Vehicle System Dynamics

Vol. 44, No. 6, June 2006, 431447

Critical review of handling diagram and understeer gradient

for vehicles with locked differential
FRANCESCO FRENDO, GIORDANO GRECO and MASSIMO GUIGGIANI*
Dipartimento di Ingegneria Meccanica, Nucleare e della Produzione,
University of Pisa, via Diotisalvi 2, 56126 Pisa, Italy

The steady-state cornering behaviour of rear-wheel drive vehicles fitted with locked differential is
critically analysed by means of simple, albeit carefully formulated, vehicle models, which allow for
a rigorous theoretical analysis. Results obtained for some classical manoeuvres, with either constant
forward speed, steer angle or turning radius, clearly show that, in the case of locked differential, the
vehicle cornering behaviour is strongly affected by the manoeuvre. As an important consequence,
the handling diagram is not unique and the understeer gradient is no longer dependent only upon the
lateral acceleration, as in vehicles equipped with an open differential. Accordingly, this study shows
that some typical tools and concepts of vehicle dynamics are indeed inadequate in the case of locked
differential.

Keywords: Cornering behaviour; Handling diagram; Locked differential; Yaw moment; Understeer
gradient

1. Introduction

The single-track model theory is typically employed for the analysis of vehicle dynamics.
Even though such a theory is based on a very simple vehicle model, it has been largely used
to understand and foresee the dynamic behaviour of real vehicles [15].
In the classical single-track model, a rear-wheel drive vehicle fitted with an open differential
and moving at constant longitudinal speed is considered. Owing to the open differential,
traction forces acting on the rear tyres are always mutually equal. Moreover, owing to some
strong assumptions, the slip angles of the tyres belonging to the same axle are assumed equal
and longitudinal slips are neglected. According to these hypotheses, a relationship between
the whole side force and the slip angle of each axle can be found, leading to a model fitted
with only two equivalent tyres. One of the most relevant results of this approach is that a single
handling curve [1], only depending upon the constructive parameters, describes completely the
steady-state cornering behaviour of the vehicle. The handling diagram is also often employed
for the experimental assessment of vehicle dynamics [3, 6].

*Corresponding author. Email: guiggiani@ing.unipi.it; Tel.: +39 050 836 618; Fax: +39 050 836 665

Vehicle System Dynamics

ISSN 0042-3114 print/ISSN 1744-5159 online 2006 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/09638230500348184
432 F. Frendo et al.

However, it is, perhaps too often, not sufficiently appreciated that some of its properties
heavily rely on the hypotheses behind the classical single-track model. Indeed, it will be
shown that, although this model can be properly used to characterize the cornering behaviour
of vehicles endowed with open differential, it becomes completely inadequate for vehicles
fitted with locked differential. In fact, when a vehicle with locked differential negotiates a
corner, the driving wheels are forced to rotate at the same angular velocity and longitudinal
slips become not negligible, considerably influencing vehicle dynamics. In order to analyse
these effects, it is necessary to formulate a four-wheel model and take into account both the
lateral and longitudinal slips of each tyre.
The effect of locked differentials and limited-slip differentials, both conventional self-
locking and actively controlled, on vehicle dynamics has been largely described by many
authors [715]. The influence of the distribution of longitudinal tyre forces, caused by such
devices, on traction and braking capabilities and directional behaviour has been investigated
by means of both numerical simulations and experimental ground tests. Moreover, Abe [7]
described by means of an approximate analytical method the effect of locked differentials on
the cornering behaviour in acceleration and in braking. Results were compared with those
obtained by means of numerical simulations, showing a good agreement.
Nowadays, electronically controlled limited-slip differentials, as well as steering and
traction control systems, may lead to substantial improvements in performance, stability and
manoeuvrability at corners [11, 12, 1421].
However, in order to deeply understand the effect of these devices on vehicle dynamics, the
development of a reference theory seems to be necessary. Such a theory, which appears not
to have been sufficiently developed yet, should have the same role the classical single-track
model theory has with respect to vehicles fitted with open differential; that means it should be
the analytical basis for the comprehension of vehicle dynamics when a locked or limited-slip
differential is employed. According to a well-established tradition in vehicle dynamics studies,
it seems profitable to continue to characterize the cornering behaviour of vehicles by means of
some typical tools and concepts, such as the handling diagram and the understeer gradient,
which are largely used in the single-track model theory, even though, as it will be shown, a
critical review of both is necessary.
In order to have a clear and general picture of the problems involved, it may be convenient
to develop simple vehicle models having few degrees of freedom but being able to represent
the most important characteristics of the motion all the same. In this paper, we deal with
vehicles fitted with a locked differential, which represents a very particular operation mode
of a limited-slip differential, but is also the easiest condition to deal with.
An approximate theoretical analysis of the effect of a nonfree differential on vehicle dynam-
ics is given in Dixon [2]; however, in that work the effects of combined slip operation of tyres
are not considered and the vehicle steady-state behaviour is not investigated in detail. An
experimental investigation of the dynamic behaviour of a rear-wheel drive vehicle with locked
differential was given in Frendo et al. [22], while preliminary theoretical models for analysing
the dynamics of such a vehicle were presented in Amato et al. [23, 24]. The effects on the
handling diagram of the yaw moment, arising from the difference between the rear longitudi-
nal forces, were investigated in those papers; in addition, the handling diagram was shown to
depend on the motion parameters, and not only on the vehicle features.
In the present paper, a general theoretical analysis of the steady-state cornering behaviour of
rear-wheel drive vehicles fitted with locked differential is proposed. A detailed description of
tyre kinematics under combined slip conditions is first developed. Two vehicle models are then
defined, having linear and nonlinear tyre properties, respectively. Such models are deliberately
simple. However, they allow us to catch the basic aspects of the dynamics of vehicles fitted
with locked differential and to perform a theoretical and systematic approach to the topic.
 Critical review of handling diagram 433

