Università degli Studi di Padova

Padua Research Archive - Institutional Repository

On the influence of tyre and structural properties on the stability of bicycles

Original Citation:

Availability:
This version is available at: 11577/3257371 since: 2018-02-15T14:52:23Z

Publisher:
Taylor and Francis Ltd.

Published version:
DOI: 10.1080/00423114.2017.1403032

Terms of use:
Open Access

This article is made available under terms and conditions applicable to Open Access Guidelines, as described at
http://www.unipd.it/download/file/fid/55401 (Italian only)

(Article begins on next page)

 Vehicle System Dynamics

Fo

On the influence of tire and structural properties on the

rP
stability of bicycles
ee
Journal: Vehicle System Dynamics

Manuscript ID Draft

Manuscript Type: Original Paper

rR

Date Submitted by the Author: n/a

Complete List of Authors: Doria, Alberto; University of Padova, Mechanical Engineering

Roa Melo, Sergio; Universidad de los Andes Facultad de Ingenieria,
ev

Mechanical Engineering Department

Keywords: Bicycle, Stability analysis, Measurement < Tyre dynamics

ie
w
On
ly

URL: http://mc.manuscriptcentral.com/nvsd
Page 1 of 38 Vehicle System Dynamics

1
2
3
4
5 On the influence of tire and structural properties on the stability of bicycles
6
7
8
9
A. Doria 1, S.D. Roa2
10
11
12 1
Department of Industrial Engineering
13
14 University of Padova
Via Venezia 1, 35131, Padova, Italy
Fo
15 2
16 Mechanical Engineering Department
17 Universidad de los Andes
18 Carrera 1 Este # 19A-40, Bogotá, Colombia
rP
19 [email protected]
20
21
22
ee
23
24 Corresponding Author:
25
Prof. Alberto Doria
26
rR

27 Department of Industrial Engineering

28 University of Padova
29 Via Venezia 1, 35131, Padova, Italy
30 e-mail: [email protected]
phone: +39 0498276803
ev

31
32 fax: +39 049 8277599
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 2 of 38

1
2
3
4
5 Abstract
6 In recent years the Whipple Carvallo Bicycle Model (WCBM) has been extended to analyze
7 high speed stability of bicycles. Various researchers have developed models taking into
8 account the effects of front frame compliance and tire properties, nonetheless, a systematic
9 analysis has not been yet carried out. This paper aims at analyzing parametrically the influence
10 of front frame compliance and tire properties on the open loop stability of bicycles. Some
11 indexes based on the eigenvalues of the dynamic system are defined to evaluate quantitatively
12 bicycle stability. The parametric analysis is carried out with a factorial design approach to
13 determine the most influential parameters. A commuting and a racing bicycle are considered
14 and numerical results show different effects of the various parameters on each bicycle. In the
Fo
15 commuting bicycle, the tire properties have greater influence than front frame compliance, and
16 the weave mode has the main effect on stability. Conversely, in the racing bicycle, the front
17 frame compliance parameters have greater influence than tire properties, and the wobble mode
18 has the main effect on stability.
rP
19
20
21 Keywords: Bicycle, stability, weave, capsize, wobble, tire, structural compliance.
22
ee
23
24 1. Introduction
25
26
rR

27 The bicycle is a very simple and smart mean of transport, but it has a characteristic that requires
28
29
30 some training of the rider and draws the attention of engineers and mathematicians: it can be
ev

31
32 unstable in certain conditions. The first mathematical model of a bicycle was developed at the end
33
34 of the 19th century, it is named the Whipple and Carvallo bicycle model (WCBM) from the names
ie

35
36 of the two scientists that independently developed the model [1]. This model is still in use for
37
w

38
39
studying the basic features of bicycle stability at low speed [2]. The WCBM consists of four rigid
40
On

41 bodies (front frame, rear frame and the wheels) connected by three rotational joints (steer and wheel
42
43 pivots) and is based on several assumptions: two disk wheels making point contact with the road, no
44
45
ly

structural compliance and slip of tires, no structural compliance of front and rear frame, stiff rider
46
47
stiffly connected to the rear frame. The WCBM has three velocity degrees of freedom (roll velocity
48
49
50 φ, steer velocity δ and rear wheel spin velocity θ ) and it is able to give insight into the modes of
51
52 vibration that characterize the low speed stability of the vehicle: the weave, capsize and castering
53
54 modes. The weave mode is a combination of steer and roll rotation that results in a snaking motion
55
of the bicycle that can be unstable below a certain speed, which is named weave speed v . The
56
57
58
59 capsize mode is dominated by the steer rotation and becomes unstable above a certain speed, which
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 3 of 38 Vehicle System Dynamics

1
2
3 is named capsize speed v . The castering mode, which is usually stable, is the typical mode of a
4
5 trailed wheel about the steer axis. An important result of stability analysis carried out with the
6
7 WCBM is that a bicycle without any control by the rider (hands off the handlebars) is characterized
8
9
10 by a self-stability speed range, in which all the above-mentioned modes of vibration are stable.
11
12 In recent years, research on bicycle dynamics continued both in the field of modeling and
13
14 simulation and in the field of experimentation. A large effort was made to extend the validity of
Fo
15
16 bicycle models to the high-speed range, which is characterized by the wobble mode. The wobble or
17
18
shimmy mode is a high frequency oscillation (3-6 Hz) of the front wheel about the steer axis, which
rP
19
20
21 can be found in motorcycles [3], landing gear of aircrafts [4] and racing bicycles [2]. It is very
22
ee
23 dangerous, because the rider is not able to control high frequency oscillations. Extended bicycle
24
25 models that hold true at high speed were presented in [5-7]. Some bicycle models including tire
26
rR

27 properties were presented in [8]. In these papers, the researchers highlighted the influence of tire
28
29
30 properties and compliances of structural elements on high speed dynamics and stability. For these
ev

31
32 reasons, experimental studies were carried out to identify the properties of bicycle tires and of
33
34 bicycle components (forks, wheels, frames).
ie

35
36 In [9], tires for racing bicycles were tested and results concerning lateral forces due to camber and
37
w

38
39
side-slip angles were presented; in [10], some tires for commuting bicycles were tested and results
40
On

41 dealing with both tire forces and moments were presented. Therefore, information about the ranges
42
43 of variation of tire properties is now available. Compliance properties of bicycle structural elements
44
45
ly

usually are taken into account in the models by introducing in the front frame a revolute joint
46
47
perpendicular to the steer axis, which defines the deformation axis of the front frame. A lumped
48
49
50 rotational spring and a lumped rotational damper represent stiffness and damping properties about
51
52 the deformation axis. These components represent in a simplified way the lateral displacement and
53
54 rotation of front wheel that are caused by the torsional compliance of rear frame, bending
55
56 compliance of front fork and wheel compliance due to spoke deformation. Limebeer and Sharp in
57
58
59
[2] gave typical ranges of the values of damping and stiffness of front frames. Values of damping
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 4 of 38

