavement Consultants and Dynatest 119
International, respectively. Results from testing with the Prima 100 (Fig. 2b) along with 120
backcalculated moduli from Dynatest’s LWDmod program are presented in this study. The 121
Prima 100 utilizes seismic velocity transducer geophones with a resolution of 1 µm and a 122
frequency range of 0.2 to 300 Hz [9]. 123
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 4
124
125
FIGURE 1 (a) Schematic of LWD, (b) picture of Prima 100 LWD. 126
127
Forward-calculation and Backcalculation 128
129 Dynatest’s LWDmod program forward-calculates deflections using Odemark’s layer 130
transformation approach together with Boussinesq’s equations. The basic assumption of 131
Odemark’s layer transformation is that the layered structure can be transformed into an 132
equivalent uniform, semi-infinite material, to which Boussinesq’s equations may be applied to 133
calculate deflections. Two critical assumptions of the Odemark-Boussinesq method are: (1) layer 134
thicknesses should be more than one-half the radius of the loading plate, and (2) moduli should 135
decrease with each descending layer by at least a factor of two [10]. Accurate backcalculation 136
with the Odemark-Boussinesq method is limited to pavement profiles with stiff layers over soft 137
layers. Using Odemark’s layer tranformation, a two-layer linear elastic system may be 138
transformed to a semi-infinite space provided that layer one is replaced by an equivalent 139
thickness (he) of material having the properties of the semi-infinite space. Assuming the 140
Poisson’s ratio is the same for all layers, the transformed equivalent thickness may be 141
determined as: 142
143
3
2
11
E
Efhhe = (2) 144
145
where h1 is the thickness of the top layer, E1 and E2 are the moduli of the top and bottom layer 146
respectively, and f is a correction factor for better agreement with exact values [11]. 147
A significant advantage of the Odemark-Boussinesq method is the ability to 148
accommodate for the stress dependent non-linearity in dynamic deflection testing by 149
incorporating a non-linear relation for the subgrade modulus. The universal non-linear model for 150
resilient modulus (Mr) that reflects stress dependence [12] is given by: 151
152 32
1
kk
r qpkM = (3) 153
154
where p is the mean normal stress (� � ��� � �� � �� 3⁄ ), q is the deviator stress, and k1, k2 and 155
k3 are best fit parameters determined by laboratory data. The k3 value is negative, typically in the 156
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 5
range of 0 to -0.2 for granular materials and 0 to -0.6 for cohesive materials [12]. As k3 becomes 157
more negative, a material exhibits greater stress softening behavior where the modulus decreases 158
with increasing deviator stress. Dynamic deflection testing, e.g. FWD and LWD testing, 159
produces a non-linear effect in a stress softening subgrade, where the subgrade stress levels 160
beneath the outer sensors are much lower than subgrade stress levels for the inner sensors. The 161
apparent subgrade modulus measured by outer sensor locations is therefore higher than the 162
apparent subgrade modulus measured by the inner sensors [13]. The variation of modulus with 163
radial distance due to dynamic deflection testing is similar to that given by k3 part of Equation 3. 164
To this end, the non-linear subgrade modulus (E2) for use in Odemark-Boussinesq calculations 165
[14] is expressed in the form: 166
167 n
apCE
= 1
2
σ (4) 168
169
where σ1 is the major principle stress from external loading, pa is atmospheric pressure, and C 170
and n are constants. As is the case with k3, a material exhibits greater stress softening behavior as 171
n becomes more negative. A material is considered linear elastic when n = 0. 172