The steady-state cornering behaviour is then investigated and the handling diagrams are
obtained in several manoeuvres. For the model with linear tyres, analytical expressions of
the understeer gradient are also obtained. In contrast to the classical definition, the handling
diagram is shown not to be unique.
Results clearly show the inadequacy of the classical single-track model to characterize
the cornering behaviour of vehicles with locked differential. It also suggests a more critical
and careful use of some typical tools and concepts, such as the handling diagram or the
understeer gradient. Moreover, the approach proposed here may represent the basis of a new
reference theory to be added to the classical single-track model theory for the analysis of
vehicle dynamics when a locked differential is employed.

2. Vehicle model

The plane motion of a rear-wheel drive vehicle fitted with a locked differential is considered.
The model has three degrees of freedom (figure 1): the longitudinal speed u, the lateral speed
v and the yaw rate r. The parameters a1 and a2 indicate the longitudinal distance between the
centre of mass and each axle, and t1 and t2 indicate the front and rear track. In figure 1, the
indexes ij identify the four wheels of the vehicle: the index i is related to the axle (1, front; 2,
rear) and the index j is related to the side (1, left; 2, right).
The steer angle of the front wheels is indicated by . For simplicity, it is assumed that
the front wheels have the same steer angle, whereas there is no steering in the rear wheels:
11 = 12 = , 21 = 22 = 0. In addition, small steer angles are assumed, i.e. 15 , thus
allowing for the following approximations: sin  , cos  1.
Longitudinal and lateral forces acting on each tyre are, respectively, named Fxij and Fyij ; Fa
is the longitudinal component of the aerodynamic drag force. Rolling resistance is neglected;
therefore, no longitudinal force acts on front tyres. The self-aligning torques on all tyres are
also neglected.

Figure 1. The vehicle model.

434 F. Frendo et al.

2.1 Equilibrium equations

Owing to all these hypotheses, the equilibrium equations of the vehicle are given by (figure 1)

m(u vr) = (Fy11 + Fy12 ) + (Fx21 + Fx22 ) Fa ,




m(v + ur) = (Fy11 + Fy12 ) + (Fy21 + Fy22 ),

t2 (1)

J r = (Fy11 + Fy12 )a1 (Fy21 + Fy22 )a2 + (Fx22 Fx21 )

2



t1
+ (Fy11 Fy12 ) ,
2

where m and J are the vehicle mass and yaw moment of inertia with respect to a vertical
axis passing through the centre of mass G. The aerodynamic drag force is Fa = SCx u2 /2,
where is the air density, S is the frontal area of the vehicle and Cx is the aerodynamic drag
coefficient.
As it is a common practice in classical vehicle dynamics [25], it is possible to neglect the
last term (Fy11 Fy12 )t1 /2. Therefore, the equilibrium equations (1) become

m(u vr) = Fx2 Fy1 Fa ,

m(v + ur) = Fy1 + Fy2 , (2)


J r = Fy1 a1 Fy2 a2 + Mz2 ,

where

Fx2 = Fx21 + Fx22 , Fy1 = Fy11 + Fy12 , Fy2 = Fy21 + Fy22 , (3)

and
t2
Mz2 = (Fx22 Fx21 ) . (4)
2

Having a locked differential requires the rear traction forces to be possibly different: Fx21  =
Fx22 . Therefore, also the yaw moment Mz2 appears in the yaw equilibrium equation. This is
the only, but quite relevant, difference between equations (2) and the equilibrium equations
of the classical single-track model [e.g., 1].
As shown by many authors [7, 8, 1115, 2224] and by real applications for both commercial
and sports cars, such a moment Mz2 strongly affects the cornering behaviour of vehicles. In
the following sections, this aspect will be analysed and discussed by means of a theoretical
and systematic study.

