Version of Record: [Link]

com/science/article/pii/S0022489819301181
Manuscript_8a65ea05c6a1beff2599208256cbabae

An overview of methods to convert Cone Index to Bevameter

Parameters

Authors: Dr. George L. Masona , John E. Salmona, Sheila McLeoda,

Paramsothy Jayakumarb, Michael P. Coleb, and William Smithb

Center for Advanced Vehicular Systems, Mississippi State University

a
Center for Advanced Vehicular Systems (CAVS), Mississippi State University, Mississippi State, MS 39762 USA

b U.S. Army TARDEC, Warren, MI, 48397-5000

Abstract

This paper reviews experimental methods for the conversion of cone index measurements

to bevameter parameters in support of vehicle soil/tire/track interactions for two general soil

types, sand and lean clay. The accurate prediction of traction, motion resistance, and sinkage of

tire/tracks off-road requires estimates of soil strength. Equipment used in the measurement of

soil strength to support predictions of off-road mobility include the bevameter and the cone

penetrometer. The portability of the cone penetrometer and rapid estimates of spatial/temporal

variability in all terrain conditions make it an invaluable tool. The bevameter, a less portable

tool, is used for the mechanical analysis of soils. The bevameter measures parameters defining

soil strength in terms of cohesive modulus of soil deformation (kc), frictional modulus of soil

deformation (kφ), exponent of soil sinkage (n), cohesion (c), angle of internal friction (φ), and

the plate pressure at 1 inch (2.54 cm) of penetration (K) (Bekker, 1969). The field of

terramechanics would greatly benefit from having the ability to convert cone penetrometer data

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

© 2019 published by Elsevier. This manuscript is made available under the Elsevier user license
[Link]
in areas where bevameter parameters are difficult to collect. That ability to convert from cone

index to bevameter parameters could be used for the large sets of existing cone index data to

support determination of traction and motion resistance. This paper examines those methods for

converting cone index to bevameter plate penetration parameters kc, kφ, and n.

1. Introduction

The proper evaluation of the strength properties of a given terrain is essential to predicting

off-road vehicle mobility (e.g., net traction, gross traction, and motion resistance). Strength

properties of soils are measured by several different techniques, two of which include the cone

penetrometer and the bevameter.

The cone penetrometer was developed to determine a bearing capacity in terms of an average

cone index (CI) [lb/in2 , Pa] based in part on a rate of penetration 1.2 in/sec (30 mm/s) and a

critical depth related to the load exerted by the tire and/or track, typically 6 inches (15.24 cm).

The bevameter measures normal and shear stresses of soil to derive a number of soil parameters

(kc, kφ, n, c, φ, and K) providing a solution for the Bekker vehicle/tire interaction equation

(Wong, 1989).

The behavior of soils varies greatly under a wide variety of conditions including

composition, moisture content, density, temperature, and many other factors (Laughery, Gerhart,

Muench, 2000). For example, a three percent change in moisture content can greatly change soil

strengths, as shown in Figure 1. Figure 1 was taken from (Bekker, 1960) and appended to

include remold cone index (RCI) for a sandy clay (SC) as suggested by the sandy loam

classification and pictures by Bekker, furthermore utilizing equations correlating moisture to RCI

as defined in (Mason, 2001).

Unclassified.
Moisture Content (%) 22% 20% 19%

Angle of Internal Friction 36 38 36

(ɸ)
Cohesion (c) 0.25 0.53 0.6

Frictional Modulus of 2.2 7 9

Soil Deformation (kφ)
Cohesive Modulus of 2.5 16 20
Soil Deformation (kc)
Exponent of Soil Sinkage 0.18 0.17 0.16
(n)
Remold Cone Index 30 40 46
(RCI)
Figure 1. Consistency changes and the moisture content of a sandy loam (modified Bekker

1960).

