Downloaded from https://iranpaper.

ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

2024-01-2770 Published 09 Apr 2024

Revealing the Impact of Motorcycle Caster Angles on

Weave and Wobble: A Comprehensive Analysis through
Visualized Stability Testing
Hideki Sakai Kindai University

Citation: Sakai, H., “Revealing the Impact of Motorcycle Caster Angles on Weave and Wobble: A Comprehensive Analysis through
Visualized Stability Testing,” SAE Technical Paper 2024-01-2770, 2024, doi:10.4271/2024-01-2770.

Received: 01 Sep 2023 Revised: 29 Dec 2023 Accepted: 03 Jan 2024

Abstract
variable, as the steering damper adversely affects

S
ince it takes a long time to design motorcycles in handling. In addition to the caster angle, the design vari-
order to prevent weaving and wobbling, which are ables with the most significant effects in the regions for
self-excited oscillations, this paper aims to shorten changes in design variables are the trail and front tire
the design process. The reason for the extended duration radius, the moment of inertia of the front wheels, and
is that the weave and wobble strongly depend on ‘forward the relaxation length of the front wheels. As an illustration
speed’ when the design variables are altered. Therefore, of a design that utilizes these variables, this paper demon-
this paper focuses on speed and visualizes the region strates the elimination of self-excited oscillation at high
where self-excited oscillation does not occur for each speeds by adjusting the caster angle and trail.
design variable as a function of vehicle speed. The results Consequently, by concentrating on the caster angle’s
demonstrate that at high speeds, the caster angle and sweet spot where self-excited oscillation does not occur,
steering damper are the only design variables for which weaving and wobbling at high speeds can be eliminated
this region exists near the design value. Due to the in a short time. As a result, weaving and wobble at high
narrowness of these sweet spots, they become the most speeds can be eliminated in a short time by concentrating
crucial variables in the design. However, this paper solely on setting the sweet spot of the caster angle that does
considers the caster angle as the most important design not cause self-excited oscillation.

Introduction

T
FIGURE 1 Top five predicted positive changes in weave
he phenomenon of weaves and wobbles in motor- damping at 10 m/s (dots), 30 m/s (crosses), and 60 m/s
cycles [1, 2], referred to as ‘self-excited oscillation,’ (diamonds) for a 10% parameter increase compared to the
has been the subject of various studies for half a reference case (4).
century since Sharp described it in his equations of motion
in 1971 [3]. Consequently, one possesses sufficient knowl-
edge to address the oscillations. For example, the book
explores the phenomenon of the oscillations, its modeling,
and the mitigation of its effects through design modifica-
tions [4]. However, despite these insights, the absence of
a conclusive design approach to mitigate self-excited
oscillations can be attributed to the challenges inherent
in determining optimal values for the motorcycle’s design
variables. This challenge primarily arises from the signifi-
cant influence of ‘vehicle speed’ on both the magnitude
and direction of the ‘effect,’ as illustrated by the variable area of the de-excited oscillation expressed as a function
‘h’ in Figure 1. This means that the strategy to find the of ‘vehicle speed’. Utilizing this diagram, the paper identi-
optimal value requires swapping the ‘effect’ and ‘vehicle fies key design variables such as the caster angle and
speed’ shown in Figure 1. their optimal zone and provides examples of counter-
This paper proposes an approach to suppressing measures against self-excited oscillations that occur at
self-excited oscillations in motorcycle design, using the high speeds.
Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

2 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE

Vehicle Model p=
22 (M k f
2
+ Irz + Ifx sin2 ε + Ifz cos2 ε s + Mf kV )
This paper utilizes a motorcycle model linearized around i i + λi 
straight running [3]. The geometric parameters of this p= [ Mf hf k − Crxz + (Ifz − Ifx ) sin ε cos ε ]s 2 −  fy + ry  Vs
Rr 
23
model are presented in Figure 2. Upon Laplace trans-  Rf
forming the equations of motion for the Sharp model, ify
assuming all initial conditions to be zero, the following ( Mf ek + Ifz cos ε )s 2 −
p24 = sin ε Vs
Rf
matrix representation is obtained.
p25 = −lYf
AX = B (1)
where
p26 = bYr
 p11 p12 p13 p14 p15 p16 