2.2 Congruence equations: tyre theoretical slips

In order to analyse the effects of the tyre longitudinal slips on the steady-state cornering
behaviour, a detailed theoretical description of the tyre kinematics in combined slip conditions
is necessary. Let xij and yij be the longitudinal and lateral component of the theoretical slip
ij [25] of the tyre identified by the indices ij . Assuming a planar motion of the vehicle and
small , these components are given as follows
 Critical review of handling diagram 435

Longitudinal theoretical slips

(u r(t1 /2)) + (v + ra1 ) 11 R1

x11 = ,
11 R1
(u + r(t1 /2)) + (v + ra1 ) 12 R1
x12 = ,
12 R1
(5)
(u r(t2 /2)) 21 R2
x21 = ,
21 R2
(u + r(t2 /2)) 22 R2
x22 = .
22 R2
Lateral theoretical slips

(u r(t1 /2)) + (v + ra1 )

y11 = ,
11 R1
(u + r(t1 /2)) + (v + ra1 )
y12 = ,
12 R1
(6)
v ra2
y21 = ,
21 R2
v ra2
y22 = .
22 R2
In the above equations, ij is the angular velocity of the generic tyre rim and R1 and R2 are
the front and rear tyre rolling radii, which are assumed to be constant, thus neglecting their
weak dependence on the vertical load.
It may be convenient to set
ij Ri
= 1 + ij , (7)
u
which defines the nondimensional quantities ij .
Considering that under normal driving conditions, the following relations hold

|r|ti |r|ai |v| |v rai |

|ij |  1,  1,  1,  1,  1, ||  1, (8)
u u u u
the longitudinal and lateral slips may be given approximately by
 
rt1 v + ra1
x11  11 + , y11  + = 1 ,
2u u
 
rt1 v + ra1
x12  12 , y12  + = 1 ,
2u u
  (9)
rt2 v ra2
x21  21 + , y21  = 2 ,
2u u
 
rt2 v ra2
x22  22 , y22  = 2 ,
2u u

where 1 and 2 are the slip angles of the front and rear axle (figure 1). Equations (9) are
linear relationships between the longitudinal and lateral slips of each tyre and v, r, and ij .
They also show that ij do not affect the linearized lateral slips.
436 F. Frendo et al.

For a rear-wheel drive vehicle with locked differential, the angular velocities of the rear
rims are the same (21 = 22 = 2 ) and hence, according to equation (7), 21 = 22 = .
Moreover, the front tyre longitudinal slips x11 and x12 are equal to zero, which means 12 =
11 = rt1 /(2u).

2.3 Constitutive equations: tyre model

We now have to relate the tangential forces acting on each tyre to the corresponding
theoretical slip; to keep the analysis as simple as possible, an isotropic tyre behaviour is
assumed [25].
Let Ft and Fz represent the total tangential force and the vertical load acting on the generic
tyre. Neglecting any camber effect, the constitutive equation can be written in the following
form

Ft = Ft (, Fz ), (10)

where Ft is the modulus of the total tangential force Ft , is the modulus of the theoretical
slip and Fz represents the vertical load variation with respect to the static load Fz0 . The
longitudinal and lateral force components are
x y
Fx = Ft (, Fz ), Fy = Ft (, Fz ), (11)

where x and y are the longitudinal and lateral slip components.
It is customary to employ the Magic Formula [26] for the definition of Ft (, Fz )
 
Ft (, Fz ) = Df sin Cf arctan(Bf Ef [Bf arctan(Bf )]) , (12)

where Cf and Ef are suitable constants, while, as usual

 
Df = Fz = (q1 Fz + q2 )Fz = q1 (Fz0 + Fz ) + q2 Fz0 + Fz , (13)
   
Fz F 0 + Fz
Bf Cf Df = q3 sin 2 arctan = q3 sin 2 arctan z (14)
q4 q4
do depend on Fz , with q1 < 0, q2 > 0, q3 > 0, q4 > 0. The first relation shows that the
friction coefficient decreases with respect to the vertical load Fz , and the second relation
shows that the tyre stiffness Bf Cf Df increases with respect to Fz , if Fz < q4 , and then lightly
decreases as soon as Fz > q4 .
If a nonlinear response of the tyre is considered, that is, whenever large slip values are
expected, the relationship (12) will be employed. However, if small tyre slips and small vertical
load variations are assumed, it is possible to take only a few terms in the power series expansion
of equation (12) around the point ( = 0, Fz = 0), obtaining the following expression

Ft Ft 2 Ft
Ft (, Fz ) = Ft (0, 0) + + F + Fz +
0,0 Fz 0,0 Fz 0,0
z

= 0 + C 0 + 0 + kFz + = (C 0 + kFz ) + (15)

In the above equation, C = Ft / |0,0 is the tyre stiffness at ( = 0, Fz = 0) and k =
0

2 Ft /( Fz )|0,0 . Note that, if camber is considered to be negligible, all the other second
derivatives in equation (15) are zero.
According to equation (15), the total tangential force Ft is assumed to be a linear function
of the theoretical slip but with the tyre stiffness linearly dependent on the vertical load
variation Fz .
 Critical review of handling diagram 437

