Professional Documents
Culture Documents
Stability and Control of Two-Wheel Drive Tractors and Machinery On Sloping Ground
Stability and Control of Two-Wheel Drive Tractors and Machinery On Sloping Ground
This paper examines conditions under which the overturning stability and directional control
of two-wheeled drive tractors with mounted or towed implements are lost on sloping ground.
Theoretical methods are derived for predicting these conditions. Confirmation of the pre-
dictions by experiments on full size machines and models was obtained.
The theoretical methods are used to examine the suitability of some frequently used tractor/
implement combinations for use on sloping ground.
It is concluded that loss of wheel/ground adhesion, rather than the likelihood of overturning,
limits the slope on which most combinations can safely operate.
1. Introduction
The stability of a tractor against overturning on sloping ground has received much attention.
Overturning does occur on such land but it is unlikely that lack of inherent stability is the primary
cause of accidents.
It is the author’s contention that accidents on sloping ground are likely to originate from loss of
control of the tractor/implement combination followed by an increase in speed. On sloping
ground rear wheel adhesion is likely to be lost, especially during descent, and, with retention of
front wheel steering control, a dynamically unstable situation arises. If the machine slews an
overturn due to centrifugal force can occur due to a high rate of turn.’
This contention that overturning stems in the first instance from control loss is based on an
examination of the stability and control of machines operating on sloping ground. The exam-
ination, described here, was carried out by the construction of polar diagrams.
Polar diagrams were used by Reichmann and Daskalov to show combinations of slope angle,
p, and heading angle, a (Fig. 1) at which instability sets in. *s3v* The diagrams have hitherto been
confined to vehicles not acted on by external loads and to stability loss, which occurs when the
normal to ground reaction of one of the machine’s wheels becomes zero. This paper extends the
diagrams to cover instability resulting from external loading and also to take account of wheel/
ground adhesion being inadequate to withstand braking, side or traction forces. Inadequacy
results in the driver being unable to drive the tractor on an intended path and is referred to as a
control loss. The problems examined are restricted to those arising in steady state straight line
motion.
Control loss is dependent on the ground conditions. In the absence of sufficient data to provide
the general traction/side force/braking characteristic of tyres the work described here is based on
friction concepts. This approach is not unduly restrictive since serious control loss problems occur
with hard ground conditions for which friction concepts apply.5 The mathematical model
describing the control loss boundary, i.e. the critical combinations of slope and heading angles,
can easily be extended to cover soft ground conditions once wheel behaviour under these con-
ditions can be described. One feature of the polar diagram depicting the stability and control loss
boundaries of a machine is that, in the limit of infinite coefficient of friction, the two are identical-
This is readily seen if one considers that under these conditions control will only be lost when the
wheel normal to ground load becomes zero, which corresponds to the stability limit.
Extensive use is made of vector algebra to determine expressions for the forces acting on the
wheels.* For a vehicle, this technique allows simple formulation of the equilibrium equations for
‘ScottishInstitute of Agricultural Engineering, Bush Estate, Penicuik, Midlothian, Scotland
Received 16 September 1977; accepted in revised form 8 January 1978
169
OOZI-8634/78/0601~169 $02.00/O Q 1978 The British Society for Research in Agricultural Engineering
170 STABILITY AND CONTROL ON SLOPING GROUND
any geometric configuration, such as a conventional two-wheel drive tractor with a pivoted front
axle beam. The equations when solved for the wheel forces, from which the prediction of stability
and control loss are made, give rise to lengthy expressions which are evaluated by a computer
program.
Theoretical predictions of control loss are confirmed by limited tests on two full size tractors and
by model experiments on a variety of tractor configurations. The agreement is good enough to
warrant use of the techniques for examining circumstances under which control of some com-
monly used tractor/implement combinations is lost.
