Hassan Et Al. (2022)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Transportation Infrastructure Geotechnology

https://doi.org/10.1007/s40515-021-00197-0

TECHNICAL PAPER

Prediction of California Bearing Ratio (CBR) Using Index


Soil Properties and Compaction Parameters of Low Plastic
Fine‑Grained Soil

Jawad Hassan1 · Badee Alshameri1 · Faizan Iqbal1

Accepted: 28 August 2021


© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature
2021

Abstract
California bearing ratio (CBR) is a significant test for geotechnical and highway
engineering. CBR requires extensive time and physical effort, which affects the pro-
ductivity of the projects. This study aims to develop correlations to predict CBR
with index properties of fine-grained soils. Several natural soil samples were col-
lected from different areas of Islamabad, Pakistan. Geotechnical laboratory tests
of grain size analysis, specific gravity, Atterberg limits, standard Proctor, and Cali-
fornia bearing ratio were performed. Multi-linear regression analysis (MLRA) was
performed using statistical analysis software, SPSS. Multiple predictive models
for CBR were developed using SPSS in three stages, each stage with a different set
of index soil properties as input variables. Based on this study, it is observed that
CBR of fine-grained soils can be predicted with excellent accuracy using index soil
properties and compaction parameters as input parameters with high R2 values rang-
ing from 0.786 to 0.957 and significance of 0.000 for all predictive models. MLRA
models presented in this study are based on low plastic fine-grained soils and might
not be suitable to predict CBR for high plastic or course-grained soils.

Keywords California bearing ratio (CBR) · Multi-linear regression analysis


(MLRA) · SPSS · Maximum dry density (MDD) · Optimum moisture content
(OMC) · Index soil properties

* Badee Alshameri
badee.alshameri@yahoo.com; b.alshameri@nice.nust.edu.pk
Jawad Hassan
jawad.abidi10@gmail.com
Faizan Iqbal
faizaniqbal959@gmail.com
1
National University of Sciences and Technology, Islamabad, Pakistan

13
Vol.:(0123456789)
Transportation Infrastructure Geotechnology

1 Introduction

CBR can be performed in the laboratory as well as in the field. However, the field
CBR test was withdrawn by ASTM in 2018. CBR is an essential test in geotechni-
cal engineering, widely used in highway layers, earthen dams, and airport pave-
ments. However, since CBR needs 4 days to soak the specimen, it slows down
the progress of any project by a significant margin. In addition to that, every
megaproject needs hundreds of CBR test results every day which puts a consider-
able burden on material testing laboratories. Due to excessive time consumption
and physical effort required to compact the sample and transport the bulk sam-
ples to laboratory, CBR test affects the project’s productivity (Taha et al. 2019).
Because of much reliance on CBR, there is no substitute available for this test.
Considering the aforementioned points, CBR needs to be correlated with index
soil properties of soil to save time physical effort.
Numerous studies have been performed in the past to predict CBR using index
soil properties and compaction parameters, resulting in good predictive models.
Alawi and Rajab (2013) presented CBR prediction models for granular soils with
LA abrasion, grainsize parameters, MDD, and OMC using MLRA. Fikret Kurnaz
and Kaya (2019) compare different artificial intelligence like group method of data
handling (GMDH) and artificial neural networks (ANN) with MLRA to predict
CBR values of different types of soils with index soil properties and compaction
parameters as input variables. Katte et al. (2019) presented models based on single
linear regression analysis (SLRA) and MLRA to predict CBR values of granular
soils using soil index properties and compaction parameters as input variables.
Ramasubbarao and Siva Sankar (2013) predict CBR values using index soil prop-
erties and compaction parameters of fine-grained soils (majorly high plastic soils)
using SLRA and MLRA models. Zumwari (2012) used newly developed “soil state
factors” with plasticity index of fine-grained soils to predict CBR in both soaked
and unsoaked conditions. Rehman et al. (2017) present models for prediction of
CBR of granular soils, based on compaction parameters and index soil properties,
collected from different areas of Pakistan using MLRA. Singh et al. (2011) pre-
dict CBR of laboratory made samples of fine-grained soils; predictive models were
based on varying degree of compaction and moisture content along with index soil
properties. Taha et al. (2019) presents predictive models for CBR using MLRA
and ANN techniques of granular soils; gradation properties, Atterberg limits, and
compaction parameters were used as input variables. CBR prediction models based
on past literature consisting of different soil types and types of models are pre-
sented in Table 1.
Even though multiple studies have been carried on prediction of CBR for
fine-grained soils, no study is entirely based on low plastic soils. Therefore, this
study is based on the prediction of CBR values of fine-grained soils of low plas-
tic nature using index soil properties (grainsize parameters, Atterberg limits, and
specific gravity) and compaction parameters (OMC and MDD) using SPSS to pre-
dict the CBR ratio easier in fine-grained soils.