Several sources tend to agree that finding bevameter parameters from cone index values to be

inconceivable or impossible (Janosi, 1959; WES, 1964). M. G. Bekker agreed with this

conclusion in his Introduction to Terrain-Vehicle Systems, 1969; however, he included a footnote

stating:

[the reverse process is inconceivable] if only a single index value is available. For the

minimum of three ‘cone indices’ measured at various depths the kc, kφ, and n values may be

theoretically calculated by means of an iterative computerized procedure using Janosi’s

Equation 1-6, as shown recently by J. Eilers of Land Locomotion Laboratory. Experimental

verification, however, is still lacking. (Bekker, 1969)

Unclassified.
Bekker’s footnote suggested the possibility of converting cone index values to bevameter

parameters utilizing differential pressure measured by the cone with depth. An equation

developed by Zoltan Janosi, discussed later, provides a relationship between the bevameter

parameters of the plate penetration part and cone index.

2. Background

Test methods and instruments were developed over the past several decades in attempt to

measure soil properties to predict soil/tire/track interaction. Janosi states, “the bevameter is

the standard method for scientific and engineering exploration of soils and off-road vehicle

design” (Janosi, 1959). The bevameter measures multiple soil values supporting numerical and

analytical simulations of cohesive and internal frictional forces to predict tire/track soil

interaction. The cone penetrometer measures one value at a set rate of penetration for a given

unified soil classification, supporting vehicle trafficability on a “go” or “no go” basis to support

the prediction of gross and net traction. The U.S. Army elected to use the cone penetrometer due

to its simplicity and ability to quickly collect spatial information (Janosi 1959).

Both the bevameter and cone index methods use a load (P) to displace the soil at a depth (z)

at a constant rate. The measurements from the bevameter plate load test and cone penetration

tests are both used to determine sinkage of the tire, supporting computations of motion resistance

and traction. The bevameter uses two or more rectangular or circular plates of width/diameter

(b) depending the characteristics of the soil and tire (Nguyen Van et. Al., 2007). The standard

mobility cone penetrometer consists of a 0.5 or 0.2 inch (12.7 or 5.08 mm) diameter cone

depending on soil type and vehicle loading. Both the bevameter and cone have a manual and

automated system supporting data collection in the field or laboratory.

3. Bevameter

Unclassified.
 M. G. Bekker originally proposed a system of seven soil values (kc, kϕ, n, c, φ, k1, and k2) to

describe the behavior of soils under vehicles (WES, 1964). In order to predict those properties,

the bevameter was developed. The bevameter is a device that replicates the sinkage of the

vehicle as it travels over a deformable surface while there is a change in surface traction of the

tire/track. In other words, it tries to replicate loading conditions a vehicle exerts on the soil

(Bekker, 1969). The bevameter has two separate tests, the plate penetration and shear ring, which

quantify soil parameters to support a solution of the tractive effort (Figure 2). The shear test

supports the prediction of surface traction while the plate penetration tests supports the

prediction as a result of bearing capacity. Wong suggested, “the bevameter technique provides

the closest simulation of vehicle loading conditions among the various measuring techniques

presently in use,” however, the bevameter is not widely available and not as portable as the cone

penetrometer (Wong, 1989).

Unclassified.
 Figure 2. Schematic view of a bevameter type instrument (Redrawn from Bekker 1969)

3.1 Plate Penetration Test

The bevameter parameters kc, kφ, and n are determined by a minimum of two

different size penetration plate tests. Plates of different sizes (b1, b2, b3) are used to

define the pressure-sinkage relationship. Rectangular or circular plate sizes range from ⅜

of an inch to three inches (9.52 to 76.2 mm) in width or diameter respectively. Normal

loads are replicated with the set of penetration plate tests (Bekker, 1969).

The exponent of soil sinkage, n [dimensionless], is the tangent of the slope angle

from the sinkage vs. pressure relationship. Typical test results are illustrated in Figure 3

from tests conducted in Vicksburg, MS jointly by the Tank Automotive Command

(TACOM) and Waterways Experiment Station (WES) (WES, 1964).

Figure 3. Determination of n (Redrawn from WES 1964)

Unclassified.
 It was considered in the (WES, 1964) report that n could be considered a constant for

each soil type. The n value varied between 0.11 and 0.20 for both the lean and heavy

clays tested in the fluvial plain of Vicksburg, Mississippi and sandy loam ranged between

0.2 and 0.15 as illustrated in Table 1. Dry sand tested in the laboratory had n values of

1.1. Sandy loam in Michigan and Maryland areas ranged between 0.3 and 0.8. The WES

study of 1964 suggested that n did not change significantly with changes in moisture

content or density. This seemed consistent from the sets of values provided in (Bekker,

1969).

