Mousavi 2016

Page 1 of 34
Mousavi, Gabr, Borden

For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
1 SUBGRADE RESILIENT MODULUS PREDICTION FROM LIGHT WEIGHT
2 DEFLECTOMETER
3
4 S. Hamed Mousavi, Corresponding Author
5 North Carolina State University
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Otago on 11/19/16
6 Graduate Research Assistant, Department of Civil, Construction, and Environmental

7 Engineering, North Carolina State University, Raleigh, NC 27695-7908;
8 Tel: 919-995-8792; Email: smousav3@ncsu.edu
9
10 Mohammed A. Gabr
12 Professor, Department of Civil, Construction, and Environmental Engineering, North
13 Carolina State University, Raleigh, NC 27695-7908;
14 Tel: 919-515-7904 FAX: 919-515-7908; Email: gabr@eos.ncsu.edu
15
16 Roy H. Borden
18 Professor, Department of Civil, Construction, and Environmental Engineering, North
19 Carolina State University, Raleigh, NC 27695-7908;
20 Tel: 919-515-7630 FAX: 919-515-7908; Email: borden@ncsu.edu
21
22
23
24
25
26
27
28
2
Page 2 of 34

29 ABSTRACT
30 Resilient modulus has been used for decades as an important parameter in pavement structure
31 design. Resilient modulus, like other elasticity moduli, increases with increasing confining stress
32 and softens with increasing deviatoric stress. Several constitutive models have been proposed in
33 the literature to calculate resilient modulus as a function of stress state. The most recent model,
34 recommended by MEPDG and used in this paper, calculates resilient modulus as a function of
35 bulk stress, octahedral shear stress and three fitting coefficients, k1, k2, and k3. Work in this paper
36 presents a novel approach for predicting resilient modulus of subgrade soils at various stress
37 level based on light weight deflectometer (LWD) data. The proposed model predicts the MEPDG
38 resilient modulus model coefficients (k1, k2, and k3) directly from the ratio of applied stress to
39 surface deflection measured during LWD testing. The proposed model eliminates uncertainties
40 associated with needed input parameters for ELWD calculation, such as the selection of an
41 appropriate value of Poisson’s ratio for the soil layer and shape factor. The proposed model was
42 validated with independent data from other studies reported in the literature.
43
44 Keywords: resilient modulus, light weight deflectometer, subgrade, MEPDG
45
46
47
3
Page 3 of 34

48 INTRODUCTION
49 The use of resilient modulus ( M r ) has been substituted for the California Bearing Ratio (CBR)
50 in pavement design in order to consider the deformation behavior of base and subgrade layers
51 under cyclic loading condition. The magnitude of Mr depends on, soil types, its structure,
52 physical properties such as density and water content, as well as the applied stress state (Li 1994;
53 Rahim and George 2004, Liang et al. 2008). Studies have shown that same as other soil modulus,
54 resilient modulus of subgrade soils, increase by an increase of confining pressure and reduces
55 with an increase in deviatoric stress. The value of Mr is defined as the ratio of the cyclic axial
56 stress to the recoverable or resilient axial strain (NCHRP project 1-28A, 2004), as shown
57 schematically in Fig. 1 and expressed in Equation 1:
σ cyclic
58 Mr = (1)
εr
59 Where:
60 σ cyclic : cyclic axial stress
61 ε r : resilient axial strain
62
63 Figure. 1. Definition of resilient modulus
64
65 The Mr of a subgrade layer can be determined from laboratory testing following the AASHTO T-
66 307 test protocol, which uses fifteen stress combinations: five deviatoric stress levels 13.8, 27.6,
67 41.4, 55.2 and 69 kPa (2, 4, 6, 8 and 10 psi) applied at three confining pressures 41.4, 57.6 and
4
Page 4 of 34

68 13.8 kPa (6, 4 and 2 psi). Different forms of constitutive models can be found in the literature
69 that allow for computing the Mr value as a function of one, two or three stress parameters such as
70 confining pressure, deviatoric stress, bulk stress and octahedral shear stress (e.g. Dunlap 1963;
71 Seed et al. 1967; Witczak and Uzan 1988; Pezo 1993; NCHRP project 1-28A 2004). The recent
72 universal constitutive model proposed in NCHRP 1-28A (MEPDG) is presented in Equation 2:

θ τ oct
73 M r = k1.Pa.( ) k2 .( + 1) k 3 (2)
Pa Pa
74 Where:
75 M r : resilient modulus
76 Pa : atmospheric pressure
77 σ1 , σ 2 , σ 3: principal stresses
78 θ = σ1 + 2σ 3 : bulk stress
2
79 τ oct = (σ 1 − σ 3 ) : octahedral shear stress
3
80 ki : regression constants; i:1, 2, and 3
81
82 Despite the good accuracy of the use of laboratory testing to determine Mr. values, the
83 requirement of using an expensive and sophisticated device to perform the Mr test is considered
84 as a disadvantage. Several studies have been undertaken to overcome this issue by estimating Mr
85 through the development of empirical correlations with the physical properties of soil (e.g.
86 Carmichael et al. 1985; Elliott et al. 1988; Drumm et al. 1990; Farrar and Turner 1991; Hudson
87 et al. 1994). These empirical correlations eliminate the expense of resilient modulus laboratory
88 testing, but they are generally not capable of capturing the stress dependency of the Mr, or
89 simulate various stress conditions encountered in the field. Many studies have been performed
5
Page 5 of 34

