Beruflich Dokumente
Kultur Dokumente
I. Troedsson
Department of Machine Elements
Lund Institute of Tectinology
Static Load Distribution in a
Lund, Sweden
Chain Drive
A model of how to calculate the load distribution for a chain drive is presented. In the
L. Vedmar model the complete standard geometry is used without any assumptions. The rollers which
Department of Macliine Elements are in contact with the sprockets can move freely along the tooth flanks and their positions
Lund institute of Tectinology are given by force equilibrium. Since the positions of the rollers and thereby also the load
Lund, Sweden
distribution are dependent on the two connecting spans, these necessary tight and slack
.•spans have been included in the model. The elastic deformation in the chain is included
as well as the gravitational force.
Introduction predict the behaviour for a chain drive, it is necessary to model the
complete set with the sprockets and the chain. The only alternative
Chain drives are an old invention but are still considered as a
is to assume some of the position of the rollers. This is not correct,
common way of transmitting power in mechanical systems. The since their position is given by equilibrium.
major advantages of chain drives are the flexibility in selecting
A new model, that determines the load distribution without the
shaft centre, the ability to drive more than one shaft, the high
earlier assumption, is presented here. This calculation is made on
reliability and durability and the relatively cheap components. a chain drive with two sprockets; one driving and one driven. Both
However, chain drives introduce problems such as noise and sprockets are modelled according to the geometry standard DIN
vibration. Even though chain drives have been in use for a long 8187 and all the contacting rollers have been modelled as well as
time, research in the open literature on this machine element has a tight and a slack span to connect the two sprockets.
been concentrated to the last half of this century.
In the 1950s and 1960s R. C. Binder [1] and H-G. Rachner [8]
presented the first step to an analysis of forces and stresses in a Static Model
chain drive. In the early 1980s, Naji and JVIarshek [5, 6, 7]
An example of a chain transmission is shown in Fig. 1. It
complemented Binder's and Rachner's works by theoretical and
represents the simplest kind of transmission since it only includes
experimental investigation to study contact phenomena under
two sprockets and one chain. One sprocket is driving, the other one
quasi-static condition due to non-trivial geometry of the sprocket
is driven and the chain transmits the power between them.
tooth. In the 1990's Kim and Johnson [3, 4] presented a model
In order to obtain a model for determining the load distributions
using a more exact geometry.
along the sprockets and the forces along the chain, the transmission
Binder [1] and Rachner [8] assumed that the contact angle due
is divided into four separate parts. These four parts are tension
to the symmetry of each tooth was constant. But they did not span, slack span, driving sprocket and driven sprocket. The dif-
comment on how to determine it's value. Naji and Marshek [5, 6, ferent parts will, for convenience, be treated separately but, as they
7] used a simplified model of the geometry, assuming that if the affect each other, none of them can be neglected. The chain is
roller contacted the bottom of the gap between two teeth, the affected differently at each part. At the tight span most of the
contact always occured at one point. Kim and Johnson improved power is transmitted between the two sprockets. Here the tension
this model by using the full ANSI standard geometry. However, force dominates over the gravitation force. The opposite is true for
none of them have considered any elastic deformation either in the the slack span where most of the tension force comes from grav-
chain or the sprocket and none of them have used a complete gear itation. The chain part which is in contact with the sprockets, in
set with two sprockets and two connecting chain spans. contrast to the two spans, is affected by contacting forces on the
Sprocket and chain geometry has been standardised by the rollers. The compatibility between the parts must be considered
American standard ANSI B29.1 and the European standards DIN when putting the parts together in a complete model.
8180, DIN 8187 and DIN 8188. The American standard defines the To describe the model a few parameters first have to be defined.
geometry of the tooth using four arcs. The European standards use The driving sprocket will be indexed p and the driven sprocket will
five segments which are a mix of straight lines and arcs. Each arc be indexed g. It is assumed that some parameters are known such
curvature is larger than the radius of the roller. This work has been as the driving torque M,, or the load torque Mg, the number of teeth
developed from the European standard which differs from the at the two sprockets z,, and Zg, the number of rollers n and the
works by most of the above mentioned authors. centre distance a. A coordinate system is defined in the centre of
The angle of contact force on the sprocket varies a lot. This the driving sprocket as shown in Fig. 1. This is the global coor-
variation affects the distribution of contact forces and chain link dinate system, used to describe the geometry, when calculating the
tension. It is therefore important to calculate the contact angle position of the rollers and the sprockets.
exactly at each tooth to predict the load distribution. The position The two sprocket centres are assumed to be located at the same
and subsequent contact angle can be calculated from the equilib- height with the gravitational force acting in the negative
rium equation together with geometric conditions. The external y-direction. The gravitational force at the chain is assumed to act
forces and torque acting at a sprocket are affected by the driving in the roller centre. All parts of the model, both chain and sprock-
torque, load torque, gravitational force as well as the spans. To ets, are regarded as two dimensional bodies and all variations in
the direction along the rotation axis of the sprockets are neglected.