2.4 Vertical load on each tyre

There is now the need to evaluate the vertical loads acting on the four wheels of the vehicle.
Under steady-state cornering conditions (i.e. u = v = r = 0), it is reasonable to assume that
the vertical load variation of each tyre is only due to the lateral load transfer.
For a vehicle with a locked differential, the steady-state lateral load transfer of each axle must
be related not only to the steady-state lateral acceleration ay = ur, as it is common practice
in classical vehicle dynamics, but also to the yaw moment Mz2 (an aspect apparently never
discussed before). In fact, by writing the equilibrium equations about the roll axis (figure 1)
for the sprung mass and for each axle and solving the last two equations in (2) for Fy1 and Fy2 ,
it is easy to obtain the following relations
M z2 Fx Fx21
Fz1 = mB1 ay d1 = mB1 ay 22 t2 d1 ,
lt1 2lt1
(16)
M z2 Fx Fx21
Fz2 = mB2 ay + d2 = mB2 ay + 22 d2 .
lt2 2l
where the constants B1 and B2 are given by [e.g., 5]
   
1 a2 k1 1 a1 k2
B1 = d1 + (h d) , B2 = d2 + (h d) , (17)
t1 l k t2 l k
with l = a1 + a2 . According to figure 1, in the above equations, h is the height of the centre
of mass and d1 , d2 and d are the heights of the roll axis measured, respectively, at the front
axle, the rear axle and the centre of mass. The constants k1 , k2 and k are, respectively, the
equivalent front axle, rear axle and vehicle roll stiffness (k = k1 + k2 ).
Note that some lateral load transfers may occur even at zero lateral acceleration, owing to
the presence of the yaw moment Mz2 . As shown in the following sections, this is exactly what
happens in manoeuvres with constant steer angle or with constant turning radius, in which the
yaw moment Mz2 is not zero even when the lateral acceleration ay tends to zero.
Finally, the vertical loads acting on the four tyres are

Fz11 = Fz011 Fz1 , Fz12 = Fz012 + Fz1 ,

(18)
Fz21 = Fz021 Fz2 , Fz22 = Fz022 + Fz2 .

3. Vehicle model with linear tyre behaviour

In this section, a vehicle model with linear tyre behaviour is presented and discussed. The model
is based on equilibrium equations (2), linearized slips (9), linearized constitutive equations
(15) and load transfers (16). Steady-state cornering conditions are considered to investigate
the effects of the locked differential on the handling diagram and on the understeer gradient.
Some relevant effects of several types of limited-slip differentials on the steady-state cor-
nering behaviour of vehicles have been already described in the technical literature [8, 1113,
15]. However, most authors have focused their attention on the influence of the constructive
features of the differential on the cornering behaviour, without emphasising the dependence
of the handling diagram and the understeer gradient on the manoeuvre itself.
In the following paragraphs, a more general and systematic theoretical analysis of such a
problem is proposed.
Results for a more realistic vehicle model with nonlinear tyres will be presented in
section 4.
438 F. Frendo et al.

3.1 Forces acting on the axles and yaw moment

According to equations (9), the lateral slips of the tyres belonging to the same axle can be
considered equal. Therefore, the effects of the lateral load transfers Fz1 and Fz2 cancel each
other and do not appear on the expressions of the axle lateral forces defined in equation (3)
 
v + a1 r
Fy1 = Fy11 + Fy12 = 2C10 = 2C10 1 ,
u
  (19)
v a2 r
Fy2 = Fy21 + Fy22 = 2C20 = 2C20 2 .
u

On the contrary, the rear tyres do not have the same longitudinal slips, and it is therefore
necessary to consider the effect of the rear lateral load transfer Fz2 on the axle (total)
longitudinal force Fx2 . Owing to the dependence of the load transfer Fz2 on the yaw moment
Mz2 , traction forces acting on the rear wheels can be obtained by solving the following algebraic
linear system


Fx21 = x21 (C20 k2 Fz2 )

  



rt2 d2

= + C2 k2 B2 may k2 (Fx22 Fx21 )
0
,
2u 2l
(20)

Fx22 = x22 (C20 + k2 Fz2 )



  

rt2 d2

= C2 + k2 B2 may + k2 (Fx22 Fx21 )
0
,
2u 2l

whose solution is

1 
Fx21 = C 0 d2 k2 r 2 t22 + 2C20 lt2 ur 2B2 k2 lmt2 u2 r 2
d 2 k2 ) 2
4u2 (l
  
+ 4C20 lu2 4B2 k2 lmru3 4C20 d2 k2 u2 ,
 (21)
1
Fx22 = 2 C 0 d2 k2 r 2 t22 2C20 lt2 ur 2B2 k2 lmt2 u2 r 2
4u (l d2 k2 ) 2
  
+ 4C20 lu2 + 4B2 k2 lmru3 4C20 d2 k2 u2 .

