You are on page 1of 7

C.

Bagci Optimum Synthesis of Plane


Associate Professor
Mem. ASME
Mechanisms for the Generation of
In-Ping Jack Lee Paths and Rigid-Body Positions via the
Graduate Research Associate
Linear Superposition Technique
Department of Mechanical Engineering
Tennessee Technological University A method of optimum synthesis of plane mechanisms for the generation of paths and
Cookeville, Tenn. rigid-body positions is presented. The method is developed for the four-bar plane mecha-
nism with six and eight unknown dimensions. Dimensions of the optimum mechanism
are determined by minimizing the error in the loop-closure equations for N design points
on the path, along with the loop-closure equation of the linkage, where N is not limited
by the number of the unknown dimensions of the system. Design equations are linearized
by the method of linear superposition. Solution of design equations requires no itera-
tions, and it leads to a series of optimum mechanisms of different efficiency of approxi-
mation. Numerical examples are given.

Introduction earized loop-closure equations are used along with the linear sup-
erposition technique, no iteration is required. In this paper it is
Precision position techniques of mechanism synthesis have cer- applied for optimum, design of the plane four-bar mechanism for
tain limitations. The number of precisior. positions along a func- the generation of paths and rigid-body positions. The method is
tion, or a path, is limited by the number of unknown dimensions applied for optimum synthesis of plane four-bar and slider-crank
of the mechanism, and the error in the generated function, or mechanisms for function generation in [14].
path, between two precision positions is not controlled. This is a
very undesirable situation when generating functions, paths and P a t h Generation
rigid-body positions within large intervals of motion. Error mini- The four-bar mechanism at the ith design position is shown in
mization techniques [1-12] 1 permit the use of many design po- Fig. 1. Coordinates of the design points along the path are defined
sitions with a few unknown design parameters distributing the by Pj = PiVii in the OXY system, and Pi are specified in a de-
error somewhat uniformly within the design interval. Error mini- sign problem. Uji = cos dt i + sin 5, J is the unit vector posi-
mization techniques can also separate the geometric inversions of tioned by the angle di. The fixed link of the mechanism is posi-
a mechanism [7, 10], but these techniques require iterative solu- tioned by R2 = R2VCV2 and R4 = RiUai. Dimensions of the mecha-
tions. Synthesis of the plane four-bar mechanism for function nism are defined by d& = duUtiki at the ith position, a and 7 position
generation with optimal structural error using five precision the coupler point P with respect to the coupler link. Equation of
points and constraining the extreme errors occurring between pre- loop-closure for the coupler point at the ith position, using the
cision points to be equal in magnitude is given in [13], where iter- 0 0 2 A P path, is
ative solution is also required. In the following, a method of opti-
mum synthesis of plane mechanisms using many design positions R2 + d 2 . a. = P. (1)
and linear superposition technique is presented. The optimum
mechanism is obtained by minimizing the error in the loop-clo- Furthermore, the dimensions of the four-bar loop O2ABO4 must
sure equations at each design position. The method is general and satisfy the loop-closure equation
applies to all kinds of plane and space mechanisms. Since the lin-
d, + d, = d, + d. (2)

1 At the ith design position, where di = R4 - R2. Equations (1) and


Numbers in brackets designate References at end of paper.
Contributed bv the Mechanisms—DED Division of THE AMERICAN (2) are used to determine the design equations for path generation.
SOCIETY OF MECHANICAL ENGINEERS and presented at the Mech- Separate x and y components of equations (1) and (2) to have
anisms Conference, New York, October 6-10, 1974. Manuscript received at
ASME Headquarters May 22, 1974. Paper No. 74-DET-10. R2 c o s a 2 + d 2 c o s 02. + a cos ip{ — P{ cos 6{ (ia)

340 / FEBRUARY 1975 Transactions of the ASME

Copyright © 1975 by ASME


Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
DESIGN POINTS-
DESIRED PATH to half of the number of unknown dimensions of the system. In
the following, JV is much greater than the number of unknown di-
mensions and equations (3) and (4) are solved to generate the
coordinates of the design points approximately by the optimum
mechanism. The dimensions of the optimum mechanism are
those which render

EJ=EFJi\ J =1,2 (6)

Optimum Design With Six Unknown Dimensions


In this case each of equations (3) and (4) are solved for three
unknowns to render minima for Ei and E2 at JV design positions.
Such an error minimization technique assigns the same weight to
the error in the loop-closure equation at each design position. Let
a2, <*4 and B2t at N positions be specified. Then, rearrange equa-
tions (3)to have
3

Fig. 1 The four-bar plane mechanism at the fth design position when Fi.= {T,KjGJl)-KiHl.-Hm. = Q, i= 1,...,N (7)
generating a path
where Klt K2, K3 and K4 are the unknowns and defined by
R2 s i n a2 + d2 s i n 0 2 . + a s i n ipt = P. s i n 6, (it,) Kt = R2,K2 = d2,K3 = a2 - d22 - R22,KA = d2R2 = KiK2
and and
d2 c o s 0 2 . + d3 c o s 0 3 , d . = 2P{ c o s (a2 - 0j), G 2 . = 2Pi cos (0 2 . - 6 , ) , G 3 . = 1,

= R 4 c o s o>4 - R2 cos a2 + d 4 cos 04 (2a) H,. = 2 cos (of2 - 62 ) , Hm. = P?


