Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Terramechanics, 1971, Vol. 8, No. 2, pp. 49 to 60.

Pergamon Press
Printed in Great Britain.

PLASTICITY THEORY AND THE STRESS DISTRIBUTION


BENEATH WHEELS

LESLIE L. KARAFIATH*

INTRODUCTION
THE DETERMINAa'IONof wheel motion resistance under various soil loading conditions
is one of the major objectives of land locomotion theory. The two most widely used
approaches to this problem as well as the ideal approach are illustrated schematically
in Fig. 1. The basic concept of the Bekker theory is that the vertical stress beneath a
wheel at a certain depth below the original ground surface equals that obtained in a
plate sinkage test at the same depth; wheel motion resistance is then determined on
the basis of the stress distribution beneath the wheel. The major steps in this approach
are shown on the left side of Fig. 1. Waterways Experiment Station (WES) approached
the problem differently. In their method, the motion resistance is determined directly
by experiments and expressed in terms of dimensionless parameters selected on the
basis of dimensional analysis; the relationships obtained are assumed to be valid for
any combination of the dimensional variables that results in the same dimensionless
parameter. However, there are limitations imposed on the number of independent
variables in the dimensional analysis, and, therefore, a unique experimental relation-
ship can be established only for either purely cohesive or purely frictional soils. The
soil properties are characterized in the independent dimensionless parameters by the
cone index for purely cohesive soils and by the gradient of cone index for purely
frictional soils. The schematics of the WES approach are shown on the right side
of Fig. 1.
The desirability of establishing direct relationships has been recognized, but the
problem has been termed too complex for a direct approach, schematically shown in
the center of Fig. 1. For the problem of lunar locomotion the direct approach is
especially attractive, since reasonably good estimates of the soil strength parameters
are available from Surveyor and Apollo mission data, while plate sinkage or cone
penetrometer data are lacking, and controlled wheel performance tests under lunar
gravity conditions present almost insurmountable difficulties.
The purpose of this paper is to show that a direct approach using computer tech-
niques is feasible, that the theory yields results that agree well with terrestrial experi-
ments and that the incorporation of the effect of slopes presents no difficulties.

WHEEL PERFORMANCE AND STRESS DISTRIBUTION BENEATH WHEELS


It has been shown by Schuring [1] and others that a close relationship exists between
wheel performance and the stress distribution on the wheel-soil interface. Jo-Yung
Wong and Reece [2] predict wheel performance on the basis of a symmetrical stress

*Research Scientist, Grumman Aerospace Corporation, Bethpage, New York 11714 U.S.A.
49
50 LESLIE L. KARAFIATH

IROPERSTOLs
BASIC J

I
PLATESINKAGE SOIL MECHANICS CONEPENETRO- l
ESTS STRENGTHTESTS ] METERTESTS

I I I
K~,Kc,n "Y,~C, J O o GRADIENT
OF CI

[iiRAm
I
I
I
BASICTERRAIN[ TERRAIN ]
PARAMETER PARAMETERS
PARAMETERS SLOPEANGLE,~
] I I
EXP. RELATION-
STRESS STRESS
I DISTRIBUTION DISTRIBUTION SHI PS AMONG
D IMENS IONLESS
J BENEATHWHEELS BENEATHWHEELS PARAMETERS

I WHEEL
PERFORMANCE
I
F]o. 1. Schemes of various approaches to the problem of wheel performance.

distribution which varies with the differences of the cosines of the central angle and
the angle of maximum normal stress. The parameters in the distribution function
are chosen so that the distribution fits empirical data.
It should be obvious that the distribution of stresses is the key to wheel performance
calculations. Once it is known, the problem reduces to that of a free body, the wheel,
acted upon by external stresses. The resultants of these stresses, obtained by simple
integration procedures, yield the desired vertical load, drawbar pull, and torque.

