This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/authorsrights

Author's personal copy

Mechanism and Machine Theory 70 (2013) 523537

Contents lists available at ScienceDirect

Mechanism and Machine Theory

journal homepage: www.elsevier.com/locate/mechmt

Kinematic and power-ow analysis of bevel gears planetary

gear trains with gyroscopic complexity
Germano Del Pio, Ettore Pennestr , Pier Paolo Valentini
Dipartimento di Ingegneria dell'Impresa, University of Rome Tor Vergata, via del Politecnico, 1 00133 Roma, Italy

a r t i c l e

i n f o

Article history:
Received 20 February 2013
Received in revised form 12 July 2013
Accepted 21 August 2013
Available online 21 September 2013
Keywords:
Planetary bevel gear trains
Power flow
Kinematics
Willis formula
Mechanical efficiency

a b s t r a c t
In this paper the authors propose a method for the kinematic and power-flow analysis of bevel
epicyclic gear trains with gyroscopic complexity. By gyroscopic complexity, we mean the
possibility of the gear carrier to be a floating link as, for instance, in robotic gear wrists.
Thanks to the new formulas herein deduced, the methods based on the graph representation
of planetary spur gear trains have been extended to bevel gear trains. In particular, the well
known Willis equation has been modified to maintain its validity for bevel gears. The modified
Willis equation was then embodied in new power ratio expressions. Under our simplifying
hypotheses of absence of friction and constant angular speeds, it is shown that gyroscopic
torques do not enter into power flow analysis. Two numerical examples are discussed.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
In recent times different papers on the mechanics of multi degrees-of-freedom planetary gear trains have been published. The
focus of this effort was the development of systematic methods for mechanical efficiency analysis. This renewed interest is also
due to the application of epicyclic gear trains as one of the main powertrain component of hybrid vehicles. The capability to
handle torques from different power sources is an almost unique feature of this type of transmissions.
A fundamental step in mechanical efficiency analysis is the ascertainment of the amount of power flow through the meshing
gears. Although not self evident, due to power circulation, some meshing gears may sustain a power higher than the input one.
Power circulation, that usually occurs with very low transmission ratios, must be detected at the early design stages in order to
dimension properly meshing gears and lubricating methods.
Most of the contributions are related to spur gear trains. In this case the kinematics can be studied with the classic scalar Willis
equation.1 The relationship between the absolute angular speeds of bevel gear trains is not scalar and this complicates the
analysis.
This paper focuses on kinematic and power flow analysis of planetary trains with bevel gears. It can be considered as an
attempt to extend the modus operandi of the analysis methods devised for spur gear trains to bevel gear trains.
A complete review of all the scientific contributions on the topic is outside the purpose of this paper. Thus the following review
cannot be considered exhaustive.
Belfiore, Pennestr and Sinatra [1] presented a Maple procedure for kinematic and power-flow analysis of spur planetary gear
trains based on the graph based method of Pennestr and Freudenstein [2,3].

Corresponding author. Tel.: +39 0672597138.

E-mail address: pennestri@mec.uniroma2.it (E. Pennestr).
1
Following the tradition of many textbooks, the authors called Willis equation the expression which relates the absolute angular velocities of a differential
gear train. However, it should be acknowledged that this equation was well known before Rev. Robert Willis (18001875), Jacksonian Professor of Applied
Mechanics in the University of Cambridge.
0094-114X/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.mechmachtheory.2013.08.016