1 l
i
rf, s i n + do s i n Similarly dividing equation (4) by 2d 3 yields
3
JT!4 s i n a 4 ~ R2 s i n a 2 + d 4 s i n 0 4 . (2b)
Ft.= (EK/G,,1) -KSH,.' -Hm ' = 0, i=l,...,N (8)
where \£, = 8u + 7, 7 being angular dimension. Eliminate i£j in
equations (la) and (lb), and Bu in equations (2a) and (2b) to where K-i , K2', K3' and Ki are unknowns and defined by
have, respectively, -
Kt' = l/d3,K2' = Rt,K3' = W + d2 -d2 + R2 + R2
2 2 2 2
F t . = P; + R2 + tf2 - a - 2PjR 2 c o s ( a 2 - 6 {) - 2R2RA cos (a2 - a 4 ) ] / ( 2 d 3 )
- 2Pid2 cos (0 2 . - 6j) + 2R2d2 cos ( a 2 - 0 2 .) = 0, Ki = Ri/di = Ki'K2'
t = 1 AT (3) and
and G, ' = d2R2 cos ( a , ~ 92 ) , G2 ' = - c o s ( a 4 - 03 ), G 3 ' = 1,
F2. = d22 + d32 - d 4 2 + R22 + RA2 + 2d2d3 cos (0 2 - 03 ) i i i i t
* i i H, ' = d2 c o s ( a 4 - e2 ), Hm '
i i i
+ 2d2R2 c o s (a2 - 02 ) -2d 2 .R 4 c o s ( a 4 - 0 2 .)
= ~R2 c o s ( a 2 - 0 3 .) -d2 c o s (0 2 . - 0 3 .)
+ 2d%R2 c o s {a2 - 0 3 .) - 2 d 3 K 4 c o s ( a 4 - 0 3 .) - 2J22R4
Since if4 and Ki are defined in terms of the unknowns Klr K2,
cos (a 2 - a 4 ) = 0, i = 1,. . ., JV (4) and Kt', K2', respectively, the equations of compatibility
at JV design positions. Equations (3) and (4) are the design equa- A - KiK* = 0 0)
tions that must be satisfied at JV design positions. JV is not limit-
ed by the number of unknown dimensions that equation (3) or (4) and
can have. Equations (3) are solved first for some unknown dimen- X' -Ki'Ki' = 0 (10)
sions, then, values of 831 are determined for each design position must be satisfied along with equations (7) and (8), where X = K4
using equations (la) and (lb). Thus and X' = K^. This is accomplished by using the linear superposi-
P, s i n 6,. — R, s i n a , - d2 s i n 0, tion technique. Set
t a n (0 3 . + y) = - J 1 1 ? 1 iL. (5)
1
P. cos 6 f - R2 cos a2 - d2 cos 02 Kj = Xlj + mj

Ki = X'li + mi,
In this case 7 must be one of the specified dimensions. Equa-
tion (5) gives the first approximated values of BSt after R2, a2, a n d and partition equations (7) and (8) by separating terms with and
d 2 are determined. These values of Bu are used to solve equations without X. Thus, equation (7) is partitioned into
(4) for the remaining dimensions. One should note that 82i could
have been eliminated to arrive at equation (3). In this case Bsi would Ft =(HlJG,)-H,. = 0 (12)
be specified removing the control on the input crank rotation. 'i j=i ' '
Such a design equation is used for the rigid-body guidance in the
later part of the paper. When the number of design positions JV is FX = (T,mJGJ)~Hm =0 (13)
the same as the number of unknowns in each of equations (3) and
(4)—that is, when the design points are precision points—the and equation (8) is partitioned into
coordinates of the design points can be generated exactly by the 3 '
solutions of equations (3) and (4). Exact generation of the preci- F2j ={'Blj'GJ.')-Hl/ = 0 (14).
sion coordinates, however, limits the number of precision points

Journal of Engineering for Industry FEBRUARY 1975 / 341

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


F2
= ( S m / G / ) - H „ ' = 0
m (15)
i = J=I < <
According to equation (6), the optimum values of (j, mj, a n d / /
and m/ are determined to render
N
Ek = TJF,\ k= 1,2

Ek =T,Fk \ k= 1,2
W 9
minima, or to satisfy
8Ek_ N dFb Fig. 2 The path, 12 design points and the four-bar mechanism opti-
>- = T,2~^Fk, =0, L { f ^ ' 2 ' 3 (16) mized to generate the path in Example 1
dl J i=i 9?.,
« / ^V. fe = 1,2
and