FAILURE CONDITIONS IN SOIL BENEATH WHEELS


It is generally recognized that the soil is in failure condition beneath driven or
towed rigid wheels; it has also been suggested that failure conditions prevail in the
soil beneath pneumatic tires [3]. Observations of these failure conditions suggest that
there are two separate zones of failure, a forward and a backward one [4] and [5].
These failure zones are thought to be in the vertical plane of motion; in certain
cases lateral displacement was observed in the front of the wheel, suggesting three
dimensional failure conditions [6] and [7]. For the proper assessment of this situation
the following considerations apply.
The soil-wheel interface represents a discontinuity in the soil mass where a Coulomb
type equation may be assumed to express the limiting relationship between shear and
normal stresses; the strength parameters in the equation are the adhesion and the
angle of friction between wheel and soil. In the soil mass, failure is assumed to be
governed by the Mohr-Coulomb yield criterion; the strength parameters are the co-
hesion and the angle of internal friction of the soil. The strength parameters in the
Coulomb equation are generally lower for the interface than those for the soil mass;
in extreme cases they may be equal.
PLASTICITY THEORY 51

In the idealized rigid wheel problem, all applied stresses are in the vertical plane of
motion. The normal stresses are limited by the condition that the sum of the vertical
components of the resultant stresses at the soil-wheel interface must equal the load; in
this summation the share of the shear stresses is minor. The magnitude of the shear
stresses varies with the applied torque; it is the purpose of wheel performance studies
to establish relationships between torque and other performance parameters.
The vectors of the normal and shear stresses are in the vertical plane of motion.
Considering the normal and shear stresses as stresses applied to the soil mass, we can
determine the stresses in the vertical plane on the assumption that the soil is in failure
condition. Such calculations show that both principal stresses increase with the
increase of the applied shear stress.
The third principal stress is perpendicular to the vertical plane of motion; this may
be called lateral stress to distinguish it from the other stresses in the vertical plane.
According to the Mohr-Coulomb theory, the failure plane is perpendicular to the
plane in which the major and minor principal stresses lie. This is the vertical plane as
long as the smaller principal stress in that plane is lower than the lateral stress; with
an increase of the applied stresses, however, the lateral stress may become the minor
principal stress, changing the ~1, ~3" plane to an oblique one, as shown in Fig. 2. Just
before this situation occurs and at the limit when the lateral stress and the smaller
principal stress in the vertical plane are equal, failure conditions in the vertical plane
govern the stresses; with lateral failure this condition still has to be maintained, other-
wise the failure mode would revert to that in the vertical plane. Thus the effect of
lateral failure is a limitation on the applied shear stresses; at this limit the stresses can
be calculated on the basis of two dimensional failure conditions.

APPLICATION OF PLASTICITY THEORY FOR THE DETERMINATION OF SLIP LINE


FIELDS AND ASSOCIATED STRESSES
The differential equations of plasticity express the combination of the differential
equations of equilibrium with the Mohr-Coulomb failure criterion. For the coordinate
system and notations shown in Fig. 3, these differential equations are:
dz = dx tan (0 4" ~)

da 4, 2a tan q~d0 -- Y [sin (~ 4- q~) dx q- cos (e 4" q~) dz]. (1)


cos q~

The upper sign refers to the family of the slip lines corresponding to the first, and the
lower sign corresponds to the second characteristics of the differential equations.
For numerical computations, the above differential equations are replaced by
four difference equations that allow the computation of the coordinates as well as
the ~ and 0 values at a point from values known at two adjacent points previously
computed or given at the boundary. Details of the numerical procedure are given in
[8] and [9] for the case of bearing capacity calculations; to determine slip line fields
beneath the wheel, the boundary conditions at the bearing surface had to be modified
so that the curvature of the wheel as well as the shear stresses at the boundary could

*For notation, see p. 60.


52 LESLIE L. K A R A F I A T H

VERTICAL PLANE~
O-n,r, o'], o"3 "-~

I i/ i i/ r~-x \ \\\ \

'OBLIQUE PLANE
/ "Ct
°"1 ,

//// O-n

FIG. 2. Stress state beneath a wheel--applied stresses in the vertical plane: ~,,, r, principal stresses in
the vertical plane due to applied stresses ~ , % ~ ~j.

FIG. 3. Coordinate system and notations.


