You are on page 1of 26

Rev.

D 12/20/2016

Steering and Stability Notes


(Unpublished Notes to be Included in a Future Paper on Automotive Stability)

1 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

The handling, i.e. steering and stability, character of an automobile depends upon its
responses to steering and disturbance inputs. To investigate the nature of automotive steering and
stability a simplified (linear relationships, no suspension, low speeds, steady-state) mathematical
model is initially presented and then developed further. This model begins with the steering
geometry…

Steering is required to control the direction of the vehicle, and for this to occur efficiently
it is necessary that tire scrub be minimized. This means that in a steady-state turning response to
steering input the tires should follow non-intersecting arc-shaped paths about a common turn
center (“TC”). This was already the case for the pivot axle steering systems as were in common
use by the 1700’s:

Figure 1.01 – PIVOT AXLE STEERING GEOMETRY IN LOW SPEED TURN

However, the pivoting axle had a number of short-comings. The fact that the whole axle
rotated about a central point meant that the front wheels had to be of considerably smaller radius
than the rear wheels in order to obtain a short turn radius yet avoid contact with the coachwork;
this reduction in radius resulted in increased road shock input at the front axle. Also, the fact that
the whole axle rotated resulted in a decreased resistance to rollover in turning as such decreased
the effective front axle “track”. Lastly, sudden road irregularity would produce longitudinal
impact forces acting on one wheel resulting in large “kick-backs” on the drawbar 1.

1
This last short-coming was more of an annoyance for the horses than a problem for the driver, and horses had no
say in carriage design. However, some of the earliest automobiles employed “tiller” steering, and kick-back through
the tiller was definitely not found desirable by the drivers. As drivers did have some say in automotive design this
spelled the end of both tiller steering and improper steering geometry.

2 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

It was to correct these short-comings of the pivot axle steering system that “Ackermann”
steering geometry was developed. As far as is known, “Ackermann” steering geometry was
actually first developed by Erasmus Darwin. Erasmus developed his steering geometry to correct
the short-comings of the pivoting axle steering commonly found on the carriages of his time,
although some wags have suggested that it was to alleviate the disturbance of the beautiful white
gravel driveways of the gentry. In any case, the steering geometry of the carriage morphed into:

Figure 1.02 – “ACKERMANN” STEERING GEOMETRY IN LOW SPEED TURN

Note that the inner “δi” and outer “δo” steering angles of the front wheels are also
replicated near the turn center due to the geometry of the lines. Note also that those lines form a
number of triangles from which certain useful angular relationships may be found:

∆ABC:

𝒂𝒅𝒋 𝒕 + 𝒔
𝐜𝐨𝐭 𝜹𝒐 = =
𝒐𝒑𝒑 𝒍𝒘𝒃

∆AED:

𝒂𝒅𝒋 𝒔
𝐜𝐨𝐭 𝜹𝒊 = =
𝒐𝒑𝒑 𝒍𝒘𝒃

Subtracting the first relation from the second produces a relationship between the
vehicle’s inner “δi” and outer “δo” steering angles, the track “t”, and the wheelbase “lwb”:
𝒕+𝒔 𝒔
𝐜𝐨𝐭 𝜹𝒐 − 𝐜𝐨𝐭 𝜹𝒊 = −
𝒍𝒘𝒃 𝒍𝒘𝒃

A little simplification of this produces:

3 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

𝒕
𝐜𝐨𝐭 𝜹𝒐 − 𝐜𝐨𝐭 𝜹𝒊 = 𝒍 (EQ. 1.01)
𝒘𝒃

This relationship is considered a defining characteristic of what is commonly known as


“Ackermann” steering geometry. 2 Returning to consideration of the four wheel model, note that
in Figure 1.02, for the very sharp, small “R” turn depicted, the intersection of the centerline of
the inner tire and the axle line of the outer tire falls on the vehicle centerline at a point “Q”; this
is the “ideal” situation.

If the steering angles are decreased and “R” consequently increased, then the intersection
point “Q” will move off the vehicle centerline towards point “Q'”, thereby incurring increasing
location errors “ex” and “ey”. These errors are a consequence of the particular geometry of the
specific steering linkage that coordinates the turning motion of the two front wheels:

Figure 1.03 – “ACKERMANN” STEERING LINKAGE GEOMETRY

2
In 1758 Erasmus Darwin (1731-1802), grandfather of Charles Darwin (1809-1882), invented the so-called
“Ackermann” steering system in order to rectify some of the shortcomings of the “axle” steering system then in use
for horse drawn carriages, coaches, and wagons. At around 1816, Darwin’s steering innovation was apparently
independently reinvented by a German carriage manufacturer named Georg Lankensperger. A certain Rudolf
Ackermann became acquainted with Lankensperger and his invention and, acting ostensibly as Lankensperger’s
agent, obtained an English patent (number 4212) in his own name in 1818. However, the geometry of the steering
arrangement in Ackermann’s patent illustration was both extreme (control arm inclination with respect to the
longitudinal axis was depicted as asymetrical, 9 and13 degrees; Darwin’s arrangement involved symetrical angles in
the range of 23 to 30 degrees) and difficult to discern. In 1878 French carriage maker Charles Jeantaud corrected
the “Ackermann Steering” back to the original (Darwinian) concept. However, to this day the steering geometry in
question is still known as “Ackermann Steering”, although Rudolf Ackermann, despite all his scheming perfidy,
never profited from his patent. Life is not only unfair, but often ironic.