dE„
0 { J = 1,2,3
9 ' X
k = 1,2 (17)
(=1 m(
B j = JR4 s i n aA ~ R2 s i n a2 - d2 s i n 62.
Equations (16) and (17) are four sets of linear equations in un-
knowns j , mj, / and m.j'. Both equations have the same coeffi- Dt= (A2 + B? + d2-d2)/{2di)
cient matrix for a value of k. Thus, the optimum values oi'tj and
m.j, for k = 1, are given by the solution of The root with ( - ) sign in front of the radical in equation (21) is
[C\[L) = [H] (18) for the uncrossed geometric inversion, loop O2ABO4 of the four-
where [C] is the 3-X-3 symmetric coefficient matrix whose coeffi- bar mechanism shown in Fig. 1, while the (+) signed root is for
cients are given by the crossed geometric inversion, loop 02AB'Oi shown in dashed
lines in Fig. 1. Since each mechanism has two geometric inver-
sions, the solution leads to eight distinct optimum mechanisms.
•rt — t-i \Gr Gt >,r,l= 1,2,3
Efficiency of the approximation may be tested either by the
mean deviation of the generated coordinates from the desired
[L] = m2 i s t h e m a t r i x of unknowns / , and m, values, average error,
h
.h 1 "
[H] is a 3-X-2 matrix, and the coefficients of its columns are * J + \y* -yB\)i
given by
If X or by the Root-Mean-Square-Error (RMSE)
hj =T,(GJHl), h, =T,(Gj Hm), J= 1,2,3 w j 1/2
RMSE = [{£[(*„ ~xg)t2 ~yg)(2}}^)
2
i i=l ' i j=i < « + (y„
respectively. Introduce primes into equation (18) for k = 2, then
solve for the optimum values oil,/ and mj'. Different values of a2, «4 and 7 for a specific set of values of 821
After substituting K\ and K2 from equation (11) into equation will result in different mechanisms, thus permitting infinite num-
(9) the compatibility equation becomes ber, of solutions. A designer may design a series of mechanisms
2 m w varying a2, «4 and 7 and choose one which has a good efficiency
X ld2 + x(Zjm2 + ™ih ~ 1) + l' 2 = ° (19)
of approximation and good transmission angle.
Equation (19) is solved for two values of X. For each real value of Example 1. A coupler point curve shown in Fig. 2 is to be used
X equation (11) yields a set of optimum values of Kj. If equation to drive a dwell mechanism, and it is required that the portion of
(19) has no real root a different set of specified dimensions should the coupler curve between Points 1 and 5 approximate a circular
be used. Introduce primes into equation (19) and solve for two arc of center Op and radius 2.875 in. A four-bar mechanism is de-
values of X' for each value of X, which yield four sets of optimum signed to approximately generate the path using 12 design points
values of K.j', and so four sets of dimensions of the mechanism. along the path whose specified coordinates, according to the num-
Among those, one or more mechanisms may approximately gener- bers shown in Fig. 2, are given in the first three columns of Table
ate the path. Generated coordinates of a design point are comput- 1. The third column in the table gives the specified values of 621.
ed by the left sides of equations (la) and (lb) after the generated Other specified dimensions are a2 = 56 deg, a^ = - 6 deg, 7 = 23
values of (03i)g are determined from equations (2a) and (26), deg.
where subscript g stands for "generated." Thus, Solution resulted in mt = -0.6626, m2 = 2.7487, m3 = 20.5574;
tan {93\ li = 0.5480,',£2 = -0.0515,^3 = -4.7185; m i ' = -0.7467, m 2 ' =
-21.3982, m 3 ' = -20.7888; 4a' = 0.8218, l2' = 29.5278,. t3' =
Rn s i n a 4 -R2 s i n a2 -<f2 s i n 02 + d 4 s i n (#4 A, 27.5047; Xi = 4.3653, Xi' = 1.0432, X2' = 0.6310; X2 = 14.8013, X3'
(20) = 1.11612, Xi' = 0.09944, and eight geometric inversions. Geomet-
RA c o s a 4 - R2 c o s a 2 - d2 c o s 0 2 . + ^4 c o s (#4.
ric Inversion / of the mechanism defined by Xi = 4.3653 and Xi' =
where (9n)g is computed for each geometric inversion from the 1.0432 for which Kx = R2 = 1.7294, K2 = d2 = 2.5240, K 3 =
roots of the Freudenstein's displacement equation of the four-bar -0.0392; Kx = 0.1109, K2 = fl4 = 9.4122, Kz' = 7.9106, and a =
loop determined from equations (2a) and (26) after eliminating 3.0533 in., d 3 = 9.0206 in., d 4 = 4.6176 in. This mechanism is the
0 3 / .Thus, one shown in Fig. 2. Generated values of 84/, 83;, Pi, &t and the er-
0 4 .._, _ _-B, T JB,2 + A} •Di ~
rors eXi and eyt in the generated x and y coordinates of the design
(21) points are given in Table 1. Average error in the coordinates of
tan ( V
2 )
D, the design points is em = 0.00201 in. and RMSE = 0.002618 in.
where The smallest RMSE for the other mechanisms was 0.716351 in.
for Geometric Inversion 2 of the mechanism defined by X2 and
At = RA c o s a 4 -R2 c o s a2 -d2 cos B2. Xi'.

342 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 1 Performance of the optimum four-bar mechanism synthesized
in Example 1 and shown in Fig. 2

i P
i 5
i
a2. <%•> <"l'. <v„ «i>„
e
x( e
'i
(in.) (deg) (deg) (deg) (deg) (in.) (deg) (in.) (in.)

1 3.99 71.60 160.0 112.9082 6.1894 3.9905 71.5818 .001360 .000078


2 5.08 65.15 130.0 102.4669 1.0012 5.0792 65.1572 r.000911 -.000440
3 6.05 56.10 100.0' 87.7159 -1.8381 6.0511 56.0865 . .001797 .000108
4 6.76 46.00 70.0 71.2975 -2.6408 6.7616 46.0482 -.003011 . 005074
5 7.10 36.15 40.0 56.3551 -1.2444 7.1024 36.1313 .003320 - . 000440
6 6.98 27.70 10.0 47.9207 3.6317 ' 6.9800 27.6637 .002054 -.003914
7 6.26 22.60 -20.0 53.6831 13.8957 6.2605 22.5769 .001389 -.002150
R 4.94 21.40 -50.0 73.0402 25.8472 4.9382 21.3673 -.000662 -'. 00328Z
9 3.49 24.00 -80.0 93.0204 31.2475. 3.4936 24.0891 .001075 .006424
10 2.47 36.50 -110.0 107.1717 28.9473 2.4684 36.4403 .000267 -.003000
11 2.30 58.80 -140.0 115.1867 , 22.0291 2.3035 58.8530 -.000024 .004078
12 2.97 71.70 -170.0 117.2315 13.6353 2.9674 71.7052 -.001078 -.002396

Table 2 Performance of the optimum four-bar mechanism synthesized


in Example 2 and shown in Fig. 4

E
i Pi 6
i e, «3, ( 8 2 1, ) '
q
K<)
1
9
(83.)
1 <!
({,)
' g
(Pj)
1
(«<) E

*i
v.
yi
„.
9 1
g •
(in.) (deg) (deg) (deg) (deg) (deg) (deg) (deg) (in.) (deg) (in.) (in.)