LWDmod uses a Odemark-Boussinesq static analysis to forward-calculate deflections. 173
First, apparent subgrade moduli E(r) are determined from a Boussinesq equation (Equation 5) for 174
loading on a homogeneous, isotropic, linear elastic half-space where dr is the measured 175
deflection at distance r, and P is a point load representing the uniformly distributed load of the 176
LWD. At a radial distance beyond one diameter from the center of the load, a point load 177
produces approximately the same surface deflections as a distributed load [15]. Next, the non-178
linearity constant n is calculated (Equation 6) where E(r1) and E(r2) are the apparent subgrade 179
moduli determined from radial sensors at r1 and r2. The n value is essentially a measure of 180
subgrade non-linearity as E(r2) will be greater than E(r1) when testing on a non-linear subgrade 181
due to a lower stress state at r2. Then, Odemark’s transformation is modified to accommodate a 182
non-linear subgrade in which the equivalent thickness of layer one (he,1) is calculated (Equation 183
7) as follows: 184
185
rrd
PrE
π
ν )1()(
2−
= (5) 186
187
=
1
2
2
1
log2
)(
)(log
r
r
rE
rE
n (6) 188
189
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 6
n
n
a
e
p
PnC
Ehfh
23
1
13
11,
2
3)21(
−
⋅−
=
π
(7) 190
191
where f is an correction factor (= 1.05 for a 2-layer system with n ≤ -0.4, = 1.0 for n > -0.4) [16]. 192
Then, stresses are calculated with Boussinesq’s equations on the transformed system and are 193
used to calculate E2 (Equation 4). Finally, deflections are calculated as the sum of the 194
compression in the transformed layer one, plus the deflection at the top of the subgrade. The 195
compression of the transformed layer one is the difference between deflections at the top and the 196
bottom of the transformed layer one. The deflection at the top the subgrade is calculated using 197
E2. The calculated deflections are compared with measured deflections and assumed moduli are 198
adjusted in an iterative procedure until calculated and measured deflection bowls reach an 199
acceptable match [13]. 200
201
Field Test Beds 202
203 Two field test beds were constructed to investigate the ability of the LWD with radial 204
geophones to characterize layered stratigraphies. Figure 2 summarizes the geometry of test bed 205
one (TB1) and test bed two (TB2). TB1 was 4 m (13 ft) long and 2.5 m (8 ft) wide and TB2 was 206
4 m (13 ft) long and 2 m (6.6 ft) wide. TB1 was designed as a homogeneous, medium stiffness 207
soil profile with in-situ material excavated to a depth of 1500 mm (60 in) and subsequently filled 208
with medium stiff sand (SW-SM) compacted with a vibratory plate in 150 mm (6 in) lifts. TB2 209
was designed as a medium stiffness soil over soft soil profile with in-situ soft clay (CL) 210
excavated to a depth of 600 mm (24 in) and filled with SW-SM compacted with a vibratory plate 211
in 75 mm (3 in) lifts. Great care was taken to ensure vertical homogeneity of both test beds with 212
quality assurance testing conducted every 150 mm (6 in). LWD testing was conducted at two 213
locations in each test bed, herein referred to as TB1-1, TB1-2, TB2-1 and TB2-2. 214
215
216 FIGURE 2 Test bed schematics for (a) homogeneous stiff sand profile (TB1); (b) medium 217
stiff sand over soft clay profile (TB2). 218
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 7
Soil properties are summarized in Table 1. Average dry density, moisture content, DCP 219
penetration rate and surface seismic modulus were obtained from in-situ quality assurance testing 220
with the sand cone, DCP and surface seismic. The seismic modulus was determined from multi-221
channel analysis of surface wave (MASW) testing utilizing an instrumented hammer to generate 222
surface waves which were measured at 0.1 m (4 in) intervals along a 1 m (39 in) array. MASW 223