13
Table 1  Correlations for prediction of CBR by past researchers
Reference Model/equation(s) R/R2 Soil type

Fikret Kurnaz and Kaya (2019) CBR =  − 2914.53 + 28.948GC + 29.064SC + 28.812FC + 0.070LL − 0.128PI + 1.574M R = 0.83 Fine and granular
DD + 0.406OMC
Transportation Infrastructure Geotechnology

Taha et al. (2019) CBR = 53.37 MDD + 0.526 D60 + 0.303 R#4 − 61.39 R2 = 0.935 Fine and granular
Katte et al. (2019) CBR = −61.082 + 60.233(MDD) − 2.462(OMC) R2 = 0.819 Granular
Rehman et al. (2017) CBR = 6.508×D50 + 1.4×Cu + 3.970 R2 = 0.85 Granular
Taha et al. (2014) CBR = 0.025(P200)4 + 30.130(MDD) − 25.813 R2 = 0.78 Granular soils
Alawi and Rajab (2013) CBR =  − 112.4335 − 0.2856 LA − 4.7280 OMC + 98.4613 Density R2 = 0.94 Granular soils
Singh et al. (2011) MC R2 = 0.48 Fine-grained
CBR = −2.213 − 0.055 × OMC × 100 + 0.328 × Density
MDD
× 100 − 1.147 × PL
( ) ( )

2
Yildirim and Gunaydin (2011) CBR = 0.22 G + 0.045 S + 4.739 MDD + 0.122 OMC R = 0.88 Fine and granular
Breytenbach et al. (2010) CBR = 13.984 − (0.254 × PI) + (1.963 GM) – Various rock and soil types
Black (1962) CBR = SNq/10 – Inorganic cohesive

13
Transportation Infrastructure Geotechnology

2 Methodology

Thirty natural soil samples were collected from different areas of Islamabad,
Pakistan. The soil properties were tested using sieve analysis (ASTM D6913-17
2021), hydrometer analysis (D7928–17 2021), specific gravity (ASTM D854-14
2021), Atterberg limits (ASTM D4318—17e1 2021), standard Proctor (ASTM
D698 – 12e2 2021), and California bearing ratio (CBR) (ASTM D1883 – 16
2021). All the soil samples were classified according to the USCS soil classifi-
cation system (ASTM D2487—17e1 2021). It was found that most of the soil
samples in this study represent low plastic clay (CL) (21 samples) while some
samples belong to silty clay (5 samples) (CL-ML) and low plastic silt (ML) (4
samples) type of soils. The index soil properties along with compaction param-
eters, CBR values and USCS classification are presented in Table 2. Ranges of
different soil properties for the soil under study are as follows: gravel content
(0–7), sand content (7–26), fine content (72–93), LL (27–32.3), PL (17–22.6), PI
(4.9–12.4), MDD (1.924–1.996 g/cc), OMC (11.0–13.0), Gs (2.556–2.608), and
CBR (2.1–6.6).
Multi-linear regression analysis (MLRA) was performed to develop predictive
models for CBR; for this purpose, statistical analysis software SPSS was used
(Katte et al. 2019; Rehman et al. 2017). Fine content (F), gravel content (G), sand
content (S), specific gravity (Gs), liquid limit (LL), plastic limit (PL), plasticity
index, maximum dry density (MDD), and optimum moisture content (OMC) were
taken as input parameters, whereas California bearing ratio was taken as the out-
put parameter. To develop predictive CBR models, MLRA was performed in three
stages: in the first stage, index soil properties, grain size percentages along with
Atterberg limits and specific gravity, were taken as input variables. Grain size
percentages along with compaction parameters (MDD and OMC) were taken as
input variables in the second stage. Finally, in the third stage, all the input param-
eters were used to predict CBR.