Author's personal copy

524

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

Kaharaman et al. [4] proposed a modular general formulation valid for 1-dof automatic transmission composed of complex
compound planetary gear sets.
Chen and Teh [5] applied the concept of virtual-power [6] to find ready-to-use formulas for the mechanical efficiency analysis
of 2-dof gear trains. For a Simpson differential gear train, Chen [7] analyzed the sensitivity of the mechanical efficiency.
Mathis and Remond [8] proposed a unified model for the kinematic, torque and efficiency analysis of epicyclic gear trains. An
application of the method to a Ravigneaux type gear train is discussed.
Galvagno [9] discussed the influence of dynamics on the mechanical efficiency analysis of a 2-dof differential spur gear train.
Pennestr et al. [10] presented a systematic approach for the modeling and analysis of power split transmissions which include
an epicyclic gear train. The method is a refinement of the one proposed by Pennestr and Freudenstein [2,3,11].
However, less common are the power flow and efficiency analyses of gear trains with bevel gears.
Day et al. [12] proposed a matrix method for the kinematic analysis of planetary bevel gear trains using the concept of the
fundamental circuit. However, the method is limited to gear carriers rotation about a fixed axis.
Freudenstein et al. [13] extended the concept of fundamental circuit to the analysis of geared robotic wrists. This category of
wrists, enumerated by Belfiore [14], has gyroscopic complexity when the gear carrier is not adjacent to the frame link.
Further contributions on planetary geared robotic wrists are due to Tsai [15] who hinted that the motion of a bevel-gear-type
end-effector can be described by an equivalent open-loop chain. The analysis equations follow by considering relative rotation
between every two adjacent links in the equivalent open-loop chain and coaxial conditions.
Litvin and Zheng proposed a matrix method for the kinematic analysis of differential trains with bevel gears [16].
Gupta and Ma [17], extended the tabular superposition method to derive the relations among the three wrist joint variables
and three coaxial actuation variables.
Nelson and Cipra [18] proposed a systematic graph-based matrix method for kinematic, power-flow and mechanical efficiency
suited to the complete solution of bevel gear sets as well as planar epicyclic sets.
Uyguroglu and Demirel [19,20] applied oriented linear graph techniques toward the kinematic and static moment analysis of
robotic bevel-gear trains.
Staicu proposed recursive matrix relations for kinematics and dynamics analysis of different orienting bevel gear train [2123].
The relations are particularly useful for inverse dynamics.
Laus et al. [24] combined graph and screw theory for the analysis of mechanical efficiency of bevel gear trains with complex
architecture. In particular Davies' equations [25] have been applied to gear trains. The analysis equations are deduced in a way
similar to the one followed for electrical networks. The mechanical analog of the electrical resistance is introduced to take into
account power losses.
Chen [26] introduced the constraint equations for kinematics and power flow analysis. The method is very systematic and can
be potentially implemented in a general purpose code. The use of constraint equations for kinematic and dynamic analysis of
planetary gear trains was also explored by Mantriota and Pennestr [27] by means of multibody dynamics approach.
The classical tabular method, based on Willis equation, cannot be applied to complex planetary bevel gear trains.
In this paper, we propose a scalar equation between the absolute angular velocities of the simplest bevel gear train. By means
of this equation, the method of fundamental circuits [28,29] can be extended to the kinematic analysis of planetary bevel gear
trains.
Moreover, the power ratios through the bevel gears and gear carrier forming a fundamental circuit are herein deduced. The
result is also novel. In fact, such ratios include, as a particular case, those deduced by Pennestr and Freudenstein [2,3] for
planetary spur gear trains.
By means of these ratios, graph based methods of power flow analysis of planetary gear trains can include the presence of
bevel gears.
The paper is divided into three parts. The first two parts are dedicated to methods of kinematic and power flow analysis,
respectively, and the third one to applications.

k
R(b)

R(a)

Fig. 1. Labeled graph representation of a fundamental circuit. G: Gear pair; R(a), R(b): revolute pairs with different axes; i and j, gear wheels; k gear carrier
(transfer vertex).

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

525

2. Kinematic analysis
The mechanism is considered in a right-hand Cartesian coordinate system fixed to its frame member. Stationary conditions are
assumed.
The topology of a gear train can be meaningfully identified by means of a labeled graph where links are represented by vertices
and kinematic pairs by edges, respectively. Edges are labeled according to the type of kinematic pair, i.e. R: revolute pairs and G:
gear pair. A letter within parentheses specifies the axis level of the revolute pair.
This approach allows us to locate fundamental circuits in the planetary gear train (Fig. 1). The fundamental circuit identifies the
simplest gear unit, that is two meshing gears connected through a gear carrier (see Fig. 1).
A planetary gear train can be considered as an assembly of elementary gear units each of them associated with a fundamental
circuit. By identifying the fundamental circuits one recognizes such units and the application of the analysis equations can be
made in a very systematic manner (e.g. [28,29,10]).
The gears of a generic fundamental circuit are denoted with the letters i and j and the gear carrier with k. The frame member of
a mechanism is denoted with F.
The vectors will be interpreted as a multiplication of an algebraic scalar with a generic unit vector b
aba . It is imposed that
b
aba b
aab .
With reference to the mth fundamental circuit (see Fig. 2), the following angles are introduced:
m angle between the axis of relative rotation of the two bevel gears and the X axis (see Fig. 6);
qF angle between the absolute angular velocity vector of a generic link q and the X axis;
m and m, angles between X axis and the axes of bevel gears i, j, respectively. These are also the anomalies of the torques
!
!
T im and T jm acting on the gear wheels i and j of the mth fundamental circuit (see Fig. 7);
!
m, angle between the X axis and the gear torque T km acting on the gear carrier of the mth fundamental circuit (see Fig. 7).
The angles are measured positively counterclockwise.
Moreover, the following nomenclature is introduced:
!
!
ab is the module of the relative angular velocity of link a with respect to link b; it is imposed that ab ba ;
qF is the module of absolute angular velocity of a generic link q.
In particular, denoted by Nj and Ni, respectively, the number of gear teeth of the bevel gears i and j:
N