4 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

It is the steering arm inclination with respect to the longitudinal, i.e. the angle “α”, which
is the most critical geometry parameter. In a classical “Ackermann” steering arrangement the
steering arms form an angle “α” with the longitudinal so that the projected arm centerlines
intersect at the vehicle centerline on the rear axle. This means that for correct “Ackermann”
steering the steering arm angle(s) “α” will tend to be in the range of around 23 to 30 degrees. The
angle “α”, track “t”, and the wheelbase “lwb” constitute the principal parameters in determining
the steering “behavior”; that is to say, determining the changing relationship between the steering
angles “δi” and “δo” as a turn is implemented 3.

This relationship between the inside of turn wheel steering angle “δi” and the outside of
turn wheel steering angle “δo” is plotted in the next figure. Note that if the steering arms are
parallel (“α = 0”), which is not “Ackermann” steering, then the steering angles “δi” and “δo” are
always equal. For the theoretically correct “Ackermann” steering the outside steering angle “δo”
is less than the inside steering angle “δi”, with the difference growing with the sharpness of the
turn. However, in actual practice, the relation between “δi” and “δo” tends to be somewhere
between that of parallel steering arms and that of perfectly “Ackermann” angled steering arms 4:

Figure 1.04 – INSIDE vs. OUTSIDE STEERING ANGLE RELATION IN TURNING

As an automobile is turned not only is there a variation in the relation between “δi” and
“δo”, but also the error values of “ex” and “ey” vary as the point “Q” moves with respect to the
theoretically (“Ackermann”) perfect error plot line “M-F” as shown in the next figure:

3
The linkage and analysis shown is appropriate for a beam axle. Independent suspensions require a much more
complicated steering linkage, and thus the analysis is also much more complicated but similar in execution and
result.
4
This illustration and the following illustration are both drawn from J.Y. Wong’s Theory of Ground Vehicles, page
366 and page 367, respectively. However, if for racing the intent is to have the steering linkage operate so that the
“δo” vs. “δi” plot mirrors the behavior of the “ideal” “ψo” vs. “ψi” plot, then it would seem to this author that the two
non-parallel curves would be on the other side of the parallel arm plot line and curving upward (“anti-Ackermann”).

5 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Figure 1.05 – “ACKERMANN” STEERING LINKAGE ERROR VARIATION IN TURNING

Now that the steering and steering linkage geometry has been dealt with somewhat, it is
time to consider the related matter of yaw (directional) stability which traditionally been
investigated through the use of the “bicycle model” 5. The “bicycle model” is considerably
simpler than a four wheel model; thereby making an analysis easier and the results can still be
extended to apply to the four wheel case. The bicycle model situation at low speed (no “slip
angles”) can be depicted:

Figure 1.06 – BICYCLE MODEL AT LOW SPEED

5
Today much more complex models are available, but generally require a computerized analysis to be useful. Note
that for the simplified bicycle model each tire represents both the two tires as would be on each axle.
Therefore, the individual tire characteristics such as a tire’s “Cornering Stiffness” must be doubled to
produce the proper values to be used in a “bicycle model” analysis.

6 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Note that the steering angle of the front wheel is again reflected in the angle at the vertex
of the lines “TC-to-B” and “TC-to-C”. The turn radius “R” times the vertex angle “δ” (in
radians) produces an arc length “S”:

𝑺=𝑹𝜹
If it is postulated that the turn radius must be equal to, or greater than, 24 feet or so, then
the arc length “S” is approximately equal to the wheelbase, which produces another defining
characteristic of Ackermann steering geometry:

𝑺 ≈ 𝒍𝒘𝒃

𝒍𝒘𝒃 = 𝑹 𝜹

𝐥𝐰𝐛
𝛅= (EQ. 1.02)
𝐑
By a similar chain of reasoning the body attitude angle “β” can be established:
𝒍𝒓
𝜷= (EQ. 1.03)
𝑹
Now consider how the automotive steady-state turning action at a “high” (significant)
speed produces a number of effects which result in a new situation which may be depicted:

Figure 1.07 – BICYCLE MODEL AT HIGH SPEED


First and foremost there is now significant centrifugal force acting outward from the
turning center “TC” and through the vehicle “CG”; this force is of the magnitude “W ay” where
“W” is the weight of the automobile and “ay” is the lateral acceleration in gravity units. The
magnitude of this lateral acceleration is determined:

7 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

𝑽𝟐
𝒂𝒚 = (EQ. 1.04)
𝑹𝒈

Where:

ay = the lateral acceleration of the vehicle (g’s).


V = the velocity of the vehicle (ft/sec).
R = the turn radius (ft).
g = the standard value of gravity (32.174 ft/sec2).
To counter this centrifugal force there are two centripetal forces acting as the result of
traction, one force at each of the two tires:

𝑭𝒚𝒇 = 𝑾𝒇 𝒂𝒚 (EQ. 1.05)

𝑭𝒚𝒓 = 𝑾𝒓 𝒂𝒚 (EQ. 1.06)

Due to these lateral traction forces acting on the tires there are corresponding “slip
angles” 6 at the front “ψf” and rear “ψr” tires which were not present earlier in the depiction of
low speed (“ay ≈ 0”) turns. These slip angles are the quotient of the lateral force on each tire
divided by each tire’s “cornering stiffness” 7:

𝒅𝑭𝒚𝒇
𝝍𝒇 = 𝑭𝒚𝒇 �
𝒅𝜶𝒇

𝒅𝑭𝒚𝒓
𝝍𝒓 = 𝑭𝒚𝒓 �
𝒅𝜶𝒓

For convenience in notation the “cornering stiffness” values are usually replaced
symbolically by “Cs”:

𝝍𝒇 = 𝑭𝒚𝒇 ⁄ 𝑪𝒔𝒇

6
“Slip angle” is a misnomer, like the term “shock absorber” when used to refer to a damper. The tires don’t actually
“slip” but, through a process of continual deformation at the road contact patch with rotation, give the impression of
slippage; the only real slippage occurs in a small aft portion of the total tire/road contact area. It is for this reason the
great automotive dynamicist Donald Bastow suggested the use of the term “drift angle” in lieu of “slip angle”
7
If the lateral force-slip angle function is linearized by the fitting of a straight line to the original shallow curve, then
a simplified lateral force/drift angle relationship (the slope of the fitted line) can be obtained. This quantity is
commonly called the “Cornering Stiffness” and symbolized as “Cs” (sometimes it is also called the “Cornering
Power” of the tire, which is yet another misnomer like “shock absorber” or “slip angle”). Linearized tire relationship
values such as “Cs” are often used in directional stability determinations and for various simulations and studies of
automotive dynamic behavior. When normalized by the normal load “N” the “Cornering Stiffness” becomes the
“Cornering Coefficient” “Cc” which is expressed in units such as “lb/deg/lb”.