1 6.00 43.00 0.0 -10.0 120.8023 95.9755 -10.4948 -0.4948 5.9942 43.0175 -.005483 -.002604
2 5.00 45.60 6.0 -4.0 144.0289 120.0051 -3.1454 6.8546 5.0176 45.6261 .010680 .014169
3 3.88 43.00 19.0 9.0 173.8708 142.0879 9.1333 19.1333 3.8827 43.0373 .000255 .003701
4 3.10 28.25 44.0 34.0 -146.3824 151.2692 33.7286 43.7286 3.0969 28.0835 .001502 -.009404
b 3.30 14.00 67.0 57.0 -113.7467 147.3705 56.7706 66.7706 3.3027 13.8417 .004769 -.008216
6 3.75 9.45 80.0 70.0 -93.5064 140.7731 69.9933 79.9933 3.7499. 9.4297 .000149 -.001322
7. 4.40 9.13 ,90.0 80.0 -72.7630 127.9552 80.2189 90.2189 4.3918 9.1939 -.008837 .003543

Table 3 Performance of the optimum four-bar mechanism synthesized


in Example 3 and shown in Fig. 5

i P
i 6
i E
i H, "K'g '(N>q <•,,), (P,)
1 1 q ' E
*i E
*i
9
(in.) (deg) (deg) (deg) (deg) - .(deg) (deg) (in.) (deg) (in.) (in.)

1 12.0000 60.0000 90.00 45.00 -92.6013- 72.5299 44.6289 12.0114 59.7093 .058401 -.020694
2 12.5300 56.0368 88.00 43.00 -84.8251 69.0173 42.9705 12.4881 55.8533 .009741 -.057150
3 13.1149 52.4109 85.50 40.50 -77.8130 65.1995 40.6006 13.0725 52.2892 -.003886 -.050580
4 13.7477 49.1066 82.50 37.50 -71.5329 61.2831 37.7502 13.7098 49.1072 -.024917 -.028558
5 14.4222 46.1021 79.25 34.25 -65.4867 57.1433 34.3899 14.4070 46.1205 -.013912 -.007770
6 15.1327 43.3728 75.50 30.50 -59.8320 53.0540 30.7806 15.1109 43.4556 -.030860 -.000889
7 15.8745 40.8934 71.50 26.50 -54.0335 48.7919 26.7398 15.8582 40.9139 -.016048 -.006370
. H 16.6433 38.6391 67.50 22.50 -47.7943 44.3282 22.1968 16.6593 38.4664 .043781 -.029279
9 17.4356 36.5868 62.50 17.50 -41.333) 40.0658 17.5102 17.4533 36,3165 .063140 -.056678
10 18.2483 34.7150 57.50 12.50 -33.9369, 35.8837 12.4458 18.2864 34.4005 .088249 r. 060953
II 19.0788 33.0045 52.50 7.50 -25.8618 32.3621 7.5425 19.0823 33.0014 .003555 .001078

Optimum Design With Eight U n k n o w n Dimensions j = 1,2,3,4


Let d2, R2, «2, a and d3, diy #4, on be the unknown dimensions
in equations (3) and (4), respectively. Rewrite equation (3) to and
have
4 1U= T,Gjti„, J= 1,2,3,4
3
' • i=i ' •
Fi = (EKJGJ) -K,H^ ~K,Hm.-H„. = 0, j = 1,. . .,iV
' r-1 ' ' ' '
respectively.
(22) Two compatibility equations that must be satisfied along with
equation (22) are
where

Kt = d2,K2 = R2 s i n a2, K3 = R2 cos a2, •Kits (24a)


KA = a2 - R2 - cl2, t a n a2 = K2/K3, KXK2 (24b)
*2
K5 = R2d2 c o s a2 = K{K3,K% = R2d2 s i n a2 = KlK2

Gj = 2 P ; cos (92. - 6 , . ) , G 2 . = 2P f s i n 6;, \2LA, + X,A2 + A, = 0 (25a)


G 3 . = 2P f cos 6 „ G4 =1
i i A2 i?i X2B2 + B3 0 (25b)
,2
H, = 2 c o s B2,, Hm. : 2 sin H„ =P,<
l
i where Ai = m1m2, A2 = A1A4 + A5, A3 = Ai2A6 + AiA7 + Aa, Ai
Let 7f5 = A! and K6 = A2 and = fams + mifo, A5 = nxm3 + m1n3; As = tiU, A1 = nti3 +.fan3 -
1/Ag = rei7i3, Bi = mim2, B2 = A1B4 + B5, B3 = Ai2fl6 + A1B7 +
K, H., \2mj + iij, J = 1,2,3,4 (23)
Bs, B4 = t1m2 + miC2, B5 = n i m 2 + mjn 2 - 1, B6 =Cife, B7 =
I'j, mj and n.j are given by the solution of equation (18) in which "lfe + «2ti, Bg - n\n2. Equations (25a) and (256) are solved si-
[C] becomes a 4-X-4 matrix, [L] and [H] are 4-X-3 matrices. The multaneously for Ai and A2. Thus, using Sylvester's Dialytic
third columns of [L] and [H] are given by Methodof Elimination [19], Ai is given by ., . ... :. . ,•

Journal of Engineering for Industry FEBRUARY 1975 / 343

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 4 Performance of the optimum four-bar mechanism synthesized
in Example 4 with eight unknown dimensions, e m = 0.023164 in.

i P
i S
i «i e3j «'l\ ('**>, (
°'i>n
(
Vn
E E
yi
' 9
(in.) ((leg) (cleg) (deg) (deg) (deg) (deg) . (In.) (deg) (in.) (in.)