3 Results

Stepwise and enter methods were adopted in SPSS to perform MLRA. MLRA
was performed in three stages with different combinations of variables so that
whatever soil properties are available, the best fit model can be used accordingly.
In the first stage, gravel content (G), sand content (S), fine content (F), specific
gravity (Gs), liquid limit (LL), plastic limit (PL), and plasticity index (PI) were
selected as input parameters for prediction of CBR. SPSS presented five models
in this stage. Pictorial representation of MLRA is presented in the form of a flow-
chart in the Fig. 1.

13
Table 2  Laboratory test results data along with USCS classification
Sample name Grainsize analysis Atterberg’s limit Specific gravity OMC MDD CBR USCS classification
(g/cc)
Gravel Sand Fine LL PL PI

TP-1 0 24 76 30 18 12 2.565 11.8 1.988 4.1 CL


TP-2 0 12 88 28.4 17 11.4 2.573 11.7 1.982 3.1 CL
TP-3 0 9 91 29.8 18.1 11.7 2.57 12.7 1.965 2.3 CL
TP-4 0 11 89 30.2 18.1 12.1 2.567 12.2 1.976 2.8 CL
TP-5 0 7 93 30.1 18.1 12 2.571 12.3 1.938 2.1 CL
TP-6 0 9 91 29.9 18.8 11.1 2.582 12.6 1.955 2.4 CL
TP-7 0 26 74 29.5 20.2 9.3 2.59 11.6 1.996 4.8 CL
Transportation Infrastructure Geotechnology

TP-8 0 8 92 31.2 19.3 11.9 2.562 12.3 1.958 2.6 CL


TP-9 0 9 91 32.2 20.1 12.1 2.556 12.1 1.964 2.6 CL
TP-10 0 10 90 31.2 19.3 11.9 2.566 12.2 1.952 2.5 CL
TP-11 0 10 90 31.1 18.7 12.4 2.571 12.4 1.955 2.8 CL
TP-12 0 11 89 30.5 18.8 11.7 2.565 12.4 1.945 2.9 CL
TP-13 0 14 86 29.4 18.4 11 2.58 13 1.924 2.5 CL
TP-14 0 9 91 31.4 19.4 12 2.575 12.2 1.962 3.2 CL
TP-15 0 8 92 32.3 20.1 12.2 2.577 12.1 1.97 3.5 CL
TP-16 0 11 89 30.4 18.8 11.6 2.58 12.7 1.942 3 CL
TP-17 0 13 87 29.8 18.6 11.2 2.571 12.3 1.932 2.8 CL
TP-18 0 10 90 30.9 19.1 11.8 2.558 12.4 1.96 3.4 CL
TP-19 0 12 88 29.9 18.4 11.5 2.56 12.8 1.934 2.9 CL
TP-20 0 9 91 31.5 19.5 12 2.562 12 1.968 3.7 CL
TP-21 5 18 77 28.2 21.9 6.3 2.582 11.3 1.974 5.6 CL-ML
TP-22 7 21 72 27 22.1 4.9 2.6 11.5 1.968 5.3 ML
TP-23 4 19 77 28 22.2 5.8 2.595 11.8 1.972 5.5 ML
TP-24 3 17 80 29.3 21.4 7.9 2.593 12 1.964 4.8 CL

13
Table 2  (continued)
Sample name Grainsize analysis Atterberg’s limit Specific gravity OMC MDD CBR USCS classification
(g/cc)
Gravel Sand Fine LL PL PI

13
TP-25 6 22 72 27 21.3 5.7 2.602 11 1.996 6.6 CL-ML
TP-26 5 18 77 28.1 22.4 5.7 2.605 11.4 1.982 6.2 ML
TP-27 7 19 74 27.7 22.6 5.1 2.603 11.4 1.994 6.5 ML
TP-28 4 17 79 28.2 21.9 6.3 2.604 11.3 1.976 5.7 CL-ML
TP-29 5 16 79 28.3 22.1 6.2 2.605 11.4 1.976 5.8 CL-ML
TP-30 7 20 73 27.2 21.9 5.3 2.608 11.2 1.992 6.4 CL-ML
Transportation Infrastructure Geotechnology
Transportation Infrastructure Geotechnology