90 over the past two decades to model the stress dependency of the resilient modulus by predicting
91 the coefficients of constitutive model, NCHRP Project 1-28A 2004, using basic soil properties.
92 For example, Yau and Von Quintus (2002), Rahim and George (2004), and Nazzal and
93 Mohammad (2010) each proposed different models to estimate the constitutive model
94 coefficients (k1, k2, and k3) based on the physical properties of soils. Another approach is the use
95 of expedient in-situ approaches such as light weight deflectometer, LWD, and dynamic cone
96 penetrometer, DCP, testing to estimate Mr. For example, White et al. (2007) and Mohammad at
97 el. (2008) have proposed correlations to estimate the Mr of subgrade soil by LWD at one specific
98 confining pressure and deviatoric stress (confining pressure = 41.4 and 13.8 kPa and deviatoric
99 stress = 69 and 41.4 kPa, respectively). Since the Mr depends on the confining pressure and
100 applied deviatoric stress, these models become inapplicable for cases with different stress states.
101 Work in this paper presents a model to compute Mr. values of A-4 (ML, SM, and CL) and A-7-5
102 (MH) soils from LWD data at any stress state by estimating the MEPDG recommended
103 constitutive model parameters (k1, k2 and k3). The proposed model is based on the use of LWD
104 measured data, the ratio of applied stress to the measured surface deflection, rather than on the
105 use of LWD-estimated Mr, in order to minimize uncertainties involved with the selection of
106 Poisson’s ratio and shape factor input parameters. The validity of the proposed model for A-1-b
107 (SP) and A-6 (CL-ML) soils, is examined by using data presented in the literature.
108 BACKGROUND
109 The LWD is a portable falling weight deflectometer for measuring in-situ modulus of soil
110 (Fleming 2007). Compared to the falling weight deflectometer (FWD), the LWD is cheaper and
111 more convenient to perform. The device used in this study was a Prima 100, as shown in Fig. 2,
112 and consisted of 10 kg falling weight, which can induce 15-20 ms pulse load up to 450 kPa, with
6
Page 6 of 34

113 its 20-cm diameter plate (radius = 10 cm). A geophone is used to measure surface deflection,
114 right at the center of the plate load. Surface deflection and applied load are monitored and
115 recorded through Prima 100 software. Fig. 3 shows an example of applied load and surface
116 deflection for one drop.
117
118 Figure 2. Prima 100 sketch (after Vennapusa, 2008)
119
120 The in-situ modulus is calculated based on Boussineq’s static elastic half space theory by
121 assuming a homogeneous isotropic soil layer (Fleming 2007). Therefore Poisson’s ratio and
122 shape factor are assigned as input parameters to the software to calculate the modulus per
123 Equation 3 (Fleming 2007).
f .(1 −ν 2 ).σ .r
124 ELWD = (3)
δ
125 Where:
126 ELWD : surface Modulus (MPa)

127 σ : applied stress (kPa)
128 δ : surface Deflection (µm)
129 f : shape factor
130 ν : Poisson’s ratio
131 r : radius of plate (mm)
132
133 Figure 3. Example of recorded applied stress and surface deflection during LWD
134 testing
7
Page 7 of 34

135 As previously mentioned, there are few empirical correlations to approximate the Mr of subgrade
136 soil from LWD measurements. Although the main advantage of these models is that they can
137 capture actual moisture and density conditions of the soil layer, they are mostly limited to one
138 specific stress state. White et al. (2007) proposed the model presented in Equation 4 to predict Mr
π
139 of subgrade soil from ELWD with an assumed Poisson’s ratio of 0.35 and shape factor of and 2
140 for cohesive and cohesionless soils, respectively, at a confining pressure = 41.4 kPa (6 psi) and
141 deviatoric stress = 69 kPa (10 psi).
( E LWD + 45.3)
142 Mr =
1.24
143 (4)
144 With M r , ELWD in MPa
145 Mohammad et al (2008) presented the model in Equation 5 in order to estimate Mr from LWD
π
146 data by assuming a Poisson’s ratio of 0.4 and a shape factor of for cohesive soil and 2 for
2
147 cohesionless soils, at a confining pressure 13.8 kPa (2 psi) and deviatoric stress 41.4 kPa (6 psi).
148 These stress values were chosen to represent the stress state at the top of the subgrade layer
149 under standard single axle loading of 80 kN (18 kips) and tire pressure of 689 kPa (100 psi) with
150 a 50-mm asphalt wearing course, 100-mm asphalt binder course and 200-mm aggregate base
151 course (Mohammad et al. 2008; Rahim and George 2004; Asphalt Institute 1989).
0.18
152 M r = 27.75 × ELWD
153 (5)
154 With M r , ELWD in MPa
8
Page 8 of 34