the bevel gear ratio Rm j is always negative unless the semi-vertex angle of one bevel gear is greater than 90;
Ni
the unit vector b
aik associated with a gear rotation axis is directed outward from the center of spherical motion (see Figs. 2 and 3).
In a graph representing a bevel gear train, we may distinguish two types of fundamental circuits. With reference to the gear
arrangements shown in Fig. 4, for the sake of brevity, we shall define them as Type A and Type B [30].
Type A fundamental circuits are those in which one of the gears and the gear carrier both rotate about a fixed axis (see Fig. 4).
Type B fundamental circuits are those in which the bevel gears do not share revolute joints with the frame link.
When type B circuit is present in the graph, the bevel gear train will have gyroscopic complexity. The analysis of
planetary gear trains with type B circuits requires further conditions [15] on angular velocities of links connected by
revolute pairs.
For the purpose of kinematic analysis, for each fundamental circuit, the following equations can be written:
ij b
aij ik b
aik jk b
ajk ;

1a

ai

ai

ajk

j
Y

m
m

X
Fig. 2. Externally meshing bevel gears. Rm b 0.

Author's personal copy

526

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

aij
k

ajk
j
aik

m m

X
i

Fig. 3. Internally meshing bevel gears. Rm N 0.

jk

ij
j

ik iF

kF
F

Type A

ij

ik

jk

j
k

kF
F

Type B
Fig. 4. Type A (top) and Type B (bottom) fundamental circuits. The orientation and modules of vectors are represented qualitatively.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

527

ik
Rm ;
jk

1b

iF b
aiF ij b
aij jF b
ajF :

1c

Eq. (1c) is required for fundamental circuits that do not contain the frame member, while Eqs. (1a) and (1b) have to be written
for every fundamental circuit.
The unit vectors of the rotation axes of the relative motion between the fundamental circuit links can be expressed as
follows
b
aij
b
aiF


cos m
;
sin m

cos iF
sin iF


;

b
aik

b
ajF


cos m
;
sin m

cos jF
sin jF


;

b
ajk

b
akF


cos m
;
sin m

cos kF
sin kF


:

Manipulating Eqs. (1a)(1c) and taking into account Eqs. (2) and (3), the angle m is computed by means of the following
expression
tan m

Rm sin m sin m
:
Rm cos m cos m

One can test the correctness of Eq. (4) for the simple case shown in Fig. 5, where Rm = 1, m = 180 and m = 90. Form
Eq. (4) one obtains tan m = 1, or m = 135.
A visual inspection is required to determine whether iF, jF, and kF need to be computed by means of Eqs. (1a), (1b), and
(4) or whether they can be directly obtained by inspection from the mechanism drawing.
The solution of the system of equations formed by Eqs. (1a), (1b), (1c) and (4), written for each fundamental circuit of the
planetary gear train, gives the magnitudes and directions of the unknown angular velocities.
Although geared wrists are spatial mechanisms, our algebraic treatment is planar. In fact, for a given fundamental circuit, in a
! !
!
unit of two bevel gears i, j and a gear carrier k, the relative angular velocities ij, ik, and jk, must lie all on the same plane (say
in Fig. 4), as stated by Eq. (1a).
! !
!
!
Also ij , iF , and jF must lie on the same plane (say 1), in fact, adding kF at left and right sides we obtain
!
!
!
!
!
ik kF ij jk kF ;

which is equivalent to Eq. (1c).

!
! ! ! !
For type A fundamental circuits, ik , ij , jF , jk and kF lie on the same plane, therefore and 1 necessarily coincide. For
type B fundamental circuits, the two planes may coincide or may form a constant angle. However, in current industrial solutions

m =135
m=90

m =180

Rm= -1

Fig. 5. Simple test case.