PLASTICITY THEORY 53

be taken into account. This required a different sequence of operations at the boundary
than in the bearing capacity case; the curved boundary was approximated by a
polygonal one, and the quantities 0, x, z, a were determined in that order.
The accuracy of the numerical computations can be increased by an iteration pro-
cedure for each point; in the computations presented herein one iteration was applied,
as recommended by Sokolovskii [9]. Although further iteration would have improved
the accuracy of the calculations, in some instances the iteration is very slow, requiring
considerable computer time. For the purpose of this paper it was not deemed necessary
to apply multiple iterations.
The numerical computation of the slip line field geometries and associated stresses
was performed on an On-Line-Systems time sharing system. This system is based on
Digital Equipment Corporation PDP-10 processors. The coordinates of the slip line
fields obtained in the computer program were plotted automatically and electronically
on the display tube of a Computer Displays, Inc., Advanced Remote Display System.
The figures showing the slip line fields were photographically reproduced from the
display.

LIMITATIONS
There are limitations to the application of plasticity theory for the determination of
failure conditions beneath wheels. The theory is based on the differential equations
of equilibrium and the Mohr-Coulomb failure criterion. The equilibrium equations,
as used in the development of plasticity theory, do not contain inertia terms. Conse-
quently, the solutions obtained by plasticity theory are strictly valid only for the static
case and approximately valid where inertia forces are negligible.
The application of the Mohr-Coulomb failure criterion in the plasticity theory also
requires qualifications. In the context of plasticity theory the Mohr-Coulomb failure
criterion applies only if pore pressures are negligible. If the Mohr-Coulomb failure
criterion expresses an undrained strength condition, the formal application of plasticity
theory with undrained strength parameters no longer yields physically acceptable
results because the slip lines in such a solution intersect each other at an angle that
corresponds to the undrained angle of friction. The angle between slip lines depends
on the drained angle of friction, as theory and experiments show.
A further limitation imposed by the Mohr-Coulomb yield criterion, as applied in
the present theory, is that it assumes a linear relationship between shear and normal
stress. This limitation, however, is not inherent in the theory. Sokolovskii [9],
Szymanski [10], and others have shown that nonlinear strength relationships can be
adapted in the theory and numerical computations.
Kinematic admissibility of the slip line fields is not considered a limitation to the
applicability of plasticity theory because in soils that compress significantly under
wheel loads, the deformations adjust themselves to the conditions required by statically
admissible solutions.

COMPARISON OF EXPERIMENTAL AND THEORETICAL RESULTS


The stress distributions determined by the computer program on the basis of
forward and backward slip line fields were compared with experimental results ob-
tained in the laboratory. Experiments where only normal stresses were measured were
54 LESLIE L. KARAFIATH

not suitable for this purpose since shear stress is a principal variable in the theory
influencing the distribution of normal stresses. Another factor restricting the use of"
available experimental results is the lack of data on the shear strength of soil.
Figures 4 and 5 show a comparison of measured stresses, taken from Figs. 17 and 19
in [11], and computed ones. The pertinent slip line fields are shown in Figs. 6 and 7.
Forward of the wheel the soil is in its original state and is compacted by the stresses
developing in the forward slip line field. In the backward zone the soil is already in a
compacted state. This difference in the soil conditions affects the shear strength of the

- - MEASURED

. . . . COMPUTED

- 1500

/ I
- IOOO
¢ = 40' ~'
_,a
\
\
\

-500

-30 -20 -10 O 10 20 30 40 50 °

FIG. 4. Comparison of measured and computed stresses, driven wheel. Measured stresses from
Ref. [11], Fig. 17.

soil and must be accounted for by using appropriate shear strength parameters for the
forward and backward field. The shear strength parameters used in the computer
program for the determination of stresses in Figs. 4 and 5 are shown in the figures;
they agree well with the shear strength, which was determined by triaxial tests to be
q~ = 38 ° for the loose and q~ = 41 ° for the dense state in the soil mechanics laboratory
at G r u m m a n on the test sand obtained from the Land Locomotion Laboratory where
Sela performed his experiments.
It should be noted that the above friction angles are significantly different from that
obtained by Sela in Bevameter tests (q~ ~ 24°). The reason for this discrepancy is that
lateral failure limits the magnitude of applicable shear stresses in the Bevameter test,
just as in the case of wheels; therefore, Bevameter tests (apart from effects of geometry)
PLASTICITY T H E O R Y 55

- - MEASURED

I . . . . . . COMPUTED

-2000

E
g
,..J

-1000

-a
- 10 0 10 20 30 40 5do
a

FIG. 5. Comparison of measured and computed stresses, driven wheel. Measured stresses from
Ref. [11], Fig. 19.

