30

force between the surfaces, and is the relative velocity of surface a

with respect to surface b.

To model the effects of friction in the pin connections, the friction

circle concept is employed. This concept is illustrated in Figure 3-3.

In Figure 3-3, points A and B represent the centers of the bearing and

journal, respectively, who's radii are r and R. Since the percent differ­

ence between the two radii is quite small, they are assumed to be equal

for the force analysis. The reaction force, F, is the force exerted by

the journal on the bearing. The reaction force has components normal to

the surface, F^, and tangential to the surface, Fy. The tangential force

is a friction force that opposes the relative motion of the bearing with

respect to the journal, w. The tangential force produces a moment about

B that is R Fy. The line of action of the force is tangential to a smaller

circle centered at B with a radius Rp. A moment is produced that is equal

to RpF. By equating the moments the following relationship is developed:

(3-13)

Noting that Fy = yFn then eq. (3-13) can be reduced to

(3-14)

where Rp is the friction circle radius. Note that this radius is constant

as long as the bearing radius (R), and the coefficient of friction (y) re­

main constant. A friction torque exerted by member i on member j

(T. .) is then found to be equal to

'vJ

(3-15)
FijI RF s9"<"j - “i>

where F . . is the reaction force of member i acting on member j, and <u. and
•J 1
FIGURE 3-3 - ILLUSTRATION OF THE FRICTION CIRCLE CONCEPT
 32

u)j are the angular velocities of the two members. Figure 3-3 is redrawn

in a rectangular coordinate system with the reaction force moved to the

center of the journal (see Figure 3-4). The reaction is represented as

a force couple. The force is represented by its components parallel to

the real axis ( F ^ ) and the imaginary axis (F ^.). The dimensionless

friction torque is:

-F 2 , -F
sgn (i (3-16)
T j = -(F X1J + Fyij 2 >1 / 2 *F “i*

To complete the dynamic force analysis, the members of the mechanism

are shown as free bodies (Figure 3-5). The reaction forces are represented

by their x and y components, while the torques are shown as being positive

in a counter-clockwise direction. The inertia forces and torques are re­

placed by their D'Alembert equivalents, therefore, the sum of forces and

the sum of the torques are zero for each free body. Since this is a

planar mechanism, three independent equations can be developed for each

link. They are the sum of the forces in the x direction equals zero, the

sum of the forces in the y direction equals zero, and the sum of the mo­

ments equals zero. Also the sum of the forces in the x and y directions,

and the sum of the moments equals zero for each pin. Using this fact, the

following general equations are developed:

F . . = -F ••
xij xji

F . . = -F .. (3-17)
yij yji

TU = -TJF

Analyzing the slider crank linkage, shown as free bodies in Figure

3-5, the following equations are developed for the crank, noting that it

is a uniform disk rotating about its center at a constant angular velocity:

 33

Imaginary Axis

FIGURE 3-4 - BEARING FORCE IN A RECTANGULAR COORDINATE SYSTEM

 34

Figure 3-5B - FREE-BODY DIAGRAM

OF THE SLIDER

FIGURE 3-5A - FREE-BODY DIAGRAM

OF THE CRANK

FIGURE 3-5C - FREE-BODY DIAGRAM OF THE CONNECTING ROD

 35

Ti|vj ~ T j2 - T^2 R|_2"*i^(®2^ ^ x 32 - ^ L 2 <’’ ^ " * ^ 2 ^ ^"y32 (3-18)

(3-19)
Fx 12 _Fx 32

(3-20)
Fyl2 " "Fy32

Note that Tj^ is the external input torque applied to the crank. Equilib­

rium equations for the slider are

(3-21)
Fx 34 + FL + Fx 14 " M4 A41 0
lu^

iu *

o
(3-22)
+

II
CO

rH

where is the external load and Fx ^ is the friction force the frame ex­

erts on the slider given by

^xl4 = “ lFyl4ly s9n(^41) (3-23)