1
2
3 and stiffness were given by Klinger et al. as well [7]. Recently, Doria et al. [11] carried out series of
4
5 experimental tests to investigate the contribution of wheel, fork and chassis to bicycle compliance;
6
7 therefore, ranges of variation of compliance parameter of bicycle are now available.
8
9
10 A systematic analysis of the effects on stability of tire and structural properties is useful for
11
12 designing safe and manoeuvrable bicycles. Even if the two tires are assumed to have equal
13
14 properties and a linear tire model is adopted, at least six tire properties have to be considered:
Fo
15
16 cornering stiffness, camber stiffness, self-aligning stiffness, twisting stiffness, overturning stiffness
17
18
and relaxation length. Even if structural compliance is represented by means of the deformation
rP
19
20
21 axis, at least three parameters have to be considered: position of the deformation axis and values of
22
lumped stiffness and damping. In the previous studies a whole parametric analysis was not carried
ee
23
24
25 out. In [5] and [2] some results dealing with the effect of structural compliance on open loop
26
rR

27 stability (hands off the handlebars) were presented, but a fixed location of the deformation axis was
28
29
30 considered. In [6] some results regarding the effect of cornering stiffness on the open loop stability
ev

31
32 were presented and the effect of the position of the deformation axis was studied as well. The
33
34 effects on the open loop stability of parametric variations in cornering stiffness, relaxation length
ie

35
36 and camber stiffness were considered in [8], but different models were adopted for studying the
37
w

38
influence of the various properties, and the comparison with other results is difficult.
39
40
On

41 Results reported in [6] and [11] highlight that bicycles with different geometrical and inertial
42
43 properties may have different stability properties. Therefore, it is difficult to generalize the effects
44
45
ly

of tires and structural properties found in a particular bicycle to other bicycles. Different bicycles or
46
47 at least different types of bicycles have to be considered to perform a more systematic study.
48
49
50 Nowadays, there is a database of bicycle properties from experimental measurements carried out by
51
52 Moore, et.al. [12] that makes easier a parametric analysis of different bicycles.
53
54 This paper aims at analyzing the influence of tires and compliance properties on the open loop
55
56 stability of bicycles (without the control from the rider). Stability indexes are defined for evaluating
57
58
the self-stability condition determined by the weave, capsize and wobble modes. The indexes are
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 5 of 38 Vehicle System Dynamics

1
2
3 used for investigating the effect of tires and structural properties in a parametric analysis carried out
4
5 with a factorial design approach [13]. The ranges of variation of the properties are defined
6
7 according to the results of experimental tests reported in the literature. The influence of the various
8
9
10 properties is ranked making use of correlation coefficients with the stability indexes [14], and the
11
12 most influential properties are found. The analysis is performed for two bicycles: a commuting and
13
14 a racing bicycle, in order to understand the different stability properties of these types of bicycles.
Fo
15
16
17
18
2. Mathematical model
rP
19
20 The mathematical model for the stability analysis is based on the model presented by Klinger et.al
21
22 with rider hands off the handlebars [7]. The model consists of the linearized equations of motion of
ee
23
24 the bicycle represented by rigid bodies linked by rotational joints at the steer axis, deformation axis
25
26
and wheel pivots and a linear tire model. The mathematical model has 5 velocity degrees of
rR

27
28
29 freedom (DOF) corresponding to the lateral velocity of the rear contact point y , the yaw velocity
30
rotation around the vertical axis ψ, the roll velocity around the horizontal axis φ, the steering
ev

31
32
33
velocity δ and the deformation velocity about the deformation axis β (as shown in Fig. 1).
34
ie

35
36 The tire model represents the lateral tire forces, the self-aligning moments, the twisting torques and
37
w

38 the overturning moments as linear functions of the side-slip and camber angles.
39
40
=   +    =   +   
On

41   ;   (1)
42
43
44 In Eq. (1), F and F are the front and rear tire lateral forces, F and F are the front and rear
45
ly

46 vertical forces, α and α the front and rear side-slip angles, γ and γ are the front and rear camber
47
48 angles and C" and C# are the cornering and camber stiffness of both tires.
49
50
51 The yaw moments of the tires (about the axis perpendicular to the ground) include both the moment
52
53 due to side-slip (self-aligning moment) and the one due to camber (twisting torque) [10].
54
55
56 $ = −&  + &    ; $ = −&  + &    (2)
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 6 of 38

1
2
3 In Eq. (2), M and M are front and rear yaw moments, and C(" and C(# are self-aligning
4
5 stiffness and twisting stiffness.
6
7
8
The overturning moment (about the longitudinal axis parallel to the ground) is proportional to the
9
10 camber angle as shown in Eq. (3), where M) and M) are the front and rear overturning moments
11
12 and C() is the overturning moment stiffness.
13
14
$* = −&*  ; $* = −&* 
Fo
15   (3)
16
17
18 Some authors have shown that the transient tire behavior plays an important role in stability [2, 3].
rP
19
20 For including the transient phenomena and the delay in the development of tire forces, transient
21
22 side-slip angle α′ and transient camber angle γ′ are included in the model. The transient side-slip
ee
23
24 angle is described by Eq. (4) in which α ′ and α ′ are the transient side-slip angles of front and rear
25
26
tires respectively, and σ" is the relaxation length due to side-slip (the same for both tires).
rR

27
28
29
30 -. 0 -. 0

/ 
+ 0 =  ; 
/ 
+ 0 =  (4)
ev

31
32
33
34 The transient camber angle is described by Eq. (5) in which γ ′ and γ ′ are the transient camber
ie

35
36 angles of front and rear tires respectively, and σ# is the relaxation length due to camber (the same
37
w

38 for both tires), which in most cases is a small fraction of the relaxation length due to side-slip σ"
39
40
[3]. In the following, σ# is set to 0.1σ" .
On

41
42
43
44
-4 -4
0 + 0 =  ; 0 + 0 = 
(5)
45
ly

/ /
46
47
48 3. Analyzed bicycles
49
50
Since different types of bicycles can exhibit different stability features, the stability analysis is
51
52 focused on two types of bicycles: a commuting bicycle and a racing bicycle. The commuting
53
54 bicycle is the well-known benchmark bicycle presented by Meijaard et.al [1]. Despite the authors do
55
56 not explicitly mention the benchmark bicycle as a commuting bicycle, when comparing its
57
58 parameters with the set of parameters of different bicycles presented in [12], the benchmark bicycle
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 7 of 38 Vehicle System Dynamics

1
2
3 can be classified as a commuting bicycle. The racing bicycle is the Bianchi Pista presented in [12],
4
5 which is a track bicycle made of steel. A summary of the geometric and the inertial properties of
6
7 both bicycles is presented in Table 1.
8
9
10
11
12 4. Range of variation of tire properties
13 To carry out a parametric analysis of the effect of tire properties on stability, a low value, a high
14
Fo
15 value and a nominal value of each property have to be defined.
16
17
18 The ranges of the tire properties are based on the experimental results presented in [9] and [10]. The
rP
19
20 tires tested in [9] are racing tires (ERTRO XX-622 or equivalent 700C size). The data of cornering
21
22 and camber stiffness of three tires is taken into account: “DT1” (ERTRO 19-622), “DT2” (ERTRO
ee
23
24 23-622) and “DT3” (ERTRO XX-622 with unknown width). The tires tested in [10] are commuting
25
26
tires (ERTRO XX-622 or equivalent 700C size) with widths 35 and 37 mm. The data of cornering
rR