Author's personal copy

528

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

(e.g. [3133]) of robotic geared wrists, all Type B fundamental circuits have revolute axes on the same plane (see Fig. 4).
Therefore, in the present treatment the planes and 1 are always considered coincident.
In any case, the relationships between angular velocities are not altered by the positions of these planes in space or by the pose
of the end-effector. This justifies the analytic treatment in a plane.
In other words, the relationships between the modules of angular velocities do not change if the plane changes its position
in space.
Our conclusions are consistent with the findings of other investigators (e.g. [33,19]).
For these reasons a planar description of angular speeds of the links belonging to a fundamental circuit has been herein
adopted.
3. A generalized form of Willis equation
The previous equations can be manipulated in order to obtain a scalar generalized Willis equation valid for bevel gears (see
Fig. 6):
ik Rm jk ;

aik iF b
aiF kF b
akF ;
ik b

jk b
ajk jF b
ajF kF b
akF :

From these last two equations follow

ik iF

b
b
aiF  b
a b
x
x
kF kF
;
b
b
b
aik  x
aik  b
x

9a

jk jF

b
b
ajF  b
x
a b
x
kF kF
:
b
b
ajk  b
ajk  b
x
x

9b

Combining Eqs. (6), (9a) and (9b) one obtains

cim iF cjm Rm jF ckm Rm 1kF 0;

10

where
cim

b
aiF  b
x cos iF

;
b
aik  b
x cos m

11a

cjm

b
ajF  b
x cos jF

;
b
ajk  b
x cos m

11b

ckm

b
x b
x
a b
a b
Rm kF kF
b
ajk  b
aik  b
x b
x

Rm 1

Rm cos kF

1
1

cos m cos m

Rm 1


:

11c

Eq. (10) is a generalized form of the Willis equation. This equation can be applied for the kinematic analysis of planetary bevel
gear trains, including those with gyroscopic complexity. The deduced expression allows the extension of the graph based method

iF

ij

jF

m
iF

jF

- jF
Fig. 6. Nomenclature.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

529

of Freudenstein and Yang [28,29] to bevel gear trains. For each fundamental circuit, Eq. (10) must be particularized. Solving the
system of the obtained equations, the unknown angular velocities are systematically computed.
4. Power-ow analysis
The gear train is analyzed under steady state conditions, neglecting power losses and inertia forces. The convention for
! !
!
measuring torque vectors is the same as outlined for angular velocities. Let us denote by T im , T jm and T km , respectively, the
torques applied on the gear wheels i and j and the gear carrier k of the mth fundamental circuit.
Taking into account only the torque components in the or 1 plane, the equilibrium condition of the mth gear unit
yields
!
!
!
T im T jm T km 0 :

12

Since the torques on the gears are applied through revolute joints, we can state that the direction of a torque vector applied on
a gear coincides with the axis of its revolute joint.
For power flow analysis we are interested in the generalized forces (torques) that do work. Under the hypotheses of negligible
friction, gyroscopic torques in geared wrists do not do work.
Let us consider a type A fundamental circuit (see Fig. 8). The gyroscopic torque is acting on gear j (satellite) since gear i rotates
about a fixed axis.
!
!
The absolute angular velocity kF of the gear carrier can be interpreted as the precession velocity and jk as the spin velocity.
The absolute angular speed of gear j is
!
!
!
jF kF jk
and its local components, in a Cartesian frame system attached to the gear, are
x

jF jk kF cos ;
y
jF kF sin ;
z
jF 0 ;
where is the angle between the axes of the revolute pairs.

Gyroscopic torque
is orthogonal to
the plane

jk

Tkm

kF

y
Fig. 8. Nomenclature.

Author's personal copy

530

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

Assuming constant angular speeds, the application of NewtonEuler equations (e.g. [34]) gives
M x 0;
My 0


x
y
M z J y J x jF jF ;
where Jy and Jx are moment of inertia of the gear j about axes x and y, respectively.
 
!
Hence, the gyroscopic torque Mg f 0 0 Mz gT is always orthogonal to the plane containing the angular speeds kF and
!
jk . Consequently, assuming absence of friction in the kinematic pairs, the gyroscopic torque will not produce any work and
should not be taken into account in any power balance condition.
!
The externally applied torque vectors in the equilibrium Eq. (12) lie all in the plane. Since the gyroscopic torque vector M g is
orthogonal to this plane, it does not appear in the Eq. (12).
A similar reasoning can be applied to the case of Type B fundamental circuits.
For the purpose of power flow analysis each fundamental circuit is represented by means of a box with three nodes, one for
each link. Boxes are connected through nodes (shared links). Powers Pim, Pjm and Pkm are positive when entering the mth box,
negative when leaving it.
The following torque equilibrium and power balance equations hold for every fundamental circuit (see Fig. 7)
T im cos m T jm cos m T k;m cos m 0;

13

T im sin m T jm sin m T k;m sin m 0;

14

!
!
P im T im  iF T im iF cos iF m ;

15



!
!
P jm T jm  jF T jm jF cos jF m ;

16

!
!
P km T km  jF T km kF cos kF m ;

17

P im P jm P km 0;

18

cim iF cjm Rm jF ckm Rm 1kF 0;

19

!
where m is the angle formed by T km with the abscissa axis, and Pim is the share of power on gear i flowing through the mth
fundamental circuit.
In a graph representation of a gear train, a link corresponds to a vertex. By analogy with the Kirchhof current law, the sum of
powers through a node must be zero. Hence, for the generic qth moving member the following power balance condition can be
written,
N
X

P qk 0;

20

k1

where N is the number of adjacent members.