 p21 p22 p23 p24 p25 p26 
 p31
= ( M h + M h) s
f f r

p p32 p33 p34 p35 p36 

A =  31 
 p41 p42 p43 p44 p45 p46  p=
32  (
 Mf hf k − Crxz + Ifz − Ifx sin ε cos ε  s )
p p52 p53 p54 p55 p56 
 51   i i + λi 
+  Mf hf + Mr h + fy + ry V
 p61 p62 p63 p64 p65 p66   R f Rr 

X t = v 1 r ϕ δ Yf Yr  p=
33 (M h f f
2
+ Mr h2 + Irx + Ifx cos2 ε + Ifz sin2 ε s 2 )
Bt = 0 0 0 τ 0 0 
(
− Mf hf + Mr h g )
ify
p34= ( M eh + I sin ε s 2 +) (
cos ε Vs + tZf − Mfeg )
(M + M ) s
f f fz
p11
= f r
Rf

p35 = 0
(
p12 = Mf ks + Mf + Mr V )
p36 = 0
p13
= ( M h + M h) s
f f r
2

p41 = Mfes
p14 = Mfes2
 i 
p42 = ( M ek + I
f fz )
cos ε s +  Mfe + fy sin ε  V
R
p15 = −Yf  f 
ify
p16 = −Yr
p43= ( M eh + I
f f fz sin ε s2 − ) Rf
(
cos ε Vs + tZf − Mfeg )

p21 = Mf ks ( )
p44 = Ifz + Mfe2 s2 + Ks + tZf − Mfeg sin ε ( )
p45 = tYf
FIGURE 2 Motorcycle Model.
p46 = 0

Cf 1
p51 = −
V

Cf 1
p51 = −
V

p53 = Cf2

tC
p54 = f 1 s + Cf 1 cos ε + Cf2 sin ε
V
Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE 3

FIGURE 3 Root loci plots of weave and wobble as a function

σf of V (3.6V= 1 to 300[km/h]); if any roots have non-negative real
p55 =
− s−1
V part, the self-excited oscillations occur.

p56 = 0

Cr1
p61 = −
V

bCr1
p62 =
V

p63 = Cr2

p64 = 0

other hand, wobbling occurs at higher speeds, i.e., the

p65 = 0
region indicated as ‘self-excited oscillation 2’ in the figure.
σ
p66 =
− r s−1
V
Non-Occurrence Region of
How to Find Self-Excited Self-Excited Vibrations for
Oscillations Each Design Variable
The presence of self-excited oscillations is determined by
the signs of the real parts of the solutions of the charac- How to Depict Areas Where Self-
teristic equation. The characteristic equation in equation Excited Vibrations Does Not
(1) is expressed as follows: Occur
A =0 Figure 4a illustrates the range, as a function of the vehicle
(2)
speed V, marked as ‘to be set,’ where self-excited oscil-
Among the solutions to the characteristic equation lation does not occur. By identifying ε values for which
solved for s, their conjugate complex roots signify oscil- the real part of certain conjugate complex roots becomes
lations. When the real parts of these roots are negative, zero for every 0.1[m/s] increase in V, I determined the
self-excited oscillations are absent as the excitation dimin- boundary between the zone where self-excited oscillation
ishes and ultimately reaches zero. On the other hand, if does not occur (indicated as ‘Area to be set’) and the
these real parts are positive, the oscillation’s amplitude region where it does occur.
intensifies over time and becomes unbounded [5]. This
constitutes self-excited oscillations. Hence, at the
boundary where the transition from the occurrence of Principle Design Variables of Front
self-excited oscillations to their absence happens, the real
parts of the roots are zero, resulting in imaginary roots.
Frame
Self-excited oscillations also occur in Equation (1) as Some of the design variables related to the front frame,
shown in Figure 3. This figure provides an illustration of as shown in Figure 4, have narrow “sweet spots” for the
how its roots vary across the complex plane with respect prevention of self-excited oscillations in the design. For
to vehicle speed. In this depiction, only the roots charac- instance, in Figure a, the caster angle, and in Figure h, the
terized by positive imaginary parts are displayed among steering damper, exhibit optimal zones where self-excited
the conjugate complex numbers. The roots of the weave oscillation does not occur near their nominal values.
produce self-excited oscillation because the real part is However, steering dampers can negatively impact
positive in the region from vehicle speed 0 to about 20 handling, making it necessary to primarily focus on the
[km/h], marked as ‘self-excited oscillation 1’ in the figure. optimal value for the caster angle.
However, this paper does not address such low-speed In Figure 4, the “effect” shown in Figure 1 is mani-
self-excited oscillation. At higher speeds, the real part is fested in the small slope of the boundary line near the
negative, preventing the occurrence of the weave. The design nominal value. The closer the slope is to horizontal,
real part of the Wobble root is negative up to about 200 the more pronounced the change in the no excited oscil-
[km/h], thus preventing self-excited oscillation. On the lations region—indicating higher sensitivity. The two
Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