It is furthermore assumed that the two sprockets are working in the
Contributed by the Mechanisms Committee for publication in the JOURNAL oi'
MECHANCAL DESIGN. Manuscript received Jul. 1998; revised May 1999. Associate same plane and trictional effects are neglected in the model. Since
Technical Editor: G. S. Chirikjian. the model is assumed to be strictly static, all inertia forces are
402 / Vol. 121, SEPTEMBER 1999 Copyright © 1999 by ASIWE Transactions of thie ASI\/IE
Nomenclature
a = Centre distance ^6011 = Bottom radius T) = Coordinate along the tooth flank
C = Chain link stiffness R,ii, = Tip radius (non dimensional)
F = Force / = Pitch 6 = Chain angle
g = Gravitational constant x, y = Coordinates ^ = Angle to previous roller
h = Distance z = Number of teeth
k = Distance a = Pitch angle Subscripts
m = Mass of the chain and roller j3 = Angle to following roller (• = Roller number
M = Torque 7 = Pressure angle g = Driven sprocket
n = Number of rollers r = Contact angle p = Driving sprocket
R,i = Pitch radius
As seen in Fig. 6, the curves stay very close to each other in the
interval 4 - 6 . It is because the rollers centre only moves a very
short distance in the radial direction relative to the sprockets
centre, and is therefore still close to the pitch radius. An opposite
effect could be seen in the Fig. 5 where the curve differs a lot in
the same interval, since the roller ; + 1 lands on a greater pitch
radius when the chain pitch increases and vice versa. It could also
be noted that the possibility of two solutions occuring seems to 0 1 2 3 4 5 6
decrease with higher numbers of teeth, since the maximum deriv- n,
ative of the curve decreases. This means that the risk for two Fig. 5 Relationship between the location of the contact points of two
solutions is greatest for small numbers of teeth. The Fig. 5 showed adjacent rollers
that this did not occur for a sprocket with ten teeth. This implies
that two solutions do not exist for a normally working sprocket
with ten teeth or more.
F,+i and F^ are possible to determine if the previous tension force
When static force equilibrium of the chain and sprocket in F|, the mass m, and the coordinates for the rollers ;' - 1 and / +
contact is calculated, all friction forces and friction torques are 1 are known. Equilibrium gives the contact force
neglected. This means that four forces act at one roller; the two
tension forces, one in the previous and one in the following link, sin (i// + j3) mg cos j3
the contact force and the gravitational force. (2)
^' " ^' • s i n T r r ^ ^ sin ( F T ^
The coordinates ip^j and (pg„, are introduced to describe the
position of the two sprockets. Coordinates to describe the angle and the relation between the tension forces
between the centre of the first roller contacting each sprocket on
the tight side are also introduced as (p,,, and tpi^. The driving sin (F + i//) mg cos F
f,-n = Pi sin (F + |8) (3)
torque is called A/,, and the load torque M^. sin (F + /3)
In Fig. 7 the forces and their directions are shown. The forces
It is possible to predict the chain tension of the following link,
as long as the position of the next roller is known. This is the case
if the position of one roller is known and all the elastic deforma-
tions are neglected. Since elastic deformation exists, an iterative
method has to be used to calculate the tension force in the follow-
ing link and the thereby given coordinate for the next roller. As
(= 15.875 mm
^1=5.2 mm
r^= 12.7 mm
^M 3
ijfi
0 1 2 3 4 5 6
e„ (5)
and
F„,i, + n-mg-sin (6,,,,,,) (6)
where n is the number of teeth in the span, mg is the gravitation
force, and 0,„,, and 0,„g are the angles for start and end point.
Fig. 7 Sprocket with the contacting chain part
Slack Span Section
The slack side is quite different from the tension span, since the
shown in Eq. 3 it is necessary to know the force in the previous tension force is very small here. The major part of the tension force
chain link. It is obvious that the only links where the force can be comes from the gravitation force in the chain. This side can
calculated from the compatibility are the first and the last one at the therefore not be assumed to be a straight line. When calculating the
sprocket. These are the link which connects the first roller at the positions and tension forces in this span, it is assumed that the half
sprocket to the last roller at the tight span, and the link which mass of the links and the mass of the roller are lumped to a point
connects the last roller in contact with the sprocket to the first one at the centre of the roller here as well. A method of how to
at the slack side. This force could be calculated from geometric calculate the coordinates for the slack span, shown in Fig. 9, is
parameters, moment equilibrium and compatibility. The parame- described below.