155 EXPERIMENTAL PROGRAM
156 A series of laboratory and in-situ LWD tests were performed to evaluate subgrade soil modulus
157 properties of four 4.88-m (16-ft) wide by 15.2-m (50-ft) long test sections located in the
158 Piedmont area, North of Greensboro, North Carolina. In this case, LWD testing was conducted in
159 the field to monitor the variation of subgrade modulus across the test sections. As shown in Fig.
160 4, the LWD tests were carried out at four locations within each test section; those locations were
161 offset 1-m (3.3-ft) away from boreholes, from which Shelby tubes were obtained. In parallel, a
162 laboratory testing program including resilient modulus testing and physical properties
163 characterization was performed on undisturbed samples retrieved from within the LWD
164 influence zone, (a depth = 1.5~ 2 diameter of the LWD loading plate, as was specified by
165 Mooney and Miller 2007; Khosravifar et al. 2013; Senseney et al. 2016), as presented in Table
166 1, and indicated in Fig. 1.
167
168 Figure 4. Location of resilient modulus specimens and LWD tests (dimensions in cm).
169
170 MATERIALS
171 Basic index property tests, including grain size distribution, Atterberg limits, and specific gravity
172 were conducted on the specimens after the Mr tests were completed. As shown in Table 1 and
173 Fig. 5, the site soils were classified as A-7-5 (MH), A4 and A4-a (SM, ML and CL,
174 respectively). The high plasticity specimens, A-7-5, were taken from the deeper depths, and
175 correspond to natural soil, while the low plasticity soils (A4 and A4-a) were located at shallower
176 depths and corresponded to compacted fill.
9
Page 9 of 34

177
178 Table 1. Summary of Sample Properties

179
180 Figure 5. Grain size distributions of resilient modulus samples
181
182 LWD MEASUREMENTS
183 Subgrade modulus values were measured at the test site by LWD (with 20 cm plate) following
184 the ASTM E2583-07. To do so, the plate is located horizontally on the surface, and first three
185 seating drops were considered to provide full contact between the plate load and the subgrade;
186 which followed by another three drops to obtain data for estimation of the surface modulus,
187 ELWD . The ELWD values were calculated using Equation 3 and assuming a Poisson’s ratio of
π
188 0.35. A shape factor of was selected for the MH soil, and 2 for the ML, SM, and CL soil,
2
189 (Mooney and Miller 2007; White et al. 2007). Two LWD test stations were located about 1 m
190 apart on each side of the boreholes, from which samples for resilient modulus testing were
191 collected, as shown in Fig. 4. At each station, 3 ELWD were measured. The Standard deviation
192 and coefficient of variation of the six LWD measurements corresponding to the resilient modulus
193 laboratory specimen are presented in Table 2. The COVs are less than 5%. Hence for comparison
194 sake, the average of six field measured ELWD values was used to estimate the Mr from LWD. The
195 calculated Mr values were compared to the laboratory-measured Mr values for specimens
196 retrieved from the corresponding borehole within the LWD effective zone, as summarized in
197 Table 2.
198 As previously mentioned, the assumption of Poisson’s ratio and shape factor can lead to various
199 estimates of ELWD. In order to overcome the ambiguities with which values to use, the ratio of
10
Page 10 of 34

σ
200 applied stress to surface deflection, , from LWD direct measurements; which is representative
δ
201 of soil layer elasticity, is directly used herein in the proposed model development, instead of
202 computing the ELWD by Equation 3.
203 The applied stress to surface deformation ratios from the LWD measurements are summarized in
204 Table 2 and are shown associated with the Specimen Number upon which the subsequent Mr
205 tests were performed.
206 LABORATORY TESTING: RESILIENT MODULUS
207 Resilient modulus tests were performed on undisturbed soil specimens following AASHTO T-
208 307. Resilient modulus at each load combination was computed as the ratio of the cyclic axial
209 stress to average resilient axial strain for the last 5 of the 100 applied load cycles. Laboratory
210 results from each specimen were imported into MATLAB in order to evaluate fitting coefficients
211 for the MEPDG universal constitutive model, as presented in Equation 2. The calculated k1, k2,
212 and k3 values are provided in Table 2 along with the respective coefficient of determination (R2).
213
214 Table 2. Summary of LWD Measurements and M r Model Parameters
215
216 EFFECT OF LWD INPUT PARAMETERS
217 Although input parameters such as Poisson’s ratio and shape factor can be approximated for
218 particular cases from values presented in literature, it is difficult to select appropriate values for
219 these input parameters for site specific conditions. Poisson ratio values between 0.2 to 0.5 are
220 recommended in the literature (Bishop 1977). Various shape factors are recommended for
11
Page 11 of 34

