Académique Documents
Professionnel Documents
Culture Documents
Computer-Aided
&sign, Vol. 30. No. 3, pp. 199-204, 1999
0 1999 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
0010~4495/99i$19.00+oo.oo
INTRODUCTION
Traditional
computer-aided
design has tended to focus
largely on recording the shapes of proposed designs, as
specified by the designer. Some significant advances have
now been made in the direction of allowing shapes to be
generated automatically: either through parametric design ,
or by allowing users to specify design rules to be applied
under specific, well-defined conditions2. A more long-term
line of research at the Cambridge Engineering
Design
Centre is attempting to tackle the problem of generalised
embodiment design: that is, the fleshing out of conceptual
solution to almost any type of mechanical problem such that
the components
are automatically
assigned dimensions
which will guarantee their ability to carry out their intended
functions.
The details of the approach, the problems which it has
raised, and the current solutions to those problems have
been extensively
reported elsewhere3-.
The technique
essentially involves describing the parts of a proposed
design in terms of commonly
understood
engineering
199
Open architecture
kinematic analysis:
A L Johnson
of a mechanism
from the
KINEMATIC INTERFACES
Designers who have attempted to specify three-dimensional
kinematic models will be well aware of the difficulty of
creating models which are not over- or under-constrained.
This difficulty is clearly made worse when (as in this case)
there is a requirement for the kinematic model to be
generated automatically from the initial, component-oriented,
description.
The approach to kinematic interfaces adopted in Promech
addresses this problem in two ways. Firstly, the kinematic
pairs between bodies are not specified explicitly by the user,
but are defined implicitly from the relevant features of the
two parts involved in the pair. For instance, if an offset
section of a crankshaft (i.e., a short length of shaft) is
connected to the big end of a connecting rod, a revolute
joint is automatically
implied: the combination
of a
cylindrical shaft and a cylindrical hole whose depth is not
negligible means that the joint must either be cylindrical or
revolute; and the crank webs at each end of the section of
shaft imply that the connecting rod cannot slide along the
shaft, making the joint revolute.
Secondly, the software has been written to make it easy to
relax some of the constraints implied by the joints, without
the need to re-specify the features of either part. For
instance, the connecting rod mentioned above might be
re-declared by the software to have a very short hole through
it, if the longer hole had the effect of over-constraining
it:
this effectively converts the revolute joint into a spherical
joint, without the need to redefine any physical aspect of the
bodies involved in the joint.
The kinematic pairs which Promech can currently handle
are made up from all possible combinations of the features
shown in Figure 1. A shaft element can be plain, or it can
have a circlip (to prevent axial motion) and/or a keyway (to
prevent rotation about its own axis). Likewise, holes can
either be shallow or deep, and in addition can either
The combination
rules are
be circular or slotted.
Groove
for Circlip
Keyway
08 nil
6 w
PLANAA
w
+A
$R
@A
$A
+A
@T
SPHERICAL
CPS
$A
a:
cD
Figure 1
200
4A
CYLINDRICAL
@p
Feature combination
REVCUTE
4s
PRISMATIC
+p
Figure 2
Basic kinematic
constraints
Symbol
Description
Sliding constraint:
Axial constraint:
Lateral constraint:
Crossing constraint:
Parallel constraint:
Twisting constraint:
= 0
(S - H)S2
(S - H)S3
= 0
= 0
(S - H).Hl = 0
(S - H).H3 = 0
(S - H) X H2)Sl
Sl.H2
Sl.H3
Driving constraints
In addition to the static kinematic constraints implied by the
joints between the various parts of the mechanism, so-called
driving constraints
must be specified to give it motion.
Promech currently allows two types of driving constraint,
which together can cover virtually all motions found in
practice: a rotary driver, which twists the shaft vectors S2
and S3 about the shaft axis vector Sl, and a linear driver
which moves the shaft reference point S along the shaft axis
vector. These two motions can be used to simulate rotary
input and linear actuators, respectively: in conjunction, they
can be used to simulate a screw. Since virtually all helical
joints in real mechanisms are part of a drive system, this
facility adequately overcomes the lack of a helical joint
amongst the static constraints.
No of constraints
(S - H)Sl
Vector condition
2
= 0
= 0
= 0
S2.H3 = 0
Sl.H3
= 0
Vs.VH= e.