Bekker defines the K value as the measured plate pressure at 1 inch (2.54 cm) of

penetration. K has different values for each size plate, and is considered to be a

composite of kc, kφ, and b (width of plates) as shown in the algebraic example in Figure 4

(WES, 1964). This is better explained in Figure 3, where the subscript of the K along the

y-axis is the width of each plate and the values of each K is where the three lines

intercept the y-axis. In this example, two of those intercepts are 150 and 160 psi,

correlating with the two K values in the algebraic solution in Figure 4. The cohesive

modulus of soil deformation as defined by Bekker, kc [lb/in(n+1) , kN/mn+1], and the

frictional modulus of soil deformation, kφ [lb/in(n+2), kN/mn+2], can be found graphically

or algebraically by a solution of simultaneous equations. An algebraic example of solving

for the values is shown in Figure 4.

. = + . = +

150 = + 160 = +
1.25 1.00

187.5 = + 1.25

− 160 = + 1.00

Unclassified.
 27.5 = 0.25

= 110

= 50

Figure 4. Algebraic example of kc and kφ

It is generally understood that for clays, the size of the plate compared to the size

of the footing does not change the results; however for sands, test plates smaller than the

footing tend to overestimate the sinkage/load relationship and give conservative factors

of safety. There is reason to assume that for wheels operating on sand the bevameter

plate load tests may operate in the same manner.

The unit conversion from English to metric units for kc and k is complex.. The

units for k and k are measured in units of force (F) per unit length (L) and force (F) per

unit length (L) squared respectively. The units for k and k are in terms of n defined by

the slope of the line derived during testing (WES, 1964). Equation 1 presents the

conversion of k and k from English units to metric.

: !" = # + %& !
"
(1)
$

()* + ()* +
∴ = $+",-. /01 , % = $+",-. /03

3.2 Shear Test

The shearing ring of the bevameter finds the surface shear strength parameters of

c, ϕ, and the slip coefficient K (or k1 and k2). An annular shear ring or plate is placed on

the soil with an applied normal load and rotated at a constant rate. The shearing ring finds

the shear stress vs. shear displacement relationship. A shear ring can have an outside

Unclassified.
 diameter of 13.5 inches (34.29 cm) and an inside diameter of 10.5 inches (26.67 cm). The

torque applied to the shear ring can be around 2000 lb-in (225.96 N-m). “The test is

performed at a range of normal loads to determine the Coulomb shear strength

parameters (Equation 2) corresponding to the soil/metal or soil/rubber shear” (Shoop,

1993). Similar to the plate load tests, values of c and φ are found graphically and

algebraically by the solution of simultaneous equations using the following equation

(Bekker, 1960).

4 = c + p tanφ (2)
Where:

4 is shear strength, lb/in2 (Pa)

c is cohesion, lb/in2 (Pa)

p is normal load, lb/in2 (Pa)

φ is the angle of internal friction, degrees

The shear strength vs. soil deformation graph in Figure 5 presents two different soil

relationships, a brittle soil, which is typically granular soil (e.g., sand) and a plastic soil

such as (e.g., clay). For brittle soils that display a decay in the stress-deformation graph

after a peak shear strength, the slip coefficients k1 and k2 can be determined either

algebraically or graphically.

This K-parameter can be found by using Equation 3.

τ = (c + p × tanφ) (1-e-j / K) (3)

Where:

τ is shear strength, lb/in2 (Pa)

c is cohesion, lb/in2 (Pa)

Unclassified.
 p is normal load, lb/in2 (Pa)

φ is the angle of internal friction, degrees

j is deformation, in (m)

K is the slip coefficient for plastic soils

The K parameter is the distance between the y-axis and the point of intersection of

two straight lines tangent to both the sloped and horizontal portions of the shear strength

vs. deformation curve as shown in Figure 4. (Bekker, 1969). Soils that do not display a

rapid decay or “hump” and represent an ideal plastic material can be defined by a single

K parameter (Figure 5) (Bekker, 1969). It is important to note that this K parameter is

different from the K value used to find kc and kϕ in the plate penetration part of the

bevameter.