Fig. 7. Torque equilibrium: Nomenclature.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

531

Manipulating these equations, one obtains the power ratios2

P jm
P im



jF
cos kF m
cos m Rm cjm cim
Rm 1ckm cos m
iF
cos iF m



cos kF m

 Rm 1ckm cos m
cos m cim iF Rm cjm
jF

P km

P jm

21a

cos jF m

cos iF m

 cim cos m
cos m Rm 1ckm kF Rm cjm
jF

cos jF m



:

cos iF m
cos m Rm cjm jF Rm 1ckm
cim cos m
kF

21b

cos kF m

For each fundamental circuit, Eq. (18) and either Eq. (21a) or (21b) need to be included in the power-flow analysis system of
equations, together with link power balance Eq. (20).
Eqs. (21a) and (21b) represent an extension to bevel gears of the power ratios deduced by Pennestr and Freudenstein [2,3] for
the case of spur gear trains.
The expressions (Eqs. (21a) and (21b)) may look cumbersome, however, once the kinematic analysis is completed, the angles
m, m can be obtained by visual inspection.
The angle m follows from the equation
tan m

sin m Rm sin m
cos m Rm cos m

22

deduced by combining Eqs. (13), (14) and

T jm
Rm :
T im

23

5. Numerical examples
5.1. Kinematic analysis
The proposed method is applied first to planetary gear trains with type A and then to those with type B circuits. In particular,
the Humpage reduction gear (see Fig. 9) and the Cincinnati Milacron gear wrist (see Fig. 10) will be considered.
For brevity, the units of measurement have been omitted from the text.
Example 1. Humpage reduction gear
The Humpage reduction gear has only type A circuits
As summarized in Table 1, the graph of the gear train contains three fundamental circuits.
For each fundamental circuit, Eqs. (1a), (1b) and (1c) are particularized:
Fundamental circuit 1
23 b
a23 24 b
a24 34 b
a34 ;
24
56
;
20
34
2 F b
a2 F 23 b
a23 3 F b
a3 F :
Fundamental circuit 2
a31 34 b
a34 14 b
a14 ;
31 b
34
76
:
56
14
Fundamental circuit 3
a35 34 b
a34 54 b
a54 ;
35 b
34
35
;
24
54
b
3 F a3 F 35 b
a35 5 F b
a5 F :

In our derivations Eq. (13) is used instead of Eq. (14).

Author's personal copy

532

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

The value 2F = 1 rad/s is prescribed. The angles summarized in Table 2 are preliminary obtained. Moreover, also by
inspection 1F = 2 = 180, 2F = 1 = 0, 4 = 2 = 180, 5F = 3 = 180. The angles 1, 2, and 3 are unknowns and
need to be computed by means of Eq. (4):
R1 sin
R1 cos
R sin
tan 2 2
R2 cos
R sin
tan 3 3
R3 cos
tan 1

1 sin 1
14:71B;
1 cos 1
2 sin 2
105:30B;
2 cos 2
3 sin 3
102:11B:
3 cos 3

Therefore, the unit vectors relevant for our analysis are:


cos 2 F
;
 sin 2 F 
cos 5 F

;
 sin 5 F
cos 1

;
sin 1

b
a2 F
b
a5 F
b
a34


cos 2
;
 sin 2 
cos 1
b
a23
;
 sin 1 
cos 3
b
a35
;
sin 3

b
a31
a3 F b


cos 2
;
 sin 2 
cos 1
b
a24
;
 sin 1 
cos 3
b
a54
:
sin 3

b
a14
a4 F b

One can solve the system of eight equations in the eight unknowns formed by particularizing Eqs. (1a), (1b) and (1c) for each
fundamental circuit. The solution is herein reported:
 
 
1
1
!
: 2 F
0 
0


0:264
0:067
!
0:254; b
a3 F
: 3 F
0:965
0:245




1
0:208
!
b
b
41 0:208; a4 F a14
: 4 F
0 
0



1
0:014
!
0:014; b
a5 F
: 5 F
:
0
0

2 F 1;
3 F
4 F
5 F

b
a2 F

The solutions are in agreement with those given by Nelson and Cipra [18] for the same mechanism. Their system of equations
was
8! 9 2
3
2
3