based methods typically produce moduli greater than moduli determined from conventional 224
methods due to the low strains associated with MASW testing [16]. In the laboratory, the triaxial 225
secant modulus was determined from a consolidated, drained triaxial test using 150 mm (6 in) 226
tall, 70 mm (2.8 in) diameter samples. 227
228
TABLE 1 Summary of soil properties from TB1 and TB2 229
230 parameter Medium stiffness sand Soft clay
USCS classification SW-SM CL
AASHTO Classification A-1-b A-6
Liquid limit N/A 17
Plasticity index N/A 11
Average dry density 1988 kg/m3 (124 pcf) 1890 kg/m
3 (118 pcf)
Average moisture content 3.5% 17%
Average DCP penetration rate 7 mm/blow 22 mm/blow
Average low strain seismic modulus 263 MPa (38,000 psi) 48 MPa (7000 psi)
Triaxial secant modulus 63 MPa (9100 psi) 9 MPa (1300 psi)
231
Testing using the Prima 100 LWD with a 300 mm load plate and radial sensors was 232
conduced on both test beds. The test protocol consisted of testing on two points with three initial 233
10 kg (22 lb) pre-load drops. Measurements were then taken using 10 kg drop weights for 3 234
drops with radial sensor spacing (r) at 300/600 mm (12/24 in), followed by 3 drops with r = 235
450/750 mm (18/30 in). The same protocol was then used for 15 kg (33 lb) and 20 kg (44 lb) 236
drop weights. Testing was first conducted on the surface of TB1 to establish a baseline modulus 237
for the medium stiff sand. Then, testing was conducted on each 75 mm (3 in) lift of medium stiff 238
sand in TB2 in order to assess the capability of LWD to characterize a known two-layer system 239
with increasing depth of a medium stiff soil over a soft subgrade. Results from 10 kg and 15 kg 240
drop weight loading are presented in the following section; results from 20 kg loading are not 241
presented. It was determined that the LWD is not robust enough to support the 20 kg drop weight 242
because the device jumps and rocks during loading with such a heavy weight. Because of the 243
rocking motion, data from 20 kg loading was inconsistent and deemed unreliable, especially 244
deflection data from the center geophone. 245
246
RESULTS 247
248
Radial Deflections 249
250 LWD testing on medium stiffness granular soil produced a measureable deflection bowl to a 251
radial sensor spacing (r) of 750 mm. For both test beds, a decrease in surface deflection with 252
increasing radial sensor spacing was evident with the rate of decay dependent on soil modulus 253
and layering. Measured deflections on points 1 and 2 were nearly identical for both test beds; 254
therefore, only results from TB1-1 and TB2-1 are presented. Measured deflections on the 255
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 8
homogeneous SW-SM profile of TB1-1 are presented in Figure 3a for r = 300/600 mm with 10 256
kg and 15 kg loading. TB1-1 deflections indicated a gradual rate of decay with an increase in 257
radius. The change in deflection from r = 0 to r = 150 mm radial spacing is constant and equal to 258
d0 because the plate is assumed to be rigid. Deflections for the two-layer profile, TB2-1, are 259
presented in Figure 4a and 4b for r = 300/450/600/750 mm with 10 kg loading. The deflections 260
on TB2 with 15 kg loading are not presented here since the contours of the 15 kg deflection 261
bowls were similar to the contours of the 10 kg deflection bowls. Because the LWD is limited to 262
two radial sensors per measurement, measurements on TB2 were recorded for r = 300/600 mm, 263
followed by measurements for r = 450/750 mm. TB2-1 deflections exhibited a steep rate of 264
decay from r = 150 to r = 450 mm at shallow h1 of 240 mm, 310 mm, 385 mm (9 in, 12 in, 15 265
in). The steep decrease is an indication of weak CL material below layer one that caused high 266