1 12.0000 60.0000 90.0 45.0 -92.7802 62.8436 44.4321 12.0601 59.8672 .054237 .038042
2 12.5300 56.0368 88.0 43.0 -89.2975 59.8068 42,5307 12.5439 55.8444 .042661 -.012055
3 13.1149 52.4109 85.0 40.0 -86.1709 56.6632 40.3467 13.0732 52.4280 -.028533 -.030672
4 13.7477 49.1066 82.5 .37.5 -82.8328 52.8330 37.4681 13.7294 49.0045 .006486 -.029870
' 5 14.4222 46.1021 79.5 34.5 -79.6609 48.7387 34.1774 14.4301. 45.9708 .029269 -.017279
6 15.1327 43.3728 76.0 31.0 -76.6324 44.4117 30.6232 15.1578 43.2714 .036612 -.002292
7 15.8745 40.8934 72.0 27.0 -73.7095 39.8910 26.5445 15.9018 40.8469 .029072 .008124
8 16.6433 38.6391 68.0 23.0 -70.7192 34.9892 22.0728 16.6910 38.5523 .053009 .010024
9 17.4356 36.5868 63.0 18.0 -67.8659 30.1634 17.5175 17.4660 36.5465 .023722 .002318
10 18.2483 34.7150 57.5 12.5 -64.9317 25.1526 12.6630 18,2408 34.6857 -.000835 -.011891
11 19.0788 33.0045 52.5 7.5 -61.7343 19.8455 7.3452 19.0798 32.9127 .017502 -.025093

Table 5 Performance of the optimum four-bar mechanism synthesized


to generate the Joukowski airfoil profile in Example 5

i P
i
6
i
0
2f (.„)„ 1*M\ <v, ( « • )
% \
On.) . (deg) (deg) (deg) (deg) (in.) (deg) (In.) (in.)

1 12.4200 .0000 4.2000 78.0850 40.5003 12.4171 -.1058 -.002919 -.022932


2 8.2008 13.3245 64.8700 78.6002 24.2457 8.2122 13.4645 .006504 .022012
3 4.5136 32.4225 104.8500 85.4529 24.9137 4.5122 32.4340 -.001667 .000013
4 1.6548 62.6607 145.8200 93.2279 32.7934 1.6446 61.7129 .019377 -.021754
b .'1803 -33.6901 186.5600 99.6717 45.7679 .1629 -38.6342 -.022755 -.001705
6 1.5532 -33.1785 219.2400 103.1448 57.8511 1.5435 -32.7434 -.001710 .015123
5.1088 -3.3665 268.8899 5.1089 -3.3767 .000011
;
8 8.8351 1.9456 303.4500
103.3619
96.4760
72.1656
69.9339 8.8360 1.9728 .000757
-.000916
.004172

more mechanisms, at most 36 mechanisms and 72 geometric in-


At A2 0
versions. The number of optimum mechanisms can be increased
0 At 2 ^3
by re-assigning values of 7 and 821. Refer to Example 4 for the il-
= S \tkQk (26) lustration of optimum design with eight unknowns.
B B
i 2 0 *=o
0 B, B. Rigid-Body Guidance
B,
Fig. 3 shows a rigid-body positioned at its ith position by the
When the terms in Qi are substituted, it can be shown that Q4 coupler link of a four-bar mechanism. The rigid-body is posi-
vanishes, and equation (26) reduces to a third-order polynomial tioned by positioning the body-line PQ. Then, a rigid-body posi-
resulting in at most three real values of Xi. The acceptable values of tion is generated by generating the coordinates of the body-point
X2 for each Xi are the common roots of equations (25a) and (25b). P and the angular coordinate £ of the body line, where | = 03 + t,
Since equation (26) could.have also been written for X2, at most and e is an angular dimension that defines the angular position of
three distinct real values of X2 exist. the body-line with respect to the coupler link. The angles 7 and t
To compute the remaining dimensions of the system, rewrite can be used as two of- the unknowns of the system, t can be a
equation (4) to have specified dimension, if the angular orientation of the coupler link
A with respect to the rigid-body is of concern. In this case the de-
sired values of dai are specified by 83; = 1/ - t, and the problem
' J,l 1 ' > ' of optimum synthesis for rigid-body guidance reduces to the prob-
z = l,...,iV (27)

where

Ki' = d3,Kz = -R4 cos a 4 , K 3 ' = Rt s i n a 4 ,


V = d 2 2 + d 3 2 + fi22 + Rt''~d4\ 1 RIGID-BODY

J.-V
K5' = diRi s i n a 4 =.Ki'K3',K^ = d3RA c o s a 4 = K^K2 ,
tan a4 = K3'/K2',