0:067 0:933

0:245 0:245

< 3F =

2 F b
!
x4
0
0 5
y
24
4 F 4 0:208 0:792 5 2 F b
1 F
:! ;
1F
0:014 0:986
0
0
5F
with 2F = 1 and 1F = 0.
Alternatively, one could apply the proposed modified Willis Eq. (10). In this case the coefficients (Eqs. (11a)(11c)) must be
computed in advance
c21
c41

cos 2 F
1;
cos 1
cos 4 F cos 4 F
R1

cos 1

cos 1

R1 1

c31

cos 3 F
0:528;
cos 1

1:737;

cos 3 F
cos 1 F
0:528;
c12
1;
cos 2
cos 2
cos 4 F cos 4 F
R2

cos 2
cos 2
0:273;

R2 1
cos 3 F
cos 3 F

0:528;
c53
1;
cos 3
cos 3
cos 4 F cos 4 F
R3

cos 3
cos 3
0:221:

R3 1

c32
c42
c33
c43

Since there are three fundamental circuits, the modified Willis equation must be written three times:
I) 2F ( 0.258)(2.8)3F + ( 1.737)( 2.8 1)4F = 0.
II) (0.528)3F (1)( 1.357)1F + ( 0.273)( 1.357 1)4F = 0
III) (0.528)3F (1)( 1.458)5F + ( 0.221)( 1.458 1)4F = 0.
The solution of the system is 3F = 0.254, 4F = 0.208, and 5F = 0.014.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

533

Example 2. Cincinnati Milacron gear wrist (Table 3)

In this analysis, due to the presence of type B circuits, for the computation of the cim, cjm and ckm coefficients, the following
kinematic conditions must be taken into account
f41 g 43

 


 
1
cos
1
32
21
;
0
sin
0


f51 g 52

f31 g 32

cos
sin
cos
sin

25

21

 
1
;
0

26

21

 
1
:
0

27

Writing the Willis equation for each fundamental circuit yields:

Fundamental circuit 1
c41 41 R1 c51 51 c31 31 R1 1 0;

28

where
c41 41 43 32 cos 21 ;

c51 51 52 21 ;
 cos 
c31 31 R1 1 R1 32 21 32 cos 21 :
cos
Fundamental circuit 2
c52 51 R2 c62 61 c22 R2 121 0;

29

where
21
;
cos
c62 1;


1
:
c22 R2 1 R2
cos
c52 51 52

Fundamental circuit 3
c33 31 R3 c73 71 c23 R3 121 0;

30

where
32
21
;

31 31 cos
c73 1;
1
:
c23 R3 1 R3
cos
c33

Solving the system of equations formed by Eqs. (28), (29) and (30) one obtains:
32 R3 71 21 ;

31

43 R1 R3 R2 21 R2 61 R3 71 ;

32

52 R2 61 21 :

33

5.2. Power-ow analysis

In this section the power-flow analyses of the Humpage reduction gear and of the Cincinnati Milacron geared wrist by means
of the proposed method are discussed.

Author's personal copy

534

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

Example 3. Humpage reduction gear

For each fundamental circuit, Eqs. (18) and (21a) are written:
I) Since the link 1 is the driving link, let P21 = 1.
P 21 P 31 P 41 0;
P 31
0:5
P 21
II) Since link 1 is the frame, P12 = 0 and only Eq. (18) is required
P 32 P 12 P 42 0:
III)
P 33 P 53 P 43 0
P 53
0:114:
P 33

The Eq. (20) is written for each member, except the driving, driven and frame links:
Gear 3) P31 + P32 + P33 = 0.
Gear 4) P41 + P42 + P43 = 0.

3 N=56

F=1

N=24

2 N=20

5 N=35

1 N=76

P32

3
5

P53

P43
4

P31

P33

P21

P42

P41

Fig. 9. Kinematic structure of the Humpage reduction gear and power flow graph.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

535

The solution of the system of previous equations is

P 21 1:000;

P 31 0:500;

P 41 0:500;

P 32 9:272;

P 42 9:272;

P 33 8:772;

P 43 9:772;

P 53 1:000:

Example 4. Cincinnati Milacron gear wrist

The following numerical data are prescribed or known from kinematic analysis:

gear ratios: R1 = 1, R2 = 1, R3 = 0.4;

driving angular velocities: 21 = 1, 61 = 1, 71 = 0.1;
known angles: 1 = 0, 1 = 45, 2 = 225, 2 = 180, 3 = 225, 3 = 180;
computed angles: 1 = 157.5, 2 = 22.5, 3 = 21.524, 1 = 22.5, 2 = 157.5, 3 = 113.476;
angular velocities of link 3: 37 = 0.694, 32 = 0.360, 31 = 0.788, with 31 = 18.854;
angular velocities of link 4: 43 = 2.360, 45 = 4.361, 41, with 41 = 171.040;
angular velocities of link 5: 56 = 3.696, 52 = 2, 51 with 51 = 30.361.