deflections at the center geophone, followed by significant decreases in deflections with an 267
increase in radius. As h1 increased, measured deflections decayed at a more gradual rate and 268
began to match the 10 kg measured deflections from TB1-1. 269
Measured deflections indicate that TB1 was representative of a homogenous, isotropic, linear 270
elastic half-space. This is supported by the Boussinesq equation for deflection on the surface a 271
homogenous, isotropic, linear elastic half-space (see Equation 5). With an assumed modulus of 272
63 MPa, Equation 5 was used to calculate the theoretical radial deflections presented in 273
conjunction with measured deflections in Figures 3a, 4a and 4b. The theoretical deflections 274
between r = 150 and 300 mm are labeled invalid because the assumption of P (Equation 5) 275
representing a uniformly distributed load is only valid one diameter and beyond from the center 276
of the load. Equation 1 was used to generate the theoretical deflections between r = 0 and 150 277
mm (Figures 3a, 4a and 4b) with a modulus of 63 MPa and A = 2. Assuming the load plate is 278
perfectly rigid, theoretical deflections beneath the 150 mm radius plate are equal. Measured 279
radial deflections from TB1 showed good correlation with theoretical radial deflections. Figures 280
3b, 4c and 4d show apparent subgrade moduli E(r) calculated at radial distances based on 281
measured deflections. If a given test bed was an ideal homogeneous, isotropic, linear elastic 282
halfspace all the moduli would be the same. In fact, the calculated E(r) in Figure 3b closely 283
match the assumed theoretical modulus of 63 MPa, indicating TB1 is nearly homogeneous. 284
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 9
285 286
FIGURE 3 (a) TB1-1 deflections at radial distances for 10 kg and 15 kg; (b) TB1-1 moduli 287
at radial distances for 10 kg and 15 kg. 288
Measured deflections indicate TB2 with h1 < 400 mm (16 in) was not representative of a 289
homogeneous, isotropic, linear elastic halfspace. Figure 4a shows that measured deflections with 290
h1 < 400 mm do not match theoretical deflections. A mismatch was expected because TB2 was 291
vertically heterogeneous with E1 = (5-7)E2. As h1 increased, the degree of homogeneity 292
measured by the LWD in TB2 increased (Figure 4b). At an h1 = 620 mm (24 in), measured 293
deflections converge on the theoretical deflection bowl. Also, Figure 4c shows that calculated 294
E(r) do not match the assumed theoretical modulus of 63 MPa at h1 < 400 mm, while Figure 4d 295
shows that calculated E(r) start to converge on the assumed theoretical modulus of 63 MPa at h1 296
> 400 mm. 297
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 10
298 299
FIGURE 4 TB2-1 deflections at radial distances for 10 kg and (a) h1 = 240 mm, 310 mm, 300
385 mm; and (b) h1 = 460 mm, 535 mm, 620 mm; TB2-1 moduli at radial distances for the 301
same h1 thicknesses (c-d). 302
303
Backcalculation Results 304
305 Moduli were backcalculated from measured deflections and compared to laboratory 306
triaxial moduli with the caveat that stress levels in the lab do not necessarily represent those 307
experienced in the field owing to the varying stress state with depth. For this study, the 308
comparison soil moduli for layer one and two (E1 and E2) were estimated to be 63 MPa and 9 309
MPa based on consolidated, drained triaxial testing. The triaxial E1 and E2 were dependent on the 310
mean normal stress of the chamber (p) and applied deviator stress (q) as evidenced in Equation 3. 311
When E1 was measured during triaxial testing of the SW-SM material, mean p and q were 38 kPa 312
(5.5 psi) and 24 kPa (3.5 psi). When E2 was measured during triaxial testing of the CL material, 313
mean p and q were 34 kPa (4.9 psi) and 12 kPa (1.7 psi). A separate study measured in-situ 314
