You are on page 1of 10

T&DI ©ASCE 2014 813

Correlation between Subgrade Reaction Modulus and CBR for Airport


Pavement Subgrades

K. Tuleubekov1*
1
SRA International, 1201 New Road, Suite 242, Linwood, NJ 08221; PH (609) 601-
6800 ext. 250; FAX (609) 601-6803; email: Kairat_Tuleubekov@sra.com

D. R. Brill2
2
Federal Aviation Administration, William J. Hughes Technical Center, ANG-E262,
Atlantic City International Airport, NJ 08003.
*
Corresponding Author

ABSTRACT

Subgrade characterization plays a key role in the thickness design of airport


pavements. From a mechanical point of view, a subgrade soil represents a highly
diverse heterogeneous structure. The usual practical way to deal with heterogeneous
composites is to average their mechanical properties over the material volume. In the
current study, the particular soil parameters under consideration are Young’s modulus
E, subgrade reaction modulus k (k-value), and California Bearing Ratio (CBR).
The Federal Aviation Administration’s (FAA) FAARFIELD thickness design
program embeds a simple mathematical correlation between k-value and CBR. This
correlation is conservative in the sense of over-estimating the subgrade reaction for a
given CBR value, which encourages designers to unnecessarily increase the design
thickness of rigid pavements. On the other hand, the commonly used PCA
conversion formula overestimates k-value. Both formulas are not reliable for low-
strength soils.
The objective of this work is to present an updated conversion formula relating the k-
values and CBR for lower CBR soils. A new conversion formula is proposed based
on the analysis of available plate load and CBR tests for a range of subgrade soils.
Results are shown to be consistent with linear regression of experimental data
obtained during tests in the FAA’s National Airport Pavement Test Facility.

INTRODUCTION
From the mechanical point of view, subgrade soil represents a highly diverse
heterogeneous structure. Mechanical properties of soil vary both spatially and in
time. With current technology, to building a numerically efficient time-spatial model
of such materials is practically impossible, although there are particle-based
approaches to tackle the problem in smaller spatial scales (Oñate and Owen 2011,
Hoover 2006). Generally, the only practical way to deal with heterogeneous
composites is to average their mechanical properties over the material volume. As a
result of such a homogenization procedure, the researcher obtains effective subgrade

1
T&DI ©ASCE 2014 814

properties that depend on the particular type of soil under consideration. A


homogenization approach demands as large a sample size as possible to get the most
representative value for the soil property being considered. In the current study, the
the soil properties under consideration are Young’s modulus E, subgrade reaction
modulus k (k-value) and CBR.

PROBLEM FORMULATION
An existing correlation between k-value (k in pci) and CBR, used in FAA airfield
pavement thickness design (Advisory Circular 150/5320-6E, FAA 2009) is:

0.7788
⎛ 1500 CBR ⎞
k =⎜ ⎟ (1)
⎝ 26 ⎠

Equation (1) was obtained by substituting the following empirical relationship


between Young’s Modulus E (in psi) and CBR:

E = 1500 CBR (2)

into a correlation formula used for FAA rigid pavement design (Parker et al., 1977):

E = 26k 1.284 (3)

and solving for k.


Since Eq. (1) is conservative for estimating subgrade reaction modulus (Figure 1),
designers are encouraged to overdesign rigid pavement thickness. This is particularly
true since the FAARFIELD design procedure is more sensitive to the foundation
properties than the previous FAA design method based on Westergaard theory, as
given in the canceled AC 150/5320-6D. On the other hand, a commonly used
conversion formula developed by Packard (Barker and Alexander, 2012) for the
Portland Cement Association (PCA):

k = 53.438 CBR0.5719 (4)

apparently overestimates the k-value, as can be seen in Figure 1. Note that Eq. (4)
matches the conversion adopted by PCA (Packard, 1973) very closely up to CBR 10.
The test data plotted in Figure 1 are from the FAA’s National Airport Pavement
Testing Facility (NAPTF) located at the Atlantic City, International Airport, New
Jersey.

2
T&DI ©ASCE 2014 815

250
k, FAARFIELD conversion k, PCA Conversion
k, psi/in, from CC6 k, psi/in, from CC2
k, psi/in, from CC4
200

150
k, pci

100

50

0
0 2 4 6 8 10 12
CBR

Figure 1. NAPTF Test Data vs. Conversion Formulas

It must be emphasized that CBR is a measure of local soil shear strength, while k
measures foundation reaction in the elastic range only. Therefore, a simple
mathematical formula that converts from one to the other, and that is valid for all soil
types, is unrealistic without considering other soil test parameters. Both formulas (1)
and (4) must be used with great caution where corroboratory field data are lacking.
Figure 1 shows that there is considerable variability in the correlation even under the
highly controlled conditions prevailing at the NAPTF. Nevertheless, direct
conversion formulas such as Eq. (1) continue to be used in design because of their
great convenience. In the current study, we investigate ways to improve the
correlation for use in rigid pavement design. The objective of this study is to update
the existing AC 150-5320-6E conversion formula (1) between k-value and subgrade
CBR, reducing some of its over-conservatism.