4 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE

FIGURE 4 Stable zone of the design parameters of front frame.

Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE 5

FIGURE 5 Stable zone of the design parameters of rear frame.

variables closest to horizontal are the trail ‘t’ and the The length ‘a’ in Figure a and the height ‘h’ in Figure c
moment of inertia ‘Ifz’ of the front frame around the z-axis. represent the zones for the best and worst prevention
of self-excited oscillations, respectively, near the nominal
values. However, this effect is not as pronounced as that
Principle Design Variables of Rear of the caster angle. The slopes of the boundaries around
Frame the design nominal values for all the design variables
related to the rear frame are significantly steeper
As shown in Figure 5, the rear frame design variables do compared to those in Figure 4d.
not have a significant effect on self-excited oscillations.
Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

6 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE

Principle Design Variables of Tires Design Variables Effective in

Only the front tire significantly affects self-excited oscil- Preventing Self Excited Vibration
lations. All design variables for tires shown in Figure 6 Six design variables effective in preventing self-excited
lack an optimal value near the nominal value. The slopes oscillations were identified in the previous chapter. Among
of the boundaries near the design nominal values are them, only the caster angle ‘ε’ and the damping factor ‘K’
significantly smaller for variables such as ‘σf,’ ‘Rf ’ and ‘ify.’ of the steering damper have significant optimal zones near

FIGURE 6 Stable zone of the design parameters of tires.

Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE 7

FIGURE 7 Demonstration of setting the principal design

parameters based on Figure 4.
References
1. Doria, A., Favaron, V., Taraborrelli, L., and Roa, S.,
“Parametric Analysis of the Stability of a Bicycle Taking
into Account Geometrical, Mass and Compliance
Properties,” International Journal of Vehicle Design 75, no.
1-4 (2017).
2. Passigato, F., Eisele, A., Wisselmann, D., Gordner, A. et al.,
“Analysis of the Phenomena Causing Weave and Wobble
in Two-Wheelers,” Applied Sciences 10, no. 19 (2020).
3. Sharp, R., “The Stability and Control of Motorcycle,”
Journal of Mechanical Engineering Science 13, no. 5 (1971).
4. Cossalter, V., Motorcycle Vibration Modes and Stability
(London: Lulu.com, 2006), 241-282
5. Karnopp, D., “Automobile” in Vehicle Dynamics, Stability,
and Control (Boca Raton: CRC Press, 2016), 42-54
their nominal values. The parameters with boundary slopes 6. Foale, T. and Willoughby, V., Motorcycle Chassis Design:
closest to horizontal near the design nominal values are t The Theory and Practice (London: Osprey Publishing
for the front frame and ‘σf,’ ‘Rf ’ and ‘ify’ for the tire. Limited, 1984), 38

