Professional Documents
Culture Documents
ScienceDirect
Yingjun Jiang a,b, Louis Ngai Yuen Wong b,*, Jiaolong Ren a,c
a
Key Laboratory for Special Area Highway Engineering of Ministry of Education, Chang'an University, Xi'an 710064,
China
b
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore
c
School of Transportation, Southeast University, Nanjing 210096, China
Article history: In order to better understand the mechanical properties of graded crushed rocks (GCRs)
Available online 21 February 2015 and to optimize the relevant design, a numerical test method based on the particle flow
modeling technique PFC2D is developed for the California bearing ratio (CBR) test on GCRs.
The effects of different testing conditions and micro-mechanical parameters used in the
Keywords: model on the CBR numerical results have been systematically studied. The reliability of the
Graded crushed rocks numerical technique is verified. The numerical results suggest that the influences of the
Particle flow modeling loading rate and Poisson's ratio on the CBR numerical test results are not significant. As
California bearing ratio such, a loading rate of 1.0e3.0 mm/min, a piston diameter of 5 cm, a specimen height of
Numerical test 15 cm and a specimen diameter of 15 cm are adopted for the CBR numerical test. The
Micro-mechanical parameters numerical results reveal that the CBR values increase with the friction coefficient at the
Mesoscopic mechanism contact and shear modulus of the rocks, while the influence of Poisson's ratio on the CBR
values is insignificant. The close agreement between the CBR numerical results and
experimental results suggests that the numerical simulation of the CBR values is promising
to help assess the mechanical properties of GCRs and to optimize the grading design. Be-
sides, the numerical study can provide useful insights on the mesoscopic mechanism.
2015 Periodical Offices of Chang'an University. Production and hosting by Elsevier B.V. on
behalf of Owner. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
Attoh-Okine, 2008; Taskiran, 2010). The CBR test was originally degree of correlation coefficient. A reliable estimation of CBR is
developed by the California Department of Transportation and thus difficult to obtain by using the conventional statistical
was subsequently incorporated by the Army Corps of Engineers methods. Besides, in most studies, CBR was expressed in
for the design of flexible pavements (Al-Amoudi et al., 2002; terms of only one or two soil properties groups, with reference
Attoh-Okine, 2004; Duncan-Williams and Attoh-Okine, 2008; to soil parameters including plasticity, gradation and
Taskiran, 2010). It has become so globally popular that it is compaction properties. For example, PL belongs to plasticity
incorporated in many international standards (ASTM D group, while optimum moisture content is in compaction
1883e14, 1999). However, engineers always encounter group. The reason for this undesirable correlation is due to
difficulties in obtaining representative CBR values for design inadequate representation of soil properties groups, existence
(Al-Amoudi et al., 2002; Attoh-Okine, 2004; Duncan-Williams of complex relationships among the parameters and
and Attoh-Okine, 2008; Taskiran, 2010). Due to limited budget unpowerful methods of calculations (Taskiran, 2010).
and occasional poor project planning, GCRs investigation data Most of the above-mentioned models were developed based
are insufficiently obtained in many cases (Al-Amoudi et al., on statistical correlations, often inadequately supported by
2002). On the other hand, laboratory CBR test is often laborious mechanics-based analysis. Besides, more attention has been
and time consuming. The test results are also significantly placed on soil than on GCRs. Research on the effects of
influenced by the sample disturbance and laboratory testing composition and structure of GCRs on the CBR can help raise the
conditions (Al-Amoudi et al., 2002; Link et al., 1999; Taskiran, GCRs performance and improve the material design. The CBR of
2010; Yildirim and Gunaydin, 2011). Therefore, the GCRs depends on the quality and size of aggregate, aggregate
development of reliable prediction models might be useful to gradation and MDD, etc. However, it is very challenging to
supplement or replace some of the CBR tests, and to validate perform an in-depth and comprehensive investigation of the
the experimentally determined CBR values. properties of GCRs simply by directly observing and recording
Some attempts of such prediction model development were the movement of the GCRs in response to loading in the labo-
found in the literature. The effects of soil types and character- ratory CBR study. Particle flow modeling is one of the most
istics on CBR values have been studied (Agarwal and Ghanekar, rapidly developing numerical methods, which overcomes the
1970; Black, 1962; Degraft-Johnson et al., 1969; Stephens, 1990; limitation associated with the macroscopic continuum
Taskiran, 2010). Prediction of CBR of fine grained soils by artifi- assumption of relevant conventional numerical techniques.