Figure 5. Definition of Slip Coefficient K for plastic and brittle soils (Bekker, 1969)

4. Cone Penetrometer

Unclassified.
 The Waterways Experiment Station (WES) of the U.S. Army Corps of Engineers developed

the cone penetrometer during the Second World War. The cone penetrometer, as shown in Figure

6, is a device typically consisting of a ⅝ inch (1.58 cm) diameter rod, a 30-degree or 60-degree

right circular cone having base area of 0.5 in.2 (323 mm2) or 0.2 in.2 (130 mm2) respectively and

1.5 inches (3.81 cm) tall. The recommended rate of penetration is about 1.2 inch/s (30 mm/s)

(Wong, 1989). The cone index is the force required to sink the cone divided by the cone’s base

area. The reading from the dial on the penetrometer is twice the value of the force required to

penetrate the soil (Janosi, 1959). It was developed for the purpose of providing military

intelligence and reconnaissance with a simple, portable field device for predicting the soil

trafficability on a “go” and “no go” basis. In other words, it is used to determine if the soil is

capable of supporting certain vehicles without the vehicle becoming immobilized. The cone

penetrometer is lightweight, portable, and reliable for determining basic trafficability (Mason,

2015; Wong, 1989).

Unclassified.
 Figure 6. A cone penetrometer (Shoop, 1993)

5. Theory

Equation 4 is used in the foundation of Bekker’s bevameter tests in which a plate is

considered to penetrate a soil to depth z [in.] under pressure p [psi]. The empirical curve p(z) can

be fitted with the following equation, where k is the modulus of inelastic deformation and n is the

exponent of sinkage (Bekker, 1969).

5 ≈ 7" (4)

The K value in Equation 4 was very sensitive to the form of the test plate, which is

unacceptable for making generalizations of soil parameters. Equation 4 was later developed into

the following Equation 5, in which kc and kφ are cohesive and frictional moduli of deformation,

and (b) is the diameter of smaller loading plate (Bekker, 1960).

"
5 = [# & + % ]7 (5)

Thus, soil is defined in a load penetration test by three parameters, kc , kφ, and n, which define

empirically the vertical stress-strain relationship (Bekker, 1969).

In the relating of this equation to the cone penetrometer, the pressure (P) applied by a

cone or plate is the total force (W) divided by the area of the plate or cone. The cone is shown in

Figure 7 as a series of plates with varying diameter (dDx). In order to find the total force (W), it

is necessary to consider the equilibrium of the elemental truncated cone indicated by the cross-

hatching. The vertical active force (dW) equals the sum of the pressure acting against the mantle

of the cone element (Equation 6):

:; = <= :> (6)

Where (dA) is the area of the annulus shown at the bottom of Figure 7. The increment (dA)

can be expressed as Equation 7.

Unclassified.
 :@=
:> = ?[@= − :@= ]
2
A
∴ :; = <= @= :@= (7)

Figure 7. Notation and Equilibrium of Forces (Redrawn from Janosi 1959)

Where:

α is the vertical angle of the cone, 15 degrees

D is the diameter of the cone, 0.8 inches

@= is the diameter of the top of the truncated cone

:@= is the change in the diameter of the truncated cone with respect to dz

dA is the area of the annulus

dz is the height of the truncated cone

H is the height of the cone, 1.5 inches

Unclassified.
 px is the soil resistance force

zx is the height from the base of the cone to the height of a given diameter

z is the depth of the top of the cone to the surface

(CDE )3
Since ≈0

Substituting in Equation 7 we have Equation 8:

"A CDE
:; = ( + % )7= @= (8)
DE

From Fig. 7

@ G
=
@= 7 + G − 7=

Solving for Dx:

@
@= = (7 + G − 7= )
G

And differentiating Dx with respect to zx

:@= @
=−
:7= G

Giving:
A D D
:; = [ + % H (7 + G − 7= )]7=" (− ):7= (9)
H

The total force, W, is found by integrating Equation 9 between the limits (z) and (z+h):

AD JKH D D
;=− I [ + % H (7 + G − 7= )]7=" (− ):7=
H J H

Giving:

