Determination of California Bearing Rati

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Determination of Califorrnia Bearing Ratio Through

Material Index Properties


1
Samar A. Taha, B.Sc., 2Sherif M. El-Badawy, Ph.D. Alaa M. Ali, Ph.D.
Public Works Engineering Dept. Faculty of Engineering Civil Engineering Dept.
Mansoura University Misr Higher Institute for Engineering and Technology
Mansoura, Egypt Mansoura, Egypt
e-mail: 1samar_civil_eng2010@yahoo.com, e-mail: allaawy@yahoo.com
2
sbadawy@mans.edu.eg

Abstract—In Egypt, California Bearing Ratio (CBR) is the major and implemented in the AASHTOWare Pavement ME Design
input parameter used to characterize the strength of the unbound since its first release [1]. These relationships are as follows:
materials and subgrade soils for pavement structural design.
Despite its simplicity, the CBR test is tedious and laborious. For cohesionless materials:
Thus, this paper focuses on the development of a simple but CBR = 28.091 (D60) 0.3581 (1)
reliable and rational regression model to predict the CBR of the
unbound materials as a function of the index material properties. For cohesive materials:

CBR 
A historical database containing measured CBR values along
75
1  0.728( wPI )
with index material properties was collected from Mansoura (2)
University Highway Engineering Laboratory. The index
properties in the database are gradation, liquid limit (LL), plastic
limit (PL), material diameter at which 60 percent of the material wPI = P200 . PI (3)
is finer (D60), soil classification according to American
Association of State Highway Transportation Officials
Where D60=diameter at 60% passing from the grain size
(AASHTO), maximum dry density (MDD), and optimum distribution, mm; wPI=weighted plasticity index; P200=
moisture content (OMC). The database contained granular as percent passing No. 200 U.S. sieve, decimal; and PI = plasticity
well as fine-grained materials with CBR values in the range of index, percentage.
10% to 95%. Analysis of the data showed that the CBR values It should be noted that, these relationships are questionable
are greatly influenced by percent passing No. 200 sieve and because of the way they were developed. Only seven and
MDD. A simple model predicting the CBR as a function of these
eleven data points were used to establish Eqs (1) and (2),
parameters was developed using optimization techniques. The
respectively. Moreover, the data points used for the
goodness of fit statistics of the measured and predicted CBR
values using the proposed model indicates a good prediction
development of these relationships were typical ranges and
accuracy and minimal bias. values of the CBR and Atterberg limits based on the material
class either according to AASHTO or Unified systems soil
Keywords-CBR; index properties; dry density; liquid limit; classification systems rather than laboratory measured values.
Proctor; soaked Many other correlations between CBR and index soil
properties were found in the literature. Some of these
I. INTRODUCTION correlations are summarized in TABLE 1. One can surmise from
California Bearing Ratio (CBR) is a measure of the relative the data shown in the table that most of these models are for
resistance of the unbound granular base/subbase materials and fine materials. Few models are for coarse materials. In
subgrade soils to uniaxial penetration. In Essence, it is a addition, some of the cited models are based on very limited
relative measure of the strength (stiffness) of the unbound materials while others are based on a good number of
materials. The CBR can be measured either in the laboratory or materials.
in the field. This test was developed by the California Division The correlations found in the literature are based on simple
of Highways around 1930s and was subsequently adopted by regression analysis between the CBR either soaked or
many countries around the world. In Egypt, CBR is the major unsoaked and the index soil characteristics and compaction
input parameter used to characterize the strength of the properties such as liquid limit (LL), plastic limit (PL),
unbound materials and subgrade soils for pavement structural maximum dry density (MDD), optimum moisture content
design. Despite its simplicity, the CBR test is tedious and (OMC), etc. Some recent correlations were based on the use of
laborious. Thus, many studies have been conducted to estimate Artificial Neural Networks (ANNs) such as the ones given
the CBR from the physical and compaction properties of the in [12], [13] and [14].
materials. The most important relationships found in the
literature are the ones developed at Arizona State University
TABLE 1. SUMMARY OF THE LITERATURE MODELS CORRELATING CBR AND INDEX MATERIAL PROPERTIES