stresses generated by LWD loading on a test bed with 470 mm (18.5 in) of SP-SM sand over a 315
CL subgrade [17]. Table 2 presents measured stresses from that study where p and q were 316
calculated assuming an at rest earth pressure coefficient (Ko) of 0.5 and where �� � �� , �� �317
�� � ��. Table 2 suggests that p and q from the SW-SM triaxial test correspond to stresses at a 318
depth of approximately 200 mm (8 in) from LWD loading. According to Equation 3, the 319
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 11
comparison E1 would be larger if derived from a lower stress state than the stress state observed 320
at 200 mm. Table 2 also suggests that the p and q from the CL triaxial test correspond to stresses 321
at a depth of approximately 250 mm (10 in). Again, the comparison E2 would be larger if derived 322
from a lower stress state than the stress state observed at 250 mm. 323
324
TABLE 2 In-situ stresses from LWD loading where �� � ��, �� � �� � �� 325
326
depth (mm) ��(kPa) p (kPa) q (kPa)
0 125 84 42
190 70 47 23
240 50 34 17
310 35 23 12
385 25 17 8
460 18 12 6
535 15 10 5
327
LWDmod demonstrated the ability to accurately characterize the moduli and nonlinearity 328
of the stiff over soft layered system in TB2. Backcalculation results from TB2-1 are presented in 329
Table 3 where measured deflections are shown for r = 0, 300 and 600 mm along with the 330
backcalculated deflections from LWDmod. LWDmod attempts to minimize the root mean square 331
(RMS) of the absolute difference between measured and backcalculated deflections. E1 and E2 332
are the backcalculated moduli for layer one and layer two respectively. Percent error represents 333
the percent discrepancy from the comparison moduli. C and n (Equation 4) are the non-linear 334
subgrade modulus constants. ELWD (Equation 1) is also provided. The most accurate moduli 335
backcalculation results were generated in a range of h1 = 385 mm (15 in) to h1 = 535 mm (21 in). 336
The soft clay of layer two caused underestimation of E1 at shallow h1. At deep h1, it is proposed 337
in the next section that the LWD with radial sensors has reached the depth of influence or 338
measurement depth, resulting in an overestimation of E2. The value of n increases with 339
increasing h1, indicating the radial sensors measure more of the non-linear CL at shallow h1 and 340
measure more of the linear SM-SW at deep h1. 341
342
343
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 12
TABLE 3 TB2-1 Backcalculation results for 10 kg and 15 kg 344
345
parameter h1 = 245 mm 310 mm 385 mm 460 mm 535 mm 620 mm
10 kg
d0 (measured) (mm) 1047 854 442 444 441 439
d300 (measured) (mm) 276 239 190 149 117 94
d600 (measured) (mm) 28 52 59 58 53 42
d0 (back) (mm) 1111 875 464 454 439 433
d300 (back) (mm) 198 225 163 143 117 97
d600 (back) (mm) 31 53 62 60 53 41
RMS (%) 17.9 3.9 9 3.5 0.5 1.9
E1 (MPa) [% error] 23 [63.5] 32 [49.2] 67 [-6.4] 61 [3.2] 60 [4.8] 56 [11.1]
E2 (MPa) [% error] 6 [33.3] 7 [22.2] 10 [-11.1] 15 [-66.7] 37 [-311.1] 77 [-755.6]
C (MPa) 2 2 2 4 26 77
n -0.82 -0.71 -0.68 -0.52 -0.16 0
ELWD (MPa) 24.6 30.9 59.6 59.3 62.4 62.3
15 kg
d0 (measured) (mm) 1575 1304 786 699 689 688
d300 (measured) (mm) 407 355 328 253 184 150
d600 (measured) (mm) 28 63 93 83 79 75
d0 (back) (mm) 1671 1331 836 762 691 667
d300 (back) (mm) 263 323 284 231 183 159
d600 (back) (mm) 30 64 99 87 79 72
RMS (%) 21.2 5.5 9.5 6 0.6 4.3
E1 (MPa) [% error] 23 [63.5] 33 [47.6] 60 [4.8] 59 [6.35) 60 [4.76] 60 [4.76]
E2 (MPa) [% error] 5 [44.4] 7 [22.2] 9 [0.0] 12 [-33.3] 29 [-222.2] 71 [-688.9]
C (MPa) 2 2 2 3 16 71
n -1 -0.86 -0.77 -0.69 -0.3 0
ELWD (MPa) 24.8 31.6 54.6 58.1 63.4 64.4
346
Measurement Depth 347
348 The measurement depth of the LWD with radial sensors was found to be greater than the 349