3. Experimental results
Experimental confirmation of predicted safety limits on machines poses investigators with
problems. Tests on full scale machines to determine limits of safe performance on sloping ground
H. B. SPENCER 171
KEY TO SYMBOLS
A
B
denotes tractor wheels
C
D
BT denotes trailer wheels
CT i
Qr
br vector position of wheel/ground contact points of tractor, m
Cl
d, I
vector position of wheel/ground contact points of trailer, m
:;: I-
Pt vector position of drawbar hitch point on tractor, m
Ptl vector position of drawbar hitch point on trailer, m
Ai
Br
vector component of tractor wheel forces acting at wheel/ground contact points, kg
Ci 1
D, J
CT, vector component of trailer wheel forces acting at wheel/ground contact points, kg
Pk vector component of drawbar hitch load, kg
component of unit vector describing the hinge axis of the front axle, m
;: component of a position vector defining a point on the front axle hinge axis, m
component of the vector giving the position of the centre of gravity of the front axle, m
component of the vector giving the position of the centre of gravity of tractor excluding
front axle, m
Sl component of the vector giving the position of the centre of gravity of the complete
tractor, m
heading angle, degrees
; slope of land (% slope = 100 x tan /?), %
P coefficient of sliding friction
& permutation symbol
i,i, k subscripts taking values 1, 2, 3
G, vector component of gravity force for tractor, kg
CT, vector component of gravity force for trailer, kg
have a high probability of severe accidents occurring involving irreparable damage to expensive
machinery or endangering life and limb of operators. Previous investigators have used only one
full scale machine in their trials.6 Some investigators have restricted testing to small scale models
in order to obtain confirmation of their mathematical models.‘* * The approach used here is to
use limited full size trials supplemented by a wide range of tests with small scale models.
k7;
Troller co-ordinate system
Fig. 2. Co-ordinate system for conventional tractor, with hinged front axle, and trailer
for full size machines. Full size manually controlled experiments to determine control loss can
only be conducted in safety for the direct descent case. The requirements are that the slope should
not be very long and that a long flat run out area be available at the bottom.
Trials were made with two different medium horse power tractors, X and Y, to determine the
gradient on which control was lost during direct descent of hard grass slopes. One of the tractors
was in the unballasted condition (X). The other (Y) was abnormally ballasted so that control was
lost on a relatively moderate slope. Dimensions of tractors X and Y used in these trials are given
in Appendix B. The tractors were driven down successively steeper slopes until control was lost.
Engine braking, not wheel braking, was used and when wheel/ground adhesion was insufficient
for equilibrium the tractor slid bodily downhill with the drive wheels rotating. The suddenness
of the loss of control was a feature of these failures. The driver reported that no warning signs
were available, such as noticeable wheel slip. This was confirmed from films taken during the
trials.
It would therefore appear that under hard ground conditions the increase in braking wheelslip
attributable to increase in slope under constant engine braking is so small as to be not noticeable
by the driver until wheel ground adhesion is insufficient for equilibrium and control is lost. This
observation, no noticeable slip, is confirmed by results obtained from testing automobile tyres on
hard surfaces.* In the idealized case of a pure friction model with a rigid wheel on hard ground
there is no slip at the wheel/ground contact point until the friction force necessary for equilibrium
exceeds the maximum available. The dangers inherent in this wheel behaviour are obvious.
H. B. SPENCER 173
Fig. 3. Theoretical control and stability loss diagrams for tractor X (lef) (- , control loss: - --
stability) and tractor Y (right) (- - - - -, control loss; - - -, stability). Experimentally determined control,
loss points at a = 180”; ?? , tractor X; x , tractor Y
line PP’. Further details and examples of the stability of freely rolling bodies are given by
Reichman. Their stability limits are usually of only academic interest as uncontrollable condi-
tions arise before these limits are reached. This can be seen in Fig. 3 where the control loss
boundaries, for tractors X and Y, lie well within the stability boundaries. In general, control loss
boundaries are not straight lines. The points on the control loss boundary represent conditions
for which controlled straight line motion is not possible and a condition can be any one of those
given in Appendix A.