Fig. 1  Flowchart of three stages of MLRA

3.1 First Stage Results

First stage models and their coefficient of correlation (R), coefficient of determi-
nation (R2), and the significance (α) of the model are presented in Table 3.
Out of the five models presented in Table 3, model number 4 is recommended.
The selection of the recommended model is based on the R2 value, fewer vari-
ables, and the significance of the model. Table 4 shows significance of the input
variables in every predictive model, for model 4 it is evident that fine content (F)
Table 3  CBR prediction models with grain size, Atterberg limits, and specific gravity as input variables
Model no Equation R R2 Sig. (α)

1 0.964 0.929 0.000


( )
CBR = −31.059 + 0.248(G).101(S) + 0.102(LL) + 0.237(PL) + 9.852 Gs
2 CBR = 8.727 − 0.491(PI) 0.932 0.868 0.000
3 CBR = 13.472 − 0.317(PI) − 0.076(F) 0.949 0.900 0.000
4 CBR = 6.285 − 0.095(PI) − 0.096(F) + 0.336(PL) 0.961 0.924 0.000
5 CBR = 5.030 − 0.113(F) + 0.426(PL) 0.960 0.921 0.000

13
Transportation Infrastructure Geotechnology

Table 4  Significance of input variables from models 1 to 5


Model no Significance of input parameters (α)
C F S G LL PL PI Gs

1 0.343 – 0.000 0.065 0.568 0.155 – 0.417


2 0.000 – – – – – 0.000 –
3 0.000 0.000 – – – – 0.00 –
4 0.040 0.000 – – – 0.008 0.341 –
5 0.063 0.000 – – – 0.000 – –

C, constant

and plastic limit (PL) have good correlation with CBR as they are statistically sig-
nificant (α < 0.05). However, plasticity index (PI) is not a statistically significant
variable for CBR (α > 0.05). Figure 2 compares the observed CBR value vs the
CBR value predicted by model number 4. As it is evident, most of the observed

Fig. 2  CBR observed vs CBR predicted by model 4 from Table 3

13
Transportation Infrastructure Geotechnology

Table 5  CBR prediction models with grain size and compaction parameters as input variables
Model no Equation R R2 Sig. (α)

6 CBR = 3.043 + 0.511(G) 0.910 0.829 0.000


7 CBR = 18.344 + 0.314(G) − 1.246(OMC) 0.949 0.900 0.000
8 CBR = 20.188 + 0.210(G) − 0.898(OMC) − 0.069(F) 0.964 0.930 0.000
9 CBR = −7.743 + 0.300(G) + 0.064(S) + 8.772(MDD) − 0.581(OMC) 0.966 0.934 0.000
10 CBR = 39.415 − 2.342(MDD) − 2.571(OMC) 0.887 0.786 0.000

values fall under the ± 10% margin, which shows that the predictive model is
reliable.

3.2 Second Stage Results

In the second stage, grain size analysis (G (%), S (%), F (%)) along with compaction
parameters (OMC and MDD) were selected as input parameters for the prediction of
CBR. SPSS presented five models. Second stage models and their coefficient of cor-
relation (R), coefficient of determination (R2), and the significance of the model (α)
are presented in Table 5.
Out of the five models presented in Table 5, model number 8 is recommended.
Even though model 8 has relatively lower R2 value, it is selected as the recom-
mended predictive model since all the variables involved in model 8 are statistically
significant (α < 0.05) as shown in Table 6. From Table 6, it can be concluded that
fine content, gravel content, and OMC correlate well with CBR. Figure 3 compares
the observed CBR value vs the CBR value predicted by model number 8. As it is evi-
dent, most of the observed values fall under the ± 10% margin, which shows that the
predictive model is reliable.