No. of
Equation R2 Remarks Country Reference
Materials
CBR = 28.09 (D60) 0.358 7 0.84 Coarse Grained Soils

CBR 
75 USA [1]
1  0.728( wPI )
11 0.67 Fine Grained Soils

CBRu=17.009-0.0696(PI)-0.296(MDD) Error =
+0.0648(OMC) -2.5% South
12 Fine Grained Soils Gujarat, [2]
CBR=43.907-0.093(PI)-18.78(MDD) Error = India
-0.3081(OMC) -2.5%

CBR = 2- 16 log (OMC) + 0.07 L.L 48 N.A N.A India [3]

MDD = 19.2 (CBRtop / OMC) 0.05


57 N.A Fine Grained Soils Malaysia [4]
MDD = 18.5 (CBR bottom / OMC) 0.05
Fine grained, cohesive Victoria,
CBR = 4.5 [(20-GI)/18]2 N.A N.A [5]
soils with CBR ≤ 20% Australia
CBRu = 0.031(LL) + 83.19
CBRu = 0.8 (PL) + 65.31
CBRu = 10.43 (SG) + 56.19
CBRu = 8.66 (MMD) + 65.88 Osogbo,
8 NA lateritic soil (A-2-4) [6]
CBR = 0.22 (LL) + 28.87 Nigeria
CBR = 1.04 (PL) + 13.56
CBR = 9.42 (SG) + 10.91
CBR = 50.28 (MMD) -70.22

N.A
Mean CBR = - (0.25) LSP + 124 0.53 N.A N.A [7]

Log10(CBR) = 0.29(GM) - 0.024( PI) +1.23 N.A N.A Base course material N.A [7]
CBR=2.1 (eGM)–23Log [(LSP) (P425)0.7]+54 0.69
CBR=96.3–17.8Log[(LSP)(P425)0.7]–
0.69
28.7Log(P200)
CBR=97.7–17.1Log[(PI)(P425)0.5]–
0.66 South
30.7Log(P200) N.A [5]
Africa
CBR=119.6–33Log[(LL)0.7(P425)0.3]–
0.62
33.2Log(P200)
CBR=80.5–32.3Log[LSP(P425)0.7] 0.59
CBR=90– 47.4Log (P200) 0.59
 CBRTop 
0.18

OMC  5.73 
 No. of Blows  26 0.45 N.A Malaysia [8]

CBR=0.064(F)+0.082(S)+0.033(G)- 0.069(LL)
25 0.92 Fine Grained Soils India [9]
+0.157(PL)-1.810(MDD)-0.061(OMC)
CBR= -0.889(WLM)+45.616
N.A 0.979 Fine Grained Soils N.A [10]
where, WLM= LL (1 – P425/100)
CBRu=(8.44-16.1PI)[Fi+488PI2-314PI+45] Fine Grained Soils
24 N.A Sudan [11]
CBR=(1.44-4.23PI)[Fs+264PI2-56PI-5] (Silty Clay)
CBR = California Bearing Ratio (soaked), %; D60 = Diameter at 60% passing from grain size distribution, mm; wPI = Weighted
Plasticity index; MDD=Maximum dry density; CBRtop=CBR values at top face of sample; CBRbottom = CBR values at bottom face of
sample; OMC=Optimum Moisture Content; GI=group index; CBRu = unsoaked CBR; LL = Liquid limit; PL = Plastic limit; PI=Plasticity
index; SG = Specific gravity; LSP = Linear shrinkage P200 = passing No. 200 U.S. sieve, %; GM = grading modulus; P425 = passing
sieve size 0.425 mm; F=Fines, %; S=Sand, %; G=Gravel, %; Fi=initial state factor; Fi=soaking state factor.
However the use of the literature models to predict the CBR for The distribution of the soaked CBR values for the entire
materials different from the original materials used to develop database is shwon by the histogram in Fig. 1. The figure
the model is risky and might lead biased estimates with indicates that the largest portion of the data has a CBR in the
significant errors. For example, Datta and Chottopadhyay [15] range of 25 to 45% which is suitable for embankment filling
tried to use correlations given by Vinod and Cletus [16] and and subbase course layers according to the Egyptian
Patel and Desai [2] using soil properties reported by Roy, et specifications.
al., [17] and found significant scatter and bias in the predicted
CBR. Thus, before adopting any correlation calibration of the 35
models based on local materials is essential. Otherwise a new 30
model should be developed based on the actual data.