The predicted control loss boundaries for the two tractors agree reasonably well with the
experimentally determined points at a = 180”, the direct descent direction. The values of
p (0.725) and R (0.05 x tractor weight) used in the predictions were measured by considering the
equilibrium of the tractors when towed over the ground in a locked and unlocked wheel condition.
One feature these diagrams reveal is the ability of the tractors to make controlled ascents of
slopes they cannot safely descend.
aluminium surface which could be angled to the horizontal plane. The model tyres were manu-
factured from a silicone rubber which gave them a friction coefficient of the order of 0.7 on the
table. The model was about & full size but no attempt was made to represent any particular
tractor to scale although centre of gravity locations relative to ground and rear axle were geo-
metrically similar to those on full size machines. The prime purpose was to see how well the
theory of tractor behaviour could predict model behaviour.
Experiments on the tilting table were carried out by placing the model at successive heading
angles a and at each steadily increasing the slope of the table surface /I until sliding or overturning
occurred. The friction coefficient was determined by measuring /? for a = 180” with all wheels
locked, since p = tan j? at onset of sliding. The value of R was determined by towing the model
with the surface level.
Fig. 5. Comparison of experimentally determined control loss conditions and theoretical predictions for the model
tractor with same dimensions but (left) high centre of gravity and (right) lower centre of gravity. 0, x , experimental
observations; -, theoretical control loss boundaries: ~ - -, theoretical stability boundaries
176 STABILITY AND CONTROL ON SLOPING GROUND
Fig. 6. Comparisons of experimentally determined control loss conditions and theoretical predictions for the model
tractor having a forward centre of gravity position but (left) with a wide wheel track and (right) a narrow wheel track.
0, X, experimental observations: -, theoretical control loss boundaries; - - -, theoretical stability
boundaries
weights above the centre of gravity position. Both configurations had rearward centre of gravity
positions. It is typical, for these configurations, that the control loss and stability boundaries
coincide at a = 0”. On this heading the front wheel loads become zero on theoretical slopes of
58 % and 68 % respectively.
Fig. 6 shows experimentally determined control loss boundaries for the two-wheel drive
configuration model having a forward position of centre of gravity but with two different track
widths. The left-hand part of the diagram is for a configuration with a wide track and the right-
hand part a narrow track, otherwise dimensions are the same. With the wide track configuration
no stability problems occur at a = go”, as evidenced by the slope on which sliding occurs, i.e.
H. B. SPENCER 177
/l = 68 %, and stability loss occurs, i.e. p = 93 %. The configuration with the narrow track loses
stability at a = 90” as evidenced by the coincidence of the control loss and stability boundaries
where /3 = 50 %. For both configurations the minimum safe operating slope occurs for a = 180”.
The slight difference between the control loss boundaries for a = 0” and 180”is due to differences
in the measured value of ,u at the times the experiments were conducted. The safe descent slope,
for these models is of the order of 25 %, whereas they have the ability to climb slopes of 42-45 %.
Fig. 7 shows control loss boundaries determined experimentally for the model two-wheel drive
tractor with trailer. The tractor configuration was the same as for Fig. 5, right-hand part. The
trailer had the same weight and dimensions for each experiment but the left-hand part of Fig. 7
is for a forward position of the trailer centre of gravity and the right-hand part for a rearward
Fig. 7. Comparison of experimentally determined control loss conditions and theoretical predictions for the model
tractor and trailer. Tractor dimensions as for Fig. 5 (right). Trailer with forward centre of gravity position (left) and
rear centre of gravity position (right). ??