PROPOSED APPROACH
The correlation between CBR and elastic modulus E is sought in the following form:
E = γ CBR (5)
where γ is the empirical parameter. As can be seen, equation (5) represents an
anticipated replacement of equation (3). In the light of relationships between the two
equations (5) and (2), the updated conversion formula can be written as:
⎛ γ CBR ⎞
0.7788

k =⎜ ⎟ (6)
⎝ 26 ⎠
The optimal value of factor γ is the one that gives the best fit to the available test
data for the k-to-CBR relationship (6).

3
T&DI ©ASCE 2014 816

EXPERIMENTAL DATA
Test results from the NAPTF (shown in Figure 2) were analyzed as a source of
experimental data. The NAPTF is a full-scale pavement test facility. Typically, tests
at the NAPTF are designated by a construction cycle (CC) number. Data for this
analysis were obtained during 3 rigid pavement construction cycles: CC2, CC4 and
CC6.

Figure 2. National Airport Pavement Testing Facility. (Photo credit: FAA)

Measurements of modulus of subgrade reaction were taken at the top of the subgrade
during construction of each of the three construction cycles. The measurements of
subgrade reaction were obtained from plate load tests following AASHTO T 222,
using the vehicle pictured in Figure 2 as the load reaction. Tables 1, 2 and 3
summarize the k-moduli from plate load tests for CC2, CC4 and CC6, respectively.
Station numbers are in hundreds of feet, and values for “North” and “South” refer to
locations at 15-ft. offsets left and right, respectively, of the test pavement centerline.
In all cases, the soil was Dupont clay, CL/CH.

Table 1. Modulus of Subgrade Reaction (k) for NAPTF CC2


Station No. k (pci), North k (pci), South
3+70, Test Item MRC 133.78 127.89
4+60, Test Item MRG 161.97 136.02

4
T&DI ©ASCE 2014 817

Table 2. Modulus of Subgrade Reaction (k) for NAPTF CC4


Station No. k (pci), North k (pci), South
3+62 132.01 132.19
4+52 169.78 165
552 130 158.01

Table 3. Modulus of Subgrade Reaction (k) for NAPTF CC6


Station No k (pci), North k (pci), South
3+50 132.8 145.80
4+30 157.86 165.62
5+70 127.19 136.8

ADDITIONAL DATA AND ANALYSIS


After collecting all the k-values from all three NAPTF construction cycles and
plotting them against the corresponding CBR values, it is seen (Figure 1) that the
resulting plot has too narrow a CBR range to make an informative regression
analysis. To extend the range of CBR values, NAPTF data were augmented with
additional data points (Figure 3) from the US Army Engineering Research and
Development Center (ERDC) (Barker and Alexander, 2012).

300
250
200
150
k, pci

100
50
0
0 5 10 15 20
NAPTF ERDC CBR

Figure 3. Additional Data from ERDC.

This extended set of data {CBRi , ki }in=1 was considered in a log-log scale and a linear
regression was performed with respect to those data (Figure 4), providing the
following result:
k = 29.8827 CBR0.7643 (7)

The coefficient of determination for the obtained regression (7) is as follows:

Sexp
R2 = = 0.61
Stot
where:

5
T&DI ©ASCE 2014 818

2
n⎛1 n ⎞
S exp = ∑ ⎜⎜ ∑ log k j − (29.8827 + 0.7643 log CBRi ) ⎟⎟ = 2.99
i =1 ⎝ n j =1 ⎠
is the explained sum of squares and
2
n ⎛1 n ⎞
Stot = ∑ ⎜⎜ ∑ log k j − ki ⎟⎟ = 4.89
i =1 ⎝ n j =1 ⎠
is the total sum of squares.
1000

100
k, pci

10
Linear Regression NAPTF ERDC

1
1 10 CBR 100

Figure 4. Linear Regression in log-log Scale.

Now recall parameter γ from equation (6). By minimizing the quadratic norm of the
misfit between curves given by equations (7) and (6) on the set of measured CBR
data:
2
n ⎡⎛ γ CBRi ⎞0.7788 ⎤
∑ − 29.8827 CBRi ⎥ → min
0.7643
⎢⎜ ⎟
⎢⎝ 26 ⎠
i =1 ⎣ ⎦⎥ γ

it is found that the optimal γ = 1935.48 . Inserting that value of γ into (6), the desired
conversion formula is obtained:
k = 28.6926 CBR0.7788 (8)
In Figure 5, it can be seen that formula of Equation (8) agrees with the linear
regression (cross markers) obtained from the experimental data.

6
T&DI ©ASCE 2014 819

300

250

200
k, pci
150

100

50
CBR
0
0 5 10 15 20
Linear Regression NAPTF
ERDC Proposed update
k, FAARFIELD conversion k, PCA Conversion

Figure 5. Comparison of Equation (8) with Linear Regression of Test Data.