AD MD (JKH)/03 J /03 (JK )J /01

;=− { [(7 + G)"K − 7 "K ] + [ + − ]} (10)
H "K H ("K )("K ) "K "K

D D
From Fig. 7: = OP Q or = 2OP Q
H H

Unclassified.
The minus sign in Equation 10 can be neglected since it only means that W and the reaction

force act in opposite directions so that we are given Equation 11:

(JKH)/03 J /03 (JKH)J /01

; = ?OP Q{ [(7 + G)"K − 7 "K ] + 2 % OP Q[ + − (11)
"K ("K )("K ) "K "K

As previously mentioned, the dial of the cone penetrometer indicates a cone index reading that is

twice as large as the force, (W), required for penetration of the soil layer. If we denote the cone

index by CWES then

RSTU = 2;

The dimensions of the standard cone are:

H = height of the cone, 1.5 inches (38.1 mm)

V °
α= = 15°

2 π tanα = 1.625

2 tanα = 0.5175

Giving Janosi’s CI Equation:

(JK . )/03 J /03 (JK . )J /01

RSTU = 1.625{"K [(7 + 1.5)"K − 7 "K ] + 0.517 % [("K )("K ) + "K
− "K
]} (12)

The cone index, as defined by Janosi’s equation, is predicted as a function of the kc, kφ, n,

and the sinkage in inches, z (Janosi, 1959). Bevameter and cone penetrometer tests were run

concurrently under identical soil conditions. Numerous experiments were performed to check the

validity of the above equation. According to Janosi, the test results indicate that the equation is

accurate (Janosi, 1959).

6. Discussion

A computerized tool, like Microsoft Excel’s Solver, was used to run iterative procedures

to solve for kc, kφ, and n. The iterative solution was used to convert from cone index to

Unclassified.
bevameter values, varying in method depending on soil type. The method of predicting the

Bekker surface cohesion “c” and angle of internal friction “φ”, measured from the bevameter

shear ring, were not investigated in this project. However, additional research published by

Rohani and Baladi (1981) and Perkins, Meier, and Farr (1992) derived the mechanical properties

of the soil in terms of c and φ from cone index.

6.1 Comparison of Cone Index to Bekker Parameters

As previously discussed, most references on this topic state that predicting kc, kφ, and n, from

CI is impossible. Using Bekker’s suggestion in Introduction to Terrain-Vehicle Systems, an

iterative procedure using Janosi’s CI equation, comparable soil values, equations from reports

using similar soils, and field-measured cone index equations were used to predict values for kc,

kφ, and n. As previously mentioned, c and φ values were not found directly from CI values in

this project.

Various sets of data from several sources were used to give common values for CI, kc, kφ,

and n. Data sets with kc, kφ, and n values with corresponding CI values were of particular

interest. The following tables were used as reference.

Unclassified.
 Table 1. Sets of Values for Soils and Moisture Contents (Bekker 1969)

Moisture kφ kc c φ
Content % [lb/in(n+2)] [lb/in(n+1)] n (lb/in2) (degrees) Type/Location
13 7 5 0.80 1 29
11 6 11 0.90 0.7 20
Sandy Loam, Michigan (Strong,
23 20 5 0.70 1.6 25 Buchele)
23 27 15 0.40 1.4 35
21 38 14 0.40 2.5 22
32 1.2 0.7 0.50 0.75 11
31 1.2 1.5 0.40 0.8 15
30 0.1 7.5 0.40 0.9 23 Sandy Loam, Maryland (Hanamoto)
29 2.7 1.6 0.60 2 26
3 2.2 0.60 2.1 26
26 6.8 5.3 0.30 2 22
0 3.9 0.1 1.10 0.15 28 Dry Sand (Land Locomotion Lab)
45 13 14 0.30 0.8 17
47 8 24 0.60 1.1 14
38 16 12 0.50 0.6 13
185 3 3 1.00 0.5 11 Clayey Soil (Thailand)
55 5 7 1.00 0 11
55 14 7 0.70 0.3 10
43 4 22 0.90 0.26 10
25 140 45 0.13 10 34
30 65 25 0.12 7 22
Heavy clay (WES)
35 30 14 0.13 5 14
40 10 7 0.11 3 6
22 120 45 0.20 10 20
24 80 30 0.17 7 18
26 45 20 0.17 5 15
Lean clay (WES)
28 30 10 0.16 4 12
30 20 8 0.16 2.5 11
32 10 5 0.15 2 11
22 3 7 0.20 0.2 38 Sand Loam (LLL)