8 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

𝝍𝒓 = 𝑭𝒚𝒓 ⁄ 𝑪𝒔𝒓

Because of these slip angles the geometry of the turning situation has changed; the slip
angle at the rear tire “ψr” is in addition to the original vertex angle “δ” resulting in a larger total
angle “θ” at the vertex. This is because the rear slip angle defines the direction of the velocity
vector at point “C”, and the radial line from the “TC” to any point on the vehicle must be
orthogonal to the velocity vector at that point. Also, the front slip angle “ψf” represents a
“backing off” from the steering angle “δ” so that the angle that previously was just “δ” is now
“δ- ψf”. As a result of these changes the turn center “TC” no longer lies on a lateral projection
of the rear axle line, but has moved considerably forward as shown in Figure 1.07. The new
vertex angle is now also equal (approximately) to the wheelbase divided by the turn radius as per
Figure 1.06/Eq. 1.02:

𝐥𝐰𝐛
𝛉=
𝐑
From Figure 1.07, the components of angle “θ” may be substituted for that angle:
𝐥𝐰𝐛
(𝛅 − 𝛙𝐟 + 𝛙𝐫 ) =
𝐑
This equation may rearranged to establish a “high” speed steering characteristic that is
complementary to the “low” speed Ackermann steering characteristic of Eq. 1.02:

𝒍𝒘𝒃
𝜹= + 𝝍𝒇 − 𝝍𝒓 (EQ. 1.07)
𝑹
And, the related steady-state yaw angle “β” of the vehicle, as depicted in Figure 1.07,
may now be defined as well 8:

𝒍𝒓�
𝜷= 𝑹 − 𝝍𝒓 (EQ. 1.08)

Where, for both these equations, the symbolism is:

δ = the steering angle (radians).


lwb = the vehicle wheelbase (ft).
R = the turn radius (ft).
ψf = the front slip angle (radians).

8
The term “vehicle side slip angle” seems to be preferred to the term “yaw angle” by the SAE, possibly because
“β” is constant during steady-state, but there is still an angular yaw velocity “ω” (differing reference systems used!).
The sign convention of SAE J670 2008 (“FIGURE 3”) on page 17 indicates the “β” angle is “positive” when the
turning vehicle velocity vector “V” lies to the “inside” of the vehicle centerline as shown in Figure 1.07.

9 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

ψr = the rear slip angle (radians).


β = the yaw angle at the vehicle CG (radians).
lr = the longitudinal distance from CG to rear axle (ft).
Automotive yaw (directional) stability is basically a matter of the slip angle relationship
front to rear. If the slip angle at the front tire(s) is greater than the slip angle at the rear tire(s)
(“ψf - ψr > 0”) then the vehicle can be said to be “understeering”, which is considered to be a
stable condition. If the slip angle at the front tire(s) is lesser than the slip angle at the rear tire(s)
(“ψf - ψr < 0”) then the vehicle can be said to be “oversteering”, which is considered to be an
unstable condition. Of course, it would be unwise to take a single point in a vehicle’s speed
(lateral acceleration) range and use it as identifying the vehicle’s general stability condition over
its entire speed range; that’s why certain more general indicators of overall stability have been
developed 9.

One of those more general indicators is a quantity called the “Understeer Gradient”
(a.k.a. “Understeer Coefficient”, which is a misnomer), symbolized as “Kus” and expressed in
radians or degrees, which presents a more general way of classifying a vehicle’s stability
condition than comparing specific speed point drift angle values “ψf” and “ψr”. If a vehicle is
stable (understeering or US) then “Kus > 0”, if neutral (NS) then “Kus = 0”, or if unstable
(oversteering or OS): “Kus < 0”. This coefficient can take various forms, of which one of the
simpler is:

𝑾𝒇 𝑾
𝑲𝒖𝒔 = �𝒈 𝑪 − 𝒈 𝑪𝒓 � (EQ. 1.09)
𝒔𝒇 𝒔𝒓

This equation comes about through the simple comparison of the front to rear slip angles
as already discussed:

Kus ≈ ψf - ψr
The comparison becomes more general by substitution of the cornering stiffness
expressions for the slip angles:

𝑲𝒖𝒔 ≈ 𝑭𝒚𝒇 �𝑪𝒔𝒇 − 𝑭𝒚𝒓 ⁄𝑪𝒔𝒓


Now the expressions for the front and rear lateral forces may be substituted in place of
those forces:

𝑲𝒖𝒔 ≈ 𝑾𝒇 𝒂𝒚 ⁄𝑪𝒔𝒇 − 𝑾𝒓 𝒂𝒚 ⁄𝑪𝒔𝒓

9
The stability of a vehicle can change over the course of its speed range or as fuel is burnt off. Racing cars are often
designed to be initially oversteering at low speed, and then to become understeering at higher speeds.