The coefficients of the Willis equation are c41 = 0.988, c51 = 1.220, c31 = 1.142, c52 = 1.220, c62 = 1.000, c22 = 1.207,
c33 = 1.338, c73 = 1, and c23 = 1.690.

P52

P62

P6i

P41 4

P51

1
5

P2i

P31

22
2

Pout

P33

P23

P73

P7i

Fig. 10. Kinematic structure and flow graph of the Cincinnati Milacron gear wrist.

Table 1
Fundamental circuits (F.C.) of the Humpage gear train.
m

Gear i

Gear j

Gear carrier k

Rm

1
2
3

2
3
3

3
1
5

4
4
4

56
20
76
56
35
24

Author's personal copy

536

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

Table 2
Rotation axes angles within each circuit.
m

1
2
3

0
60
60

60
180
180

Table 3
Fundamental circuits (F.C.) of the Cincinnati Milacron geared wrist.
m

Gear i

Gear j

Gear carrier k

Gear ratio

1
2
3

4
5
3

5
6
7

3
2
2

R1
R2
R3

For each fundamental circuit, Eqs. (18) and (21a) are written:
I)
P 41 P 51 P 31 0
P 51
1:677
P 41
II)
P 52 P 62 P 22 0
P 22
1:707
P 62
III)
P 33 P 73 P 23 0
P 23
7:678:
P 73
When the power balance condition is applied at each node link, the following equations are obtained:
P 73 P 7i 0;
P 62 P 6i 0;
P 22 P 23 P 2i 0;
P 41 P out 0;
P 52 P 51 0;
P 33 P 31 0
Imposing Pout = 1, then the solution of the system formed by the previous equations gives: P41 = 1, P51 = 1.677,
P52 = 1.677, P31 = 0.677, P33 = 0.677, P22 = 1.057, P23 = 0.599, P62 = 0.619, P73 = 0.078, P2i = 0.459, P6i = 0.619,
P7i = 0.078.
It should be observed that in the power-flow analysis of multi degrees-of-freedom gear trains, the output power is imposed
and the input powers are computed.
According to our results, under the prescribed kinematic conditions, link 7 is driven, whereas links 6 and 2 are driving.
6. Conclusions
In the paper novel procedures of kinematic and power flow analysis of planetary bevel gear trains with gyroscopic complexity
are presented. All the methods are based on the graph representation of the gear train. This feature allows a systematic approach
to the analysis.
Innovative features of this work are:
a scalar equation that relates the absolute angular velocities of a basic bevel gear train;
the equations that express the power ratios in a basic bevel gear train.
These equations have, as particular case, the corresponding equations valid for spur gears.