G i / = 2 [ f i 2 cos (a , ) + d2 cos (e 2 - 63 )]»


i i i
G2 ' = - 2 [ d 2 cos 92 + R2 cos a 2 ] ,

G 3 .' = - 2 [ d 2 s i n 02 + R2 s i n a2] , G 4 i 1, H, = 2 s i n 0, ,

Hm, = 2 c o s 0, , H„=- 2d2R2 c o s ( a 2

>Letting K5' = Xi', _K6' = X2' and defining Kj' in the form given
by equation (23) with primes.^./, mj' and n/ are again comput-
ed by equation [18] by using the primed values. The compatibil-
ity equations are the same as in equations (24) and (25) and Xi' is
given by the same cubic in equation (26) after introducing
primes. For each set of dimensions d2, R2, a and a2, one or more Fig. 3 The four-bar plane mechanism at the ith design position when
sets of dimensions d3, dit R$ and cu will exist, resulting in one or generating rigid-body positions

344 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


lem of optimum synthesis for generating the coordinates of the -0.1795; Xi = -12.66, Xi' = 26.18, \2' = -9.820; X2 = 8.022, X3'
body-point P using the specified values of d3t. In the following e is = 1.685, X4' = —0.2857. Out of the four optimum mechanisms
a specified dimension. and eight geometric inversions, Geometric Inversion I of the
Eliminate 62t in equations (la) and (16) to have mechanism defined by X2 and X3' approximates the rigid-body
positions best. Dimensions of this mechanism are K\ = R2 =
Ft = d22 -a2 -R2 + 2R2Pt cos (6, - a2)
i 4.1153 in., K2 = a = 1.9493 in., K3 = -13.1860 and d2 = 2.7476
+ 2aP{ c o s (6 3 + y - 5j) - 2aR2 cos (0 3 . + y - a2) • in., Ki' = 0.2544, K2' = fl4 = 6.6253 in., K3' = 3.3678 and d3 =
-P,2 = 0, i = 1,...,N (28) 3.9313 in., d 4 = 1.9113 in. Generated coordinates of the rigid-
as the first set of design equations similar to those given by equa- body, generated displacements of the links and the error in the
tion (3). The second set of design equations, equations (4), remain coordinates of the body point P are given in Table 2. This mecha-
the same as in path generation, and so are their solutions for the nism is shown in Fig. 4. For this mechanism, em = 0.005331 ins,
systems with six and eight unknown dimensions. and RMSE = 0.006755 in., and the average error in the angular
For the system with six unknowns, equation (28) takes the form position of the rigid-body is ee = 0.3156 deg. The smallest em for
the other optimum mechanisms was 0.315280 in. for Geometric
of equation (7), in which
Inversion I defined by Xi and Xi'.
Kt = R2, K2 = a, K3 = d22 - a2 - R22, K4 = a.R2 = K{K2, ..
Example 3. A rigid-body is to be guided td experience rolling
motion with sliding along a straight line such that the rigid-body
Gt = 2P{ c o s (6, - a2), G 2 . = 2Pt cos (0 3 i + y - 5 , . ) ,
experiences angular acceleration at the early part of rolling as the
G3.= l, point of contact translates at a constant velocity. The coordinates
t
2
of the point of rolling contact, body-point P, are given in the first
H,, = 2 c o s (6 3< + y - a2), Hm, = P three columns of Table 3 for 11 design positions. Desired angular
where R2, a and d2 are the unknown dimensions. The solution for positions of the rigid-body are given in the fourth column, and
the unknown dimensions follows the same procedure as of path with the specified value of e = 45 deg, desired values of 83i are
generation. The values of 82t to be used in equation (9) are com- given in the fifth column of the Table. Other specified dimensions
puted by the components of equation (1). are «2 = 54 deg, «4 =—15 deg, and 7 = 35 deg.
For the system with eight unknowns, equation (28) takes the Among the eight geometric inversions, Geometric Inversion I of
form of equation (22), in which the mechanism defined by X2 = 76.68 and X4' = 0.6754 positions
Kx = a, K2 = R2 s i n a2, K3 = R2 c o s a 2 , the rigid-body at the design positions best with an average error
of em = 0.030704 in. in the coordinates of the point of contact and
Kt = d2 -a2 -R2, tan a2 = K2/K3
an average angular error of en = 0.3396 deg in the orientation of
the rigid-body. Dimensions of this mechanism are Ki = R2 =
K§ = aR2 c o s a2 = KtK3, K§ = aR2 s i n a2 = KtK2,
7.8721 in., K2 = a = 9.2321 in., K3 = -121.3593 and d2 = 5.0836
in., K-i! = 0.0543, K2' = Rt = 12.4289 in., K3' = 4.8015 and d3 =
G{. = 2Pt c o s (e 3 . + y - 6j), G2 = 2P. s i n 6 f , 18.4019 in., d4 = 18.2778 in. The generated coordinates of the
rigid body and the displacements of the links are given in the re-
G 3 = 2P, cos 5 „ G 4 j = 1,
maining columns of Table 3. This mechanism, path of point of
contact and some rigid-body positions are shown in Fig. 5. The
Ht, = 2 c o s (0 3 + y), # m . = 2 s i n (0 3 . + y), # „ . = P, 2
smallest em for the other geometric inversions was 0.274038 in. for
where a, d2, i?2 and «2 are the unknown dimensions. Geometric Inversion defined by Xi = -217.2 and Xi' = 0.9548.
Example 2. A rigid-body is to be guided through seven posi- Example 4. The same rigid-body positions specified in Exam-
tions shown in Fig. 4. Desired coordinates of the rigid-body are ple 3 were generated using eight unknown dimensions. d3t and t
given in the first four columns of Table 2. Specified dimensions have the same specified values as in Example 3. Solution resulted
are a2 = 20 deg, «4 = 10 deg, 7 = 20 deg, and t = 10 degl Then, in a series of optimum mechanisms. Some of the geometric inver-
desired values of 631 are known and given in the fifth column of sions: generate the rigid-body positions with smaller error com-
Table 2. Solution resulted in mi = 4.9360, m2 = 0.4140, m 3 = pared to the mechanism in Example 3. Geometric Inversion I de-
-23.2073,^1 = -0.1057,4 2 = 0.1914,^3 = 1.2493, m i ' '= -0.0255, fined by Xi = 41.6603, X2 = 69.1634, Xi' = 36.5820 and X2' =
m2' = 2.4449, m3' = -2.8543, V = 0.1523, t2' = 0,1590,J5 3 ' = 1282.8524 resulted in the smallest error em - 0.022341 in. and e«