ters which are known from the compatibility are the tension side Equilibrium gives
force, the slack side force and the driving torque. The moment
equilibrium gives the compatibility equations Fj sin (pj = Fj_, sin £Pi_|
M,, = /^o.,,^i,,, cos (cpi.p - e,„,p) - F„^i,R„j, cos ((p„,,, + e,,,„„,) F, cos <p, + mg = F,_i cos (p;_i (7)
Combining these two equations give
- mg-Y, Ri.p sin (<P;,„) mg sin (pi
F„ = 1, 2,
' ' sin (<p, - (p,_,) '
^ « = -Po.gRi.s COS (9,,, - 0,„,j,) + F„,gi?„,s cos (ip„,j + 9j.„„,)
mg sin (p,_
n 1, 2, (8)
sin (cp, - <p,_,)
+ mg • 2 Ri.« sin (ip,J (4)
This means that there are two equations for each roller, except
for the one at the ends.
where /?,,, and R,j, are the radii to the first contacting roller and
R„j, and R„,g are the radii to the last contacting roller at the
sprocket. The position of the rollers can be calculated numerically
using an iterative method when solving Eqs. 4.
AJC;
/i = 2 sin (f, - E ^ = + Ri.p cos^ (p,J • sin (pi.j + {Ri,p sin ip,,,, + a)
^y, Fo.p ( i + «) . \ y 2
/2 = E COS 9/ - 2 ^= 0 n\ ti„„i, + ~^+ 2-Co .^ mg sin cp,,,,,,} ] + Rl.S
number of rollers in contact with the sprockets is correct, the same 200
checks are made as on the tight span. The forces on the last roller
which contact each sprocket can now be calculated properly from
the geometric compatibility. If the contact force on any of these 150
rollers is negative, the number of rollers at the slack span is
• ^,l.f
increased by one and a new span has to be calculated from the ^
100
second last contacting roller at the sprocket. Then the forces on the
second last roller are calculated as if this is the last one and the sign r;
of the contact force is checked again and so on until the contact
50
force is positive. A check to see if the first or last roller at the span r; ^
contacting any of the sprockets is made as well. In order to get a
stable solution the limit is given that the number of rollers only is ^
allowed to increase or decrease by one in every recalculation of the
slack span. When the slack side has passed the checks, a possible
geometric solution is obtained. Now the tension force at the first -50 1 2
link given by the slack span calculation is compared to the tension 3 4 5 6 7 9 10 11 12
(p/(2a)
force in the same link given by the calculation of the sprocket. The
same is made for the last link at the span. This is defined as Fig. 11 Roller centre radii and contact angles
^F=\F,.„-F„J
AF., (13) where the contact normals are nearly radial, and therefore do not
transmit any power. It can also be seen that none of the rollers
were F,,,,, and F„_g are the forces in the link after the last contacting contact the opposite flank of their tooth space. At the driving
roller at the sprockets while F,,,,, is the force in the first and F„,, the sprocket there is also a power transmitting zone as well as a less
last link at the slack span. These forces are all dependent on TJ,,, power transmitting zone, even if the last zone is smaller then that
and T)i,j,. It is obvious that AF,, and AF^ have to be zero when the on the driven sprocket. However the last roller transmits some
solution is correct. To find the value of TJ,,,, and TJ, ^ where AFp power since the contact normal is not exactly radial. It is also
and AFg are zero the Newton-Raphson method in two dimensions noticeable that the angle between the first and the last load taking
has been used. When calculating dFJd'r]^, dF^/drj^, dFJdr],, and roller at each sprocket is in the same range. This could be a
dFJdi],, a difference equation was used. A problem occurred similarity to a belt drive. If the driving torque increases, the last
where the Newton-Raphson method did not converge when getting rollers will climb and contact the tooth at a greater radius so that
close to the zero point. This appears to be an inflexion point and more power can be transmitted. If the driving torque increases
the derivative dF^/drj,, and dFJdT]^ approached zero at the same even more, the contacting rollers will climb out on an even greater
time as the derivative dFp/drjg and dFJdt],, increased and became radius and the distance of the slack span will increase as well as the
dominant. An other problem was that the derivative given by the force in the slack span. If the necessary force at the slack side
difference equation was not sufficient when getting close to zero. cannot be exceeded, the last roller will start to jump over the teeth.