(1)
(6eA x v,b(v,
whence (MA X
x vspe,
or (Vu
+ (6eB x v,)) = 0
(2)
_
A
orAS=e
_ 64
(3)
(4)
Figure
201
Open architecture
J
Joint type:
Shaft in short
round hole
IbA
Previous
constraints
i
Current
constraints
ALIGN(
ALIGN(S3) +
Figure 4
(3
(6)
(7)
where d contains
bodies, whence
$=A-
e,
velocities
of the
(8)
e,
(10)
where A- IS
again the same as for the previous calculations.
Thus, once the position analysis has been carried out,
velocities and accelerations are then both calculated exactly
without the need either for further matrix inversion or for
202
INERTIA LOADS
Once the linear and angular accelerations of each body have
been found, the resulting inertia (dAlembert) loads on the
mechanism can be calculated. It can readily be shown from
first principles I2that these loads can be expressed as a force
of - MaGacting though the centre of gravity plus a couple
of - J(? - 0 X (Jt9), where:
M is the mass of the body
aG is the acceleration of its centre of gravity and
(9)
or
I xx
-Ix,
- IXY
IYY
- Ixz
- IYZ
- Ixz
-I,,
Izz
.
I
J = RJRT,
where J is the inertia matrix in the bodys local co-ordinate
system, and R is the transform matrix which maps the
bodys co-ordinate system into the global system.
When these loads have been calculated, they can be
combined with any other external loads applied to the
mechanism, and used to calculate forces and couples at
the interfaces between the bodies. This process involves
using the same constraint information as was used in the
kinematic analysis to create and solve a set of simultaneous
equilibrium equations for the bodies. The method for doing
this is described in detail elsewhere16.
Figure
Kinematic
probe.
204
REFERENCES
Roller, D., et al., Dimension-driven Geometry in CAD-a Survey.
Theory and Practice of Geometric ModeRing. North Holland, 1989.
Wagner, M. R., Understanding the ICAD System. ICAD Inc, 1990.
Johnson, A.L., Designing by functions. Design Studies, 1991, 12, No.
6.
I.
8
9.
10
Il.
12.
13.
14.
15.
16.
17.
Thornton, A. C., A support tool for constraint processes in embodiment design. In Proc. 6th Int. Conf on Design Theory and Methodology (DTM 94), ASME, 1994, pp. 231-239.
Thornton, A. C. and Johnson, A. L., Constraint specification and
satisfaction in embodiment design. In Proc. Int. Conf on Eng. Des.,
ZCED 93, The Hague, Netherlands. Hem&a, Zurich, 1993, pp. 13191326.
Yao, Z. and Johnson, A. L., Formalising knowledge for constraint
satisfaction in the design nrocess. In ICED 95. Prague. WDK, V
Hubka, Hemista, Zurich: 1995.
Yao, Z. and Johnson, A. L., On estimating the feasible solution space
of design. Computer-Aided Design, 1997, 29, 649-655.
Martin, E., Getting Started with the ACIS 30 Toolkit. Schemers, 1995.
Mechanica Applied Motion Users Manual, Rasna Corporation, San
Jose, 1995.
Gelsey, A., Automated reasoning about machines. Artificial Intelligence, 1995,74, l-53.
Joskowicz, L. and Sacks, E. P., Computational Kinematics. Artificial
Intelligence, 1991, 51, 381-416.
Haug, E.J., Computer-aided Kinematics and Dynamics of Mechanical
Systems Vol.1; Basic Methods, Ch. 9. Allyn and Bacon, 1989.
ADAMS Users Manual, Mechanism Dynamics Inc., Ann Arbor,
Michigan, 1987.
Kramer, G. A., Solving Geometric Constraint Systems. MIT Press.
Cambridge, MA.
Shigley, J. E. and Uicker, J. J., Theory of Machines and Mechanisms.
McGraw Hill, New York, 1980.
Johnson, A. L., Mechanism analysis by matrix reduction. Part I:
Statics. Proc. R. Sot. Lond. A, 1992, 439, 485-495.
Blessing, L. T. M. et al., Applying Systematic Design: the Flight
Refuellinn Probe Proiect. Cambridge EDC nublication CUED/CEDC/TR48, 1997.