APPLICATIONS
The new correlation between CBR and subgrade reaction modulus can be used
effectively in rigid airfield pavement design. Following are examples run using the
FA’s standard FAARFIELD 1.3 pavement thickness software. In these examples, the
traffic mixture input consisted only of a single aircraft: B777-200 ER with gross taxi
weight 634,500 lbs, 425 annual departures, and zero percent annual growth.
In the first case (new rigid design), assume that based on soil tests, design CBR=5.
The existing AC 150/5320-6E correlation formula (1) yields k=82.4 pcf, while the
new correlation (8) yields k=100.5 pcf. Using those k-values as comparative inputs to
the FAARFIELD new rigid pavement design procedure produces the results in Figure
6. All else being equal, the computed PCC surface thicknesses are 15.60 inches for
the current conversion formula, versus 14.96 inches by the proposed formula.
Assuming that concrete thicknesses would be specified to the nearest 0.5 in. per AC
150/5320-6E, using the proposed formula in this case would result in a savings of 0.5
in. of concrete thickness.

7
T&DI ©ASCE 2014 820

Figure 6. FAARFIELD New Rigid Design Comparison (CBR 5).

A similar computation for CBR 10 is shown in Figure 7. In Figure 7, the comparison


is 13.45 in (current) versus 12.31 in. (proposed), resulting in a reduction in specified
concrete thickness of one inch.

Figure 7. FAARFIELD New Rigid Design Comparison (CBR 10).

An even more pronounced difference in designed thicknesses for the two conversion
formulas is apparent for the second case, hot-mix asphalt (HMA) overlay design over
existing rigid pavement. While this result may be somewhat surprising, it should be
recalled that in FAARFIELD 1.3 HMA-on-rigid overlay design, the HMA design
thickness is based on protecting the receiving PCC layer from deterioration, not on
limiting reflection cracking failure of the HMA. Thus, the same sensitivity to
subgrade modulus that is evident in new rigid design also applies to overlay design.
In Figure 8, on the left, the k-value calculated from Eq. (1) is used, while on the right,
the k-value is calculated from Eq. (8). In both cases subgrade CBR=10. In this
comparison, the proposed CBR-k conversion would produce an HMA overlay
thickness thinner by 25% than the existing FAARFIELD conversion.

8
T&DI ©ASCE 2014 821

Figure 8. FAARFIELD HMA-on-Rigid Overlay Design Comparison (CBR 10).

Another application of the presented new formula of CBR-to-k correlation is


estimation of subgrade strength for PCN ratings. Once the subgrade’s k-value is
obtained (e.g. by the back-calculation procedures discussed in FAA AC 150-5370-
11A), then the CBR of the subgrade can be estimated from the proposed formula.

CONCLUSIONS
An improved formula for CBR-to-k conversion is proposed that agrees with
correlations obtained from soil tests. The data set consists of plate load and CBR tests
from the FAA NAPTF for clay soils in the CBR 7-8 range, augmented with published
data from US ERDC for low- to medium-CBR soils. The conversion preserves the
power form of the Packard formula with different-valued constants. Based on the
analysis presented in this paper, the proposed conversion could serve as an update for
the CBR-to-k conversion recommended in AC 150/5320-6E, and could be
implemented in future revisions to FAARFIELD. We emphasize that this formula, as
in any other correlation formula, should be used only when there is a lack of field test
data supporting a particular k-value. This study illustrates that the new conversion
formula, which is less conservative than the existing FAARFIELD conversion, results
in a more economical (thinner) pavement design.

ACKNOWLEDGMENTS/DISCLAIMER
The work described in this paper was supported by the FAA Airport Technology
R&D Branch, Michel J. Hovan, Manager. The contents of the paper reflect the views
of the authors, who are responsible for the facts and accuracy of the data presented
within. The contents do not necessarily reflect the views and policies of the FAA.
This paper does not constitute a standard, specification or regulation.

APPENDIX
Conversion from English to metric units:
1 lb = 0.453 kilogram; 1 in = 2.54 centimeter;
1 psi = 6894.75 Pa; 1 pcf =16.02 kilogram / m3;

9
T&DI ©ASCE 2014 822

REFERENCES
Barker, W.R. and Alexander, D.R. (2012). “Determining the Effective Modulus of
Subgrade Reaction for Design of Rigid Airfield Pavements Having Base Layers.”
ERDC/GSL TR-12-X.

Federal Aviation Administration (2009). “Airport Pavement Design and Evaluation.”


Advisory Circular 150/5320-6E, Office of Airport Safety and Standards, Washington,
DC.

Hoover, W. G., (2006). “Smooth Particle Applied Mechanics: The State of the Art.”
World Scientific Publishing Co.

Oñate, E. and Owen, R. (2011). “Particle-Based Methods: Fundamentals and


Applications.” Springer.

Packard, R. C. (1973), Design of Concrete Airport Pavement, Engineering Bulletin,


Portland Cement Association, Skokie, Illinois (reprinted 1995).

Parker, F., Jr., W.R. Barker, R.C. Gunkel, and E.C. Odom (1977), “Development of a
Structural Design Procedure for Rigid Airport Pavements,” Federal Aviation
Administration, Technical Report No. FAA-RD-77-81.

10

You might also like