27
28
29 and camber stiffness, and of self-aligning and twisting stiffness of these tires is taken into account:
30
“T1” (ERTRO 37-622), “T2” (ERTRO 37-622) and “T3” (ERTRO 35-622).
ev

31
32
33 To find the linear properties of these tires, a fitting procedure based on the Magic Formula is
34
ie

35 performed [15]. Since the maximum side-slip and camber angles investigated during the tests are
36
37
w

38 not large enough to give information about the shape and curvature of the curves, the fitting is based
39
40 on a simplified version of Magic Formula. The simplified Magic Formula is presented in Eq. (6),
On

41
42 where X is a tire angle (side-slip or camber angle accordingly), Y(X) a tire force or torque
43
44 component (front or rear), F is the vertical force (front or rear), and B, C and D are the coefficients
45
ly

46
47 of the Magic Formula.
48
49 :(X)
50 = ; ∙ =>?[ ∙ ABA? (C ∙ X)] (6)

51
52
53
54 The curves of the normalized side-slip force (lateral force divided by the vertical force) of the six
55
56 tires are shown in Fig. 2a, and Table 2 presents the product BCD corresponding to the cornering
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 8 of 38

1
2
3 stiffness [15]. The curves of the normalized camber force of the six tires are shown in Fig. 2b, and
4
5 Table 2 presents the product BCD corresponding to the camber stiffness.
6
7 The curves of the normalized self-aligning moment (self-aligning moment divided by the vertical
8
9
10 force) of the three tires presented in [10] are shown in Fig. 2c, and the curves of the normalized
11
12 twisting moment (twisting moment divided by the vertical force) are shown in Fig. 2d. Table 3
13
14 presents the absolute value of the product BCD for the self-aligning torque stiffness and the twisting
Fo
15
16 moment stiffness.
17
18
For the best knowledge of the authors, there is no data in the literature about the overturning
rP
19
20
21 moment of bicycles tires. Therefore, the overturning moment stiffness (C() ) is assumed equivalent
22
ee
23 to the crown radius [16]. The crown radius is approximated as the half of the nominal tire width
24
25 given by the manufacturer. Table 2 summarizes the estimated crown radius of the various tires.
26
rR

27
28
Experimental values of cornering and camber stiffness come from two independent laboratories that
29
30 used different test machines. These values are compatible, even if the camber stiffness of racing
ev

31
32 tires resulted systematically smaller than the one of tires for commuting bicycles. For parametric
33
34 analysis, the nominal values of these properties are set equal to the mean values reported in Table 2.
ie

35
36 Cornering stiffness shows maximum positive and negative variations with respect to the mean value
37
w

38
39 of +24.7% and -26.7% respectively, whereas camber stiffness shows maximum variations of
40
On

41 +26.6% and -32.2%. For this reason, in the parametric analysis, the high values of cornering and
42
43 camber stiffness are set equal to 125% of the nominal values, and the low values are set equal to
44
45
ly

75% of the nominal value, see Table 4. The choice of equal ranges of variations of these parameters
46
47
48
minimizes the introduction of systematic errors caused by different ranges of variation of the
49
50 various parameters in the correlation analysis [14].
51
52 Values of self-aligning moment and twisting torque come from the tests reported in [10]. Table 3
53
54 show rather small variations with respect to the mean value. The nominal values of self-aligning
55
56 stiffness and twisting stiffness are set equal to the mean values. In order to avoid underestimating
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 9 of 38 Vehicle System Dynamics

1
2
3 the effects of yaw torque components on bicycle stability, the ranges of variations around the
4
5 nominal value are set equal to ± 25% for self-aligning and twisting torque stiffness as well.
6
7 Table 2 shows that the maximum variations in overturning moment stiffness are +22.5% and -
8
9
10 37.1%, hence, also for this parameter variations of ±25% with respect to the mean (nominal) value
11
12 are considered.
13
14 The range of the relaxation length due to side-slip σ" is set according to the model presented in
Fo
15
16 [17]. Experimental results referring to the tires of high wheel scooters (which are similar to bicycle
17
18
tires) showed that the relaxation length in normal tire operations can be approximated as two times
rP
19
20
21 the relaxation length due to the contact patch σ (Eq. (7)). The value of σ can be calculated as in
22
Eq. (8), in which a is the half length of the contact patch and m is the ratio of the adhesion zone
ee
23
24
25 length to the total length of the contact patch (Eq. (9)). C" is the cornering stiffness, μ is the tire
26
rR

friction coefficient, F is the vertical load and α is the side-slip angle. For normal tire operation
27
28
29
30 conditions (i.e. μ~1, α~2°), the second term in the right-hand side in Eq. (9) becomes negligible
ev

31
32 and m ≅ 1. In this way, for bicycle tires similar to those presented in [9] and [10], the relaxation
33
34 length can be approximated as the contact patch length.
ie

35
36
37 The contact patch length of tire “T2” measured in [10] was about 0.1 m in nominal operation
w

38
39 conditions (inflation pressure 4 bar, F = 400 N). In the framework of this research the contact
40
On

41 patch length of a racing tire (size ERTRO 23-622) was measured. A length of 0.08 m was measured
42
43 with inflation pressure 7 bar and F = 400 N . The values of relaxation length of side-slip force
44
45
ly

46
reported in [9] are in the range 0.04-0.06 m. For the above-mentioned reasons, the nominal value of
47
48 σ" is set to 0.075 m and the range 0.05-0.09 m is adopted, which corresponds to variations of ±25%
49
50 with respect to the nominal value.
51
52
53 P ≅ 2PQ (7)
54
55
56
57 PQ = AR (8)
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 10 of 38

1
2
3 
R =1− BA?()
(9)
4
3T 
5
6
7 5. Range of variation of structural properties
8
The range of front assembly stiffness k V is basically defined according to the values presented in
9
10
11 [11], which were identified by means of experimental tests on the wheel, the fork and the frame. In
12
13 addition, the values presented in [2] and [7] are taken into account. The range of front assembly
14
damping cV is defined according to the values reported in [2] and [7]. Finally, the range of the
Fo
15
16
17
18 height of the deformation axis (hY ) is defined according to the value reported in [11], which
rP
19
20 presents a calculation based on an equivalent 1-DOF front assembly model taking into account the
21
22 stiffness of the wheel, fork and frame. Table 4 presents the range of variation of the tire and
ee
23
24 structural properties. It should be noted that the high value is 125% the nominal value, whereas the
25
26
low value is 75% the nominal value.
rR

27
28
29
30
ev

31 6. Stability behavior of the two bicycles

32 The stability behavior of the two bicycles is analyzed using the nominal values of tire and structural
33
34
properties (Table 4). Figure 3a and Fig. 3b show the real parts of the eigenvalues as functions of the
ie

35
36
37 longitudinal speed for the commuting bicycle and the racing bicycle respectively. Modes of
w

38
39 vibration with negative real parts of the eigenvalues correspond to asymptotically stable motions,
40
On