Fig. 4 Rigid-body positions and the four-bar mechanism optimized in Fig. 5 Path of point of contact, same of the rolling rigid-body positions,
Example 2 and the four-bar mechanism optimized in Example 3

Journal of Engineering for Industry FEBRUARY 1975 / 345

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


= 0.30822 deg. Dimensions of this mechanism are K% = a = introduces constraint for the continuity of the motion of .the
8.0734. in., K2 =8.5668, K3 = 5.1602, K 4 = -128.0939, a2 = -mechanism, so it tends to separate the geometric inversions of'a
58.9375 deg, R2 = 10.0009 in., d2 = 6.0913 in., and Ki' = d 3 - = mechanism. In the solution with six unknown dimensions,:values
70.2169 K2' = 18.2704, K3' = 0.5210, K4' .=. 1450.7319, « 4 = of a2 and « 4 may be estimated, especially for path generation,
1.6334 deg, fl4 = 18.2778 in., d4. = 62.8535 in. Although this observing the dimensions of a mechanism in a coupler curve atlas
mechanism generates the rigid-body positions with smaller error, which generates a path similar to the one desired. Efficiency of
its dimensions are much larger than the dimensions of the mech- the approximation can be increased by respecifying the coordi-
anism optimized in Example 3. It also has very small transmis- nates of all or some of the design points, or design positions of the
sion angle, 4.9441 deg, within the zone of motion, while it is larger rigid-body. This is justified since, in general, other units con-
than 22 deg in the mechanism of Example 3. The next smallest nected to the mechanism in a machine are designed after the di-
error is ;e m = 0.023164 in. with eo = 0.413454 deg and smallest mensions of the mechanism are determined.
transmission angle of 12.5 deg, generated by Geometric Inversion The method of optimum synthesis presented here has immedi-
I of the mechanism whose dimensions are Ai = 32.8191, A2 = ate, extension to apply for the optimum synthesis of multiloop
98.2795, Kt = a = 5.3234 in., K2 ==.18.4618, K 3 = .5.7288, K4-.= plane mechanisms, single and multiloop spherical and spatial
-227.0576, « 2 = 72.8808 deg, R2 = 19.4618 in., d2 •= 13.4180 in.., mechanisms for the generation of functions, paths and rigid-body
and Xi' = -22.9740, X2' = 569.1532, Ki' = ds = 37.1997 in., K2' positions. '
= 15.2999, K3'. = -9.6179,.K 4 ' = 895.1796, « 4 = 357.6885 deg, RA
= 15.3124 in., d 4 = 34.8035 in. Displacements of this mechanism, References
generated coordinates of the rigid-body and the errors in the gen- ' 1 Levitskii, N. I., "Application of the Least-Square Method to-Mecha-
erated coordinates are given in Table 4. In this example, loosing nism Design," Akad. Nauk. USSR, Institut Mashinouede a. Trudy Seminar
control on « 2 and ai by using system with eight unknowns did Po Theorii Masin i Mechanismou, Vol. 5, No. 17* 1948, pp. 40-68.
not result in much gain as far as the efficiency of the approxima- 2 Levitskii, N. I., and Sarkasian, ,Y., "On the Special Properties, of
Lagrange Multipliers in the Least Square Synthesis of Mechanisms," Jour-
tion and the quality of transmission were concerned. ndl of Mechanisms,.Vol .'3, No. 1,1968, pp.3-10. .:;
Example 5. This example is included here upon the recom- 3 Levitskii, N,. I,, and, Sahbazyan, K. H,, "The Synthesis of Spatial
mendations of the reviewers to compare the results with those ob- Four-Link Meckahisms with Lower Pairs," Akad. Nauk. SSSCJ, Trudy
tained by the relaxation and gradient combination given in refer- Sem. Teorri Masin i Mechanizmov, Vol. 14, No. 54,1954, pp. 5-24.
4 Han, C. Y:,""A General Method for the Optimum Design of Mecha-
ence [9]. Thus, a four-bar plane mechanism is designed to gener- nisms," Journal of Mechanisms, Vol. 1,1966, pp. 301-313.
ate the Joukowski airfoil profile using the same coordinates of 5 Tull, H. G., and Lewis, D. W., "Three-Dimensional Kinematic Syn-
eight design points along the path given in Table 1 of reference thesis," JOURNAL OF ENGINEERING FOR INDUSTRY, TRANS.
[9], where the^ninth design point is the same as the first design ASME, Vol. 90, No. 3, Series B; Aug. 1968, pp. 481-484.
6 Johnson, H.L., "Path Generation in Space," ASME Paper No. 65-
point. Coordinates of the design points are given in Table 5. Since WA/MD-i3.
the data for the specified input crank positions are not given in 7 Bagci, C, "Minimum Error Synthesis of Space Mechanisms for the
reference [9], they are approximately measured from its Fig. 5 Generation of Constrained and Unconstrained Screws," JOURNAL OF
and given in the fourth column of Table 5. The mechanism is op- ENGINEERING FOR INDUSTRY, TRANS. ASME,. Vol. 93, No. 1, Series