Hence, the rear longitudinal force Fx2 and the yaw moment Mz2 are given by

Fx2 = Fx21 + Fx22

C20 d2 k2 r 2 t22 2B2 k2 lmt2 u2 r 2 + (4C20 lu2 4C20 d2 k2 u2 )
= , (22)
2u2 (l d2 k2 )
t2 C20 lrt22 u + 2B2 k2 lmru3 t2
Mz2 = (Fx22 Fx21 ) = . (23)
2 2u2 (l d2 k2 )

Finally, let R = u/r = u2 /ay be the distance between the instantaneous centre of rotation
of the vehicle and the longitudinal vehicle axis. Note that in steady-state conditions R is also
the turning radius. Considering that under normal driving conditions | |  1 and |r|ti /u  1,
 Critical review of handling diagram 439

the following approximate expressions for the rear longitudinal force Fx2 and the yaw moment
Mz2 can be obtained

r C
Fx2  2C20 C ay = 2C20 ay , (24)
u R
 
A A
Mz 2  + C ay = + C ay , (25)
u2 R

where A = C20 t22 /2 and C = k2 B2 mt2 . It is worth noting that the yaw moment Mz2 not only
depends on the steady-state lateral acceleration ay but also on other variables defining the
vehicle motion, such as the longitudinal speed u, the variable and the turning radius R.
If the influence of the yaw moment Mz2 on the lateral load transfers Fz1 and Fz2 is
neglected in equations (16), we will find exactly the same approximate expressions shown in
equations (24) and (25) for the rear longitudinal force Fx2 and the yaw moment Mz2 itself.
Hence, for the present model, such a dependence has almost no effect on the vehicle dynamics.

3.2 Steady-state cornering behaviour

The steady-state equations of motion can be obtained by substituting expressions (19), (24)
and (25) in equations (2), with u = v = r = 0

 
r v + a1 r 1

mvr = 2C2 C ur 2C1
0 0
SCx u2 ,

u u 2

   
v + a1 r v a2 r
mur = 2C1
0
+ 2C2
0
, (26)

u u

     

v + a1 r v a2 r

0 = 2C1
0
a1 2C2
0
a2 +
A
+ C ur,
u u u2

which, in a more compact form, can be rewritten as

f (u, v, r, , ) = 0,
1
f2 (u, v, r, , ) = 0, (27)


f3 (u, v, r, , ) = 0.

According to equations (26) or (27), five quantities fully describe the vehicle motion. Two of
them need to be assigned, whereas the other three can be obtained by solving the algebraic
equations (26). In some cases, it may be convenient to use different variables, such as the
lateral acceleration ay or the turning radius R, by considering that ay = ur = u2 /R = r 2 R.
The first equation in (26) can be solved with respect to , obtaining the nonlinear relation

mvr + C(r/u)ur + 2C10 ( (v + a1 r)/u) + (1/2)SCx u2

= , (28)
2C20

that shows that is strongly affected by the vehicle motion conditions. Moreover, we see that
even with linear tyres, the overall vehicle model is nonlinear.
440 F. Frendo et al.

In order to obtain analytical expressions of the understeer gradient, the last two equations
in (26) can be rewritten in the following form

may = 2C10 1 + 2C20 2 ,
(29)
0 = 2C10 1 a1 2C20 2 a2 + Mz2 ,

which provide the slip angles 1 and 2

ma2 ay Mz2 ma2 (A/u2 + C )

1 = 0
= ay ,
2lC1 2lC10
(30)
ma1 ay + Mz2 ma1 + (A/u2 + C )
2 = 0
= ay .
2lC2 2lC20
Accordingly, we obtain the important result
 
l F
= 1 2 = K0 + G + 2 ay , (31)
R u
where G and F are constants given by

1 C10 + C20 1 C10 + C20

G= 0 0
C= k2 B2 mt2 ,
2l C1 C2 2l C10 C20
(32)
1 C10 + C20 1 C10 + C20 2
F = 0 0
A= t2 ,
2l C1 C2 4l C10

and K0 is the understeer gradient of the classical (i.e. linear, with open differential) single-track
model
m C20 a2 C10 a1
K0 = . (33)
2l C10 C20
For a compact rear-wheel drive passenger car, the following numerical values were found:
K0 = 0.00333 s2 /m = 1.87 /g, G = 0.0163 s2 /m and F = 0.519 m. For the same car,
under normal driving conditions, the term G is quite small.
Equation (31) shows that in a vehicle with a locked differential, the difference 1 2
depends not only on the lateral acceleration ay but also on other variables which characterize
the vehicle motion, such as the longitudinal speed u and the variable . This result has strong
practical relevance.
As shown in the following sections by means of some classical manoeuvres (with constant
forward speed, constant steer angle and constant turning radius), this aspect strongly affects
the handling diagram and the understeer gradient K, which now has to be defined as the partial
derivative  
l
K= = (1 2 ). (34)
ay R ay

3.2.1 Manoeuvres with constant forward speed. In these manoeuvres, the forward speed
u is assigned and kept constant. Some handling curves obtained in these manoeuvres for the
very same vehicle are shown in figure 2, and the plots of the variable (u, ay ) and the yaw
moment Mz2 (u, ay ) are shown in figure 3.
Quite remarkably, in figure 2, we no longer have a single handling curve, as predicted by
the classical theory for vehicles with open differential, but a different curve for each velocity.
 Critical review of handling diagram 441

Figure 2. Handling diagrams obtained with linear tyre behaviour in the manoeuvres with constant forward speed.