Unclassified.
 Table 2. Selected Excerpt of Bevameter Sinkage from WES 1964 and WES 1964a

Bevameter Sinkage
Lean Clay WES 1964 Yuma Sand WES 1964a
Average n kc kφ Average n kc kφ
Cone [lb/in(n+1)] [lb/in(n+2)] Cone [lb/in(n+1)] [lb/in(n+2)]
Index of 0 Index of 0
to 6 in. to 6 in.
[psi] [psi]
207 0.365 75 133 34 0.91 1.4 9.7
216 0.365 75 133 38 0.95 -0.6 9
210 0.365 75 133 36 0.88 0.8 10.3
211 0.365 75 133 37 0.97 0.8 9.1
218 0.365 75 133 42 0.86 2.8 8.6
201 0.365 75 133 15 0.79 0.6 7.2
158 0.193 15.7 69.5 46 0.76 0.3 15.8
152 0.193 15.7 69.5 40 0.89 0.6 11.3
158 0.193 15.7 69.5 61 0.78 3.1 17.9
124 0.218 21.4 57.3 60 0.62 1.4 19.7
128 0.218 21.4 57.3 44 0.94 0.3 10.2
120 0.218 21.4 57.3 25 0.89 1.3 6.9
99 0.211 26.6 31.1 23 1.02 -0.1 6.3
99 0.211 26.6 31.1 24 0.81 1.7 7.8
100 0.211 26.6 31.1 48 0.62 4.1 13.2
87 0.185 9.5 32.8 42 0.66 1.3 13.9
87 0.185 9.5 32.8 40 0.64 2.5 13
84 0.185 9.5 32.8 50 0.62 -3.5 19.2
62 0.146 0 44.2 43 0.7 0.1 14.8
60 0.146 0 44.2 35 0.66 1.9 12.6
69 0.146 0 44.2 23 0.89 0 7.6

Rula and Nuttall generated a set of predicted and measured cone indices based on a

“rearranged form of Janosi’s equations”. The equations listed for the prediction of cone index

include Equations 13-16 (Rula Nuttall, 1971).

YZ2
RX = ( + )
1
% (13)
YZ1 C

Where:

Unclassified.
 ("K1)("K2) (Z]1)/02
[1 = ∗ (14)
2H / Z/02 ]("K2)ZK("K1)

("K2) (Z]1)(Z/01 ]1)

[2 = 2
∗ Z/02]("K2)ZK("K1) (15)

H
[= J
=1 (16)

Where:

H = cone length, from tip to base, 1.5 in. (38.1 mm)

z = cone sinkage measured at the cone base, in.

d = diameter, 0.8 in. (20.3 mm)

This method was compared to Janosi’s equation (Equation 12) to test the similarity of

solutions given by two methods. It was found that when using the Rula and Nuttall method to

predict CI, the equations produced an answer that was double the actual CI values. In other

words, we were only able to replicate their data when using their exact equations and dividing

the final answer by two. This new CI equation was able to produce similar CI values to the

existing CI equation that was used for the purpose of this paper; however, the kc, kφ, and n values

that were generated were not comparable. It was decided after the comparison that the cone

index equation from Janosi’s 1959 paper was the desired method and worked better with other

data sets.

Two different soil types were investigated, sand and lean clay. The prediction of kc, kφ,

and n for sand was more complex than that of clay. The methods for each soil type are explained

in the following sections.

6.1.1 Sand

For the prediction of kc, kφ, and n for sand, three methods were used. All three of the

methods used an iterative procedure to solve for one or more unknown variables using

Unclassified.
 Janosi’s CI equation (Equation 12). Janosi’s CI equation contains four variables: z, kc, kφ,

and n. The depth of sinkage z was held at a constant of 6 inches (15.24) for all three

methods. For the first method, n was held as a constant set equal to one and kc was set

equal to zero, assumed from typical values for dry sand. The kφ value was then solved

using an iterative solution. The second method held the n value constant at one, and used

a linear relationship between kc and cone index (Equation 17) from the Measuring Soil

Properties in Vehicle Mobility Research (WES, 1964a) report for the kc value.