, x , experimental observations: -, theoretical control loss boundaries;
--- , theoretical stability boundaries
position. As can be seen from the figure the effect of the forward centre of gravity of the trailer
is to transfer load to the tractor draw bar causing a stability loss at a = O”, occasioned by both
tractor front wheels leaving the ground. The rearward movement of the trailer centre of gravity
removes the stability loss at a = 0” but it decreases the safe descent slope of the combination, at
a = 180”, from /I = 32 % to j? = 24 %, as shown in Fig. 7, right-hand part.
in pitch or roll is as likely to happen as sliding the instability of the model was observable.
Control and stability boundaries calculations for the full size tractors (Fig. 3) have the same
forms as those established for the model (Figs 5, 6 and 7) and, in addition, the single observed
value available for each tractor agrees with that predicted. The calculating methods used to
obtain the boundaries are considered to be validated to an extent sufficient to allow them to be
used to study, at least descriptively, the effects of slope on tractor/implement combinations.
Four cases are examined, typical of tractor/implement combinations with which incidents have
been known to occur. The cases examined are a two-wheel drive tractor with
(1) a flat roller,
(2) a rear mounted bale handler,
(3) a fully mounted fertilizer spreader, and
(4) a single axle silage trailer.
In each of the cases examined ~1was taken to be 0.7 and R equal to O-05x tractor weight.
A comparison is also made between a two- and a four-wheel drive tractor.
Fig. 8. Control loss diagrams for tractors with rollers. Left, tractor X: - - -, tractor alone; -
tractor with 3 ton roller and no ballast. Right, tractor Y: - - - (I), tractor alone; - - - - -, tractor witi
4 ton roller and no ballast; - - - (2), tractor alone with rear wheel ballast; -, tractor with 4 ton roller
and rear wheel ballast
H. B. SPENCER 179
reduction in a safe descent slope is due to a roller not increasing tractor rear wheel normal to
ground reaction by load transfer to the tractor draw bar, whilst increasing the overall mass to be
braked during descent. The lack of load transfer suggests that rear wheel ballast is required to
improve the safe descent slope. The effect of adding rear wheel ballast is shown in Fig. 8 (right-
hand part). On tractor Y alone it improves the safe descent slope from 39 % to 50% and on
tractor Y with the 4 ton roller 17 % to 29 %. A roller also reduces the ascent slope which can be
climbed but this effect is less of a safety hazard than the descent cases, in which a dangerous
increase of speed occurs. The magnitudes of the reduction in an ascent case, a = 0”, are easily
read off Fig. 8. At heading angles other than a = 0” or 180” the situation improves and steeper
slopes can be negotiated. For safety, however, operation should be restricted to slopes less than
the minimum safe operating slope.
Fig. 9. Tractor Y with fully mounted bale handler. Left, control loss diagram for tractor alone; right, control loss
diagram with front end ballast and loaded with 42 bales (- - ---), control loss diagram for 21 bale load
(- - - - - -) and stability boundary where different from control loss boundary (-)
180 STABILITY AND CONTROL ON SLOPING GROUND
a= 9o” : 90”
Fig. 10. Tractor Y with fully mounted fertilizer spreader. L.ef, stability boundaries; right, control loss boundaries;
- - -, fully laden spreader: -, empty spreader
Fig. 11. Tractor with 70% full siiage trailer. --“- -, stability boundary of tractor with trailer:
---- .-- , stability boundary of trailer; -, control loss boundary of tractor with trailer
Fig. 12. Experimental demonstration of dtyerence be#ween two-wheel drive andfour-wheel drive configurations. Left:
0, four-wheel drive tractor alone; IJ, tractor with trailer. Right: x , two-wheel drive tractor alone; + , tractor with trailer
182 STABILITY AND CONTROL ON SLOPING GROUND
5. Conclusions
A tractor will lose stability and start to overturn when the normal to ground component of a
wheel load becomes zero. The driver will also lose control when the wheel/ground adhesion is
insufficient for equilibrium, in the ground plane, to be maintained. The combinations of slope
and direction of travel which result in these occurrences can be calculated by vector mechanics
and the boundaries of safe operation represented by polar diagrams. The validity of the method,
applied to a two-wheeled tractor, was checked, in the first instance, by reference to downhill
slopes on which rear wheel adhesion was insufficient for equilibrium for two different tractors.