41 whereas modes with positive real parts of the eigenvalues correspond to unstable motions. There are
42
43
noticeable differences between the two bicycles. In the commuting bicycle, the weave mode is
44
45
ly

46 unstable at low speed, becomes stable at 7.7 m/s and remains stable up to 24.4 m/s, then it becomes
47
48 unstable again. In the racing bicycle, the weave mode is unstable at low speed and becomes stable
49
50 at 9.2 m/s remaining stable for the whole speed range here analyzed. In both bicycles, the capsize
51
52 mode remains stable; nonetheless, it is possible to note that the real negative eigenvalues of the
53
54
55 racing bicycle are larger (in modulus) than the ones of the commuting bicycle. This result suggests
56
57 that the capsize mode in the racing bicycle is more damped than in the commuting bicycle and
58
59 therefore it is more stable with benefits in terms of controllability [2, 5]. In both bicycles, the
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 11 of 38 Vehicle System Dynamics

1
2
3 wobble mode is stable, but the wobble mode of the racing bicycle is less damped than the wobble
4
5 mode of the commuting bicycle.
6
7 For comparison purposes, Fig. 3c and Fig. 3d present the stability features of the commuting and
8
9
10 the racing bicycles calculated by means of the WCBM. Some remarks can be made.
11
12 First, in the WCBM the wobble mode does not appear because front-assembly compliance and tire
13
14 model are not included. Second, the WCBM shows similar results for both bicycles especially at
Fo
15
16 high speed, whereas large differences between the two bicycles are found with the extended model.
17
18
Third, when using the WCBM, the damping of the weave mode of the commuting bicycle increases
rP
19
20
21 as speed increases; conversely, when using the extended model, the damping of the weave mode of
22
the commuting bicycle tends to decrease at high speed. This feature is also present when only front-
ee
23
24
25 assembly compliance is included in the model [11]. Finally, the capsize mode becomes unstable
26
rR

27 using the WCBM while it is always stable using the extended model.
28
29
30
ev

31
32 7. Definition of stability indexes
33 To evaluate the stability features, five indexes are defined. The indexes are based on the real part of
34
ie

35 the eigenvalues of the weave, capsize and wobble modes. The concept of stability area is introduced
36
37
w

38 as the area formed by the real part of an eigenvalue curve and the speed axis, when the real part is
39
40 negative. The magnitude of the stability area quantifies the damping of the mode in the stability
On

41
42 range. Therefore, a large stability area indicates good behavior in terms of stability and
43
44 controllability. The first three indexes are the weave stability area, the capsize stability area and the
45
ly

46
47
wobble stability area. The fourth index is the self-stability area, which corresponds to the
48
49 intersection of the stability areas of the three modes. The self-stability area quantifies the damping
50
51 of the system when the bicycle is in the self-stable condition (all the modes are stable). Figure 4 and
52
53 Fig. 5 present the stability areas for the commuting bicycle and the racing bicycle respectively,
54
55 considering the nominal parameters of Table 4. The fifth index is the self-stability speed range,
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 12 of 38

1
2
3 which corresponds to the range of speeds, between the weave speed v and the capsize speed v , in
4
5 which the bicycle is in the self-stable condition.
6
7 Table 5 presents the values of the areas and the speed range showed in Fig. 4 and Fig. 5. The areas
8
9
10 are calculated performing a numerical integration.
11
12
13
14 8. Numerical results
Fo
15
The parametric analysis is performed using a factorial design approach with nine factors (six tire
16
17
18 properties and three structural properties), each factor has two levels of variation (low and high
rP
19
20 values shown in Table 4). Therefore, 2Z = 512 combinations are evaluated and the five stability
21
22 indexes are calculated for each combination. The first approach is an analysis of variance
ee
23
24 (ANOVA) using the self-stability area index as the response variable. The function “anovan” of
25
26
rR

27 MATLAB was used for performing the ANOVA. Table 6 summarizes the results of the ANOVA
28
29 for the commuting and the racing bicycles. The sum of squares of the main effects (the variability
30
ev

31 from the nine factors) is shown in the first row, the sum of squares of the non-additivity residuals
32
33 (the variability from all the interactions up to ninth order) is shown in the second row, and the total
34
ie

35
sum of squares (the measure of overall variability) is shown in the third row. From these results, it
36
37
w

38 is possible to conclude that, even if there is a variability from factors interactions, most of the
39
40 variability is due to the main effects. For the commuting bicycle, 69.8% of the variability in the
On

41
42 self-stability area is due to the main effects, whereas for the racing bicycle, 91.2% of the variability
43
44 is due to main effects.
45
ly

46
47 A similar analysis was performed using the self-stability speed range as the response variable. In
48
49 this case 83.8% and 73.0% of the variability was due to the main effects for the commuting and the
50
51 racing bicycle respectively.
52
53 Since most of the variability in the stability behavior is due to the variation of the main effects (nine
54
55
factors), a linear statistical model without interactions can be adopted for the analysis of results.
56
57
58 Outputs of the parametric simulations are analyzed using linear correlation coefficients (Pearson
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 13 of 38 Vehicle System Dynamics

1
2
3 correlation coefficients) between each stability index (I) and each factor (F). The correlation
4
5 coefficients ρ^ (Eq. 12) are obtained calculating for all the factorial combinations: the variance of
6
each index (σ_^ ) and the variance of each factor (σ_ ) as shown in Eq. 10; and the covariance
7
8
9
10 between each factor and each index (σ^ ) as shown in Eq. 11. The correlation coefficients give
11
12 information about the linear relation between two variables and therefore can be used to quantify
13
14 the influence of a factor on a stability index. A correlation coefficient with large magnitude (up to
Fo
15
16
one) indicates a strong influence of the factor on that index. A positive correlation coefficient
17
18
rP
19 indicates a stabilizing effect while a negative correlation coefficient indicates a destabilizing effect.
20
21 Figure 6 shows the correlation coefficients for all the factors and stability indexes of both bicycles.
22
ee
23 e e
1 1
= a(bc − b )̅ _ ; P_ = a( − i )_
24
P`_ c (10)
25 ? ?
cfg cfg
26
rR

27
28
29 e
1
P` = a(bc − b )̅ ( − i)
30
c (11)
?
ev

31
cfg
32
33
34
ie

35 P`
j` =
P` P
36 (12)
37
w

38
39
40 The correlation coefficients of the stability area of the weave mode exhibit different trends in the
On

41
42 two bicycles; principally, cornering and camber stiffness have opposite effects and significant
43
44
differences in magnitude. This behavior mainly takes place because the weave mode becomes
45
ly

46
47 unstable at around 25 m/s for the commuting bicycle while for the racing bicycle it becomes
48
49 unstable at higher speed (not shown in the analyzed speed range). Figure 7a shows the behavior of
50
51 the weave mode when only C" is varied while the other parameters are fixed at the nominal values.
52
53 It can be seen that, if C" increases, the stability area of weave increases for the commuting bicycle
54
55
56 and decreases for the racing bicycle. Figure 7b shows the behavior of weave when only C# is
57
58 varied while the other parameters are fixed at the nominal values. If C# increases, for the
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 14 of 38