When developing the equation for the connecting rod, note that the only

mass associated with this member is that of a uniform rod. Its center of

mass is located halfway between the pins, so R ^ = R^/2. The linear ac­

celeration of the center of mass are given by eqs. (3-10, 11). The inertia

of a uniform rod about its center of mass is

I = ~ L3 pit D2/4 (3-24)

For our case, D equals L/10; then eq. (3-24) is rewritten as:

T - D 9 (3-25)
13 " 4800 k L3

The mass of the connecting rod is

M3 = p tt RL33/400 (3-26)
 36

The equilibrium equations are then

— — — — — — 3 _ _ o

T23 + T43 ‘ *3 a3 + M3 A x 3 ~T~ sin03 “ M3 Ay 3 COS03 +

Fy43 RL3 cos03 " Fx43 RL3 sine3 0

(3-29)

Fx 43 " M3 Ax 3 + Fx 23 0 (3-28)

(3-29)
Fy43 " M3 Ay3 + Fy23 " 0

The following substitutions or rearrangements of equilibrium equations are

going to be substituted into the equilibrium equations (3-18, 21, 27)

4?'
II

B1 A41

B2 ^3 Sx3

= m3
B3 \3

B4 = I 3 “3

r lx ■ RL3/ 2

X = F (3-30)
x34

Y = F
y34

2 x vr2 1 / 2
z = f 34 = (X" + Y )

X Fx 34 + B2

Y = Fy34 + B3
—2 —2
1/2
Z = (X + Y )
 37

After a few simplifications, the above equations become:

T in = -Z (Cx - C2 ) + Rl 2 (Y cose2 - X sine2 ) (3-31)

F(x,Y) = z c2 + z c3+ cose3 (-B3 r lx - RL3 y ) +

sin0 3 (B2 Rlx + X RL3^ " B4 (3-32)

0 = G (X,Y) = X + Fl + |Y| C4 - B 1 (3-33)

Where the "C" terms are parts of the friction model and equal

R-i o i /o ((jjo) (3-34)

1 { / + d 1/2 2

(3-35)
! ^ 2 + ^ 1 / 2 Sgn ^ 3 “ w 2 ^

(3-36)
R3 ^ 2 + ^ 1 / 2 Sgn ^

C4 = -y sgn(V4 1 ) (3-37)

The solution to eqs. (3-31 through 33) will be shown in a later section.

3-3 Dynamic-Force Model with Journal-Bearing Masses

The difference between this analysis and the preceding analysis is

that the journal-bearings have a mass associated with them. The journals

or pins are attached to the crank and slider while the connecting rod and

frame have the bearings attached. The pins have a radius R. and a length

of R^/IO. The bearing dimensions are assumed to be a length R^/IO, an

inside diameter R^, and an outer diameter of 1.5 R^. The bearing, journal,

and links have the same material density, p. The crank is again assumed to
 38

be rotating at a constant angular velocity about its center, which is its

center of mass. Because of this assumption, eq. (3-31) is the same whether

the journal-bearings have mass or not. Therefore, this section focuses on

the connecting rod and slider. Figure 3-6 shows a physical representation

of the connecting rod and slider.

The mass of the connecting rod is the mass of the rod plus the mass

of the two bearings. The mass of the rod is

M 31 = (*L3 " l -5 R2 " 1 ' 5 *3} "P (2 ^ )2 (3-38)

and the mass of bearings 2 and 3 are

((1.5 R2 ) 2 (3-39)

yjp ((1-5 R3 ) 2 - R32 )ttp (3-40)

The mass of the connecting rod is then

M3 . H 3 1 + H3 2 + M3 3 (3-41)

The center of mass is located a distance P ^ fr om the center of bearing 2.