Frequency, %
25
20
II. OBJECTIVES
15
This paper focuses on the development of a simple but reliable
and rational regression model to predict the CBR of the unbound 10
materials used for road construction projects in the Delta region in 5
Egypt.
0

III. DATA COLLECTION CBR Value, %


For the purpose of the study, a database including different
materials used for pavement construction projects in the Delta
Figure 1. Histogram of California Bearing Ratio values in the database
region in Egypt was collected. The database was collected from
technical reports prepared at Mansoura University Highway
and Airport Engineering Laboratory during the period of 2009 IV. MODEL DEVELOPMENT
to 2013. The majority of the materials contained in the database In order to develop a model correlating the soaked CBR
are nonplastic cohesionless materials used as fill for road with the soil index properties, the literature models were first
embankments as well as subbase and base courses. This tried with the collected database. However, results show very
historical database contains measured CBR values, under poor correlation and significant bias. Thus, the scatter diagrams
soaked condition, along with index material properties and of the soaked CBR and each of the index soil properties were
compaction characteristics. The database contains the drawn and presented in Fig. 2 through 6. Fig. 2 to 4 show
following: relatively strong linear correlations between the CBR and
 Particle size distribution and D60.
MDD, P200 and P4, respectively. This is also indicated by the
 Atterberg Limits (liquid limit: LL, plastic limit: PL,
values of the determination coefficients (R2) which ranged
between 0.345 and 0.548 as shown on the figures. On the other
 Soil classification according to the AASHTO system.
and plasticity index: PI). hand, Figures 5 and 6 show very high scatter and very poor

 Modified Proctor parameters (Maximum Dry Density


correlation with D60 (R2 = 0.101) and LL (R2 = 0.025).
A multiple regression modeling was then tried using the
 Soaked CBR values
MDD, and Optimum Moisture Content: OMC).
solver tool within Microsoft Excel and the goodness of fit
statistics according to the conceptual criteria proposed by
The descriptive statistics of the parameters contained in the Pellinen, and shown in TABLE 3 was used to select the best
database are summarized in TABLE 2. model [18]. This criterion is based on the coefficient of
determination (R2) and the standard error divided by the
standard deviation of measured CBR values about the mean
TABLE 2. Descriptive Statistics of the Database ParameteRS
measured (Se/Sy). The R2 is simply the square of
the correlation coefficient between the measured and predicted
MDD OMC LL P4 P200 D60
Parameter CBR
(t/m3) (%) (%) (%) (%) (mm) CBR. Higher R2 values indicate higher accuracy (less scatter).
The Se/Sy is an indicator for the relative improvement in
Mean 36.39 1.85 10.51 23.94 87.31 1.96 0.77 accuracy. Smaller Se/Sy values indicate better accuracy. These
Median 30.00 1.82 11.00 24.00 96.30 0.80 0.90 statistical parameters are calculated with respect to the line of
equality which is a line with a slope of unity and intercept of
Mode 30.00 1.79 12.90 24.70 100.00 0.30 0.90
zero. Eqs (4) through (7) were used to calculate the goodness
Standard
19.84 0.18 2.40 3.96 22.20 2.18 0.34
of fit statistical parameters used to judge the model
Deviation accuracy [18]:
Range 84.20 0.76 12.82 21.00 100.00 6.52 1.40
Minimum 10.80 1.47 1.18 12.00 0.00 0.08 0.10
Maximum 95.00 2.23 14.00 33.00 100.00 6.60 1.50

No. of
Data 51 51 47 51 51 51 51
Points
100 100
y = 85.668x - 121.92 y = -9.6872x + 43.807
80 R² = 0.5481 80 R² = 0.0252