FIo. 6. Slip line fields for the computation of Fro. 7. Slip line fields for the computation of
stresses shown in Fig. 4. stresses shown in Fig. 5.

are representative of the strength that may be developed at the wheel-soil interface
rather than of the internal friction of soil.
Figure 8 shows the stress distribution on a towed wheel measured by Onafeko and
Reece [12] and computed on the basis of slip line fields shown in Fig. 9. For the
computation of the forward field, q~ = 34 ° and c ~ 0.1 lb/in 2 were used, which is
slightly higher than the strength determined by Bevameter tests. In the backward field
= 31 ° and c = 0.02 lb/in 2 were used to obtain stress values close to the measured
one.
There are two reasons why the measured stresses agree with computed stresses when
a lower friction angle is used in the backward field. One is that a skidding wheel fitted
with triangular lugs may actually loosen up the soil; the other is that with the triangu-
lar lug arrangement, the shear stresses measured in skid might not have been accurate,
as the authors state in [12].
56 LESLIE L. KARAFIATH

- - MEASURED
. . . . . COMPUTED

O','t"
2000

1500

~1000

500

I
"10

-500
FIG. 8. Comparison of measured and computed stresses, towed wheel. Measured stresses from
Ref. [12], Fig. 13 (a).

Fl~. 9. Slip line fields for the computation of stresses shown in Fig. 8.

The above examples show reasonable agreement between measured and computed
stresses; they also indicate the significance of, and the need for, complete documenta-
tion of the soil strength properties in land locomotion experiments.

CALCULATION OF WHEEL PERFORMANCE


To compute the soil-wheel interface stresses on the basis of the development of
forward and backward slip line fields, the following variables and conditions have to
be considered.
Soil properties:
(1) Shear strength parameters (c and ~) both in the forward and in the backward field.
These are entered in a given computation as constants.
(2) Inclination of the resultant stress at the interface to the normal (8). This may be
entered as a constant for a given problem or as a function of central angle or slip.
(3) Unit weight of soil (7).
PLASTICITY THEORY 57

Wheel geometry:
(1) Wheel radius.
(2) Entry and rear angles (ae, ~tr), subject to constraints.
(3) Angle of separation (0tin) of the forward and backward slip line field.
Constraining conditions:
(1) Adjoining normal and shear stresses computed from the forward and backward
field, respectively, must be equal.
(2) The sum of the vertical components of the interface stresses equals the load.
(3) The sum of the components of the interface stresses parallel to the surface equals
the drawbar pull for a driven wheel and the towing force for a towed wheel.
The numerical solution of the differential equations represents the complex inter-
relationship of all variables. They yield a unique solution, if such exists, and if two of
the three central angles defining the end points of the slip line fields are known. The
constraining conditions are then met by trial and error procedure. The variation of the
three central angles in wheel performance tests has been analyzed and experimental
relationships with other wheel and soil characteristics established [2] and [11]. In the
discussion below, considerations applied in this study for the choice of the angle of
separation are presented.