measurement depth of conventional LWD testing. In conventional LWD testing, the depth of 350
influence or measurement depth reflected in ELWD has been shown to range from 1.0 to 1.5 times 351
the plate diameter (for a = 200 and 300 mm) [6, 8, 17]. An inspection of the ELWD data for TB2 352
(Table 3) and the corresponding d0 data (Figure 5) suggest the ELWD and d0 plateau at h1 = 385 353
mm for 10 kg and at h1 = 460 mm for 15 kg. These measurement depths (1.2a for 10 kg and 1.5a 354
for 15 kg) are consistent with those found in previous studies. The measurement depth for LWD 355
with radial sensors was found to be 535 mm or 1.8a for both 10 kg and 15 kg. Beyond the depth 356
of 1.8a at h1 = 620 mm, measured deflections from TB2 for r = 300 and 600 mm match 357
measured deflections from the homogeneous SW-SM profile in TB1 for r = 300 and 600 mm. 358
The matching deflections indicate that the radial sensors no longer measured the CL layer two 359
and only measured the SW-SM layer one. In addition, n = 0 at h1 = 620 mm, indicating the radial 360
sensors are only measuring the linear SW-SM material. Interestingly, radial deflections did not 361
plateau in this study (Figure 5). However, it is proposed that radial deflections would plateau at 362
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 13
h1 > 620 mm because measured deflections at h1 = 620 mm match theoretical deflections of a 363
homogeneous, isotropic, linear elastic halfspace with E = 63 MPa (Figure 4b). 364
365
366 367
FIGURE 5 TB2-1 deflection versus h1 for (a) 10 kg and (b) 15 kg. 368
The LWD with radial sensors was able to measure deeper than conventional LWD testing 369
because the radial geophones measure vertical surface deflections caused almost entirely by 370
deeper material. For a homogeneous, isotropic, linear elastic halfspace, the vertical deflection at 371
z = r is approximately equal to the vertical deflection at r [15]. Figure 6 displays deflection (dz) 372
versus depth based on Boussinesq analysis (Equation 8) 373
�� ������
���√����� 2�1 # $ �%�
&��%�' (8) 374
where E = 63 MPa, ν = 0.35, r is the radial offset and z is the depth. Figure 6 indicates that dz is 375
representative of material deformation below the 45º line because there is little change in 376
deflection from the surface to the 45º line. The 45º angle would be shallower for the layered 377
system is this study, however, the same principles apply. For the materials in this study, the sand 378
below the outer geophones is not affected by the LWD-induced stress, so the sand does not 379
experience deformation. The clay below the outer geophones is affected by the LWD-induced 380
stress that is spread through the sand. The clay deforms due to this stress and the geophones 381
measure the deflection at the surface. 382
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 14
383 384
FIGURE 6 Vertical deflection versus depth for r = 0, 300 and 600 mm on a homogeneous, 385
isotropic, linear elastic halfspace. 386 387
Radial Sensor Spacing 388
389 As compared to other configurations, the r = 300/600 geophone configuration produced 390
the most accurate moduli backcalculation results and this configuration captures deflections 391
critical to the backcalculation process on unbound materials. Results from two h1 thicknesses 392
(Table 4) show that the r = 300/600 exhibits the lowest percent error. More importantly, r = 393
300/600 characterizes the non-linearity of the CL layer two with the largest n value. Previously 394
presented results (Table 2) indicated the CL layer is highly non-linear. An r = 300 mm is the 395
recommended radial spacing for the inner sensor. For backcalculation on unbound materials, the 396
inner sensor is critical because it captures the steepness of the deflection bowl. In the Odemark-397