60 60
CBR, %

CBR, %
40 40

20 20

0 0
1.40 1.60 1.80 2.00 2.20 2.40 0 0.5 1 1.5 2
Maximum Dry Density (t/m3) D60, mm
Figure 2. Relationship between CBR and maximum dry density Figure 6. Relationship between CBR and diameter at 60 percent passing

TABLE 3. CRITERIA FOR GOODNESS-OF-FIT STATISTICAL PARAMETERS [18]

Criteria R2 Se/Sy

Excellent  0.90 ≤0.35


Good 0.70 – 0.89 0.36 – 0.55
Fair 0.40 – 0.69 0.56 - 0.75
Poor 0.20 – 0.39 0.76 – 0.89
Very Poor ≤0.19  0.90

 ŷ  yi 
n
2

Se 
Figure 3. Relationship between CBR and percent passing No. 200 sieve
i 1
i
(4)
np
100
 y  yi 
y = -0.5446x + 84.024 n
2

Sy 
80 R² = 0.3452
i 1
i
(5)
n 1
60
S 
CBR, %

R  1  e 
2

S 
(6)
 y 
2
40

20
 n  1   Se 

 1   
2

 n  p   S y 
(7)

2
0 R adj

0 20 40 60 80 100
Passing No. 4 Sieve, % where Se=standard error of estimate; Sy=standard deviation;
Figure 4. Relationship between CBR and percent passing No. 4 sieve yi=measured value; ŷi =predicted value; yi = average of yi,;
n=number of data points; p=number of regression coefficients,
R2=coefficient of determination; and R2adj =adjusted coefficient
of determination.
A number of different models have been tried and the best
goodness of fit statistics were found using the following model:
CBR= 0.025(P200)4+30.130(MDD)-25.813 (8)
Where CBR=soaked California Bearing Ratio, %;
P200=passing No. 200 U.S. sieve, %; MDD= maximum dry
density according to modified Proctor method, (t/m3).
This model have an R2 of 0.785, R2adj of 0.776, and Se/Sy of
0.463. Comparison between laboratory measured and predicted
Figure 5. Relationship between CBR and liquid limit soaked CBR using the proposed model is shown in Fig. 7 along
with the line of equality and the goodness of fit statistics. The
figure and the goodness of fit statistics show good estimates of REFERENCES
the soaked CBR.
[1] ARA., “Guide for Mechanistic-Empirical Design of new and
100 rehabilitated pavement structures, NCHRP 1-37A Final Report,
Goodness of Fit
Appendix CC-1: Correlation of CBR values with soil index properties”
Se/Sy 0.463 ERES Consultants Division, Transportation Research Board, National
80
R2 0.785 Research Council, Washington, D.C., 2004
Predicted CBR, %