π
221 different scenarios. The shape factor can be varied between for a rigid loading plate, 1.33 and
2
222 2.67 for parabolic contact stress distribution in cohesive soils and granular soils, respectively,
223 and 2 for a uniform contact stress distribution (Terzaghi and Peck 1967; Mooney and Miller
224 2007; Vennapusa and White 2007; Prima 100 software). The uncertainties associated with
225 selecting these parameters can induce significant variation in the value of calculated modulus, as
226 illustrated in Fig. 6. For the case shown, the computed ELWD for sample H1-1 can change from
227 110 up to 220 MPa by changing the shape factor from 1.33 to 2.67 at a given Poisson’s ratio of
228 0.2. In addition, it can be observed that the effect of Poisson’s ratio becomes more pronounced
229 with increasing shape factor, and can produce up to a 30% change in the computed elastic
230 modulus value.
231
232 Figure 6. Effect of Poisson ratio and shape factor on E LWD (specimen H1-1).
233
234 EXISTING MODELS
235 As previously mentioned, White et al. (2007) Mohammad et al (2008) proposed empirical
236 correlations to approximate the Mr of subgrade soil from LWD measurements at one specific
237 confining pressure and deviatoric stress.
238 The performance of these correlations in estimating the laboratory-measured Mr values from the
239 current study are both shown in Fig. 7. It can be seen that both correlations have generally
240 overpredicted the measured values; however the Mohammad et al. correlation underestimates the
241 Mr of the more highly plastic soils.
12
Page 12 of 34

242
243 Figure 7. Laboratory-measured M r vs. that predicted from existing models
244
245 DEVELOPMENT OF LWD CORRELATION

246 As previously noted, the existing models for subgrade Mr determination from LWD data are
247 limited to a specific stress level. A change in pavement structure layer thickness, axial load, and
248 tire pressure can lead to changes in stresses within the layers. In order to overcome this
249 restriction and eliminate uncertainties associated with selecting appropriate Poison’s ratio and
250 shape factor values, the ratio of applied stress to surface deflection as measured during LWD
251 testing was used. The coefficients of the MEPDG model are functionally related to the elastic
252 modulus (k1), stiffness hardening (k2) and strain softening (k3) behavior of the soil (Yau and
253 Quintus, 2002). Accordingly, the proposed model was developed to correlate k1, k2, and k3 to the
σ
254 ratio of obtained from LWD measurements, with the advantage of having the ability to
δ
255 estimate Mr. at other stress levels once these parameters are defined. The new model was
256 developed from regression analyses on the laboratory and field measurement data from the
257 cohesive (A-7-5) and cohesionless (A-4a) soils.
258 MEPDG COEFFICIENTS FROM LWD
259 Multilinear regression analyses was performed on three quarters of the data set to develop a
260 model to calculate resilient modulus indirectly at any stress level from LWD data through
261 estimating the MEPDG formula coefficients. The proposed correlation is presented in Equation
262 6, with the definition of constants presented in Table 3. Figs. 8(a-c) shows the calculated
263 coefficients, k1, k2, and k3, from curve fitting of the laboratory results versus the ratio of the
13
Page 13 of 34

264 measured applied stress to the surface deformation in the field. The proposed model predicts the
265 k1, k2 and k3 coefficients with a coefficient of determination ( R2) of 0.71, 0.82, and 0.55.
σ
266 ki = C1 + C2 ( )
δ
267 (6)
268 i :1, 2,3
269 Table 3. Constant Coefficients of Developed Model

270
271 Figure 8. (a) Computed k1 vs. σ/δ, (b) Computed k2 vs. σ/ δ, and (c) Computed k3 vs. σ/δ.
272 Following AASHTO T-307, laboratory resilient modulus test is performed at 15 different stress
273 level, 5 deviatoric stress, 13.8, 27.6, 41.4, 55.2 and 69 kPa (2, 4, 6, 8 and 10 psi), at 3 confining
274 pressure, 41.4, 57.6 and 13.8 kPa (6, 4 and 2 psi); which provides 15 resilient modulus values
275 corresponding to each stress level. Hence for each specimen, 15 resilient moduli are measured at
276 different stress level. Fig. 9 shows the laboratory measured vs. predicted Mr, for ¾ of the
277 samples at 15 different stress levels. The analyses results, illustrated in Fig. 9, show that the
278 proposed model is able to compute the laboratory-measured Mr with a coefficient of
279 determination ( R2) of = 0.83.
280 Figure 9. Laboratory-measured M r vs. that computed prediction
281
282 Model Validation
283 Fig. 10 shows laboratory-measured vs. model-computed resilient modulus values using the
284 remaining quarter of the data set which was not used in the initial statistical correlations, at 15
14
Page 14 of 34

285 different stress level. The best-fit line for the data plotted shows that the proposed model slightly
286 underestimates resilient modulus by 7% with a coefficient of determination of 0.83. The
287 performance of the proposed model was also evaluated by utilizing data available from two other
288 studies by White et al. (2007) and Mohammad et al. (2008). Data from White et al. (2007)
289 included LWD measurements as well as laboratory-measured Mr data at a confining pressure =