A more extensive comparison was made by calculating the complete control loss boundaries for
6 different model tractor configurations and comparing these with boundaries obtained by
experiment. Operation on hard ground, to which friction concepts apply, was assumed through-
out. Reasonable agreement between predicted and observed results was obtained.
The methods were used to examine the behaviour on slopes of four widely used tractor and
implement combinations. For all the cases examined a control loss limited the slope on which
operation was safe, except for the case of the 42 bale handler which so grossly overloaded the
tractor that loss of stability occurred on very low slopes.
Operation on slopes steeper than the minimum safe operating slope is possible by travel along
the contour, e.g. a = go”, 270”. For this direction of travel stability loss may occur before
control loss particularly with equipment having a high centre of gravity, e.g. a nearly full silage
trailer. Turning from the direction a = 90” or 270” to a downhill direction can be a dangerous
manoeuvre. In general the safe descent slope decreases between a = 90”, 270” and a = 180
with the minimum value occurring at a = 180”. This reduction can be very large for some tractor
implement combinations of which a tractor with a heavy roller is a common example.
The stability boundaries computed take no account of rough ground so it is to be expected that
the actual boundaries will occur at smaller values of /I than are indicated in the diagrams.
Other risks identified with a two-wheel drive tractor are:
(a) the ability to climb slopes it cannot safely descend, and
(b) the decrease in the safe descent slope with a fully mounted fertilizer spreader as emptying
proceeds.
Limited experiments with a model four-wheel drive tractor demonstrated their capability of
operating on steeper slopes than those of comparable two-wheel drive machines.
H‘. B. SPENCER 183
REFERENCES
Appendix A
Methods of determining wheel loads
A. 1. Wheel loads on tractor with no external imposed loads
The method of determining the wheel loads on a two-wheel drive tractor not acted upon by
external loads (Fig. 2) is given by Reichmann .2 His expressions for the wheel loads are the basis
of the work reported and are tabulated here for reference.
The notation for wheel loads and the tractor geometry used in this paper is identical to that
used by Reichmann. The co-ordinates are a body fixed rectangular Cartesian system with the
three axes denoted by the numbers 1, 2 and 3 (Fig. 2). Forces are denoted by capital letters, i.e.
the vector force on wheel A is denoted by Al where the three components of the force are AI, A,,
A,. Position vectors are denoted by lower case letters, i.e. the position of the wheel-ground
contact point for wheel A is a, (i = 1, 2, 3).
1
‘3 = (d2_dz) Et,9el (d,_cj) {[‘%ikel Wr-~,) bkmak)
Assuming that there is no rolling resistance and that the differential lock is not engaged:
A,=B,=O,Cz=D,= -$G,.
On the assumption that Al, B,, Cl and D1 are directly proportional to the normal loads on the
wheels, then
B = (Al+Bl)B
’ (A,+B,) ”
C = (C,+Ddc D = (CI+DI)D
’ (C*+D,) ” ’ (C,+ D3) ”
To determine the effect of a trailer or implement on the stability and control of the tractor the
additional wheel loads due to a generalized loading Pk of the tractor at a point pk needs to be
derived.
G1=O,&=O,;il=O.
The equation of force equilibrium is
Ai+B1+Ci+D1+Pt = 0.
Moment equilibrium is expressed by
a&,-‘& + si,kbj& + ‘%,k@k + QkdJDk + EtJkhPk = 0.
Solution of the above equations for the wheel loads due to PI, follows that of Reichmann.a
After manipulations solutions for the wheel loads are
H. B. SPENCER 185
C, = Dz = -0.5 Pz, A, = B, = 0.
C, = D, as a consequence of the action of the differential unit.