This distance is

Rlx= [(0)M3 2 +(1.5 R2 +(R l 3 - 1.5 R2 - 1.5 R3 )/2)M3 1 + RL3 M 3 3]/ M3 (3“42)

The inertia about the center of mass is given by

I 3 = ((1.5 R2 ) 4 - R24 ) \ yg- p + M 3 2 Rx +

£
(yjp) 2 7 tt(RL3 - 1.5 R 2 - 1.5 R3 ) 3 + M 3 1 ((RL3 - 1.5 R2 - 1.5 R3)/2 +
13

1.5 R2 - + ((1.5 R3 ) R3 ) 7 io p + M33 ^RL3 " ^ (3-43)

FIGURE 3-6A - PHYSICAL SHAPE OF THE CONNECTING ROD

FIGURE 3-6B - PHYSICAL REPRESENTATION

OF THE SLIDER AND PIN
 40

The mass of the piston is now the slider mass plus the pin mass, or
fT
M4 ‘ = M4 + P7T ^ (R3 ) 2 (3-44)

Substituting these values into B^, B^, B^, R^x of eq. (3-30) equations

of the form of eqs. (3-31 through 33) are obtained for the dynamic-force

analysis with journal-bearing masses.

3-4 Force Solutions

Since eq. (3-32, 33) are in a nonlinear form to solve for the reaction

force components X and Y, an iterative technique is used. Eqs. (3-32, 33)

are equal to zero and shall be represented by F(X",Y) and G(X,T), respect­

ively. The following series approximation of F(X,Y) and G(X,Y) are made

(Newton Method)

8 F(X„,YJ „ 8 F(Xn ,Y„)

n n' , tt
0 = FCX.Y) . F CXn , Y n ) + ~ n_ " ( X n+1 - X n ) + (Vl " V
aY
(3-45)

3G(X „ , Y J _ 9G ( X . Y J
n n'
0 = G(X,Y) = G(Xn ,Yn ) + — - (- Xn^+ 1 -
- »X nJ + -------- ( Y n+1 - Y n )
3Y
(3-46)
j_i_ _ _
where Y and Y are the nLr approximation of the actual values of X and Y
n n
while G(Xn ,Yp) and F(Yn ,Yn) are the nth approximation of G(X,Y) and F(X,Y).

Eqs. (3-45, 46) are then solved and the following iterative equations are

obtained

9G( X , Y ) 9F(X ,Y )

X. + (3-47)
n+ 1 3F(xn , Y ) a G (x n , Y n ) 3 F ( x n ,Y n ) a G (x n ,Y n )

3Y 9X aX aY
 41

_ _ 3F(X ,Y ) _ _ 3G( X , Y _ )
G( X , Y J --------n _ J l _ . p( X , Y „ ) --------° - J 1 -
n n' n' tv
9X 9X
= Y + (3-48)
n+ 1 n
9F ( X n , Y n ) 9G ( X n , Y n ) aF(Xn,Y n) 9 G( X n , Y n )

9Y 9X 9X 9Y

Initial guesses of X and Y are required, then eqs. (.3-47, 48) are used

until X ,. - X" and T ... - Y become close enough to zero. The values of
n+ 1 n n+ 1 n 3
Xn , and Yn are then substituted into eqs. (3-30, 31) for X and Y to find
_ _I
an approximate value of T ^ , and the bearing forces Z and Z . Values of
_I
Tj^, Z, and Z are calculated for equally spaced rotations of the crank.

An outline of the method is now detailed below.

1. Complete the kinematic analysis and calculation of all

constants for the given crank position.

2. Assume initial values X^, Y^.

_ 1 1 I
3. Calculate initial values of • V„ • Z„ • Zn*
4. Calculate Fn = F(Xn ,Yn), and Gn
= G < V Y n>

I
!><C
l>;C

z c2 + z c3 + cos e3 ( - b 3 r l x - r L3 y ) +
Li_

II

si n 03 ( B2 R l x + X Rl 3 ) - ( 3- 49)

G ( X n , T n ) = X + F l + | T| C4 - Bj ( 3- 50)

9F 9F 9G 9G
n, n, n, n
Calculate
9Y 9X 9Y

_1 _1
3Fn (3-51)
■ = Rl 3 si n e3 - C2 X / Z + C3 X 11
9X

_ _1 _1
9Fn
• = r L3 c o s e3 - c2 Y / Z + c3 Y / z ( 3- 52)
9Y

3G
n - 1
9X 1
 42

C4 if Y > 0
0 if Y = 0 (3-53)

-f4 if Y < 0

6. Calculate Xn+1> Yn+1 by eqs. (3-47, 48)

7. If Xn+^ - Xn and Yn+j - Yn are within some interval bounding zero,

go to Step 9.