10 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Dividing through by the gravitational constant for both right hand terms gets the result
per “g”, and that result is termed the Understeer Gradient (“Kus”) as shown before as Equation
1.09:

𝑾𝒇 𝑾𝒓
𝑲𝒖𝒔 = � − �
𝒈 𝑪𝒔𝒇 𝒈 𝑪𝒔𝒓

Calculation of the steering angle “δ” required at the front wheel(s) to make a steady-state
turn of radius “R” at speed “V” (or at a lateral acceleration “ay”) is another characteristic of
vehicle handling. Obtaining the proper formulation begins with Equation 1.07:
𝒍𝒘𝒃
𝜹= + 𝝍𝒇 − 𝝍𝒓
𝑹
Making appropriate substitutions for the slip angles this expression becomes:

𝒍𝒘𝒃 𝑭𝒚𝒇 𝑭𝒚𝒓


𝜹= + −
𝑹 𝑪𝒔𝒇 𝑪𝒔𝒓

A few more substitutions may now be made:

𝒍𝒘𝒃 𝑾𝒇 𝒂𝒚 𝑾𝒓 𝒂𝒚
𝜹= + −
𝑹 𝒈 𝑪𝒔𝒇 𝒈 𝑪𝒔𝒓

𝒍𝒘𝒃 𝑾𝒇 𝑾𝒓
𝜹= +� − �𝒂
𝑹 𝒈 𝑪𝒔𝒇 𝒈 𝑪𝒔𝒓 𝒚

A final substitution produces the desired formulation with the steering angle “δ” as a
function of “Kus” and “ay”:

𝒍𝒘𝒃
𝜹= + 𝑲𝒖𝒔 𝒂𝒚 (EQ. 1.10)
𝑹
An even more useful form for this expression is readily obtained by substituting for “ay”
the equation “V2/R” for centripetal acceleration in a turn of constant radius:

𝒍𝒘𝒃 𝑽𝟐
𝜹= + 𝑲𝒖𝒔 (EQ. 1.11)
𝑹 𝑹
This equation gives the steering angle “δ” as a function of the speed “V” around the turn
radius “R”, and can be further simplified as:

11 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

𝜹 = �𝒍𝒘𝒃 + 𝑲𝒖𝒔 𝑽𝟐 �⁄𝑹 (EQ. 1.12)

These equations relate how the steady-state steering angle “δ” is affected by the
wheelbase “Lwb”, the turn radius “R”, the Understeer Gradient “Kus” (and all that which
comprises that gradient), and the lateral acceleration level “ay” or the velocity “V”. If a vehicle
has a relatively large “Kus” value it is apparent that any small change in velocity will require a
fairly large correction at the steering wheel in order to keep the vehicle on course around the
skidpad. However, if a vehicle appears close to being neutral in behavior with a small “Kus”
value, then it will require only mild steering corrections to stay on course. However, if “Kus” is
negative then it may be very difficult to stay on course, and may even require a “counter-
steering” technique in a hard turn which runs counter to human intuition.
Note also the role wheelbase “lwb” plays in this regard. The greater the wheelbase then
the greater the necessary steering angle will be. This is one reason why “stretch limos” are never
utilized in SCCA gymkhanas and slalom competitions; it would be too difficult to keep up with
all the large steering changes 10. The bottom line here is that vehicles which are directionally
unstable, and/or require large steering inputs to keep on course, are going to be difficult, if not
impossible, to control under circumstances requiring precise and quick steering.
With further regard to the subject of directional stability, there are a number of quantities
other than just “Kus” that have been developed in order to quantify the degree of stability or
instability present in an automotive design.
For instance, there is the “Characteristic Speed”, which provides a means of comparing
designs with respect to the degree of understeer present 11. Mathematically, it is the speed at
which the steering angle “δ” required to make a turn of radius “R” is equal to “2 lwb/R” (that is,
the steering angle is twice that determined by the low speed “Ackermann” steering geometry
“lwb/R”) 12:

𝟓𝟕.𝟑 𝒈 𝒍𝒘𝒃
𝑽𝒄𝒉𝒂𝒓 = � (EQ. 1.13)
𝑲𝒖𝒔

It has been stated that traditional highly oversteering “…North American cars have a
characteristic speed as low as 40 mph…” 13. The “Vchar” is the point at which steering response
reaches a maximum for the understeering vehicle, and can go to infinity (!) for a highly
oversteering vehicle .

10
Also a long wheelbase generally corresponds with greater yaw inertia, which is another adverse maneuverability
factor, although it is really the related yaw inertia about the instantaneous center of rotation which is of concern.
11
Ergo, all vehicles so compared must be understeering.
12
Gillespie, Fundamentals of Vehicle Dynamics, pg. 204.
13
Milliken and Milliken, Race Car Vehicle Dynamics, pg. 181.

12 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

There is also a stability factor known as the “Critical Speed”, which is a speed at which a
vehicle can switch from a controllable, but oversteering initial condition, to a completely
unstable (very oversteering) condition 14:

−𝟓𝟕.𝟑 𝒈 𝒍𝒘𝒃
𝑽𝒄𝒓𝒊𝒕 = � (EQ. 1.14)
𝑲𝒖𝒔

Note that there now is a negative sign under the square root; this is in recognition of the
fact that an oversteering vehicle will have a negative “Kus”. When the vehicle under study does
not have a negative “Kus” then this particular stability factor is not applicable (“…there is no
critical speed for an understeer car…a neutral steer car does not have a real critical
speed…”).
Lastly there is “Static Margin”, which is a determinant like “Kus” for classifying a
vehicle’s stability condition as stable (understeering): “SM > 0”, neutral: “SM = 0”, or unstable
(oversteering): “SM < 0”. However, the Static Margin is a dimensionless fraction with the
advantage of relating the longitudinal center of gravity (“LCG”) placement to the “Neutral Steer
Point” (“NSP”) for that condition (“LCG – NSP = SM ˟ lwb”) 15:

𝒍𝒓 𝑪𝒔𝒓 −𝒍𝒇 𝑪𝒔𝒇


𝑺𝑴 = (EQ. 1.15)
𝒍𝒘𝒃 �𝑪𝒔𝒇 +𝑪𝒔𝒓 �

So far the concern has been with steering and steady-state turning performance, but every
turning action has three phases: initiation of the turn, turning, and termination of the turn. Only
during the middle phase can a steady-state condition be achieved; the turn initiation and
termination are both transient states. On a winding road, when weaving in and out of traffic, or
especially on a slalom test course, a vehicle may spend more time in a transient state than in
steady-state. Therefore, the transient state is every bit as important as the steady-state, perhaps
even more so. However, a complete analysis of the transient state is more complex than that of
the steady-state, and is beyond the scope of these notes, but may be dealt with at a later time.

14
Gillespie, Fundamentals of Vehicle Dynamics, pg. 205. Ergo, all vehicles for which a Critical Speed is calculated
must be oversteering. Note that in Equations 1.13 and 1.14 a radians to degrees conversion factor of “57.3 deg/rad”
is included in the numerator under the square root sign. One must always take care that there is a consistency of
units in any calculation.
15
This is all very similar to aircraft pitch stability theory wherein the relationship between the center of gravity and
the center of pressure is expressed in terms of percent of the “Mean Aerodynamic Chord” (MAC). Typical
passenger car “SM” values may range between +0.05 to +0.07 (per Gillespie, pg. 209).

13 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Although “Ackermann” steering geometry has traditionally been used for passenger cars
it is not perfect due to its mechanical nature, hence the geometric “error” plot of Figures 1.02 and
1.05, and also may not be the best geometry for ultra-high performance applications 16.
Automobiles designed for racing may employ “Anti-Ackermann” geometry as they generally
perform at lateral acceleration values so high that the inner front tire is often totally without
normal load; in such a case only the steering of the highly loaded outer tire is of importance…

An automotive design firm by the name of “OptimumG” has put forward an interesting
concept for the development of a steering geometry for maximum lateral acceleration
performance. Essentially the proposed method for determining the optimum relationship of inner
wheel steering angle to outer wheel steering angle is to run a computer simulation of a vehicle up
to its maximum lateral acceleration, and note the relationship of the inner wheel slip angle to the
outer wheel slip angle. With this relationship in hand, the steering linkage can be designed so
that the inner to outer steering angle relationship up to the maximum lateral acceleration will be
similar, and thus supposedly more conducive to forming the desired slip angles without
hindrance from the steering linkage.

This certainly sounds reasonable, but does seem to present some problems in practice.
The “OptimumG” people used extensive tire test data 17 fitted via their proprietary “OptimumT”
software to a Pacejka (a.k.a. “Magic Formula”) tire model, which in turn is to be used in
conjunction with an MS-Excel or MATLAB platform for simulation. Since this author does not
have all these resources as was available to the “OptimumG” researchers, a more simplistic but
reasonable simulation approach was undertaken to explore the concept as presented.

Using the methodology described elsewhere 18, this author ran an MS-Excel simulation of
two “high-performance” sports car designs. The tire data used came from published sources and
was far from complete, but that only affects the absolute accuracy of the projected maximum
lateral acceleration levels, not the correctness of the basic method. The baseline vehicle was a
1958 Jaguar XK150S, which reached a simulated maximum lateral acceleration of 0.56 g’s on a
100 foot radius skid pad. That seems reasonable, given the nature of the early 6.00×16 bias ply
tires used on the Jaguar. The other “advanced concept” vehicle was derived from the baseline
vehicle, with a wider track, lower c.g., greater roll stiffness, and slightly “stickier” and “stiffer”

16
Nathaniel William Troupe, in his paper “Optimization of Steering Geometry for Ultra-High Mileage Vehicles”
(August 2011, Pennsylvania State University) suggests the possible use of servo motors to precisely control the
steering angle at each wheel instead of a mechanical linkage. However, he also acknowledges that increased cost
and weight may present obstacles to employing such a possibility, and that it is not certain that a projected 1.1%
improvement in fuel economy would actually be realized. However, this author believes that Troupe’s idea is
certainly worthy of further investigation.
17
The tire data used by the OptimumG researchers was collected by the Calspan Corporation for the SAE Formula
Tire Testing Consortium, presumably for use by the various teams involved in the SAE Student Formula racing
competition.
18
Wiegand, Brian; “Mass Properties and Advanced Automotive Design”; SAE Paper 3602, May 2015.

14 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

tires. This vehicle reached a simulated maximum lateral acceleration of 0.71 g’s on a 100 foot
radius skid pad.

A problem presented itself early in applying the “OptimumG” concept for determining
the ideal steering geometry for maximum lateral acceleration; by the time maximum lateral
acceleration was achieved both “test” vehicles had completely unloaded their inner front wheels
in simulation. Thus there was no inner front wheel slip angle to compare to the outer front wheel
slip in order to establish a steering angle relationship (note that this unloaded inner tire behavior
is exactly the reason why “Anti-Ackermann” geometry is actually used by some race cars as
noted earlier). To prevent this from being the dead end result of an investigation of a promising
concept, it was decided to look closely at the slip angle relationships in an interval before the
inner tire lateral traction force begins to decrease.