1
2
3 commuting bicycle there is a negligible variation in the stability area of weave, whereas for the
4
5 racing bicycle there is an appreciable and negative variation. The arrows in Figure 7 show the effect
6
7 of increasing the magnitude of the corresponding parameter.
8
9
10 On the one hand, for the commuting bicycle twisting torque & , cornering stiffness  and front-
11
12 assembly stiffness kl have the largest effect on weave; increases in  and kl have a stabilizing
13
14 effect whereas increases in & have a destabilizing effect. On the other hand, for the racing bicycle
Fo
15
16
17 twisting torque & , camber stiffness  , cornering stiffness  and front-assembly stiffness kl
18
rP
19 have the largest effect on weave, the first three with a destabilizing effect and the last with a
20
21 stabilizing effect.
22
ee
23
24 The correlation coefficients of the stability area of the capsize mode exhibit similar trends in both
25
26 bicycles; nonetheless, for the commuting bicycle, there is a stronger influence of some tire
rR

27
28 properties than for the racing bicycle. The twisting torque C(# , the camber stiffness C# and the
29
30
cornering stiffness C" have the largest effect on capsize. There is a stabilizing effect when
ev

31
32
33 increasing C(# and C# and a destabilizing effect when increasing C" .
34
ie

35 The correlation coefficients of the stability area of the wobble mode exhibit a similar trend in both
36
37
w

bicycles. In general the structural parameters have a larger effect on the wobble mode than the tire
38
39
40 properties. The most influential parameters are front-assembly damping cV and deformation-axis
On

41
42 height hY . Figure 8 shows the wobble behavior when these parameters are varied one at a time. The
43
44 general effect is a displacement of the curve of the real part of wobble, which modifies damping.
45
ly

46
Front-assembly damping cV has a stabilizing effect whereas hY has a destabilizing effect.
47
48
49 The correlation coefficients of the self-stability speed range exhibit some differences between the
50
51 two bicycles; the greatest difference is in the cornering stiffness C" coefficient. This difference can
52
53 be explained taking into account the different influence of this tire property on the weave stability,
54
55
56
which is shown in Fig. 7a. In general, for the commuting bicycle, the self-stability speed range is
57
58 mainly influenced by the weave behavior while for the racing bicycle it is mainly influenced by the
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 15 of 38 Vehicle System Dynamics

1
2
3 wobble behavior. It is worth highlighting that considering the 512 combinations of parameters, the
4
5 wobble mode becomes unstable only in six combinations in the commuting bicycle while it
6
7 becomes unstable in 210 combinations in the racing bicycle. For the commuting bicycle, the
8
9
10 parameters that chiefly influence the self-stability speed range are the twisting torque (destabilizing
11
12 effect), the cornering stiffness (stabilizing effect) and the front-assembly stiffness (stabilizing
13
14 effect). For the racing bicycle, the most influential parameters are twist-axis height hY
Fo
15
16 (destabilizing) and front-assembly damping cV (stabilizing).
17
18
rP
19 The correlation coefficients of the self-stability area show some differences between the two
20
21 bicycles, which are is mainly due to the different behaviors of the weave and wobble modes. In the
22
ee
23 commuting bicycle, the weave mode strongly influences the self-stability area while the wobble
24
25 mode has no influence in many cases. In the racing bicycle, the wobble mode strongly influences
26
rR

27
28 the self-stability area since it is less damped and it is unstable in many cases. On the one hand in the
29
30 commuting bicycle, the parameters that mainly influence the self-stability area are camber stiffness
ev

31
32 C#, front-assembly stiffness k V and self-aligning moment C(" . This behavior is related to the
33
34
weave and the capsize modes. On the other hand in the racing bicycle, the parameters that mainly
ie

35
36
37 influence the self-stability area are deformation axis height hY , cornering stiffness C" and front-
w

38
39 assembly damping cV , which are the parameters that chiefly influence the wobble mode of this
40
On

41 bicycle. Figure 9 shows the variation in the self-stability area when only deformation axis height
42
hY is varied: the variation for the commuting bicycle is 2.1% while for the racing bicycle it is
43
44
45
ly

46 84.4%.
47
48 For the best knowledge of the authors, few researchers have performed an analysis of the influence
49
50 of tire properties and front-assembly compliance on bicycle stability. The paper presented by
51
52
Bulsink et. al. [18] analyzed numerically the influence of tire properties on the weave and capsize
53
54
55 modes of a regular bicycle with low entry; results concerning the effect of tire properties on the
56
57 weave speed agree with the results here presented for the racing bicycle. Bulsink et. al. concluded
58
59 that the main contribution to weave speed variation is the twisting torque C(# (with a destabilizing
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 16 of 38

1
2
3 effect corresponding to an increment in the weave speed), this result is confirmed in the present
4
5 paper. Coming to the capsize mode, Bulsink et.al. stated that a tire model and a passive rider
6
7 damping keep the capsize mode always stable, also this result is confirmed in the present paper.
8
9
10 The paper presented by Plöchl et.al [6] refers to a trekking bicycle and shows that the increase in
11
12 the cornering stiffness of front wheel has a large negative effect on wobble stability and negligible
13
14 effects on weave and capsize stability. The effects on wobble and capsize are in agreement with the
Fo
15
16 ones found in the present paper. The same authors stated that an increase in hY leads to a decrease
17
18
in wobble stability, also this result agrees with the ones found in the present paper.
rP
19
20
21 Several authors reported results of parametric analyses of the effect of tire properties on motorcycle
22
ee
23 stability [3, 16, 19], but the comparison with the results of the present paper is rather difficult, since
24
25 front and rear tires are different in motorcycles and the effects of front and rear tire properties were
26
rR

27 considered separately. Only it is worth highlighting that the large negative effect of front tire
28
29
30 cornering stiffness on wobble stability described in [3] is in agreement with the results of Figure 6.
ev

31
32
33
34 9. Conclusions
ie

35
The influence of some parameters on the stability of a bicycle is investigated. The analysis is
36
37
w

38 performed using a mathematical model previously presented in the literature. The model consists of
39
40 the linearized equations of motion of the bicycle represented by rigid bodies linked by rotational
On

41
42 joints and a linear tire model. Compliance properties of bicycle structural elements are taken into
43
44 account in the model by introducing in the front frame a deformation axis perpendicular to the steer
45
ly

46
47
axis.
48
49 The stability of the system is analyzed by solving the complex eigenvalue problem. The
50
51 methodology of analysis consists of a series of parametric simulations varying the tire and structural
52
53 properties with a factorial design approach. The range of variation of the parameters is chosen
54
55
considering experimental and theoretical data presented in the literature. In order to quantify the
56
57
58 effect of the variations in the parameters, the area formed by the real part of the eigenvalues and the
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 17 of 38 Vehicle System Dynamics