R2Adj 0.776 [2] Patel, R. S. and Desai, M. D., "CBR predicted by index properties of soil
60 for alluvial soils of south Gujarat," In Proc. Indian Geotechnical
Conference, IGC, Vol. I, 2010, pp. 79-82.
40 [3] Agarwal, K.B. & Ghanekar, K.D; “Prediction of CBR values from
plasticity characteristics of soil,” In Processing of 2nd South East Asian
Conference on Soil Engineering, Singapore, July 11-15 (1970).
20
[4] Mak W.K.; “California Bearing Ratio correlation with soil index
Line of Equality properties”, In 16th South East Asia Geotechnical Conference, KLCC,
0 Kuala Lampour, 7-8 August 2006.
0 20 40 60 80 100 [5] I. J. Breytenback, "The ralationship between index testing and California
Laboratory Measured CBR, % Bearing Ratio values for natural road construction materials in South
Africa,” M.S. thesis, Departement of Geology, Faculty of Natural and
Figure 7. Laboratory measured and predicted soaked CBR using the proposed Agricultural Sciences, University of Pretoria, South Africa, April 2009.
model [6] Afeez Adefemi BELLO, “Regression analysis between properties of
subgrade lateritic soil,” Leonardo Journal of Science, Issue 21, July-
To evaluate the adequacy of the proposed model, an December 2012, pp. 99-108.
analysis of variance (ANOVA) was performed on the [7] Kleyn,S.A, "Pssible development in pavement foundation design,” The
regression results. This analysis was conducted with the null South African Instituation of Civil Engineers, 5(12), 1955, pp. 286-292.
hypothesis is that the CBR is not related to the P200 and MDD [8] Qotrunnada Binit Abdul Rahman; “Correlation between California
and the alternate hypothesis is that the CBR is related to both Bearing Ratio Results and physical properties of soils,” Faculty of Civil
variables. The results of the ANOVA test are summarized in Engineering, Universiti Teknologi, Malaysia, April 2010.
Table 4. [9] Ramasubbarao, G. V., and Siva Sankar, G., "Predicting soaked CBR
value of fine grained soils using index and compaction
characteristics," Jordan Journal of Civil Engieering, Vol. 7, No. 3, 2013,
TABLE 4. ANOVA RESULTS pp. 354-360.
[10] Vinod, P. and Reena, C., "Prediction of CBR value of lateritic soils
df SS MS F P-value using liquid limit and gradation characteristics data," Highway Research
Journal, IRC, Vol. 1 (1), 2008, pp. 89-98.
Regression 2 15624.29 7812.144 92.39217 3.49E-17
[11] Zumrawi, M., "Predcition of CBR value from index properties of
Residual 48 4058.601 84.55418 cohesive soils," In Proceedings of the Annual Conference of
Total 50 19682.89 Postgraduate Studies and Scientific Research 17-20 February 2012,
Friendship Hall, Khartoum, Vol. 1, pp. 111-117.
df = degrees of freedom; SS=sum of squares; MS=mean square; F = F statistic.
[12] Yildrim, B. and Gunaydin, O. "Estimation of CBR by soft computing
Since the P-value is 3.49E-17, which is much less than the systems," Expert Systems with Applications, ELSEVIER, 38, 2011, pp.
6381-6391.
significance value of 0.05, the null hypothesis is rejected. In
[13] Roy, T. K., Kuity, A., and Roy, K., S., "Predcition of soaked CBR for
other words, there is a relation between the CBR and the subgrade layer by using Arificial Nueral Network model," In
variables proposed in the model. Proceedings of the International Symobsioum of Engineering under
Uncertainity, Safety Assemsment and Management S. Chakraborty and
G. Bahattacharya eds. Springer, India, 2003, pp 1195-1206.
V. SUMMARY AND CONCLUSIONS
[14] Venkatasubramanian, C. and Dhinakaran, G., "ANN Model for
A historical database containing measured CBR values Predciting CBR from Index Properties of Soil," International Journal of
along with index material properties was collected from Civil and Structural Engineering, Vol. 2, Issue 2, 2011, pp. 614-620.
technical reports prepared at Mansoura University Highway [15] Datta, T., and Chottopadhayay, "Correlation between CBR and Index
Engineering Laboratory. The index properties in the database Properties of Soil," Paper No. A-350, In Proceedings of Indian
Geotechnical Confernce, December 15-17, 2011.
are gradation, LL, PL, PI, D60, AASHTO soil class, Modified
Proctor MDD, and OMC. Analysis of the data showed that the [16] Vinod, P. and Cletus Reena, "Prediction of CBR value of lateritic soils
using liquid limit and gradation characteristics data," Highway Research
CBR values are greatly influenced by percent passing No. 200 Journal, Vol. I, No. 1, 2008, pp. 89-98.
U.S. sieve and MDD. No strong correlation was evident [17] Roy, T. K., Chattopadhyay, B. C. and Roy, S. K., "Prediction of CBR
between the CBR and the index properties in the database. A from compaction characteristics of cohesive soil," Highway Research
simple but reliable and relatively accurate regression model Journal, July-Dec., 2009, pp. 77-88.
predicting the CBR as a function of these parameters was [18] Pellinen, T. K., "Investigation of the use of dynamic modulus as an
developed using optimization techniques. The goodness of fit indicator of hot-mix asphalt performance," Ph.D. dissertation, Arizona
statistics of the measured and predicted CBR values using the State University, Tempe, Arizona, 2001.
proposed model indicates a good prediction accuracy and
minimal bias (R2 of 0.785, R2adj of 0.776, and Se/Sy 0.463).

You might also like