Boussinesq method with a 300 mm load plate, a sensor could not be placed any closer than r = 398
300 mm because the point load assumption is only valid beyond a radial distance of one 399
diameter. Due to the 1 µm resolution of the geophone, an r = 600 mm is the recommended 400
spacing for the outer sensor. The smallest deflection measurement recorded in this study was 13 401
µm, measured by the r = 750 mm sensor, which is starting to approach an unacceptable signal to 402
noise ratio of 10 percent. The smallest deflection measurement recorded by the r = 600 mm 403
sensor was 22 µm, well within an acceptable signal to noise ratio. 404
405
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 15
TABLE 4 TB2-1 backcalculation results by radial sensor configuration for 10 kg 406
and h1 = 385 mm and 460 mm 407
408
radial sensor spacing (mm)
parameter 300/600 450/750 300/450 600/750
h1 = 385 mm
E1 (MPa) [% error] 67 [6.3] 65 [(3.2] 71 [12.7] 62 [-1.6]
E2 (MPa) [% error] 10 [11.1] 14 [55.6] 11 [22.2] 23 [155.6]
C (MPa) 2 5 3 14
n -0.68 -0.46 -0.54 -0.26
h1 = 460 mm
E1 (MPa) [% error] 61 [-3.2] 51 [-19] 55 [-12.7] 56 [-11.1]
E2 (MPa) [% error] 15 [66.7] 22 [144.4] 17 [88.9] 18 [100]
C (MPa) 4 11 6 7
n -0.52 -0.33 -0.45 -0.41
409
CONCLUSIONS 410 411
1. LWD testing with a 300 mm diameter load plate on medium stiffness granular materials 412
produced a measureable deflection bowl to a radial sensor spacing of 750 mm. The 413
measured deflection bowl may be investigated to determine the degree to which a soil 414
system is homogeneous, isotropic and linear elastic. The measured deflection bowl on 415
TB1 indicated the one-layer, medium stiff sand profile was homogeneous, isotropic and 416
linear elastic. The measured deflection bowl on TB2 at h1 < 400 mm indicated the two-417
layer stiff over soft profile was not homogeneous, isotropic and linear elastic. 418
2. When limited to a stiff over soft layered system, the LWD with radial sensors 419
demonstrated the ability to accurately backcalculate layered moduli. Backcalculated 420
moduli closely matched laboratory determined moduli from triaxial testing at a similar 421
stress state as in the field. The companion LWDmod program found the clay subgrade in 422
the two-layer profile to be highly non-linear. 423
3. The measurement depth for the LWD with radial sensors was found to be 1.8 times plate 424
diameter versus the measurement depth of conventional LWD testing of 1.0 to 1.5 times 425
plate diameter. The LWD with radial sensors was able to measure deeper than 426
conventional LWD testing because the radial geophones measure vertical surface 427
deflections caused almost entirely by deeper material. 428
4. As compared to other configurations, the r = 300/600 mm geophone configuration is 429
recommended for unbound materials because it produced the most accurate moduli 430
backcalculation results and captures deflections critical to the backcalculation process. 431
432
Acknowledgements 433
434 The authors thank the Air Force Research Lab for funding this study. John Siekmeier (MnDOT) 435
and Roger Surdahl (FHWA) are acknowledged for providing testing equipment critical to this 436
study. The authors also thank Gabriel Bazi, Norman Facas and Caleb Rudkin for helping to 437
analyze and process LWD data.438
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 16
REFERENCES 439
440 [1] Siekmeier, J. A., D. Young, and D. Beberg. Comparison of the Dynamic Cone 441
Penetrometer with Other Tests During Subgrade and Granular Base Characterization in 442
Minnesota. Nondestructive Testing of Pavements and Backcalculation of Moduli ASTM 443
STP 1375, Vol. 3, S. Tayabji, and E. Lukanen, Editors, ASTM International, West 444
Conshohocken, PA, 2000, pp. 175-188. 445
446
[2] Livneh, M., and Y. Goldberg. Quality Assessment During Road Formation and 447