1
2
3 speed axis is used as an index of the stability of the system. Correlation coefficients are used to
4
5 identify the parameters that have the largest influence on stability.
6
7 The stability area is a quantitative index that captures the effect of parametric variations in the
8
9
10 bicycle’s properties. The stability area is related to the damping of a vibration mode, and can be
11
12 correlated to the degree of controllability. Two bicycles of different types (commuting and racing)
13
14 are analyzed in order to highlight the differences in the nominal behavior and to study the most
Fo
15
16 influential parameters in each case. Many differences in the stability features of the two bicycles are
17
18
found. The differences are related to the different geometric and mass properties of the bicycles,
rP
19
20
21 which depend on the particular purpose of each bicycle. Therefore, a generalization of the effect of
22
the front-assembly compliance and tire properties is rather difficult.
ee
23
24
25 For the weave mode, front-assembly stiffness, tire cornering stiffness and twisting stiffness are the
26
rR

27 most influential parameters. The effects of these parameters on the weave mode are different in the
28
29
30 two bicycles. For the capsize mode, tire properties produce larger variations than the structural
ev

31
32 properties. Camber stiffness and twisting stiffness are the most influential parameters, and their
33
34 effects on the capsize mode are similar in both bicycles. For the wobble mode, deformation axis
ie

35
36 height, cornering stiffness, relaxation length due to side-slip and front-assembly damping are the
37
w

38
most influential parameters. The effects on the wobble mode of these parameters are similar in both
39
40
On

41 bicycles; nonetheless, it was found that this mode is more damped in the commuting bicycle. The
42
43 self-stability area summarizes the stability properties. There are significant differences between the
44
45
ly

two bicycles here considered. The commuting bicycle is mainly influenced by the tire properties
46
47 and by the behavior of weave while the racing bicycle is mainly influenced by the front-assembly
48
49
50 compliance and by the behavior of wobble.
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 18 of 38

1
2
3 References
4 [1] Meijaard JP, Papadopoulos JM, Ruina A, et al. Linearized dynamics equations for the balance
5 and steer of a bicycle: a benchmark and review. Proceedings of the Royal Society A:
6
Mathematical, Physical and Engineering Science. 2007; 463(2084):1955-82.
7
8 [2] Limebeer DJN, Sharp RS. Bicycles, motorcycles, and models. IEEE Control Systems. 2006;
9 26(5):34–61.
10 [3] Cossalter V. Motorcycle Dynamics. Padova: Lulu.com; 2007.
11 [4] Thota P, Krauskopf B, Lowenberg M. Interaction of torsion and lateral bending in aircraft nose
12 landing gear shimmy. Nonlinear Dyn. 2009; 57: 455–467.
13 [5] Sharp RS. On the Stability and Control of the Bicycle. Applied Mechanics Reviews. 2008;
14 61(6): 60803.
Fo
15 [6] Plöchl M, Edelmann J, Angrosch B, et al. On the wobble mode of a bicycle. Vehicle System
16
Dynamics. 2012; 50(3): 415–429.
17
18 [7] Klinger F, Nusime J, Edelmann, J, et al. Wobble of a racing bicycle with a rider hands on and
rP
19 hands off the handlebar. Vehicle System Dynamics. 2014; 52(sup1): 51-68.
20 [8] Souh B. Influence of tire side forces on bicycle self-stability. Journal of Mechanical Science and
21 Technology. 2015; 29 (8): 3131-3140
22 [9] Dressel A, Rahman A. Measuring sideslip and camber characteristics of bicycle tires. Vehicle
ee
23 System Dynamics. 2012; 50 (8): 1365-1378.
24 [10] Doria A, Tognazzo M, Cusimano G, et al. Identification of the mechanical properties of
25 bicycle tyres for modelling of bicycle dynamics. Vehicle System Dynamics. 2012; 51(3):
26
405-420.
rR

27
28 [11] Doria A, Favaron V, Roa S. A DOE approach for evaluating the effect of bicycle properties on
29 stability. In: 19th International Conference on Advanced Vehicle Technologies AVT,
30 ASME International Design Engineering Technical Conferences; 2017 Aug 6-9; Cleveland,
OH. Paper 67225 to appear.
ev

31
32 [12] Moore JK, Hubbard M, Schwab A, et al. Accurate Measurement of Bicycle Parameters. In:
33 Proceedings, Bicycle and Motorcycle Dynamics 2010 Symposium on the Dynamics and
34 Control of Single Track Vehicles; 2010 Oct 20–22; Delft, The Netherlands.
ie

35 [13] Montgomery DC. Design and Analysis of Experiments. Hoboke, NJ: John Wiley & Sons;
36
37
1997.
w

38 [14] Gravetter FJ, Wallnau LB. Statistics for the behavioral sciences. [place unknown]: Cengage
39 Learning; 2016.
40 [15] Pacejka HB, Bakker E. The magic formula tyre model. Vehicle system dynamics. 1992;
On

41 21(S1): 1-18.
42 [16] Pacejka HB. Tire and vehicle dynamics. [place unknown]: Elsevier; 2005.
43 [17] Doria A, Taraborrelli L. Out-of-plane vibrations and relaxation length of the tyres for single-
44 track vehicles. Proc IMechE Part D: J Automobile Engineering. 2016; 30(5): 609–622.
45
ly

[18] Bulsink VE, Doria A, Belt vA, et al. The Effect of Tyre and Rider Properties on the Stability of
46
47
a Bicycle. Adv. Mech. Eng. 2015; 7(12): 1-19.
48 [19] Cossalter V, Doria A, Formentini M, et al. Experimental and numerical analysis of the
49 influence of tyre's properties on the straight running stability of a sport touring motorcycle.
50 Vehicle System Dynamics. 2012; 50(3): 357-375.
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 19 of 38 Vehicle System Dynamics

1
2
3 List of symbols
4
5
6 C Coefficient of magic formula. Stiffness factor [−]
7
8  Coefficient of magic formula. Shape of the curve [−]
9 ; Coefficient of magic formula. Peak value [−]
10
11  Cornering stiffness [−]
12
13  Camber stiffness [−]
Overturning moment stiffness [m]
14
&*
Fo
15
16 & Self-aligning stiffness [m]
17
18 & Twisting stiffness [m]
rP
19
20  Front tire lateral force [N]
Rear tire lateral force [N]
21
22 

Front tire vertical force [N]

ee
23

24
25  Rear tire vertical force [N]
26
bm Main body inertia [kg-m_ ]
rR

27
28 b Front wheel inertia [kg-m_ ]
29
30 bn Front-assembly inertia [kg-m_ ]
ev

31
32 b Rear wheel inertia [kg-m_ ]
33
$* Front tire overturning moment [N-m]
34
ie

35 $* Rear tire overturning moment [N-m]

36
37 $ Front tire yaw moment [N-m]
w

38
$ Rear tire yaw moment [N-m]
39
40 o Tire side-slip or camber angle (accordingly) [rad]
On

41
42 : Tire force or moment (accordingly) [N or N-m]
43
A Half-length of contact patch [m]
44
45 p
ly

Trail [m]
46
47 pl Front-assembly damping [N-m-s/rad]
48
49 ℎr Front-assembly deformation axis height [m]
50 kl Front-assembly stiffness [N-m/rad]
51
52 R Ratio of the adhesion zone length to the total length of the contact patch [-]
53
54 Rm Main body mass [kg]
55 R Front wheel mass [kg]
56
57 Rn Front-assembly mass [kg]
58
59 R Rear wheel mass [kg]
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 20 of 38