Choice of the angle of separation


The concept of forward and backward failure zones is helpful in understanding the
physical significance of the angle that separates the two zones. The role of the failure
bodies in the force system acting on the wheel is that the forward failure zone resists
the motion while the backward zone supports it by providing a forward directed
reaction. Accordingly, the components of stresses in the direction of motion on every
element of the soil-wheel interface are directed backward in the forward zone and
forward in the backward zone. It is inconceivable for the forward zone to contain an
interface element with a component in the direction of motion; such an element would
tend to join the backward failure zone. The angle that separates the forward and the
backward zone is the one at which the sign of the stress component changes from that
in the backward direction to that in the forward direction.
To find the angle at which this sign change takes place, the relationship between
shear and normal stresses at the soil-wheel interface is assumed as follows.

za----*ntan 8a (2)

where the subscript d denotes developed shear stress (or traction) and friction angle
between soil and wheel.
The component of the shear and normal stresses in the direction of motion

A ~-- za cos ct -- ~rn sinct


(3)
A = ~r. tan ~a cos a -- ~n sin a.

To find the a., angle at which sign change occurs, A is equated with zero and the
following equation is obtained:
58 L E S L I E L. K A R A F I A T H

tan fie - - tan e,~ : 0


(4)
am ~ 8d ,

the angle of separation am equals the developed friction angle. This is, indeed, what
Sela found in his experiments with dry sand.
Others found the a,, angle somewhat lower than 8d. A nonlinear relationship
between developed shear stress and normal stress would account for this phenomenon.
Unfortunately data on this relationship are insufficient to make an in depth analysis
of the problem; we intend to discuss this relationship and its relevance to slip in a
separate paper. For the presentations that follow, a m = ~d is assumed.

Method of calculation of wheelperformance


The method of calculation and the results obtained are shown in an example. The
starting point is the determination of soil strength properties, which is best done by
triaxial tests. In the example, c ---- 10 lb/ft 2 and q~ = 35 ° are assumed for the forward
field, and c = 10 lb/ft 2 and q~ = 38 ° for the backward field. The unit weight of soil is
assumed to be y = 96 lb/ft s. In actual cases the increase of friction angle and unit
weight may be directly related to the compaction effected by the wheel at various
sinkages; since the purpose here is to show the method, such refinements were not
deemed necessary. The wheel radius in the example was assumed to be 1.67 ft. The
angle of separation a,~ was assumed to equal 8a.

LOAD
...... DRAWBARPULL(DIbs)
3000

,~ 38 ° D =0
e= ~ D =50

W
(Ibs) 2000

/ ~ ~ ", ~D,200
/ ~ ', Ao\

1000
\ o "-2,,
~ o , "\. B='\ o
o ""-L ,. "Qo" . ' ~ " - - % ) 4 . 2 , D .
= , "to.. c_ _~ " o
% 16~ "x..... • '~e=32",D=68

~tl0
, I l , I , =
10 20 30
Flo. I0. Results o f wheel p e r f o r m a n c e calculations for driven wheel: © = point calculated for
level g r o u n d , e a r t h gravity; • = point calculated for sloping g r o u n d , earth gravity; • = point
calculated for level g r o u n d , l u n a r gravity.
PLASTICITY THEORY 59

In practical problems and in experiments, wheel performance parameters are usually


determined for constant load. In the present method, the load is calculated from other
variables; performance parameters for a given load can be determined by inter-
polation. It is convenient to consider 5d as the principal variable and determine per-
formance parameters for various sinkages. Results of such calculations are shown
in Fig. 10, where the full lines represent variation of loads with 5d at constant sinkage
(or entry angle). The corresponding drawbar pull values are shown by dashed lines;
both drawbar pull and load values shown refer to 1 ft width of the wheel. The actual
calculated points are indicated by circles; again the plot could be refined by increasing
the number of calculated points, but for the purpose of presentation of the method
this was not deemed necessary. Figure 11 shows a typical slip line field pertaining to
the circle marked by A in Fig. 10.

Special conditions
One of the advantages of the method presented is that it allows analysis of the effect
of any of the variables that affect wheel performance. An important variable is the
slope angle, the effect of which has not yet been analytically treated in land locomotion
theories. A detailed discussion of this problem will be presented elsewhere; here only
a sample slip line field is shown for e = 12 ° in Fig. 12. All other conditions are the
same as in the preceding problem; the calculated wheel performance parameters are
shown by a full circle marked B in Fig. 10. The comparable drawbar pull figures are
200 lb for level and 27 lb for sloping ground. If the component of the load in the
direction of slope, 133 lb, is added to the net drawbar pull, the total drawbar pull
amounts to 160 Ib, some 40 lb less than that obtainable on horizontal ground.
The method also allows the consideration of lunar gravity in wheel performance
calculations. Figure 13 shows the slip line field for lunar gravity conditions assuming
the same strength properties as in the preceding examples. The calculated wheel per-
formance data are shown by a full rectangle marked C in Fig. 10.