Author's personal copy

G. Del Pio et al. / Mechanism and Machine Theory 70 (2013) 523537

537

Thanks to the availability of these equations, graph based methods of kinematic and power flow analysis of planetary gear
trains have been extended to a new class of mechanical devices.
Although geared robotic wrists provide good load and dexterity capabilities, their application in the automation field has been
somewhat limited. To the best of the authors' knowledge, this is the first time that a power flow-analysis for this class of
mechanical devices has been discussed.
References
[1] N.P. Belfiore, E. Pennestri', R. Sinatra, Kinematic and power-flow analysis of epicyclic gear drives, MAPLETECH Maple Tech. Newsl. 1 (1994) 4755.
[2] E. Pennestr, On the Automatic Design Analysis of Gear Trains, (PhD thesis) Columbia University in the City of New York, 1990.
[3] E. Pennestr, F. Freudenstein, A systematic approach to power-flow and static force analysis in epicyclic spur-gear trains, ASME J. Mech. Des. 115 (1993)
639644.
[4] A. Kahraman, H. Ligata, K. Kienzle, D.M. Zini, A kinematics and power flow analysis methodology for automatic transmission planetary gear trains, ASME J.
Mech. Des. 126 (2004) 10711081.
[5] C. Chen, L. Teh, Theoretic study of efficiency of two-dofs of epicyclic gear transmission via virtual power, ASME J. Mech. Des. 133 (2011) 17.
[6] C. Chen, J. Angeles, Virtual-power flow and mechanical gear-mesh power losses of epicyclic gear trains, ASME J. Mech. Des. 129 (2007) 107113.
[7] C. Chen, Power flow analysis of compound epicyclic gear transmission: Simpson gear train, ASME J. Mech. Des. 133 (2011) 15.
[8] R. Mathis, Y. Remond, Kinematic and dynamic simulation of epicyclic gear trains, Mech. Mach. Theory 44 (2009) 412414.
[9] E. Galvagno, Epicyclic gear train dynamics including mesh efficiency, Int. J. Mech. Control 11 (2010) 4147.
[10] E. Pennestr, L. Mariti, P.P. Valentini, V. Mucino, Efficiency evaluation of gearboxes for parallel hybrid vehicles: theory and applications, Mech. Mach. Theory
49 (2012) 157176.
[11] E. Pennestr, F. Freudenstein, The mechanical efficiency of epicyclic gear trains, ASME J. Mech. Des. 115 (1993) 645651.
[12] C. Day, H.A. Akeel, L.J. Gutkowsky, Kinematic design and analysis of coupled planetary bevel-gear trains, ASME J. Mech. 105 (1983) 441445.
[13] F. Freudenstein, R.W. Longman, C.-K. Chen, Kinematic analysis of robotic bevel-gear train, ASME J. Mech. Transm. Autom. Des. 106 (1984) 371375.
[14] N.P. Belfiore, An atlas of remote actuated bevel-gear type wrist mechanisms of up to 9 links, Int. J. Robot. Res. 12 (5) (1993) 448459.
[15] L.W. Tsai, The kinematics of spatial robotic bevel-gear trains, IEEE J. Robot. Autom. 4 (1988) 150156.
[16] F.L. Litvin, Y. Zheng, Robotic bevel-gear differential train, Int. J. Robot. Res. 5 (1986) 7581.
[17] K.C. Gupta, R.C. Ma, On the motion of oblique bevel geared robot wrists, J. Robot. Syst. 6 (1989) 509520.
[18] C.J. Nelson, R.J. Cipra, Simplified kinematic analysis of bevel epicyclic gear trains with application to power-flow and efficiency analyses, ASME J. Mech. Des.
127 (2005) 278286.
[19] M. Uyguroglu, K. Tokad, A dynamic model for the Bendix wrist, Acta Mech. 133 (1999) 199218.
[20] M. Uyguroglu, H. Demirel, Kinematic analysis of bevel-gear trains using graphs, Acta Mech. 177 (2005) 1927.
[21] S. Staicu, Kinematics and dynamics of a planetary gear train for robotics, Sci. Bull. Univ. Politech. Buc. Ser. D 68 (2006) 312.
[22] S. Staicu, Inverse dynamics of a planetary gear train for robotics, Mech. Mach. Theory 43 (2008) 918927.
[23] S. Staicu, Matrix modelling in dynamics of a 2-DOF orienting gear train, Robot. Auton. Syst. 59 (2011) 923929.
[24] L.P. Laus, H. Simas, D. Martins, Efficiency of gear trains determined using graph and screw theories, Mechanism 52 (2012) 296325.
[25] T.H. Davies, Kirchhoff's circulation law applied to multi-loop kinematic chains, Mech. Mach. Theory 16 (1981) 171183.
[26] C. Chen, Power analysis of epicyclic transmissions based on constraints, J. Mech. Robot. 4 (4) (2012) 041004.
[27] G. Mantriota, E. Pennestr, Theoretical and experimental efficiency analysis of multi degrees-of-freedom epicyclic gear trains, Multibody Syst. Dyn. 9 (2003)
389408.
[28] F. Freudenstein, An application of Boolean algebra for the motion of epicyclic drives, ASME J. Eng. Ind. (1971) 176183.
[29] F. Freudenstein, A.T. Yang, Kinematics and statics of coupled spur-gear trains, Mech. Mach. Theory 7 (1972) 253275.
[30] E. Pennestr, On the kinematic analysis of geared robotic wrists, Meccanica 26 (1991) 155160.
[31] M.E. Rosheim, Robotic Wrists Actuators, Wiley, 1989.
[32] T.E. Stackhouse, Geared robot wrist, US Patent US 4068536 A, 1978.
[33] L.W. Tsai, Robot Analysis: The Mechanics of Serial and Parallel Manipulators, John Wiley & Sons, 1999.
[34] R.C. Hibbeler, Engineering Mechanics Statics and Dynamics, Prentice-Hall, 2001.