At a lateral acceleration level of 0.384 g’s on the 100 foot radius skid pad, the baseline
Jaguar had an “ideal” inner slip angle of 2.97 degrees and an “ideal” outer slip angle of 6.68
degrees. Under the same conditions, the modified Jaguar had an “ideal” inner slip angle of 0.65
degrees, and an “ideal” outer slip angle of 0.95 degrees 19. The Jaguar had a “body attitude angle”
(yaw angle) “β” of -0.71 degrees, while the “advanced” variant had a “β” of 1.86 degrees. The
wheelbase “l” of each vehicle was 102.00 inches and 99.33 inches, and the front tracks were
51.75 inches and 53.75 inches, respectively.

The need for all the previous information is mandated by the formulae inherent in the
“OptimumG” methodology. The first set of “OptimumG” equations are:

𝒍
𝜶𝒐 𝑮 = 𝟓𝟕. 𝟑 𝒕 +𝜷 (EQ. 1.16)
𝑹+
𝟐

𝒍
𝜶𝒊 𝑮 = 𝟓𝟕. 𝟑 𝒕 +𝜷 (EQ. 1.17)
𝑹−
𝟐
Where:

𝜶𝒐 𝑮 = the outer wheel “geometric” slip angle (deg).

𝜶𝒊 𝑮 = the inner wheel “geometric” slip angle (deg).


l = the wheelbase (in).
R = the skid pad radius (in).
t = the vehicle track (in).

19
The “baseline” Jaguar was very “understeering” in its behavior, while the “advanced” Jaguar variant was almost
“neutral” in its behavior.

15 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

β = the vehicle body yaw angle (deg).


“Plugging in” the simulation values into these equations for the Jaguar we get (in
degrees) the “geometric” slip angles:

𝜶𝒐 𝑮 = 57.3 deg/rad [102 / (1200+25.875)] - 0.71 deg = 4.06 deg


𝜶𝒊 𝑮 = 57.3 deg/rad [102 / (1200-25.875)] - 0.71 deg = 4.27 deg
“Plugging in” the simulation values into these equations for the modified Jaguar (in
degrees) for the “geometric” slip angles:

𝜶𝒐 𝑮 = 57.3 deg/rad [99.33 / (1200+26.875)] + 1.86 deg = 6.50 deg


𝜶𝒊 𝑮 = 57.3 deg/rad [99.33 / (1200-26.875)] + 1.86 deg = 6.71 deg
This brings us to the point of finding the “ideal” steering angles for the steered wheels by
using the second set of “OptimumG” equations:

𝜹𝒐 = 𝜶 𝒐 𝑮 − 𝜶 𝒐 𝒊 (EQ. 1.18)

𝜹𝒊 = 𝜶𝒊 𝑮 − 𝜶𝒊 𝒊 (EQ. 1.19)

Where:

δo = the “ideal” steering angle of the outer wheel (deg).


δi = the “ideal” steering angle of the inner wheel (deg).

𝜶𝒐 𝑮 = the outer wheel geometric slip angle (deg).

𝜶𝒊 𝑮 = the inner wheel geometric slip angle (deg).

𝜶𝒐 𝒊 = the outer wheel “simulation” slip angle (deg).

𝜶𝒊 𝒊 = the inner wheel “simulation” slip angle (deg).


Again, “plugging in” the known values into this second set of equations, for the Jaguar
we get the “ideal” (OptimumG) steering angles:

δo = 4.06 deg – 6.68 deg = -2.62 deg


δi = 4.27 deg – 2.97 deg = 1.30 deg
“Plugging in” the known values into the second set of equations for the modified Jaguar
we get the “ideal” (OptimumG) steering angles:

16 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

δo = 6.50 deg – 0.95 deg = 5.55 deg


δi = 6.71 deg – 0.65 deg = 6.06 deg
It seems immediately apparent that something is not quite “kosher” here. In both vehicle
cases the “ideal” OptimumG values of the “inside” steering angles are greater than the values of
the “outside” steering angles which is in accord with conventional Ackermann steering
geometry 20. Also, the variation in sign for the stock Jaguar “ideal” OptimumG steering angles
would also seem to indicate that the Jaguar is attempting to steer in two different directions at the
same time. Compare these OptimumG results with the “ideal” steering angles as obtained by the
author’s MS-Excel simulation for the Jaguar and modified Jaguar which are, respectively:

δo = 7.53 deg
δi = 5.76 deg
and:

δo = 4.89 deg
δi = 4.80 deg
These simulation “ideal” steering angles would seem to make a lot more sense than what
was obtained by using the “OptimumG” equations. However, it’s possible that the disappointing
“OptimumG” results was obtained by some misunderstanding on this author’s part of how those
equations were meant to be used, or maybe by a typo in the source material. What is really under
study here is the “OptimumG” concept for determining the “ideal” high performance (lateral
acceleration) steering geometry, and we can continue with that study regardless of the soundness
of Equations 1.16 through 1.19.

From this author’s proprietary (and admittedly very simplified) MS-Excel simulation of
the Jaguar XK150S lateral acceleration performance a data table of lateral acceleration in “g’s”
versus “actual” outer and inner slip angles “ψo, ψi” was obtained. To this table corresponding
values of “ideal” steering angles “δo, δi” were added as determined by the equations 21:

20
Note that Figure 1.02 for pure “Ackermann” low speed (no slip angles) turning shows this steering angle
relationship: the “inside” angle is greater than the “outside” angle; Figure 1.04 shows how the magnitude of this
difference increases as the steering angles increase.
21
Note that theses equations for “ideal” steering angles are almost identical to the OptimumG equations for the
“geometric” slip angles (Eq. 1.16 & 1.17), which is yet another reason to be suspicious of the OptimumG
methodology.