Demonstration of Symbols
Countermeasures against Self Cf1 - Front-wheel cornering stiffness (11.2[kN/rad])
Excited Vibrations Cr1 - Rear-wheel cornering stiffness (15.8[kN/rad])
By adjusting the effective design variables based on the Cf2 - Front-wheel camber stiffness (0.939[kN/rad])
stability regions shown in the figures, the self-excited Cr2 - Rear-wheel camber stiffness (1.33[kN/rad])
oscillation at high speed in Figure 4a can be eliminated.
Crxz - Rear frame inertia of product between front/rear
As a countermeasure against the oscillation only with one
axis and transverse axis (1.74[kgm2])
of the effective variables, the caster angle is changed to
the sweet spot in Figure 4a, whose root loci plot is shown Ifx - Front frame moment of inertia around front-rear axis
in Figure 7a. The oscillation at high speeds disappears, (1.23[kgm2])
but the stability margin of the weave seems to be too Ifz - Front frame moment of inertia around vertical axis
small. Therefore, as the other countermeasure due to the (0.442[kgm2])
combination of the two variables, the trail was increased Irx - Rear frame polar moment of inertia around front-rear
by 20% and the caster angle was varied with very little axis (31.2[kgm2])
trial and error, using its stability region as a reference. As
a result, some stability margin was secured, as shown in Irz - Rear frame polar moment of inertia around vertical
Figure 7b. In this case, caster is the “magic” 27deg [6]. axis (21.1[kgm2])
Consequently, by using visualized stable regions, designs K - Damping coefficient of steering damper (6.78[Nm/
to prevent self-excited oscillations can be effectively (rad/s)])
implemented. Mf - Mass of front frame (30.6[kg])
Mr - Mass of rear frame (218[kg])
Rf - Radius of front tire (0.305[m])
Conclusion Rr - Radius of rear tire (0.305[m])
V - vehicle speed [m/s]
This paper highlights that illustrating the region of non-
Yf - Side force of front-wheel [N]
excited oscillations as a function of vehicle speed simplifies
the countermeasures against self-excited oscillations, and Yr - Side force of rear-wheel [N]
further provides examples of its application. As a result, Zf - Vertical load on front wheels (-1005.3[N])
it has been clarified that the caster angle is the primary b - Horizontal distance from the center of gravity of the
design variable for mitigating self-excited oscillations. If rear frame to the rear wheel (0.480[m])
the caster angle cannot be changed, adjust the design
variables such as ‘t,’ ‘σf,’ ‘Rf,’ or ‘ify’ to match its sweet spot e - Distance from center of gravity of front frame to
to the caster angle. Setting the caster angle to its optimal steering axis (0.0244[m])
zone, either directly or indirectly, appears to be the key to f - Distance from center of gravity of front frame to point
self-excited oscillations countermeasures. B (0.0283[m])
Downloaded from https://iranpaper.ir
Downloaded from SAE International by UCLA Library - CDL, Thursday, July 04, 2024 https://www.tarjomano.com https://www.tarjomano.com

8 REVEALING THE IMPACT OF MOTORCYCLE CASTER ANGLES ON WEAVE AND WOBBLE

g - acceleration of gravity (9.81[m/s2]) s - Laplace variable

h - Height of rear frame center of gravity (0.616[m]) t - Trail (0.116[m])
i - Polar moment of inertia of the engine flywheel (iry+λi= v 1 - Sideslip velocity at point A
1.05[kgm2]) δ - Steering bar angle [rad]
Ifx - Polar moment of inertia of front frame around front- τ - Steering torque [Nm]
rear axis (1.23[kgm2])
ε - Caster angle (0.472[rad])
ify - Front wheel moment of inertia (0.719[kgm2])
λ - Gear ratio [-] (iry+λi=1.05[kgm2])
iry - Rear wheel moment of inertia (iry+λi=1.05[kgm2])
σf - Relaxation length of afront tire (0.244[m])
l - Horizontal distance from the center of gravity of the
σr - Relaxation length of a rear tire (0.244[m])
rear frame to the front wheels (0.935[m])
φ - Roll angle [rad]
r - Yaw velocity [rad/s]

© 2024 SAE International. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of SAE International.

Positions and opinions advanced in this work are those of the author(s) and not necessarily those of SAE International. Responsibility for the content of the work
lies solely with the author(s).

ISSN 0148-7191