290 41.4 kPa (6 psi) and deviatoric stress = 69 kPa (10 psi), for A-6 (CL), sandy lean clay; and A-1-b
291 (SP) soil, poorly graded sand with silt and gravel. Mohammad et al. (2007) presented LWD and
292 Mr measurements at a confining pressure 13.8 kPa (2 psi) and deviatoric stress 41.4 kPa (6 psi),
293 for A-4(CL-ML) and A-6 (CL-ML) soils. In order to be able to utilize this data from the
σ
294 literature, the ratio of values were back calculated using Equation 3. The parameters utilized
δ
295 in the back-calculation were Poisson’s ratio of 0.35 and 0.4, for White et al. (2007) and
π
296 Mohammad et al. (2008), respectively, and shape factors of for cohesive soils and 2 for
2
297 cohesionless soils, as originally reported by the authors. As shown in Fig. 11, the proposed
298 model underestimates laboratory-measured Mr by 11% with a coefficient of determination of
299 0.96.
300
301 Figure 10. Laboratory-measured M r vs. that predicted for one quarter of data set.
302
303 Figure 11. Laboratory-measured M r vs. that predicted using literature data.
304
15
Page 15 of 34

305 SUMMARY AND CONCLUSIONS
306 A model for estimating Mr on the basis of LWD data is presented in this paper. A performance
307 evaluation of existing models from the literature to assess Mr based on LWD data indicated a
308 limitation related to the ability of these model to predict Mr at only a single stress level. Using a
309 linear regression analyses approach, applied to laboratory measured Mr and field measured LWD
310 data, a new model is proposed to compute Mr at any desired stress state. The use of such a model
311 exclude uncertainties involved with the assumption of parameters needed for current LWD
312 modulus determination. In the proposed model, the ratio of applied stress to surface deformation
313 measured by LWD is used directly to compute MEPDG universal constitutive model coefficients
314 (k1, k2 and k3). Based on the results obtained in this study, the following conclusion can be
315 stated:
316 • The proposed model is capable of predicting Mr at various stress combinations with a
317 coefficient of determination of, (R2)0.83. Good agreement was obtained between the
318 LWD-predicted Mr and laboratory-measured data presented in other studies, with an 11%
319 average underprediction of Mr. values with R2 = 0.96.
320 • Examination of the ability of two existing models to predict Mr from in-situ LWD data,
321 showed a general trend toward overprediction, except for the higher plasticity soils tested
322 in this study. For this soil the Mohammad et al. correlation was seen to underestimate the
323 measured Mr. values.
324 • Input parameters needed to calculate elastic modulus by the LWD, such as Poisson’s ratio
325 and shape factor, can have a significant effect on the calculated ELWD. At a given
326 Poisson’s ratio, the ELWD can change by a factor of 2 depending on the shape factor
16
Page 16 of 34

327 selected. The effect of Poisson’s ratio was shown to increase with increasing shape
328 factor.
329 • The proposed model was demonstrated to predict the resilient modulus of A-1-b (SP), A-
330 4 (ML, SM, CL), A-6 (CL-ML), and A-7-5 (MH) soil types using the measured ratio of
331 applied stress to surface deflection from a Prima 100 LWD device, employed in this
332 study. Further work will need to be performed to evaluate the applicability of the
333 proposed approach to other soil types as well.
17
Page 17 of 34

REFERENCES
AASHTO 2003. Standard method of test for determining the resilient modulus of soils and
aggregate materials. AASHTO T307-99, Washington, D.C.
Asphalt Institute, 1989, The Asphalt Handbook. Manual. Series No. 4 (MS-4), pp. 435-437.
ASTM D2216. 2010. Standard Test Methods for Laboratory Determination of Water (Moisture)
Content of Soil and Rock by Mass.
ASTM D422-63. 2007. Standard Test Method for Particle-Size Analysis of Soils.
ASTM D4318-10. Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of
Soils.
ASTM D6913. 2009. Standard Test Methods for Particle-Size Distribution (Gradation) of Soils
Using Sieve Analysis.
ASTM D854. 2010. Standard Test Methods for Specific Gravity of Soil Solids by Water
Pycnometer.
ASTM E2583-07. 2011. Standard Test Method for Measuring Deflections with a Light Weight
Deflectometer (LWD).
Bishop, A. W., and Hight, D. W. 1977. “The value of Poisson’s ratio in saturated soils and rocks
stressed under undrained conditions”. Geotechnique 27, No. 3, pp 369-384.
Carmichael, R.F., III and Stuart, E. 1985. “Predicting Resilient Modulus: A Study to Determine
the Mechanical Properties of Subgrade Soils”. In Transportation Research Record: Journal of
the Transportation Research Board, No. 1043, Transportation Research Board, National
Research Council, Washington, D.C.,pp. 145–148.
Drumm, E.C., Boateng-Poku, Y., and Johnson Pierce, T. 1990. “Estimation of Subgrade
Resilient Modulus from Standard Tests”. Journal of Geotechnical Engineering, Vol. 116, No. 5,
pp. 774–789.
Dunlap, W.S. 1963. “A Report on a Mathematical Model Describing the Deformation
Characteristics of Granular Materials”. Technical Report 1, Project 2-8-62-27, Texas
Transportation Institute, Texas A&M University, College Station.
Elliot, R.P., Thorton, S.I., Foo, K.Y., Siew, K.W., and Woodbridge, R. 1988. “Resilient
Properties of Arkansas Subgrades”. Final Report, TRC-94, Arkansas Highway and
Transportation Research Center, University of Arkansas, Fayetteville.
Farrar, M.J., and Turner, J.P. 1991. “Resilient Modulus of Wyoming Subgrade Soils”. MPC
Report No. 91-1, Mountain Plains Consortium, Fargo, N.D.
2
Page 18 of 34