The corresponding understeer gradients Ku can be directly obtained from equations (31)
and (34) as
  
F
Ku = K0 + G (u, ay ) + 2 ay
ay u
 
F (u, ay )
= K0 + 2 + G (u, ay ) + ay , (35)
u ay
where the function (u, ay ) is given by the solution (which can be obtained employing some
software for symbolic computation) of equations (26) with respect to v, and , after having
set r = ay /u.
We see that Ku is strongly affected by u and, particularly at low speeds, it may be very
different from K0 . It is worth remarking that, with linear tyres, the classical theory predicts a
constant value K0 for the understeer gradient.

Figure 3. Variable and yaw moment Mz2 obtained with linear tyre behaviour in the manoeuvres with constant
forward speed.
442 F. Frendo et al.

Owing to the nonlinear dependence of on the forward speed u and on the lateral accelera-
tion ay , the understeer gradient does not decrease, for increasing u, exactly with the square of
the forward speed, although, in practice, the term G[ (u, ay ) + ( (u, ay )/ ay )ay ] is quite
small.
According to equation (31), the difference between the slip angles goes to zero if the lateral
acceleration goes to zero
 
F
lim (1 2 ) = lim K0 + G (u, ay ) + ay = 0. (36)
ay 0 ay 0 u2

As shown in figure 3, the variable (u, ay ) is always less than 0.012, confirming the assump-
tion | |  1. Note that the curves of do not pass through the origin if the forward speed is
positive, even if the lateral acceleration goes to zero (which requires 0).

3.2.2 Manoeuvres with constant steer angle. The steer angle is assigned and kept
constant. The handling diagrams and the plots of the yaw moment Mz2 (, ay ) are shown in
figure 4. Also, in this case, the handling curve is not unique.
After having set r = ay /u, equations (26) can be symbolically solved for the variables
u(, ay ), v(, ay ) and (, ay ). In order to analytically express the understeer gradients K , it
is convenient to introduce the relation
1 (K0 + G )ay ay
= = 2, (37)
R F +l u
and, therefore, rewrite the term 1 2 in equation (31) in the following form

l K0 + G (, ay ) l ay + F
1 2 = = . (38)
R F +l
Therefore, the understeer gradients K in manoeuvres with constant become
 

l K0 + G (, ay ) + ( (, ay )/ ay )ay
K = = l, (39)
ay R F +l

which, again, are not equal to K0 .

Figure 4. Handling diagrams and yaw moment Mz2 obtained with linear tyre behaviour in the manoeuvres with
constant steer angle.
 Critical review of handling diagram 443

Figure 5. Handling diagrams and yaw moment Mz2 obtained with linear tyre behaviour in the manoeuvres with
constant turning radius.

At low levels of lateral acceleration, the term G[ (, ay ) + ( (, ay )/ ay )ay ] in equation

(39) is quite small and the handling curves can be approximately considered straight parallel
lines.
According to equation (38), the difference between the slip angles does not go to zero when
the lateral acceleration vanishes, but it linearly increases with
[K0 + G (, ay )]l ay + F F
lim (1 2 ) = lim = . (40)
ay 0 ay 0 F +l F +l

3.2.3 Manoeuvres with constant turning radius. Finally, the turning radius R can be
assigned and kept constant. The handling diagrams and the yaw moment Mz2 (R, ay ) obtained
are shown in figure 5.
Equations (26) can 
be symbolically
solved for the variables v(R, ay ), (R, ay ) and (R, ay )
after having set u = ay R and r = ay /R. It is convenient to rewrite 1 2 in equation
(31) in the following new form
l
F
1 2 = = K0 + G (R, ay ) ay + , (41)
R R
which provides the understeer gradients KR in manoeuvres with constant turning radius
   
l (R, ay )
KR = = K0 + G (R, ay ) + ay . (42)
ay R ay
The difference between the slip angles is not equal to zero for zero lateral acceleration
 

F F
lim (1 2 ) = lim K0 + G (R, ay ) ay + = . (43)
ay 0 ay 0 R R

4. Vehicle model with nonlinear tyre behaviour

To investigate what happens in more realistic situations, we have to consider the fully nonlinear
tyre behaviour. This vehicle model is based on equilibrium equations (2), on linearized slips (9),
like the former one, and on the nonlinear constitutive equations (12). Therefore, the model is
completely and neatly defined.
444 F. Frendo et al.

Figure 6. Handling diagrams obtained with linear and nonlinear tyre behaviour in the manoeuvres with constant
forward speed.

Owing to the complexity of constitutive equations (12), only a numerical solution is now
possible for the equations governing the vehicle dynamics. All the other parameters defining
the model are the same used for the model with linear tyre behaviour and are typical of a
compact rear-wheel drive passenger car.
The present model was, therefore, used to simulate some manoeuvres with constant forward
speed, constant steer angle and constant turning radius. The handling diagrams obtained are
shown in figures 6 and 7, where they are also compared with those previously presented for
the vehicle model with linear tyre behaviour.
The plots of the yaw moment Mz2 are presented in figures 8 and 9 and again compared with
those of the model with linear tyres.
As we can see, in the manoeuvres with constant forward speed the plots obtained with linear
and nonlinear tyre behaviour have the same asymptotic trend at small lateral accelerations.
On the contrary and quite interestingly, in the manoeuvres with constant steer angle and
constant turning radius, the asymptotic trends for ay close to zero are different. This phe-
nomenon is more evident if the assigned steer angle is high and, accordingly, the assigned