1
2
3 s Front wheel radius [m]
4
5 s Rear wheel radius [m]
6 t Bicycle-cyclist set longitudinal speed [m/s]
7
8 tQ Capsize speed [m/s]
9
10 tr Weave speed [m/s]
11 u Wheelbase [m]
12
13 vm Horizontal position of center of mass of main body [m]
14
vn Horizontal position of center of mass of front-assembly [m]
Fo
15
16 w Lateral velocity of the rear tire contact point [m/s]
17
18 xm Vertical position of center of mass of main body [m]
rP
19
20 xn Vertical position of center of mass of front-assembly [m]
21  Front tire side-slip angle [rad]
22
 Rear tire side-slip angle [rad]
ee
23
24
25 0 Front tire transient side-slip angle [rad]
26 0 Rear tire transient side-slip angle [rad]
rR

27
28 { Rotation velocity around the front-assembly deformation axis [rad/s]
29
30  Front tire camber angle [rad]
Rear tire camber angle [rad]
ev

31

32
33 0 Front tire transient camber angle [rad]
34
0 Rear tire transient camber angle [rad]
ie

35
~ Steering velocity [rad/s]
36
37
w

38  Head angle [rad]

39
40 T Tire friction coefficient €– ‚
On

41
42 j` Correlation coefficient between the stability index I and the factor F
43
44 P_ Variance of the factor F
45 P`_ Variance of the stability index I
ly

46
47 P` Covariance between the stability index I and the factor F
48
49 PQ Relaxation length due to the contact patch [m]
50
P Relaxation length due to side-slip [m]
51
52 P Relaxation length due to camber [m]
53
54 ƒ Roll velocity around the horizontal axis [rad/s]
„
55
56 Yaw velocity around the vertical axis [rad/s]
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 21 of 38 Vehicle System Dynamics

1
2
3
4
5
6
7 Captions of figures
8
9
10 Figure 1: 5 DOF bicycle model.
11
12 Figure 2. Properties of the analyzed tires. (a) Normalized side-slip force (b) Normalized camber
13
14 force (c) Normalized self-aligning moment (d) Normalized twisting torque.
Fo
15
16 Figure 3. Real part of the eigenvalues. (a) Commuting bicycle and (b) racing bicycle using the
17
extended model previously presented. (c) Commuting bicycle and (d) racing bicycle using the
18
rP
19 benchmark model presented in [1].
20
21 Figure 4. Stability indexes for the commuting bicycle using the nominal parameters. (a) Stability
22 area of weave mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-
ee
23
stability speed range and self-stability area.
24
25
26 Figure 5. Stability indexes for the racing bicycle using the nominal parameters. (a) Stability area of
rR

27 weave mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-stability
28 speed range and self-stability area.
29
30
Figure 6. Correlation coefficients of the five stability indexes. (a) Commuting bicycle, (b) racing
ev

31
32 bicycle.
33
34 Figure 7. Weave mode behavior when varying only (a) cornering stiffness and (b) Camber
ie

35
36 stiffness.
37
w

38 Figure 8. Wobble mode behavior when varying only (a) front-assembly damping and (b)
39
deformation axis height.
40
On

41
42 Figure 9. Self-stability area of (a) commuting bicycle and (b) racing bicycle when varying only the
43 deformation axis height.
44
45
ly

46
47 Captions of tables
48
49
50 Table 1. Parameters of the analyzed bicycles.
51
52 Table 2. Cornering stiffness, camber stiffness and estimated overturning stiffness of the analyzed
53
tires.
54
55
56 Table 3. Self-analyzing stiffness and twisting stiffness of commuting tires.
57
58 Table 4. Range of tire properties and compliance parameters.
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 22 of 38

1
2
3 Table 5. Stability indexes for the commuting and racing bicycles in nominal conditions.
4
5 Table 6. Summarized results of ANOVA.
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 23 of 38 Vehicle System Dynamics

1
2
3
4 Captions of figures
5
6
Figure 1: 5 DOF bicycle model.
7
8
9 Figure 2. Properties of the analyzed tires. (a) Normalized side-slip force (b) Normalized camber
10 force (c) Normalized self-aligning moment (d) Normalized twisting torque.
11
12
13 Figure 3. Real part of the eigenvalues. (a) Commuting bicycle and (b) racing bicycle using the
14 extended model previously presented. (c) Commuting bicycle and (d) racing bicycle using the
Fo
15
benchmark model presented in [1].
16
17
18 Figure 4. Stability indexes for the commuting bicycle using the nominal parameters. (a) Stability
rP
19 area of weave mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-
20 stability speed range and self-stability area.
21
22
Figure 5. Stability indexes for the racing bicycle using the nominal parameters. (a) Stability area of
ee
23
24 weave mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-stability
25 speed range and self-stability area.
26
rR

27 Figure 6. Correlation coefficients of the five stability indexes. (a) Commuting bicycle, (b) racing
28
bicycle.
29
30
ev

31 Figure 7. Weave mode behavior when varying only (a) cornering stiffness and (b) Camber
32 stiffness.
33
34
ie

35 Figure 8. Wobble mode behavior when varying only (a) front-assembly damping and (b)
36 deformation axis height.
37
w

38
39 Figure 9. Self-stability area of (a) commuting bicycle and (b) racing bicycle when varying only the
40 deformation axis height.
On

41
42
43
44 Captions of tables
45
ly

46
47 Table 1. Parameters of the analyzed bicycles.
48
49 Table 2. Cornering stiffness, camber stiffness and estimated overturning stiffness of the analyzed
50 tires.
51
52 Table 3. Self-analyzing stiffness and twisting stiffness of commuting tires.
53
54 Table 4. Range of tire properties and compliance parameters.
55
56
Table 5. Stability indexes for the commuting and racing bicycles in nominal conditions.
57
58
59
Table 6. Summarized results of ANOVA.
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 24 of 38

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39 Figure 1: 5 DOF bicycle model.
40 230x219mm (96 x 96 DPI)
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 25 of 38 Vehicle System Dynamics

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28 Figure 2. Properties of the analyzed tires. (a) Normalized side-slip force (b) Normalized camber force (c)
29 Normalized self-aligning moment (d) Normalized twisting torque.
30
277x174mm (96 x 96 DPI)
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 26 of 38

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
Figure 3. Real part of the eigenvalues. (a) Commuting bicycle and (b) racing bicycle using the extended
ev

31
32 model previously presented. (c) Commuting bicycle and (d) racing bicycle using the benchmark model
presented in [1].
33
34 426x302mm (96 x 96 DPI)
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 27 of 38 Vehicle System Dynamics

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32 Figure 4. Stability indexes for the commuting bicycle using the nominal parameters. (a) Stability area of
33 weave mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-stability speed
34 range and self-stability area.
ie

35
36 308x231mm (96 x 96 DPI)
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 28 of 38

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32 Figure 5. Stability indexes for the racing bicycle using the nominal parameters. (a) Stability area of weave
33 mode. (b) Stability area of capsize mode. (c) Stability area of wobble mode. (d) Self-stability speed range
and self-stability area.
34
ie

35 313x231mm (96 x 96 DPI)

36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 29 of 38 Vehicle System Dynamics

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

Figure 6. Correlation coefficients of the five stability indexes. (a) Commuting bicycle, (b) racing bicycle.
41
42 393x398mm (96 x 96 DPI)
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 30 of 38

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
Figure 7. Weave mode behavior when varying only (a) cornering stiffness and (b) Camber stiffness.
34
ie

35 329x261mm (96 x 96 DPI)

36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 31 of 38 Vehicle System Dynamics

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34 Figure 8. Wobble mode behavior when varying only (a) front-assembly damping and (b) deformation axis
ie

35 height.
36
37 333x271mm (96 x 96 DPI)
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 32 of 38

1
2
3
4
5
6
7
8
9
10
11
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
Figure 9. Self-stability area of (a) commuting bicycle and (b) racing bicycle when varying only the
37
w

deformation axis height.