To solve this problem a method was developed where the deriv- In order to visualise the results, a rollers' centre radius and the
ative was limited to remove the instability. corresponding contact angle is shown in Fig. 11. The radius is
calculated relative to the sprocket's centre and is then compared to
the pitch radius of the sprocket. The contact angle T is calculated
Numerical Example relative to the middle of the gap between two teeth. The contact
A chain drive is studied with the number of rollers n = 40, the
number of teeth at the driving sprocket z^ = 10, the number of
teeth at the driven sprocket Zg = 20, the centre distance a =
196.8 mm, the pitch at the two sprockets t,, = tg = 15.875 mm, 250
1
the chain pitch f/,,,* = 15.95 mm, and the constant driving torque F[N]
Mp = 5 Nm. The contacting points and their normals for the
studied example are shown in Fig. 10. Since friction is neglected,
200 '\ ^ XvA;yj\. \AJ vA;
the contact forces will act along these normals. For the driven
sprocket there is one zone where the rollers take load and one zone
150
\
100
I -
11
50 \
\
\
\ ... " T " 1 .-
0
0 12 16 20 24 1
28 1
32 36 40
ip/(2a,,)
Fig. 10 The contact normals given by the model Fig. 12 Tension forces and contact forces In a complete chain
II
from the point 4 to the point 6. The curves change a lot when the 2.02
. Rachner's model
torque increases to the point where the last roller at the sprocket
takes load, since the centre radius of the last roller then can be
greater than the first. 2.00
The tension force Fi + i in the link i + 1 and the contact forces
/ \
This model •'
F,. at the roller / are shown in Fig. 12. The forces are plotted ''
against a non-dimensional coordinate which has its origin at the 1.98
first contacting roller at the driving sprocket and acts along the
driving sprocket. Between zero and one the link is not in contact
and the variation in force depends on the polygon effect. When 1.96
contact occurs the contact force increases and the stretching force 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
decreases. When the next roller gets into contact the curves change
direction. The two forces decrease until the roller leaves the
Fig. 13 Torque ratio when the driving sprocl<et is turned one pitch
sprocket and enters the slack span. Then the contact force is zero
until the roller enters the driven sprocket. The way the coordinate
is defined, the stretching force at the driving sprocket will now be
the local force F,. This is why the contact force decreases to zero distribution are affected by the pitch of the chain, the pitch of the
after the stretching force has entered the tension side at point 29. sprockets, the elastic deformation and also by the torque's on
After coordinate 29 the tension is affected by the polygon effect. driving and driven sprocket. It is possible to find solutions very
Notice that since the driving and driven sprocket do not work in close to the point where the jumping over teeth phenomena occurs.
phase, a roller will not leave and enter the driven sprocket at an Also there is a difference in the polygon effect when comparing
even coordinate in Fig. 12. the developed model to previous models. This difference is caused
From the forces shown in Fig. 12, the driven torque can be by the slack span as well as the fact that the radius to the first
calculated according to Eq. 4. The torque at the driven sprocket, contacting roller at each sprocket vary during one pitch.
calculated from this theory as well as from Rachners [3] theory, is
shown in Fig. 13. It can be seen that a roller leaves the driven
References
sprocket at 0.75. Since the pitch angle is constant and the entrance
radii Rip and /?i„, do not differ much from the pitch radii, the 1 Binder, R. C , 1956, Mechanics of Roller Chain Drive, Prentice-Hall, Engle-
wood Cliffs, New Jersey.
polygon effect cannot be expected to be affected a lot. It can be 2 Hertz, H., 1895, Gesammelte werke, Leipzig.
seen in Fig. 13 that the curves stay relatively close to each other, 3 Kim, M. S,, and Johnson, Glen E., 1992, "Mechanics of Roller Chain-Sprocket
but the difference is visible and will for instance affect the results Contact: General Modelling Strategy," ASME.
if a dynamic calculation is made. 4 Kim, M. S., and Johnson, Glen E., 1992, "Mechanics of Roller Chain-Sprocket
Contact: Observations about the Contact Phenomena and Load Distribution," ASME.
5 Naji, M. R., 1981, "On Timing Belt and Roller Chain Load Distribution," Ph.D.
Summary Dissertation, University of Houston, Houston, Texas.
6 Naji, M. R., and Marsheck, K. M., 1983, "Experimental Determination of the
The load distribution in a chain drive with two sprockets was Roller Chain Load Distribution," ASME Journal of Mechanisms, Transmission and
calculated. The method used in this paper is shown to be a Automation in Design, Vol. 105, pp. 331-338.
powerful tool to calculate the rollers' position along the chain as 7 Naji, M. R., and Marsheck, K. M., 1984, "Analysis of Roller Chain Sprocket
Pressure Angles," Mechanisms and Machine Theory, Vol. 19, No. 2, pp. 197-203.
well as the forces acting at the rollers and the sprockets. The 8 Rachner, H-G., 1962, Staalgelenkketten und Kettengetriebe, Springer-Verlag,
contacting angles between sprockets and rollers and the load Beriin.