B, Feb. 1971, pp. 165-175.
timized using eight unknown dimensions with the specified value , 8 Tomas, J., "The Synthesis of Mechanisms as a NonLinear Program-
of y,:= -47.45 deg. The optimum geometric inversion has the di- ming Problem," Journal of Mechanisms, Vol.3, 1968, pp. 119-130.
mensions R2'= 5.6294, Ri = 22.5674, d2 = 6.2146, d3 = 21,3763, rf4 9 Nechi, A. J., "A Relaxation and Gradient Combination Applied to
= 36.6258, a = 11.8539 in.; a2 = 170.2203, a 4 = -65.5431 deg. the Computer Simulation of a Plane Four-Bar Chain," JOURNAL OF EN-
GINEERING FOR INDUSTRY, TRANS. ASME, Vol. 93, No. 1, Series B,
The generated displacements, generated coordinates arid the er- Feb. 1971, pp. 113-119. - <, s, . •
rors, in the generated coordinates of the design .points are given in 10 Prasad, K. N., and Bagci, C, "Minimum.Error Synthesis of Multi-
the remaining portion of Table 5. The : average error is em = loop Plane Mechanisms for Rigid-Body Guidance," TRANS. ASME,
0.009021 in. This solution was reached within 0.89 minute by JOURNAL OF ENGINEERING FOR INDUSTRY, Vol. 96,. No. 1, Feb.
1974, pp.107-116. " ""
Xerox Sigma 6 digital computer. As far as the comparison of l i ' Bagci, C, and Parekh, K. C, "Minimum Error Synthesis of the
these results is concerned, this em cannot be compared with, the Spherical Four-Bar and Watt's Type Spherical Six-Bar Mechanisms," Pro-
error defined in reference [9], where the error is defined to be the ceedings of the Third World Congress, on the Theory of Machines and
normal distance from the generated position of the coupler point Mechanisms, Paper C-l, Vol. C, pp. 1-19, Dubrovnik-Kupari, Yugoslavia,
Sept: 13-20,1971.
to the line drawn from the desired position of the point parallel to
12 Sarkisyan, Y. L., Gupta, K. C, and Roth, B-, "Kinematic Geometry
the tangent of the coupler curve at the generated position of the Associated with the Least-Square Approximation of a Given Motion,"
design point. See Fig. 2 of [9], This certainly is a one-dimensional TRANS. ASME, JOURNAL OF ENGINEERING FOR INDUSTRY, Vol.
error and it does not include the error in the direction parallel to 95, No. 2, Series B, May 1973, pp. 503-510.
the tangent line, which can have considerable effect on the mag- 13 Rose, R. S., and Sandor, G. N., "Direct Analytic Synthesis of Four-
Bar Function Generators with Optimal Structural Error," TRANS. ASME,
nitude of the two-dimensional error em given above. Even then, (8 JOURNAL OF ENGINEERING FOR INDUSTRY, Vol. 95, No. 2, Series
em) is less than the total error given in reference [9]. Furthermore, B, May 1973, pp. 563-571.
the method of [9] requires analog computer approximation of the 14 Bagci, C, "Optimum Synthesis of Function Generators via the Lin-
gradient, and combination of relaxation method which in turn re- ear Superposition Technique," Proceedings of The First ASME Design
Technology Transfer Conference. Transactions of the ASME, Oct 6-10,
quires many iterations and adds to the computer bill. 1974, New York, pp. 529-538. -
15 Zimmerman; J. R., "Five Precision Point Synthesis of the Four-Bar
Conclusions Function Generator," TRANS. ASME, JOURNAL OF APPLIED ME-
CHANICS, Vol. 86,.,Series A, Sept. 1964, pp. 563-564.
The method of optimum synthesis of plane mechanisms for the 16 Hartenberg, ,R. S., and Denavit, J., Kinematic Synthesis of Link-
generation of path and. rigid-body positions presented in the fore- ages, McGraw-Hill, New York, 1966. '
going is general. It does not require iteration, and is easy to 17 Rao, A, V. M., and Sandor, G; N., "Extension of Freudenstein's
apply. It eliminates the searching for zones,where optimum Equation to Geared Linkages," TRANS. ASME, JOURNAL OF ENGI-
NEERING FOR INDUSTRY, Vol. 93, No. 1, Series B, Feb. 1971, pp. 201-
mechanism exist, which is required by other methods of optimum 210. ; ; ••., , .

synthesis. The solution, in general, results in a series of optimum 18 Rao, A. V. M,, and Sandor, G. N;,! "Closed Form Synthesis of Four-
mechanisms. The number of optimum mechanisms can be-in- Bar Path Generators by Linear Superposition," Proceedings of the Third
creased by assigning different values to the specified dimensions World Congress for the Theory of Machines and Mechanisms, Kupari, Yu-
goslavia, Vol. G, Sept. 13-20, 1971, pp. 383^394.
of the system, or by developing systems with more unknown di- 19 Bdcher, M., Introduction to Higher Algebra, Macmillan, New York,
mensions. Since many design positions can be used, the method 1949, pp. 195-202.

346 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like