3.3 Third Stage Results

For the third stage of MLRA, all the input variables, including grain size analysis
(G (%), S (%), F (%)), Atterberg limits (LL, PL, and PI), specific gravity (Gs), and

Table 6  Significance of input Model no Significance of input parameters (α)


variables from models 6 to 10
C F S G MDD OMC

6 0.000 – – 0.000 – –
7 0.000 – – 0.000 – 0.000
8 0.000 0.003 – 0.001 – 0.002
9 0.681 – 0.005 0.000 0.262 0.139
10 0.182 – – – 0.850 0.000

13
Transportation Infrastructure Geotechnology

Fig. 3  CBR observed vs CBR predicted by model 8 from Table 5

compaction parameters (OMC and MDD) were selected as input variables to develop
CBR predictive models. SPSS presented five models. Five predictive models based
on the mentioned set of variables and their coefficient of correlation (R), coefficient
of determination (R2), and the significance of the model are presented in Table 7.
Even though all models in the table have very high R2 values, model number
12 was selected as the recommended model to predict CBR despite its relatively

Table 7  CBR prediction models with all (index properties and compaction parameters) input variables
Model no Equation R R2 Sig. (α)

11 CBR = −33.505 + 0.112(G)


( ) + 0.059(S) − 0.037(LL) 0.978 0.957 0.000
+0.287(PL) + 6.796 Gs + 9.985(MDD) − 0.445(OMC)
12 CBR = 20.564 − 0.325(PI) − 1.121(OMC) 0.963 0.927 0.000
13 CBR = 15.102 − 0.086(PI) − 0.819(OMC) − 0.066(F) + 0.254(PL) 0.975 0.950 0.000
14 CBR = 21.664 − 0.243(PI) − 0.950(OMC) − 0.047(F) 0.968 0.937 0.000
15 CBR = 14.049 − 0.826(OMC) − 0.081(F) + 0.335(PL) 0.974 0.948 0.000

13
Transportation Infrastructure Geotechnology

Table 8  Significance of input variables from model 11 to 15


Model no Significance of input parameters
C F S G LL PL PI Gs MDD OMC

11 0.257 – 0.019 0.322 0.810 0.047 – 0.501 0.161 0.196


12 0.000 – – – – – 0.000 – – 0.000
13 0.000 0.005 – – – 0.017 0.298 – – 0.001
14 0.000 0.046 – – – – – – – 0.001
15 0.000 0.000 – – – 0.000 – – – 0.001

lower R2 value than some other models in the table. The reason for the selection
of model 12 is the fewer number of tests and variables required. Furthermore all
the input variables in model 12 are statistically significant (Table 8) which means
CBR correlates well with PI and OMC. Figure 4 presents the observed CBR val-
ues against the predicted values using model number 12 equation. This relation-
ship is very reliable since all the observed values fall under ± 10% margin.

Fig. 4  CBR observed vs CBR predicted by model 12 from Table 7

13
Transportation Infrastructure Geotechnology

4 Conclusions

The following conclusions are drawn from this study:

• Multi-linear regression analysis shows that CBR of fine-grained soils can be pre-
dicted with reasonable accuracy using index soil properties like grain size analy-
sis, Atterberg limits, and specific gravity.
• Compaction parameters (OMC and MDD) provide a comparatively weaker cor-
relation with the CBR of fine-grained soils (model no. 10). But when used with
grain size analysis or Atterberg limits, good predictive models are presented
(model nos. 11, 12, 13, and 14).
• Best predictive models are encountered when all the input variables, i.e., grain
size, Atterberg limits, specific gravity, and compaction parameters, are used to
predict CBR.
• The MLRA models to predict CBR presented in this study are based on fine-
grained soils with low plasticity. Thus, it might not predict CBR with the same
accuracy in granular soils or soils with very high plasticity.

Acknowledgement Not applicable

Author Contribution The main idea was conceptualized by Badee Alshameri, whereas the methodology
was defined by Badee Alshameri and Jawad Hassan. The data was collected by Jawad Hassan. The SPSS
and linear regression analysis were performed by Jawad Hassan and Faizan Iqbal. The original draft was
prepared by Jawad Hassan and Faizan Iqbal and reviewed and edited by Badee Alshameri. The overall
work was carried out under supervision of Badee Alshameri.

Data Availability All data, models, and code generated or used during the study appear in the submitted
article.

Declarations

Conflict of Interest The authors declare no competing interests.