= 0.15 ∗ RX − 3.20 (17)

As in the first method, the kϕ value was solved for using an iterative solution. The first

and second method do not absolutely need an iterative solution to find the only missing

variable kϕ. The third method was truly an iterative procedure. All three of Bekker’s

parameters, kc, kφ, and n, were variables. The (WES, 1964a) report had kc, kφ, and n

values with corresponding cone index values. This full range of data was also used to find

an acceptable range for each of the three Bekker parameters to vary within. The

maximum and minimum value for each parameter were expanded by five percent to give

an ideal range. Equation 12 was set equal to field measured CI values, and the three

Bekker parameters were varied iteratively within the bounds of the ideal range and the CI

equation.

All three methods were compared with the original set of data from WES 1964a

report to see how closely they matched the field measured values. Table 3 shows the

standard deviation between the measured and predicted for each method.

Table 3. Standard Deviation of three methods in sand

Standard Deviation Values:

Measured vs. Predicted

Unclassified.
 kc kφ n
Method 1 2.748 4.085 0.161
Method 2 2.770 4.250 0.161
Method 3 2.483 3.194 0.173

Method 3 had the least deviation for kc and kφ, but a slightly larger value for n

compared to the other two methods.

6.1.2 Clay

In order to generate kc, kφ, and n values for clay soil types, linear soil relationships

and Janosi’s equation (Equation 12) were used. According to Strength-Moisture-Density

Relations of Fine-Grain Soils in Vehicle Mobility Research (WES 1964), heavy and lean

clay’s kc and kφ values share linear relationships with cone index. The following

equations (Equation 18 and 19) are the linear relationships, as shown in Figures 8 and 9

(WES, 1964).

= 0.19 ∗ RX (18)

% = 0.48 ∗ RX (19)

160
140 kφ = 0.48*CI
120
100
Kφ

80
60
40
20
0
0 50 100 150 200 250 300
Cone Index

Figure 8. kφ vs. Cone Index (Redrawn from WES 1964)

Unclassified.
 80
70
60 kc = 0.19*CI
50
Kc

40
30
20
10
0
0 50 100 150 200 250 300
Cone Index

Figure 9. kc vs. Cone Index (Redrawn from WES 1964)

These equations allowed for the generation of kc and kφ values that were used in Janosi’s CI

equation to solve for the unknown n value. Since n is an exponent in Janosi’s cone index

equation, the slightest change in the hundredth decimal place had an effect on the final predicted

result. The process of finding only one missing variable did not need an iterative procedure, like

methods one and two in sand. The linear relationships between CI, kc and kφ reduced the amount

of uncertainties in the prediction in clayey soils

7. Summary

The results that were obtained from this experimental process are not proven and have

considerable room for error. This process relies on assumptions made regarding soil values from

published reports containing values for similar soils. Those typical values were measured by

different teams working over a span of several decades with no standard technique. However,

some of the data did exhibit a consistent change with moisture and showed a believable

quantitative relationship between soil values (Bekker, 1969). For future use, these methods could

Unclassified.
be applied to other locations, assuming that those soils were similar enough to the Vicksburg

heavy/lean clays, and the Yuma Sands.

The assumptions made were reasonable based on the soil type of this project, but would not

have been feasible if there was an absence of pre-existing soils data to compare with the soil type

of the given project. To create bounds for an iterative solution to work within, average values for

cone index and corresponding bevameter parameters are needed.

The bevameter parameters from the WES 1964a report, when plugged into the CI equation,

did not produce a CI value similar to the measured CI value. Some data sets produced CI values

exactly double what they actually were supposed to be, while others had no consistent deviation

from the original measured value. This brings the validity of Janosi’s equation, or the data sets

into question. In respect to the computerized iterative process, there is also considerable room for

error. The output values of the iterative solver that was used in this project were only one set of

many possible combinations of values. The variety of possible combinations were due to the

complexity of Janosi’s equation and the nature of the iterative process.