Foundation Construction: Use of Falling-Weight Deflectometer and Light Drop Weight. 448
In Transportation Research Record: Journal of the Transportation Research Board, No. 449
1755, Transportation Research Board of the National Academies, Washington, D.C., 450
2001, pp. 69-77. 451
452
[3] Fleming, P. R., M. W. Frost, and J. P. Lambert. Review of Lightweight Deflectometer for 453
Routine In Situ Assessment of Pavement Material Stiffness. In Transportation Research 454
Record: Journal of the Transportation Research Board, No. 2004, Transportation 455
Research Board of the National Academies, Washington, D.C., 2007, pp. 80-87. 456
457
[4] Vennapusa, P. K. R., and D. J. White. Comparison of Light Weight Deflectometer 458
Measurements for Pavement Foundation Materials. ASCE Geotechnical Testing Journal, 459
Vol. 32, No. 3, 2009, pp. 1-13. 460
461
[5] Adam, D., C. Adam, F. Kopf, and I. Paulmichl. Computational Validation of Static and 462
Dynamic Plate Load Testing. Acta Geotechnica, Vol. 4., 2009, pp. 35-55. 463
464
[6] Nazzal, M. D., M. Abu-Faraskh, K. Alshibli, and L. Mohammad. Evaluating the Light 465
Falling Weight Deflectometer Device for In Situ Measurement of Elastic Modulus of 466
Pavement Layers. In Transportation Research Record: Journal of the Transportation 467
Research Board, No. 2016, Transportation Research Board of the National Academies, 468
Washington, D.C., 2007, pp. 13-22. 469
470
[7] Horak, E., J. Maina, D. Guiamba, and A. Hartman. Correlation Study with Light Weight 471
Deflectometer in South Africa. Proc., 27th
Southern African Transport Conference, 472
Pretoria, South Africa, 2008. 473
474
[8] Fleming, P. R. Small-scale Dynamic Devices for the Measurement of Elastic Stiffness 475
Modulus on Pavement Foundations. Nondestructive Testing of Pavements and 476
Backcalculation of Moduli ASTM STP 1375, Vol. 3, Vol. 3, S. Tayabji, and E. Lukanen, 477
Editors, ASTM International, West Conshohocken, PA, 2000, pp. 41-58. 478
479
[9] Grontmij - Carl Bro Pavement Consultants. Prima 100 LWD: Light Falling Weight 480
Deflectometer. Kolding, Denmark, 2008. 481
482
[10] Ullidtz, P., and K. R. Peattie. Pavement Analysis by Programmable Calculators. ASCE 483
Journal of Transportation Engineering, Vol. 106, No. TE5, 1980, pp. 581-597. 484
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.
Senseney, C.T. & Mooney, M.A. 17
485
[11] Ullidtz, P., and R. N. Stubstad. Analytical-Empirical Pavement Evaluation Using the Falling 486
Weight Deflectometer. In Transportation Research Record: Journal of the Transportation 487
Research Board, No. 1022, Transportation Research Board of the National Academies, 488
Washington, D.C., 1985, pp. 36-44. 489
490
[12] Witczak, M. W., and J. Uzan. The Universal Airport Design System, Report I of IV: 491
Granular Material Characterization. Department of Civil Engineering, University of 492
Maryland, College Park, 1988. 493
494
[13] Ullidtz, P. Analytical Procedures in Nondestructive Testing Pavement Evaluation. In 495
Transportation Research Record: Journal of the Transportation Research Board, No. 496
1482, Transportation Research Board of the National Academies, Washington, D.C., 497
1995, pp. 61-66. 498
499
[14] Ullidtz, P. Pavement Analysis. Elsevier, Amsterdam, 1987. 500
501
[15] Dynatest International. ELMOD 6. Glostrup, Denmark, 2008. 502
503
[16] Ryden, N., and M. A. Mooney. Analysis of Surface Waves from the Light Weight 504
Deflectometer. Soil Dynamics and Earthquake Engineering. Vol. 29, Iss. 7, 2009, pp. 505
1134-1142. 506
507
[17] Mooney, M. A., and P. K. Miller. Analysis of Light Weight Deflectometer Test Based on 508
In-situ Stress and Strain Response. ASCE Journal of Geotechnical and Geoenvironmental 509
Engineering, Vol. 135, No. 2, 2009, pp. 199-208. 510
TRB 2010 Annual Meeting CD-ROM Paper revised from original submittal.