cial intelligence methods was studied by Taskiran (2010). A Based on the microscopic mechanical properties of the indi-
correlation between CBR and plasticity index (PI) for cohesive vidual particles, the particle flow modeling has been success-
soils was developed by Black (1962). A correlation for CBR fully used for modeling the macroscopic material deformation
using the concept of suitable index which varied with behavior in rock, soil and other civil engineering materials
plasticity and grading characteristic was suggested by Degraft- (Bardet and Proubet, 1991, 1992; Lorig et al., 1995; Shen and Yu,
Johnson et al. (1969). A correlation equation between CBR and 2011; Vu-Quoc and Zhang, 1999; Yoon, 2007; Zeghal, 2004; Zhang
either liquid limit (LL), plastic limit (PL) or PI was developed by and Wong, 2012; Zhou et al., 2009). With the rapid advance of
Agarwal and Ghanekar (1970). However, they were not able to computational efficiency, particle flow code (PFC) offers a
establish any significant correlation among these parameters. powerful tool for investigating the key attributes of GCRs, which
Instead, they obtained an improved correlation when possess a discrete particle nature and non-linear mechanical
optimum moisture content and LL were included. It was found property (Itasca Consulting Group, 2004). To our best
that the grading constant is the best parameter to describe the knowledge, although PFC has been used extensively in civil
influence of grain size distribution on CBR. In the same study, engineering analysis, the use of this technique for modeling
CBR was found to be most dependent on maximum dry the CBR test is limited.
density (MDD) and least dependent on optimum moisture In the present study, a CBR numerical test method (NTM) of
content. Using grading constant, MDD and optimum moisture GCRs using the PFC2D is proposed. The effects of testing con-
content as independent variables, several equations for CBR ditions including loading rate, piston diameter and specimen
were presented. Stephens (1990) evaluated the performance of size on the stability of CBR numerical results are analyzed.
existing models for some selected natural soils and these The optimized testing conditions for the CBR NTM are sub-
models were found to be generally unsatisfactory. Another sequently recommended. Finally, the CBR NTM is validated by
method for the estimation of CBR was presented by British laboratory tests and an analysis of the mesoscopic mecha-
Highways Agency (1994), which made use of PI for British soils nism of CBR test is performed. The investigation of the present
compacted at natural moisture content where the correlations study is based on one typical limestone which is commonly
were given in the format of a table. NCHRP (2011) suggested found and used in Ankang District in China.
some correlations for describing the relationship between soil
index properties and CBR. A best-fit equation was proposed by
NCHRP for clean, coarse grained soil. This equation was 2. CBR NTM on GCRs
limited to D60 values varying between 0.01 and 30 mm. For
D60 smaller than 0.01 mm, the recommended value of CBR 2.1. Research approach
was 5% whereas CBR value of 95% was recommended for D60
greater than 30 mm. A review of the above literature reveals The research approach of investigating the CBR values of GCRs
that satisfactory correlations could not always be obtained, is based on both laboratory physical study and numerical
even for specific soils encountered locally. Many of the study (Fig. 1). Eventually, based on a reliability analysis, a CBR
proposed correlation equations suffered from a relatively low numerical test of GCRs is developed.
j o u r n a l o f t r a f fi c a n d t r a n s p o r t a t i o n e n g i n e e r i n g ( e n g l i s h e d i t i o n ) 2 0 1 5 ; 2 ( 2 ) : 1 0 7 e1 1 5 109
2.2. Numerical model then filled up by circular particles of various sizes. The grading
of the particles is controlled by monitoring the area of the
2.2.1. Specimen modeling generated particles. Afterwards, wall c is then applied to
To construct the CBR numerical specimen, orthogonal walls a, compact the top of the specimen in order to achieve the
b, d are first defined (Fig. 2). The space bounded by the walls is dimension requirements of the CBR numerical test.
2.3. Determination of testing conditions Fig. 4 e A load-penetration curve obtained by CBR NTM.
Fig. 10 e Effect of friction coefficient at the contact on CBR. Fig. 12 e Effect of shear modulus on CBR.
simulation can be quickly determined by the linear modulus and Poisson's ratio. These parameters are dictated by
relationship of CBR and shear modulus established above. the properties of GCRs. However, the circular particles adop-
ted in the PFC2D cannot directly reflect the angular and platy
3.1.3. Poisson's ratio nature of the crushed rocks and the voids of GCRs. In order to
For the GCRs used in road base, the Poisson's ratio typically reveal the influences of these factors on the numerical model,
ranges between 0.15 and 0.35. Numerical analyses are con- calibration is performed based on those micro-mechanical
ducted to obtain the CBR values for different values of Pois- parameters according to the following procedures. Note that
son's ratio, including 0.15, 0.20, 0.25, 0.30 and 0.35, while the the specimen height and diameter are both fixed at 15 cm,
friction coefficient at the contact and shear modulus are fixed while the piston diameter is fixed at 5 cm.
at 0.35 and 8 GPa respectively. The results are plotted in Figs.