Fleming, P. R., Frost, M. W., and Lambert, J. P. 2007. “Review of Lightweight Deflectometer
for Routine in Situ Assessment of Pavement Material Stiffness”. In Transportation Research
Record: Journal of the Transportation Research Board, No. 2004, Transportation Research
Board of the National Academies, Washington, D.C., pp. 80–87.
Hudson, J.M., Drumm, E.C., and Madgett, M. 1994. “Design Handbook for the Estimation of
Resilient Response of Fine-Grained Subgrades”. Proceedings of the 4th International
Conference on the Bearing Capacity of Roads and Airfields, University of Minnesota,

Minneapolis, Vol. 2, pp. 917–931.
Khosravifar, S., Asefzadeh, A., and Schwartz, C. 2013. “Increase of Resilient Modulus of
Unsaturated Granular Materials During Drying After Compaction”. Geo-Congress 2013: pp.
434-443.
Li, D., and Selig, E. T. 1994. “Resilient Modulus for Fine-Grained Subgrade Soils”. Journal of
Geotechnical Engineering, Vol. 120, No. 6, pp 939-957.
Liang, R.Y., Rabab’ah, S., and Khasawneh, M. 2008. “Predicting moisture-dependent resilient
modulus of cohesive soils using soil suction concept”. Journal of Transportation Engineering,
134(1), pp.34-40.
Mohammad, L. N., Herath, A., Gudishala, R., Nazzal, M. D., Abu-Farsakh, M. Y., and Alshibli,
K. 2008. “Development of Models to Estimate the Subgrade and Subbase Layers’ Resilient
Modulus from In situ Devices Test Results for Construction Control”. Publication FHW-LA-
406. FHWA, U.S. Department of Transportation.
Mooney, M. A., and Miller, P. K. 2009. “Analysis of Lightweight Deflectometer Test Based on
In Situ Stress and Strain Response”. Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 135, No. 2, pp 199-208.
Nazzal, M.D., and Mohammad, L.N. 2010. “Estimation of resilient modulus of subgrade soils for
design of pavement structures”. Journal of Materials in Civil Engineering, 22(7), pp.726-734.
NCHRP. 2003. “Harmonized test methods for laboratory determination of resilient modulus for
flexible pavement design”. Final Rep. NCHR Project No. 1-28 A, National Cooperative Highway
Research Program (NCHRP), Washington, D.C.
Pezo, R.F. 1993. “A General Method of Reporting Resilient Modulus Tests of Soils: A Pavement
Engineer’s Point of View”. Presented at the 72nd Annual Meeting of the Transportation
Research Board, Washington, D.C.
Rahim, A.M., and George, K.P. 2004. “Subgrade Soil Index Properties to Estimate Resilient
Modulus”. CD-ROM. Transportation Research Board of the National Academies, Washington,
D.C.
3
Page 19 of 34

Seed, H.B., Mitry, F.G., Monismith, C.L., and Chan, C.K. 1967. “NCHRP Report 35: Prediction
of Flexible Pavement Deflections from Laboratory Repeated-Load Tests”. Highway Research
Board, National Research Council, Washington, D.C.
Senseney, C.T., Grasmick, J., and Mooney, M.A. 2014. “Sensitivity of lightweight deflectometer
deflections to layer stiffness via finite element analysis”. Canadian Geotechnical Journal, 52(7),
pp. 961-970.
Terzaghi, K., and Peck, R. B. 1967. “Soil mechanics in engineering practice”. 2nd Ed., John
Wiley & Sons, Inc., New York.

Vennapusa, P.K., and White, D.J. 2009. “Comparison of light weight deflectometer
measurements for pavement foundation materials. Geotechnical Testing Journal, Vol. 32, No. 3,
pp. 1-13,
Witczak, M.W., and Uzan, J. 1988. “The Universal Airport Pavement Design System”. Report 1
of 4, Granular Material Characterization, University of Maryland, College Park.
White, D., Thompson, M., and Vennapusa, P. 2007. “Field Validation of Intelligent Compaction
Monitoring Technology for Unbound Materials”. Report No. MN/RC-2007-10.
Yau, A., and Von Quintus, H. 2002. “Study of laboratory resilient modulus test data and
response characteristics”, Rep. No. FHWA-RD-02-051, FHWA, U.S. Department of
Transportation, Washington, D.C.
4
Page 20 of 34

Figure captions
Figure 1. Definition of resilient modulus

Figure 2. Prima 100 sketch (after Vennapusa and White 2009)
Figure 3. Example of recorded applied stress and surface deflection during LWD
testing
Figure 4. Location of resilient modulus specimens and LWD tests (dimensions in cm).
Fig. 5. Grain size distributions of resilient modulus samples