FIG. 11. Slip line field for computation of Fro. 12. Slip line fields for driven wheel on
stresses for driven wheel. Performance data sloping ground (e = 12°).
indicated by A in Fig. 10.

CONCLUSIONS
The analysis of failurc conditions beneath wheels by the theory of plasticity allows
the computation of interface stresses for various soil conditions and wheel geometries.
These stresses agree reasonably well with those determined experimentally, proving
60 LESLIE L. KARAFIATH

FIG. 13. Slip line fields for driven wheel on level ground, lunar gravity.

the validity of the theory. The interrelationships a m o n g soil strength parameters,


developed traction, load, sinkage, slope, a n d gravity can be analyzed by c o m p u t e r
techniques developed for the solution of the basic differential equations of plasticity.

NOTATION
c cohesion 6 angle of inclination of
b d r a w b a r pull resultant stress to n o r m a l
W load -c slope angle
x,z coordinates ~ friction angle
ct central angle 0 angle enclosed by m a j o r
cte entry angle principal stress and x axis
~m angle of separation ~ 45 ° - - q~/2
c~8 rear angle ~r n o r m a l stress
7 unit weight of soil ~1,2,~ principal stresses
T shear stress

REFERENCES
[1] D. SCHtrRIN~. The energy loss of a wheel. Proc. 2nd Int. Conf. Int. Soc. for Terrain Vehicle
Systems, University of Toronto Press (1966).
[2] J. Jo-Y't~NG-WoN~and A. R. REECE. Prediction of rigid wheel performance, Part I. J. Terra-
mechanics, 4 (1) (1967).
[3] L. M. KRAFT. A Quantitative Approach to the Pneumatic Tire-Soft Soil System. U.S. Army
Eng. Waterways Experiment Station M.P. No. 3-944, Vicksburg, Miss. (1967).
[4] J. Jo-Yt~N~-WONcand A. R. REECE. Soil failure beneath rigid wheels. Proc. 2nd lnt. Conf. lnt.
Soc. for Terrain Vehicle Systems, University of Toronto Press (1966).
[5] K. W. WIENDIECK.Contribution to the mechanics of rigid wheels on sand. U.S. Army Eng.
Waterways Experiment Station, Tech. Report M-68-2, May (1968).
[6] D . R . P . HETTIARATCHIand A. R. REECE. Symmetrical three-dimensional soil failure. J. Terra-
mechanics, 4 (3) (1967).
[7] E.T. VJNCENT.Pressure distribution on and flow of sand past a rigid wheel. Proc. 1st lnt. Conf.
Mechanics of Soil- Vehicle Systems, Torino ( 1961).
[8] M.E. HARR. Foundations of Theoretical Soil Mechanics, McGraw-Hill, New York (1966).
[9] V.V. SO~OLOVSKII.Statics of Granular Media, Pergamon Press, Oxford (1965).
[10] C. SZYMANSKLSome plane problems of the theory of limitingequilibrium of loose and cohesive,
non-homogeneous isotropic media in the case of a non-linear limit curve. In Non-Homogeneity
in Elasticity and Plasticity (ed. by W. Olszak), Pergamon Press, Oxford (1958).
[11] A.D. SELA. The Shear to Normal Stress Relationship between a Rigid Wheel and Dry Sand.
U.S. ATAC Land Locomotion Laboratory Report, June (1964).
[12] O. ONAFEKOand A. R.. REECE. Soil stresses and deformations beneath rigid wheels. J. Terra-
mechanics, 4 (1) (1967).
[I 3] L.L. KARAFIA33-Iand E. A. NOWATZKI.Stability of slopes loaded over a finite area. Highway
Research Board Record No. 323, November (1970).

You might also like