8. Conclusion/Recommendations

The process of predicting Bekker parameters from known cone index values has been

considered impossible by many sources. It was suggested that the process of predicting the three

unknown bevameter variables in Janosi’s cone index equation may be theoretically worked by

means of an iterative procedure if additional information is provided such as soil type. This effort

used previously found relationships and known test values to predict bevameter parameters from

cone index data using direct and indirect methods in an iterative process. The validity of this

process and its results relies on the accuracy of previously determined soils data, relationships,

and the iterative process with Janosi’s cone index equation. It is recommended for future tests

Unclassified.
that a cone penetrometer and a bevameter be used in conjunction to verify the predicted values

when compared against the actual field-measured values. This experiment only used a small

scope of data in testing the experimental methods due to the small amount of bevameter data

with corresponding CI data. In these circumstances, the results cannot be confirmed and are

inconclusive. This process should be further explored to find if a more accurate and reliable

method of predicting Bekker’s kc, kφ, and n parameters from cone index values is possible.

9. Acknowledgments

This effort was funded by the Simulation Based Reliability and Saftey (SimBRS) research
program U.S. Government under Contract W56HZV-17-C-0095-TO-001. SimBRS focuses on
efforts of university and industry consortiums in meeting the objectives of TARDEC to address
the various aspects of design and simulation of vehicular systems to improve the safety and
reliability.

10. References

Ahlvin, R.B. (1992) NATO Reference Mobility Model Edition II, NRMM II Users Guide

Bekker, M. G. (1960). Off-the-road locomotion; research and development in terramechanics.

Ann Arbor: University of Michigan Press.

Bekker, M. G. (1969). Introduction to Terrain-Vehicle Systems. University of Michigan Press

Janosi, Zoltan. (1959) Prediction of "WES Cone Index" by Means of a Stress-Strain Function of

Soils.

Laughery, S., Gerhart, G., & Muench, P. (2000). Evaluating Vehicle Mobility Using Bekker's

Equations. Warren, MI: US Army TARDEC 48397-5000.

Mason, G., Ahlvin R., and Green, J., (2001) “Short-Term Operational Forecasts of

Trafficability” TR-01-22 Engineering Research and Development Center

Unclassified.
Mason, G., Priddy, J. D., Stevens, T. M., Towne, B. W., and Osorio, J. E., (2015) “Test

Procedures for the Determination of One-Pass Vehicle Cone Index (VCI)” Conference:

Proceedings of the 13th ISTVS European Conference, Rome Italy

Nguyen Van, Nang., Matsuo, Takaaki., Koumoto, Tatsuya., and Inaba, Shigeki (2008)

Experimental Device for Measuring Sandy Soil Sinkage Parameters” Bulletin Fac.

Agriculture, Saga University. No. 93

Perkins, W. E., Meier, R. W., & Farr, J. V. (1992, May). Strength Property Estimation for Dry,

Cohesionless Soils Using the Military Cone Penetrometer [PDF]. Washington D.C.: US

Army Corps of Engineers.

Rohani, B., & Baladi, G. Y. (1981). Correlation of mobility cone index with fundamental

engineering properties of soil: Final report. Washington, D.C.: GPO.

Rula, A. A., & Nuttall, C. J. (1971). An analysis of ground mobility models (ANAMOB).

Vicksburg, MS: Waterways Experiment Station.

Shoop, S. A. (1993). Terrain characterization for trafficability. Hanover, NH: U.S. Army Corps

of Engineers Cold Regions Research & Engineering Laboratory.

Waterways Experiment Station (1964). Strength-Moisture-Density Relations of Fine-Grain Soils

in Vehicle Mobility Research Waterways Experiment Station TR 3-639, Vicksburg, Miss.

Waterways Experiment Station (1964a). Measuring Soil Properties in Vehicle Mobility

Research, Strength-Density Relations of an Air-Dry Sand; Waterways Experiment

Station TR 3-652, Report 1, Vicksburg, Miss.

Wong, J. Y. (1989). Terramechanics and off-road vehicles. Amsterdam: Elsevier.

Unclassified.