Figure 6. Effect of Poisson ratio and shape factor on ELWD (specimen H1-1).
Figure 7. Laboratory-measured Mr vs. that predicted from existing models
Figure 8. (a) Computed k1 vs. σ/δ, (b) Computed k2 vs. σ/ δ, and (c) Computed k3 vs. σ/δ.
Figure 9. Laboratory-measured Mr vs. that computed prediction
Figure 10. Laboratory-measured Mr vs. that predicted for one quarter of data set.
Figure 11. Laboratory-measured Mr vs. that predicted using literature data.
5
For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record. Page 21 of 34
Figure 1. Definition of resilient modulus

Figure 2. Prima 100 sketch (after Vennapusa and White 2009)

Page 22 of 34
Applied stress (kPa)
testing
0
30
60
90
120
150
180
400
20
Time (ms)
60
Applied stress
Surface deflection
80
0
-50
-100
-150
-200
-250
Surface deflection (μm)

Figure 3. Example of recorded applied stress and surface deflection during LWD
Figure 4. Location of resilient modulus specimens and LWD tests (dimensions in cm).
Page 24 of 34
Percent Finer, P%
0
10
20
30
40
50
60
70
80
90
100
10
CL (H5-2)
SM (H3-1)
ML (H4-2)
MH (H1-1)
1
0.1
Particle Diameter, D (mm)
0.01
Figure 5. Grain size distributions of resilient modulus samples

0.001
ELWD (MPa)
0
50
100
150
200
250
0
ν=0.50
ν=0.35
ν=0.20
1.50.5
1
Shape factor, (f)
2
2.5
Figure 6. Effect of Poisson ratio and shape factor on E LWD (specimen H1-1).
3
Page 26 of 34
Predicted Mr (MPa)
0
30
60
90
120
150
180
0
30
60
90
1-1
Measured Mr (MPa)
120
150
White et al. (2007), A-4a
White et al. (2007), A-7-5
Figure 7. Laboratory-measured M r vs. that predicted from existing models

Mohammad et al. (2008), A-4a
Mohammad et al. (2008), A-7-5
180
(a) (b)
1600 1.2
k2 = -0.90(σ/δ) + 1
1400 R² = 0.82
1.0
1200
0.8
1000
Computed k1
Computed k2
800 0.6
600
0.4
400
k1 = 1040(σ/δ) + 480 0.2
200 R² = 0.71
0 0.0
0.0 0.5 1.0 0.0 0.5 1.0
σ/δ σ/δ
(c)
-5.0
k3 = 2.80(σ/δ) - 3.8
-4.5
R² = 0.55
-4.0
-3.5
Computed k3
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.0 0.5 1.0
σ/δ
Figure 8. (a) Computed k1 vs. σ/δ, (b) Computed k2 vs. σ/ δ, and (c) Computed k3 vs. σ/δ.
Computed Mr (MPa)
0
40
80
120
160
0
40
80
Laboratory-Measured Mr(MPa)
120
Figure 9. Laboratory-measured M r vs. that computed prediction

1-1
R² = 0.83
A-4a
A-7-5
160
Computed Mr (MPa)
0
30
60
90
120
0
30
60
Laboratory-Measured Mr(MPa)
90
1-1
R² = 0.83
Mr-P. = 0.93Mr-M.
120
Figure 10. Laboratory-measured M r vs. that predicted for one quarter of data set.
Page 30 of 34
Predicted Mr (MPa)
0
40
80
120
160
40 0
80
Laboratory-Measured Mr (MPa)
120
White et al. (2007)
R² = 0.96
Mohammad et al.(2008)
Mr-Pr. = 0.89Mr-M.
Figure 11. Laboratory-measured M r vs. that predicted using literature data.

160
Table 1. Summary of Sample Properties

Sample P200% Clay%
Depth 1 γtotal 2 w %3 e4 Gs5 LL6 PL7 PI8 9 10 USCS AASHTO
No.
H1-111 79 16.9 34.0 0.98 2.70 72 48 24 80 65 MH A-7-5
H2-2 86 17.7 27.8 0.80 2.66 58 36 22 72 46 MH A-7-5
H2-1 18 18.7 14.4 0.51 2.64 15 12 4 41 16 SM A-4a
H3-1 8 18.0 16.8 0.61 2.65 20 17 3 49 19 SM A-4a
H6-1 8 18.3 17.0 0.58 2.65 13 10 3 42 14 SM A-4a
H3-2 23 18.8 16.0 0.51 2.62 18 17 2 51 16 ML A-4a

H4-1 23 18.6 21.8 0.63 2.60 20 17 3 54 20 ML A-4a
H4-2 23 19.3 15.4 0.47 2.62 19 16 3 55 20 ML A-4a
H5-1 0 19.4 13.3 0.46 2.61 22 19 2 51 20 ML A-4a
H6-2 23 18.8 21.2 0.59 2.61 19 15 4 49 19 ML A-4a
H8-1 8 18.8 16.6 0.52 2.61 19 17 2 54 14 ML A-4
H5-2 30 19.7 16.5 0.44 2.64 25 16 8 42 16 CL A-4a
1 2 3 3 4 5 6 7
cm, kN/m , natural water content, void ratio, specific gravity, Liquid limit, Plastic limit,
8
Plasticity index, 9pass sieve No.200, 10Clay< 5µm , 11Hi-j, i:borhole number, j: sample number
Page 33 of 34
Table 2. Summary of LWD Measurements and M r Model Parameters
Soil Specimen Std . Cv % σ b ,c