17 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

𝑳𝒘𝒃
𝜹𝒐 = 𝟓𝟕. 𝟑 � 𝒕𝒇 + 𝝍𝒐𝒇 − 𝝍𝒐𝒓 (EQ. 1.20)
𝑹+
𝟐

𝑳𝒘𝒃
𝜹𝒊 = 𝟓𝟕. 𝟑 � 𝒕𝒇 + 𝝍𝒊𝒇 − 𝝍𝒊𝒓 (EQ. 1.21)
𝑹−
𝟐

Where the symbolism is:


δo = The “ideal” outer steering angle (deg).
δi = The “ideal” inner steering angle (deg).
Lwb = The wheelbase (in).
R = The radius of the turn (in).
tf = The front track (in).
Ψof = The outer front slip angle (deg).
Ψor = The outer rear slip angle (deg).
Ψif = The inner front slip angle (deg).
ψir = The inner rear slip angle (deg).

From that table a plot of the Jaguar’s “actual” slip angles and “ideal” steer angles vs. the
lateral acceleration was made:

Figure 1.08 – SLIP & “IDEAL” STEER ANGLES vs. LATERAL ACCELERATION

18 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Note that up to approximately 0.33 g’s lateral acceleration the steering angle plots follow
the “Ackermann” low speed steering angle relationship of the “inside” angle being greater than the
“outside” angle. Also note that after a point of about 0.38 g’s the inside steering angle becomes less and
less a matter of significance as the inner slip angle (and hence its lateral traction force contribution)
abruptly becomes less and less. To clarify this “ideal” steering angle behavior a re-plot of Figure 1.08 is
presented without the distraction of the slip angles:

Figure 1.09 – “ANTI-ACKERMANN” INNER vs. OUTER STEER ANGLE RELATION

No mechanical linkage could replicate this relationship in its totality; an Ackermann


steering geometry linkage would suffice up to about 0.33 g’s (at least for this vehicle) after
which an increasingly (not even constant!) anti-Ackermann would be required, although at some
point after 0.38 g’s (only 0.05 g’s later!) the whole matter would become moot due to the
decreasing significance of the inner tire. The additional complexity and cost of a servo driven
steering system (a la Troupe) for a possible 1% improvement in maximum lateral acceleration
capability might be justifiable for a race car. However, for a passenger car which seldom, if ever,
would exceed 0.35 g’s in use, the continued reliance on a mechanical Ackermann linkage would
seem a wise choice even if some vestigial tire scrub were the consequence.

While suspension/steering designers seek to minimize “scrub”, they generally can’t


eliminate it completely. Therefore, designers often balance “scrub” against “toe” and “camber”
effects so as to neutralize each other. The following figure illustrates “scrub” due to lateral
suspension movement and “scrub” due to poor steering geometry:

19 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Figure 1.10 - TWO TYPES OF TIRE SCRUB

Tire “scrub” can also result from extreme “toe” and “camber” settings. Road cars are
designed to have a small amount of “toe in” to enhance directional stability; most race cars are
designed to have a small amount of “toe out” to enhance maneuverability:

Figure 1.11 - ROAD vs. RACE TOE SETTINGS

20 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Generally independent suspensions are designed to have a small amount of static negative
“camber” to maximize tire-road contact patch on the outside wheel in turning (countering the
effect of tread “lift” due to lateral distortion, and possibly creating some beneficial “camber
thrust”):

Figure 1.12 - CAMBER

“Camber Thrust” refers to a force generated perpendicular to the direction of travel of a


rolling tire due to its camber angle. Thrust is generated whenever a rotating tire goes into a
cambered position; points on the tread, which would freely follow an elliptical path projected
onto the ground, are instead forced to follow a straight path when coming into contact with the
ground due to the traction force. The total point path deviation towards the direction of the lean
is a deformation of the tire tread/carcass caused by the total force in the direction of the lean,
i.e., the “Camber Thrust”:

Figure 1.13 - CAMBER THRUST GENERATION

21 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

“Camber Thrust” is approximately linearly proportional to camber angle for small


angles, reaching its steady-state value nearly instantaneously after a change in camber angle, and
so does not have an associated “Relaxation Length” 22.

“Camber Stiffness” is a parameter used to determine the Camber Thrust generated by a


cambered tire and is influenced by tire inflation pressure and normal load. The Camber Stiffness
is the rate of change of camber force with Camber Angle (∆FY/∆ε) at constant normal force and
inflation pressure. The net Camber Thrust acts usually toward the front of the tire/ground contact
patch and so generates a “Camber Torque” such that it tends to steer a tire towards the direction
of lean (an alternate explanation for this torque is that the two sides of the contact patch are at
different radii from the axle and so would travel forward at different rates unless constrained by
traction with the pavement, but that effect is more properly termed “conicity” which is to be
discussed later).

Figure 1.14 - CAMBER THRUST vs. CAMBER ANGLE @ 1000 lb NORMAL LOAD

Donald Bastow’s Car Suspension and Handling (1980) has essentially the same
explanation of Camber Thrust generation (pp. 74-75) as previous, and relates how “old time”
vehicles generally had much higher front suspension “static camber angles” than are prevalent
today, and in the interest of decreased tire wear the suspensions were designed with a “toe-in”
setting intended to negate the associated Camber Thrust 23.

22
“Relaxation length” is a property of pneumatic tires that defines the delay between when a steer angle is
introduced and when the slip angle and cornering force reach steady-state values. It has also been defined as the
distance that a tire rolls before the lateral force builds up to 63% of its steady-state value. It can be calculated
as the ratio of cornering stiffness to lateral stiffness, where cornering stiffness is the ratio of cornering force to slip
angle, and lateral stiffness is the ratio of lateral force to lateral displacement. The camber thrust relaxation length
may be small and thus considered negligible, but this author would consider the length of the contact patch (“lc”) a
good value to use if such a length is needed…
23
Figure from: Riley, Robert Q.; Automobile Ride, Handling, and Suspension Design, Robert Q. Riley Enterprises
LLC, Phoenix, AZ. It would have been helpful if the tire designation and inflation pressure had been given.)