38
39 452x401mm (96 x 96 DPI)
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 33 of 38 Vehicle System Dynamics

1
2
3 Table 1. Parameters of the analyzed bicycles.
4
5
6
Wheelbase ‫[ ݓ‬m]
Parameter Commuting bicycle Racing bicycle
7
Trail ܿ [m]
1.02 0.989
8
Head angle ߣ [deg]
0.08 0.062
9
Rear wheel: radius ‫ݎ‬ோ [m]
18 15.81
10
Rear wheel: mass ݉ோ [kg]
0.3 0.332
11 2 1.38
12 Rear wheel: mass
13
‫ܫ‬ோ௫௫ , ‫ܫ‬ோ௬௬ [kg⋅m ]
moments of inertia 0.0603 , 0.12 0.055 , 0.076
2
14
Fo
15 Main body: position of

‫ݔ‬஻ , ‫ݖ‬஻ [m]

16 the center of mass of 0.3 , -0.9 0.296 , -1.072

Main body: mass ݉஻ [kg]

17
‫ܫ‬஻௫௫ 0 ‫ܫ‬஻௫௭
85 76.49
9.2 0 2.4 9.978 0 2.123
18
቎ 0 ‫ܫ‬஻௬௬ 0 ቏ = ൥ 0 11 0൩ ൥ 0 10.271 0 ൩
rP
Main body: mass
19
‫ܫ‬஻௫௭ 0 ‫ܫ‬஻௭௭ 2.4 0 2.8 2.123 0 2.648
moments of inertia
20 [kg⋅m ]
2

21
mass ‫ݔ‬ு , ‫ݖ‬ு [m]
Front assembly: center of
22 0.9 , -0.7 0.906 , -0.732
Front assembly: mass ݉ு
ee
23
4 2.27
24
‫ܫ‬ு௫௫ 0 ‫ܫ‬ு௫௭
[kg]
0.05892 0 −0.00756 0.098 0 −0.004
቎ 0 ‫ܫ‬ு௬௬ 0 ቏=൥
25
0 0.06 0 ൩ ൥ 0 0.069 0 ൩
Front assembly: mass

‫ܫ‬ு௫௭ 0 ‫ܫ‬ு௭௭ −0.00756 0 0.00708 −0.004 0 0.04

26 moments of inertia
rR

Front wheel: radius ‫ݎ‬ி

27 [kg⋅m ]
28 0.35 0.334
Front wheel: mass ݉ி
29 [m]
30 3 1.58
[kg]
ev

31 Front wheel: mass

32
‫ܫ‬ி௫௫ , ‫ܫ‬ி௬௬ [kg⋅m ]
moments of inertia 0.1405 , 0.28 0.055 , 0.106
33 2

34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 34 of 38

1
2
3 Table 2. Cornering stiffness, camber stiffness and estimated overturning stiffness of the analyzed
4 tires.
5
6
7
8 CORNERING STIFFNESS CAMBER STIFFNESS ESTIMATED OVERTURNING STIFFNESS
9 Deviation BCD product Deviation Estimated Deviation
Tire BCD product
10 from mean (%) from mean (%) crown radius [m] from mean (%)
11 T1 13.356 4.7 1.446 25.9 0.019 22.5
12 T2 12.483 -2.1 1.442 25.5 0.019 22.5
13 T3 13.440 5.4 1.454 26.6 0.018 15.9
14 DT1 9.350 -26.7 0.966 -15.9 0.010 -37.1
Fo
15 DT2 11.982 -6.0 0.779 -32.2 0.012 -23.8
DT3 15.908 24.7 0.806 -29.8 N/A N/A
16
Mean 12.753 - 1.149 - 0.015 -
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 35 of 38 Vehicle System Dynamics

1
2
3 Table 3. Self-analyzing stiffness and twisting stiffness of commuting tires.
4
5
6
7 SELF-ALIGNING STIFFNESS TWISTING STIFFNESS
8 Tire BCD product Deviation BCD Deviation
9 from mean (%) product from mean (%)
10 T1 0.333 6.7 0.0499 -4.0
T2 0.296 -5.1 0.0528 1.6
11
T3 0.307 -1.6 0.0532 2.4
12
Mean 0.312 - 0.0520 -
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 36 of 38

1
2
3 Table 4. Range of tire properties and compliance parameters.
4
5
6
7 Parameter Low value High value Nominal value
8 Cornering stiffness
 − 9.565 15.941 12.753
9 Camber stiffness
 − 0.862 1.436 1.149
10 Self-aligning torque
   0.234 0.390 0.312
11 Twisting torque
   0.0390 0.0650 0.0520
12 Overturning couple
  0.0113 0.0189 0.0151
Relaxation length due to sideslip    0.056 0.094 0.075
13 
14 Front assembly stiffness    2100 3500 2800

Fo

15 Front assembly damping     19 31 25
16 Height of twist axis ℎ   0.24 0.40 0.32
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
Page 37 of 38 Vehicle System Dynamics

1
2
3 Table 5. Stability indexes for the commuting and racing bicycles in nominal conditions.
4
5
6
7 Index Commuting bicycle Racing bicycle
8 Weave stability area 24.25 76.04
9 Capsize stability area 57.73 129.34
10 Wobble stability area 194.03 52.12
11 Self-stability area 20.34 24.32
Self-stability speed range 16.7 m/s 15.8 m/s
12
13
14
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd
 Vehicle System Dynamics Page 38 of 38

1
2
3 Table 6. Summarized results of ANOVA.
4
5
6
7 Commuting bicycle Racing bicycle
8 Source Sum of Degrees Source Sum of Degrees
9 of variation squares of freedom of variation squares of freedom
10 Main effects Main effects
4931.8 9 103074.1 9
11 (factors) (factors)
Nonadditivity Nonadditivity
12 2133.0 502 10004.9 502
(interaction) (interaction)
13 Total sum Total sum
14 7064.9 511 113079.0 511
of squares of squares
Fo
15
16
17
18
rP
19
20
21
22
ee
23
24
25
26
rR

27
28
29
30
ev

31
32
33
34
ie

35
36
37
w

38
39
40
On

41
42
43
44
45
ly

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
URL: http://mc.manuscriptcentral.com/nvsd