Figure 7. Handling diagrams obtained with linear and nonlinear tyre behaviour in the manoeuvres with constant
steer angle and constant turning radius.
 Critical review of handling diagram 445

Figure 8. Yaw moment Mz2 obtained with linear and nonlinear tyre behaviour in the manoeuvres with constant
forward speed.

turning radius is small. Therefore, we have obtained the important result that, in the case of
locked differential, linear tire models may be very inadequate even at very low levels of lateral
acceleration, a fact rarely acknowledged in the vehicle dynamics literature.
Actually, this phenomenon can be easily explained. In steady-state conditions, if the lateral
acceleration ay tends to zero, lateral forces acting on the front and rear axle are given by

Mz 2 Mz2
F y1 = , Fy2 = . (44)
l l
Moreover, according to the relation u = rR, the rear longitudinal slips can be rewritten in the
form x21 = [ + t2 /(2R)] and x22 = [ t2 /(2R)].
If the forward speed is kept constant, when the lateral acceleration is close to zero, the
steer angle and the yaw moment Mz2 go to zero, whereas the turning radius tends to infinity:
the trajectory of the car is therefore rectilinear. Owing to the rectilinear path of the car, the

Figure 9. Yaw moment Mz2 obtained with linear and nonlinear tyre behaviour in the manoeuvres with constant
steer angle and constant turning radius.
446 F. Frendo et al.

locked differential has no effect on vehicle dynamics. Lateral forces tend to zero and hence the
lateral slips yij go to zero. The rear longitudinal slips are also very small because the traction
forces only have to balance aerodynamic drag. Therefore, in these conditions, linear tyres are
adequate for the whole range of forward speeds of the vehicle (figures 6 and 8).
Vehicle conditions are completely different in the manoeuvres with constant steer angle and
constant turning radius. In fact, even if the lateral acceleration tends to zero, the turning radius
R is a finite quantity and hence the trajectory of the vehicle is not rectilinear. As shown in
figure 9, the yaw moment Mz2 is always negative (if > 0) at zero lateral acceleration and its
absolute value increases if the steer angle increases, that is, if the turning radius decreases.
The front lateral forces are positive, whereas the rear lateral forces are (quite surprisingly)
negative. Owing to the finite value of the turning radius R and to the presence of the locked
differential, the lateral slips and the longitudinal rear slips can be relatively high, especially
if the steer angle is high and the turning radius is small. Therefore, in these manoeuvres,
the assumption of linear tyre behaviour, which requires tyre slips ij close to zero, may be
inadequate even at low ay , as shown in figures 7 and 9.
Results show that, for the vehicle configuration considered here, the model with linear tyre
behaviour can be considered valid at small levels of lateral acceleration only if the assigned
steer angle is less than 5 or, equivalently, if the turning radius is greater than 35 m. Otherwise,
nonlinear tyres have to be employed even for very low levels of lateral acceleration ay .
All the simulations were also repeated using a simpler vehicle model, in which the depen-
dence of the lateral load transfers Fz1 and Fz2 on the yaw moment Mz2 (equation (16)) was
neglected. Results show that such a dependence has always very little effect on the variables
describing the vehicle motion, confirming the possibility to use, approximately, the classical
relation Fzi = mBi ay instead of equations (16).

5. Conclusions

In the present paper, the effect of a locked differential on the steady-state cornering behaviour
of vehicles was investigated by means of two neatly defined vehicle models having linear and
nonlinear tyre behaviour, respectively. To obtain the yaw moment Mz2 acting on the rear axle,
a detailed kinematic analysis of tyres under combined slip was developed; the longitudinal
slips were formulated on the basis of the nondimensional quantities ij . Moreover, the steady-
state lateral load transfers were shown to be dependent not only on the steady-state lateral
acceleration but also on the yaw moment Mz2 acting on the rear axle.
The handling diagrams obtained for manoeuvres conducted at constant speed, constant steer
angle and constant turning radius were discussed, showing a strong influence of the manoeuvre
itself on the steady-state behaviour and understeer gradient. Notably, the concept of a certain
degree of understeeroversteer to be associated to each level of lateral acceleration, typical of
the classical theory, has to be completely abandoned.
In addition, it was also shown that, in general, even for small lateral accelerations, the
linear tyre model is not adequate for the steady-state cornering analysis. All these aspects had
received very little attention so far.
The analysis described here represents a theoretical and systematic approach to a classical
subject, which in the past was investigated by means of numerical simulations and experimental
ground tests only. It may also be the basis of more refined, yet simple, vehicle models to be
used in the case of locked differential.
The obtained results, which appear not to be previously published in the technical literature,
also suggest a critical use of the single-track model theory for the analysis of experimental data.
 Critical review of handling diagram 447