8. Loop back to Step 3.

9. Calculate T .
in

Tin = ^ )R[_2 c o s 02 " ^ ^ L 2 sin 02 “ * - Cg) (3-54)

10. Rotate the crank to it's new position and start over from Step 1.

3-5 Objective Function

As previously stated, the objective function is a combination of the

input work and shear stress. The input work W is expressed exactly by the

following equation
r2 tt
W = (3-55)

This integral can be approximated as follows:

N _
W = E Tj^j A0 (3-56)

Where are the input torques for crank positions a distance A0 apart,

and N is the number an increment of crank rotation needs to complete the

cycle. The bearing shear stress is represented by the stress factor, 0 .

Discrete values of 6* are determined for each crank position and used to

approximate the maximum shear stress. Obviously, the accuracy of the ap­

proximation decreases with increasing AO.

 CHAPTER 4

OPTIMIZATION OF THE SLIDER-CRANK MECHANISM

The slider-crank mechanism is optimized on the basis of performance

criteria related to the required cycle input work and the maximum bearing

shear stress. These criteria tend to be competing objectives. For exam­

ple, it appears that increasing the bearing radii will decrease the assoc­

iated shear stress, but as the bearing becomes larger the friction torque

increases which increases the cycle input work. Therefore, to establish

the trade offs between these quantities the following objective function is

used:

f = w + w em (4-1)

where f is the objective function, W is the dimensionless cycle input work,

Bm is the largest of the maximum dimensionless stress factors associated

with each bearing during a complete cycle, and w is a weighting factor which

adjusts the relative merit of work and stress in the optimization process.

Throughout the remainder of this thesis the dimensionless cycle input work

and the maximum cyclic dimensionless stress factor are simply referred to

as the work and stress factor, respectively.

A brief examination of eq. (4-1) reveals the effect of the weighting

factor on the optimization process. Very large values of w tend to mini­

mize the stress factor with little regard to the work, while values of w

approaching zero tend to minimize the work with little regard to the stress

factor. Values of w between the extremes produce an optimum that reflects

the effects of both the work and stress factor.

43
 44

The design variables for the optimization are the linkage dimensions

^L2 ’ ^L 3 ’ ^ 1 ’ ^2 ’ anc* ^ 3 ' Since the stroke is held constant for all

mechanism configurations, is computed from R ^ and H by eq. (2-16).

From eqs. (3-19, 20) it can be shown that the reaction forces for bearings

1 and 2 have the same magnitude. Since the optimization minimizes the max­

imum stress factor; the stress factors for the individual bearings should

be equal at the optimum. Therefore, R^ = R,-, for the optimization process

since this makes the stress factors for these bearings equal.

Thus, the mathematical optimization problem can be stated as follows.

minimize f = W + w Bm
(4-2)
with respect to R ^ > H, R^

The optimization was evaluated with the aid of an optimization package de­

veloped by Afimiwala and Mayne [12]. The specific technique used is the

variable metric search also referred to as the Davidon-Fletcher-Powel1

method [13].