ELWD a MPa k1 k2 k3 R2
Classification No. δ
MH (A-7-5) H1-1 154 8.92 3.8 0.87 1310 0.230 -1.04 0.96
H2-2 111 4.32 1.4 0.63 1440 0.329 -2.44 0.99
SM (A-4a) H6-1 32 3.03 3.8 0.18 567 0.808 -2.11 0.96
H3-1 41 1.03 1.3 0.23 513 0.970 -3.08 0.96

H2-1 64 3.07 2.7 0.36 634 0.882 -2.50 0.94
ML(A-4a) H5-1 48 2.94 4.7 0.27 680 0.910 -2.82 0.96
H4-1 21 2.07 4.2 0.12 666 0.879 -3.28 0.94
H8-1 35 3.05 4.0 0.20 555 0.925 -2.82 0.96
H4-2 30 1.72 3.08 0.17 646 0.896 -2.44 0.97
H6-2 20 0.51 2.1 0.11 717 0.673 -4.39 0.86
H3-2 56 2.40 3.6 0.32 593 0.874 -3.09 0.95
CL(A-4a) H5-2 20 3.31 9.2 0.11 897 0.897 -4.26 0.92
a
ELWD ( MPa)
b
σ (kPa)
c
δ ( µ m)
k3
k2
k1
C1
1.0
480
-3.7
C2
2.8
-0.9
1040
Table 3. Constant Coefficients of Developed Model
Page 34 of 34

Mousavi 2016

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mousavi 2016

Uploaded by

Copyright:

Available Formats

Page 1 of 34

Mousavi, Gabr, Borden

1 SUBGRADE RESILIENT MODULUS PREDICTION FROM LIGHT WEIGHT

6 Graduate Research Assistant, Department of Civil, Construction, and Environmental

Mousavi, Gabr, Borden

44 Keywords: resilient modulus, light weight deflectometer, subgrade, MEPDG

Mousavi, Gabr, Borden

57 schematically in Fig. 1 and expressed in Equation 1:

60 σ cyclic : cyclic axial stress

61 ε r : resilient axial strain

63 Figure. 1. Definition of resilient modulus

Mousavi, Gabr, Borden

72 universal constitutive model proposed in NCHRP 1-28A (MEPDG) is presented in Equation 2:

Mousavi, Gabr, Borden

Mousavi, Gabr, Borden

116 deflection for one drop.

118 Figure 2. Prima 100 sketch (after Vennapusa, 2008)

123 Equation 3 (Fleming 2007).

126 ELWD : surface Modulus (MPa)

Mousavi, Gabr, Borden

141 deviatoric stress = 69 kPa (10 psi).

144 With M r , ELWD in MPa

154 With M r , ELWD in MPa

Mousavi, Gabr, Borden

155 EXPERIMENTAL PROGRAM

166 1, and indicated in Fig. 1.

176 depths and corresponded to compacted fill.

Mousavi, Gabr, Borden

178 Table 1. Summary of Sample Properties

182 LWD MEASUREMENTS

Mousavi, Gabr, Borden

202 computing the ELWD by Equation 3.

205 tests were performed.

206 LABORATORY TESTING: RESILIENT MODULUS

214 Table 2. Summary of LWD Measurements and M r Model Parameters

216 EFFECT OF LWD INPUT PARAMETERS

Mousavi, Gabr, Borden

230 modulus value.

234 EXISTING MODELS

237 confining pressure and deviatoric stress.

241 Mr of the more highly plastic soils.

Mousavi, Gabr, Borden

243 Figure 7. Laboratory-measured M r vs. that predicted from existing models

245 DEVELOPMENT OF LWD CORRELATION

257 cohesive (A-7-5) and cohesionless (A-4a) soils.

258 MEPDG COEFFICIENTS FROM LWD

Mousavi, Gabr, Borden

268 i :1, 2,3

269 Table 3. Constant Coefficients of Developed Model

278 proposed model is able to compute the laboratory-measured Mr with a coefficient of

279 determination ( R2) of = 0.83.

280 Figure 9. Laboratory-measured M r vs. that computed prediction

282 Model Validation

Mousavi, Gabr, Borden

286 underestimates resilient modulus by 7% with a coefficient of determination of 0.83. The

289 included LWD measurements as well as laboratory-measured Mr data at a confining pressure =

298 model underestimates laboratory-measured Mr by 11% with a coefficient of determination of

Mousavi, Gabr, Borden

305 SUMMARY AND CONCLUSIONS

319 average underprediction of Mr. values with R2 = 0.96.

323 measured Mr. values.

Mousavi, Gabr, Borden

333 proposed approach to other soil types as well.