13 and 14. (a) Obtain CBR values for three to five different sets of GCRs
A comparison of Figs. 9, 11 and 13, and Figs. 10, 12 and 14 is in the laboratory.
performed to study the degree of influence of different pa- (b) Based on the properties of GCRs, develop a preliminary
rameters on the CBR values. The general trend of influence of set of micro-mechanical parameters. Perform numeri-
friction coefficient on the CBR results is similar to that of shear cal analysis by PFC2D to determine the CBR values of the
modulus and Poisson's ratio. The CBR value increases by 34% specific set of GCRs tested in step (a).
when the Poisson's ratio increases 133% from 0.15 to 0.35. (c) Compare the CBR values obtained from laboratory study
Such a CBR percentage increase is comparable to that caused in step (a) and numerical study in step (b). If the
only by 33% increase of shear modulus from 6 GPa to 8 GPa.
The above illustrates that Poisson's ratio has a less degree of
influence on CBR numerical analyses, while the friction co-
efficient has the least influence relatively. To simplify the
parameter calibration procedures, only one single Poisson's
ratio based on the typical rock property is chosen and used
throughout the entire numerical study. Poisson's ratio will not
take part in the parameter calibration processes.
According to the Hertz model and the slip model, the key
parameters in characterizing the constitutive relationship of
the contact are friction coefficient at the contact, shear Fig. 13 e Load-penetration curves corresponding to
different Poisson's ratios.
Fig. 15 e Deformation of GCRs at different depths of piston penetration. (a) 0.80 mm penetration. (b) 1.60 mm penetration. (c)
2.50 mm penetration.
particleeparticle and particle-wall contact forces. Thicker increase of the specimen dimensions. When the specimen
lines represent a larger magnitude of contact force. height and diameter are both larger than 15 cm, the specimen
As shown, when the piston penetration depth is 0, i.e. the dimensions have insignificant effects on the CBR numerical
piston has not yet acted on the numerical specimen, the values. Based on the present study, a loading rate of
contact force among the particles are generally evenly 1.0e3.0 mm/min, a piston diameter of 5 cm, a specimen height
distributed. When the piston penetration depth reaches of 15 cm and a specimen diameter of 15 cm are recommended
1.75 mm, a remarkable force concentration occurs around the for the CBR numerical test.
piston. When the piston penetration depth reaches 2.50 mm, The micro-mechanical parameters have the following in-
the force concentration around the piston further increases, fluences on the CBR numerical results. The CBR curves
which reaches the bottom of the specimen. During the entire become gentler and the CBR values increase when the friction
loading process, although the contact forces on and close to coefficient at the contact increases. For friction coefficient at
the side walls also increase with the piston penetration depth, the contact greater than 0.7, such a variation trend is less
local force concentration is absent. The above phenomena remarkable. In contrast, the shear modulus has a far more
illustrate that the evolution of the contact force network in the influence on the CBR curves. The CBR value increases linearly
vertical direction is more prominent than that in the lateral with shear modulus. The influence of Poisson's ratio on the
direction. CBR numerical test is not significant. Parameter calibration is
The above numerical results suggest that the CBR NTM can subsequently performed by comparing the CBR value ob-
not only replicate the process of laboratory CBR test, but also tained from laboratory study and numerical study. Using the
provide more insights than those available by the conven- Ankang limestone as an example, the results illustrate that
tional laboratory CBR test. the percentage error between the numerical results and lab
results is less than 7% with an average value of 4.5%.The
reliability of the CBR numerical test is thus verified.
6. Conclusions A detailed analysis of the mesoscopic behavior of the GCRs
in the CBR test reveals the followings. When the piston
A numerical test method based on the particle flow modeling penetration increases, the change of the displacement vector
technique is developed for the CBR test on GCRs. The effects of magnitude and the contact force is more prominent in the
different testing conditions on the CBR numerical results have vertical direction than in the lateral direction. It illustrates
been systematically studied. The loading rate has a negligible that in the CBR test of GCRs, the dominant effect of the piston
effect on the numerical results. The CBR value of GCRs in- on the GCRs is reflected in the vertical direction rather than in
creases with the piston diameter, while it decreases with the the lateral direction. In addition, the disturbance of the piston
Fig. 16 e Visualization of mesoscopic contact force in the numerical specimen. (a) 0 mm penetration. (b) 1.75 mm
penetration. (c)2.50 mm penetration.
j o u r n a l o f t r a f fi c a n d t r a n s p o r t a t i o n e n g i n e e r i n g ( e n g l i s h e d i t i o n ) 2 0 1 5 ; 2 ( 2 ) : 1 0 7 e1 1 5 115
is not restricted to those shallow particles in vicinity of the Duncan-Williams, E., Attoh-Okine, N.O., 2008. Effect of geogrid in
piston. A significant disturbance effect on the deeper particles granular base strengthdan experimental investigation.
is also numerically obtained. When the penetration depth of Construction and Building Materials 22 (11), 2180e2184.