Originality Statement The work titled “Prediction of California Bearing Ratio (CBR) Using Index Soil
Properties and Compaction Parameters of Low Plastic Fine-Grained Soil” has not been published else-
where, in part, or in another form.

References
Alawi, M.H., Rajab, M.I.: Prediction of California bearing ratio of subbase layer using multiple linear
regression models. Road Mater. Pavement Des. 14, 211–219 (2013). https://​doi.​org/​10.​1080/​14680​
629.​2012.​757557
ASTM D698 - 12e2, A.: Test Methods for Laboratory Compaction Characteristics of Soil Using Standard
Effort (12 400 ft-lbf/ft3 (600 kN-m/m3)). ASTM International (2021)
Black: a Method of Estimating the California Bearing Ratio of Cohesive Soils from Plasticity Data *.
281–282 (1962)

13
Transportation Infrastructure Geotechnology

Breytenbach, I.J., Paige-Green, P., Van Rooy, J.L.: The relationship between index testing and California
bearing ratio values for natural road construction materials in South Africa. J. South African Inst.
Civ. Eng. 52, 65–69 (2010)
D1883 – 16, A.: Test Method for California Bearing Ratio (CBR) of Laboratory-Compacted Soils. ASTM
International (2021)
D2487 - 17e1, A.: Practice for Classification of Soils for Engineering Purposes (Unified Soil Classifica-
tion System). ASTM International (2021)
D4318 - 17e1, A.: Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils. ASTM
International (2021)
D6913–17, A.: Test Methods for Particle-Size Distribution (Gradation) of Soils Using Sieve Analysis.
ASTM International (2021)
D7928–17, A.: Test Method for Particle-Size Distribution (Gradation) of Fine-Grained Soils Using the
Sedimentation (Hydrometer) Analysis. ASTM International (2021)
D854–14, A.: Test Methods for Specific Gravity of Soil Solids by Water Pycnometer. ASTM Interna-
tional (2021)
Fikret Kurnaz, T., Kaya, Y.: Prediction of the California bearing ratio (CBR) of compacted soils by using
GMDH-type neural network. Eur. Phys. J. Plus. 134, 0–15 (2019). https://​doi.​org/​10.​1140/​epjp/​
i2019-​12692-0
Katte, V.Y., Mfoyet, S.M., Manefouet, B., Wouatong, A.S.L., Bezeng, L.A.: Correlation of California
bearing ratio (CBR) value with soil properties of road subgrade soil. Geotech. Geol. Eng. 37, 217–
234 (2019). https://​doi.​org/​10.​1007/​s10706-​018-​0604-x
Ramasubbarao, G.V., Siva Sankar, G.: Predicting soaked CBR value of fine grained soils using index and
compaction characteristics. Jordan J. Civ. Eng. 7, 354–360 (2013)
Rehman, A.U., Farooq, K., Mujtaba, H.: Prediction of California bearing ratio (CBR) & compaction
characteristics of granular soils. Acta Geotech. Slov. 14, 63–72 (2017)
Singh, D., Reddy, K.S., Yadu, L.: Moisture and compaction based statistical model for estimating CBR of
fine grained subgrade soils. Int. J. Earth Sci. Eng. 04, 100–103 (2011)
Taha, S., Gabr, A., El-Badawy, S.: Regression and neural network models for california bearing ratio pre-
diction of typical granular materials in Egypt. Arab. J. Sci. Eng. 44, 8691–8705 (2019). https://​doi.​
org/​10.​1007/​s13369-​019-​03803-z
Taha, S.A., El-Badawy, S.M., Ali, A.M.: Determination of California bearing ratio through soil index
properties. 4th Jordan Int. Conf. Exhib. Roads Transp. (JITC4). (2014)
Yildirim, B., Gunaydin, O.: Estimation of California bearing ratio by using soft computing systems.
Expert Syst. Appl. 38, 6381–6391 (2011). https://​doi.​org/​10.​1016/j.​eswa.​2010.​12.​054
Zumrawi, M.: Prediction of CBR from index properties of cohesive soils. Adv. Civ. Eng. Build. Mater. 1,
561–565 (2012). https://​doi.​org/​10.​1201/​b13165

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.

13

You might also like