(A,+BJ =
(d,!a,)
-
1P,(P,--d,)+P, (qy -P2Pl},
A, = (A + 4) A B = (Al+B,lB
(A,+B,) 3’ ’ (A3+B3) 3’
(Cl+ D3 =
2
&-a3),
&) ($,+d,)-PA+&
D, = e2 (c:_dl) ieel b2-l2) p3-p2 (P3-l3)1
-(C,+DJ [e2(c,-~3)-e3(c2-~2)l,
c3 = -P,-A3-B,-D,,
c = (c,+DJc D = (c,+WD
’ (C,+D,) 3’ ’ (C,fD,) 3’
To determine the total wheel forces the above forces due to P, are simply added to those for the
tractor without external imposed loading.
where p1 is the effective friction coefficient in direction 1. The coefficient is evaluated using
friction circle c0ncepts.l Sideways slide of a trailer occurs when
(BT,+CT,)3p(BT,+CT,).
Uncontrolled down slope acceleration occurs when wheel ground adhesion is insufficient at
one of the rear wheels for maintenance of equilibrium in the co-ordinate direction 2 during an
engine braked descent. Account must be taken of the rolling resistance, R, for this control loss
situation. This resistance represents losses occurring during tyre rolling and motion of the
differential unit during such control loss situations. Fig. 13 shows diagrammatically the situation
occurring during such a control loss, with down slope acceleration, under engine braking. Due
to the action of the differential, the wheel on which the adhesion is insufficient starts to rotate
backwards. If D is the rolling wheel and C the wheel on which there is insufficient adhesion then,
equating work done,
G,dx = Rdx+ 2puadxC,.
Thus the criteria for control loss on wheel C is
G&R+2p&
H. B. SPENCER 187
Fig. 14. Schematic diagram of tractor showing disposition of ballast and mounted implements
Appendix B
Details of tractor, implements and ballast used in the stability and control loss examples
A schematic diagram of a two-wheel drive tractor indicating the disposition of loads is shown
in Fig. Z4. Dimensions and values of loads are given in Table 1.
W = weight of body,
D = distance of centre of gravity of body in horizontal plane from rear axle centre line, and
H = height of centre of gravity of body from ground measured normal to ground plane.
Subscripts refer to body, e.g. W, = weight of body 1.
Subscript dejinitions
1 = tractor alone without ballast,
2 = mounted implement, unloaded e.g. bale handler unloaded or for trailed implement centre
of gravity position aft of hitch point in a horizontal plane,
3 = mounted implement, loaded e.g. bale handler loaded or for trailed implements position of
wheels or rollers aft of hitch point,
4 = ballast on front wheels,
5 = ballast mounted on frame at front of tractor, and
6 = rear wheel ballast consisting of weights and water.
TABLE I
-
WI Dl HI W, Wa De 08 HZ W4 D4 DS Ha W, HB
--
Tractor X* 2245 0.722 0.773 1.925
Tractor Yt 2425 0.739 0.841 195 2.083 (MO6 255 2.083 1.052 2677 0.840
Balehandler (21 bales)* 152.5 572 0.965 1.219 1.613 2.133
Balehandler (42 bales)* 305 1144 0.94 1.702 1.613 2.133
Tractor Y !j 3400 1.346 0.840 2.083
Fertilizer spreader (empty) 136 1.226 0.72
Fertilizer spreader (full) 543 1.226 1.177
3 ton roller 3000 2.0 20 0.38
4 ton roller 4000 2.0 2.0 0.38
Silage trailer 6245 3.9 4.7 1641
- -
*Hitchpointtakento be 0.641 m behind rear axle and 0.387m above ground
tHitch point taken to be 0.775 m behind rear axle and 0464 m above ground
tBaIeweiahtsof 27.2 kg (60 lb): from N.I.A.E. Test Report No. RT 445(1965)
$Abnormally ballasted to produce a forward centre of gravity position
All dimensions in m, and weights in kg