References
[1] Pacejka, H.B., 1973, Simplified analysis of steady-state behaviour of motor vehicles. Vehicle System Dynamics,
2, 161172, 173183, 185204.
[2] Dixon, J.C., 1991, Tyres, Suspension and Handling (Cambridge: Cambridge University Press).
[3] Gillespie, T.D., 1992, Fundamentals of Vehicle Dynamics (Warrendale: Society of Automotive Engineers).
[4] Milliken, W.F. and Milliken, D.L., 1995, Race Car Vehicle Dynamics (Warrendale: Society of Automotive
Engineers).
[5] Guiggiani, M., 1998, Dinamica del veicolo (Torino: CittStudiEdizioni).
[6] International Standard ISO 4138, 1982, Road Vehicles Steady State Circular Test Procedure (Geneva:
International Organization for Standardization).
[7] Abe, M., 1986, A theoretical analysis on vehicle cornering behaviors in acceleration and in braking. Dynamics
of Vehicles on Roads and on Tracks, Paper presented at the Proceedings of 9th IAVSD Symposium, Linkoeping,
Sweden, 2428 June, 1985, 114.
[8] Hall, L.C., 1986, International Conference on All-Wheel Drive, London (IMechE Conference Publications),
pp. 5966.
[9] Lorenz, K., Dietrich, C. and Donges, E., 1986, Einfluss des Sperrdifferentials auf Traktion und Fahrverhalten
von Fahrzeugen in Standardbauweise Teil 1. (Influence of a limited slip differential on the traction and handling
of rear axle driven cars). Automobiltechnische Zeitschrift, 88(2), 95100.
[10] Lorenz, K., Dietrich, C. and Donges, E., 1986, Einfluss des Sperrdifferentials auf Traktion und Fahrverhalten
von Fahrzeugen in Standardbauweise Teil 2. (Influence of a limited slip differential on the traction and handling
of rear axle driven cars Part 2). Automobiltechnische Zeitschrift, 88(3), 149153.
[11] Koepf, P., Escher, M., Gazyakan, Ue. and Oberhauser, M., 1990, Optimization of traction and driving stability in
2-wheel drive cars by means of electrohydraulic limited-slip differentials. Proceedings Society of Automotive
Engineers, 813818.
[12] Iwata, T., Murakami, T. and Tamura, M., 1991, Development and analysis of new traction control system with
rear viscous LSD. SAE Transactions, 100, 935945.
[13] Mastinu, G. and Battistini, E., 1993, The influence of limited-slip differentials on the stability of rear-wheel-drive
automobiles running on even road with dry surface. International Journal of Vehicle Design, 14(2/3), 166183.
[14] Motoyama, S., Uki, H., Isoda, K. and Yuasa, H., 1993, Effect of traction force distribution control on vehicle
dynamics. Vehicle System Dynamics, 22, 455464.
[15] Huchtkoetter, H. and Klein, H., 1996, The effect of various limited-slip differentials in front-wheel drive vehi-
cles on handling and traction. Transmission and Driveline Systems Symposium: Efficiency, Components, and
Materials, Detroit, JAE Paper 960717, 131141.
[16] Yamamoto, M., 1991, Active control strategy for improved handling and stability. SAE Transactions, 100,
16381648.
[17] Shibahata,Y., Shimada, K. and Tomari, T., 1993, Improvement of vehicle maneuverability by direct yaw moment
control. Vehicle System Dynamics, 22, 465481.
[18] Abe, M., Ohkubo, N. and Kano, Y., 1996, A direct yow moment control for improving limit performance of
vehicle handling comparison and cooperation with 4WS. Vehicle System Dynamics, (Suppl. 25), 323.
[19] Furukawa, Y. and Abe, M., 1997, Advanced chassis control systems for vehicle handling and active safety.
Vehicle System Dynamics, 28, 5986.
[20] Mokhiamar, O. and Abe, M., 2002, Combined lateral force and yaw moment control to maximize stability as
well as vehicle responsiveness during evasive maneuvering for active vehicle handling safety. Vehicle System
Dynamics, (Suppl. 37), 246256.
[21] Mokhiamar, O. and Abe, M., 2004, Simultaneous optimal distribution of lateral and longitudinal tyre forces
for the model following control. Journal of Dynamic Systems, Measurement and Control, Transactions of the
ASME, 126(4), 753763.
[22] Frendo, F., Ghelardi, E., Leoncini, A. and Vitale E., 2001, A lumped parameter model for the analysis of kart
dynamics. Paper presented at Florence ATA 2001: the Role of Experimentation in the Automotive Product
Development Process, 2325 May 2001.
[23] Amato, T., Frendo, F. and Guiggiani, M., 2002, Handling behaviour of racing karts. SAE Transactions, sec. 6,
111, 20962103.
[24] Amato, T., Frendo, F. and Guiggiani, M., 2004, Handling behaviour of vehicles with locked differential system.
Paper presented at Fisita 2004, Barcelona, Spain, 2327 May 2004.
[25] Pacejka, H.B. and Sharp, R.S., 1991, Shear force development by pneumatic tyres in steady state conditions: a
review of modelling aspects. Vehicle System Dynamics, 20, 121176.
[26] Pacejka, H.B., 2002, Tyre and Vehicle Dynamics (Oxford: Butterworth-Heinemann).