Hadi, M.N.S., Bodhinayake, B.C., 2003. Non-linear finite element
the piston reaches 2.50 mm, its disturbance reaches the par-
analysis of flexible pavements. Advances in Engineering
ticles located in the bottom of the specimen. Software 34 (11/12), 657e662.
Itasca Consulting Group, Inc, 2004. Particle Flow Code in 2
Dimensions (PFC2D) User's Manual. Itasca Consulting Group,
Inc., Minneapolis.
Acknowledgments Link, R.E., Pandian, N.S., Sridharan, A., et al., 1999. California
bearing ratio test simplified. Journal of Testing and
This research is supported by the Program for New Century Evaluation 27 (1), 72e75.
Excellent Talents in University (NCET-08-0749) and Funda- Lorig, L.J., Gibson, W., Alvial, J., et al., 1995. Gravity flow
mental Research Funds for the Central Universities simulations with the particle flow code (PFC). ISRM News
(CHD2012JC054). The first author thanks the Nanyang Tech- Journal 3 (1), 18e24.
NCHRP, 2011. Guide for mechanistic-empirical design for new
nological University for the opportunity to conduct research
and rehabilitated pavement structures. In: Appendix CC-1:
there as a visiting researcher.
Correlation of CBR Values with Soil Index Properties. A.R.A.,
Inc., Illinois.
Shen, S., Yu, H., 2011. Characterize packing of aggregate particles
references for paving materials: particle size impact. Construction and
Building Materials 25 (3), 1362e1368.
Stephens, D.J., 1990. Prediction of the California bearing ratio.
Agarwal, K.B., Ghanekar, K.D., 1970. Prediction of CBR from Civil Engineering in South Africa 32 (12), 523e527.
plasticity characteristics of soil. In: Proceeding of 2nd South- Taskiran, T., 2010. Prediction of California bearing ratio (CBR) of
east Asian Conference on Soil Engineering, Singapore, 1970. fine grained soils by AI methods. Advances in Engineering
Al-Amoudi, O.S.B., Asi, I.M., Wahhab, H.I.A., et al., 2002. Clegg Software 41 (6), 886e892.
hammereCalifornia-bearing ratio correlations. Journal of Vu-Quoc, L., Zhang, X., 1999. An accurate and efficient tangential
Materials in Civil Engineering 14 (6), 512e523. forceedisplacement model for elastic frictional contact in
ASTM D 1883e14, 1999. Standard Test Method for California particle-flow simulations. Mechanics of Materials 31 (4),
Bearing Ratio (CBR) of Laboratory-compacted Soils. American 235e269.
Society for Testing and Materials, West Conshohocken. Yildirim, B., Gunaydin, O., 2011. Estimation of California bearing
Attoh-Okine, N.O., 2004. Application of genetic-based neural ratio by using soft computing systems. Expert Systems with
network to lateritic soil strength modeling. Construction and Applications 38 (5), 6381e6391.
Building Materials 18 (8), 619e623. Yoon, J., 2007. Application of experimental design and
Bardet, J.P., Proubet, J., 1991. A numerical investigation of the optimization to PFC model calibration in uniaxial
structure of persistent shear bands in granular media. compression simulation. International Journal of Rock
Geotechnique 41 (4), 599e613. Mechanics and Mining Sciences 44 (6), 871e889.
Bardet, J.P., Proubet, J., 1992. Shear-band analysis in idealized Zeghal, M., 2004. Discrete-element method investigation of the
granular material. Journal of Engineering Mechanics 118 (2), resilient behavior of granular materials. Journal of
397e415. Transportation Engineering 130 (4), 503e509.
Black, W.P.M., 1962. A method of estimating the California Zhang, X.P., Wong, L.N.Y., 2012. Cracking processes in rock-like
bearing ratio of cohesive soils from plasticity data. material containing a single flaw under uniaxial
Geotechnique 12 (4), 271e282. compression: a numerical study based on parallel bonded-
British Highways Agency, 1994. Design Manual for Roads and particle model approach. Rock Mechanics and Rock
Bridges (DMRB). The British Highways Agency, Birmingham. Engineering 45 (5), 711e737.
Degraft-Johnson, J.W., Bhatia, H.S., Gidigasu, D.M., 1969. The Zhou, J., Wang, J., Zeng, Y., et al., 2009. Simulation of slope
engineering characteristics of the lateritic gravels of Ghana. stability analysis by particle flow code. Rock and Soil
In: Proceedings of 7th International Conference on Soil Mechanics 30 (1), 86e90.
Mechanics and Foundation Engineering, Mexico, 1969.