22 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

Bastow also provides an estimation formula for Camber Thrust (pg. 82) as the normal
load times the tangent of the camber angle:

𝑭𝒄𝒂𝒎𝒃𝒆𝒓 = 𝑵 × 𝐭𝐚𝐧 𝜺 (EQ. 1.22)

Which for small angles can be further approximated as:

𝑭𝒄𝒂𝒎𝒃𝒆𝒓 = 𝑵 × 𝜺𝒅𝒆𝒈𝒓𝒆𝒆𝒔 ⁄𝟓𝟕. 𝟑 (EQ. 1.23)

Using the previous camber thrust graph for “N = 1000 lb”, it is interesting to see how
closely these two approximations mimic that plot…

Table 1.01, Figure 1.15 - CAMBER THRUST APPROXIMATIONS vs. ACTUAL

Bastow provides example problems involving camber thrust (pp. 88 and 91) with an
interesting plot on page 117 of Camber Thrust (Newtons) vs. Normal Load (kgf) for various
camber angles (degrees) 24:

Figure 1.16 - CAMBER THRUST VARIATION FOR A SPECIFIC TIRE @ 29 PSI

24
Bastow doesn’t so state, but it is obvious that a steering mechanism that generates camber angles in lieu of, or in
conjunction with, steer angles might make for a very big advance in automotive steering mechanism efficacy.

23 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

“Camber Torque” tends to be counter-acted by “self-aligning torque”. Although a rolling


tire provides a certain amount of self-aligning torque through the phenomenon of pneumatic trail,
the front suspension is usually set with a certain amount of positive caster to augment that
torque. The positive caster angle “τ” in the X-Z plane enhances directional stability, but also
increases steering wheel effort due to the associated “caster trail”, symbolized as “t”, scrub:

Figure 1.17 – TRAIL SCRUB

There is also an offset actually termed the “Scrub Radius”, symbolized as “rσ”, resulting
from the pivot axis inclination in the Y-Z plane “σ” (a.k.a. the “kingpin angle”):

Figure 1.18 - SCRUB RADII, POSITIVE & NEGATIVE

24 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

These offsets affect directional stability, steering effort, tire wear, magnitude of bump
feedback, and braking stability. The result is that there is always some tire scrub compromise,
and where that compromise is made depends on the vehicle use and type.

Continuing with the steering aspect, there is the matter of “Bump Steer”. “Bump Steer” is
the tendency of a wheel on a car to steer as the wheel moves upwards relative to the body. It is
typically measured in units of steer angle per unit of upwards motion (deg/foot, or rad/meter,
etc.). Typical values are from 2 to 10 degrees per meter for IFS wheels; solid axles generally
have zero bump steer (but may still have “Roll Steer”). If bump steer can be held to less than
0.20 degree of toe out change per inch of suspension travel then it should not be a problem, but
no amount of bump steer is desirable. Excessive bump steer increases tire wear and makes the
vehicle more difficult to handle on rough roads.

In a bump steer situation, both wheels may rise together, or just one wheel may rise; note
that if both wheels on an axle line move upwards by the same amount there tends to be no net
steering effect. The linearity of the bump steer vs. lift plot is important. If it is not “linear” then
the length of the tie rod or some other parameter may need to be adjusted. Bump or roll can
result in inadvertent “toe out” steering due to poor positioning of steering pivot points and/or
poor angular orientation of steering components, or due to improper length of outer tie rods.

For instance, bump steer (and roll steer) can result from the outer tie rod being of position
and/or length such that for the same travel upward it results in greater lateral movement than the
lower suspension arm. The following illustrates both good and bad outer tie rod geometry:

Figure 1.19 - BUMP STEER GEOMETRY

“Roll Steer” is the term for the tendency on modern cars to have the IFS wheels steer
“outward” as the vehicle enters a turn and the sprung mass rolls; this produces a condition of
roll understeer. The rear suspension is usually set up to minimize roll steer, where possible. Roll

25 DOI: 10.13140/RG.2.2.12002.22728
Rev. D 12/20/2016

steer is an important part of the stability “budget” used in a “Bundorf Analysis” 25 to determine a
vehicle's total understeering characteristic.

“Compliance Steer” is due to the use of “rubber” bushings in the mounting of suspension
components. When longitudinal and/or lateral forces are generated at the ground contact area of
a tire, those forces cause deflections at the bushings. The deflections alter the alignment, and
may thereby produce in effect a “steer angle”. This compliance “steer angle” may add to or
subtract from the intended steer angle created by the use of the steering wheel.

Lastly, as mentioned earlier, there is still the matter(s) of “Conicity” and “Ply Steer”.
These terms refer to tire construction features that cause lateral forces to develop in use; the
importance of these features has increased with the rise of radial tire popularity. Generally, these
features are considered to be flaws in tire manufacture or design. Of the two, “Conicity” seems to
be the most troublesome as “Ply Steer” effects generally are small and tend to go unnoticed in
normal applications. When extreme enough, both their effects on vehicle performance are
manifested in vehicle pull and drift.

The explanation for conicity force/torque is that the two sides of the contact patch are at
different radii from the axle, and so would travel forward at different rates as the tire rotates;
therefore the large radius side goes forward faster than the small radius side forcing a turn toward
the small radius side…

Figure 1.20 - TIRE CONICITY

25
“Bundorf Analysis” is a summation of the vehicle characteristics that govern its understeer balance. The
understeer characteristics are measured in units of degrees of yaw per “g” of lateral acceleration. This analysis
technique is named after GM automotive engineer R.Thomas Bundorf.

26 DOI: 10.13140/RG.2.2.12002.22728

You might also like