4-1 Optimization Procedure

In this thesis the slider-crank mechanism is loaded as a single acting

compressor. Since this is a numerical model, numerical values must be as­

signed to all parameters. Therefore, throughout the remainder of this

thesis the following numerical values are assigned to the dimensionless

quantities used to model the compression cycle:

P„ = 7.65
e

P. = Pa = 1.93
a
 45

k = 1.4

y = 0.1

Ap = 0.35

The following numerical values or relationships are assigned to dimension­

less mechanism parameters of mass density (p), connecting rod diameter

(D), and bearing length (L).

p = 12.3

D = Rl 3 / 1 0

L = D

Two other factors need to be studied before the optimization process

can be started. The factors are the size of the increment of crank rotation

and the accuracy of the approximation of the reaction forces. The smaller

increment of crank rotation and larger accuracy in approximation of the

reaction forces improves the model's prediction of the work and stress fac­

tor. This highly accurate model is obtained at the expense of computing

time, the higher the accuracy the larger the computing time. By analyzing

the effects of single parameter variation, it was determined that a 6 ° in­

crement is sufficient. Also, by performing the iterative technique used

to determine the reaction forces four times, the error in the approximation

was less than 0.001 percent for all cases considered. These values are suf­

ficient if the optimization process is divided into two steps. The first

step is the optimization of the inline slider-crank mechanism, with the

offset equal to zero. Once this is completed and an optimum mechanism is

determined, the offset is adjusted to attempt to further minimize the

 46

objective function. This division was determined to be necessary by anal­

yzing a plot of stress factor versus offset (Figure 4-1). The curve pro­

duced is jagged and tends to smooth as the increment of crank angle rota­

tion is decreased from 6 ° to 1/2°. The stress factors as a function of H

should be the smooth curves drawn through all the points of relative maxi-

mums. This curve corresponds to an increment of crank rotation approaching

zero. To produce this curve would take an extremely large amount of com­

puting time. A smooth curve is required otherwise the optimization process

picks some false location at the base of any one of the jogs. The curves

for stress factor versus any of the other parameters were smooth. There­

fore, the optimization statement now becomes first

minimize f = W + w(3
(4-3)
with respect to , Rl = Rg, R3, H = 0

and secondly,

minimize f = W + wg
(4-4)
with respect to H

The tradeoffs that are recognized to exist for the optimization statement

given by eqs. (4-3, 4) are enumerated as follows:

1. The trade offs incurred by R^:

a. Increasing R. increases the friction torque which in­

creases the work as shown by eq. (3-18).

b. Increasing R. increases the area bearing the reaction

force which decreases the associated stress factor,
as shown by eq. (2-19).

2. The trade offs occurring for R^:

a. Increasing R^ increases the bearing area which de­

creases the ^associated stress factor.
 -.6 -.5 -.4 -.3 -.2 - 0.0
Dimensionless Offset
FIGURE 4-1 - INPUT WORK AND STRESS FACTOR VERSUS OFFSET
 48

b. Increasing R~ increases the bearing mass which

increases tne inertia forces.

c. Increasing R? increases the friction torque

which increases the work.

d. Increasing R£ increases the friction torque which

changes the reaction forces as given by eq. (3-27).

3. The trade offs existing for R^:

a. Increasing R^ decreases the associated stress factor.

b. Increasing R^ increases the inertia forces.

c. Increases in R-, increase the friction torque which

changes the reaction forces.

4. The trade offs for R ^ :

a. Increasing R ^ increases the connecting rod mass which

increase the inertia forces.

b. Increasing R. ~ reduces the magnitudes of the connecting

rod velocityL and acceleration which decreases the in­
ertia forces.

c. The bearing lengths are proportional to the connecting

rod length. Increases in connecting rod length then in­
crease the bearing area, reduce the stress factor, and
increase the bearing mass, increasing the inertia effects

d. Increases in R.~ tend to align the connecting rod with

the direction of slider motion. If they are aligned
the y component of the reaction force would be reduced
which reduces the reaction forces, and hence, the stress
factors are also reduced. Also decreasing the y com­
ponent of the forces reduces the normal force at the
sliding contact. This reduction decreases the effects
of friction acting upon the slider.

5. The trade offs that exist for H:

a. Varying IT from zero tends to increase the maximum angle

between the connecting rod and direction of slider motion
There is then an increase in the y component of the reac­
tion forces and an associated increase in the friction
effects.

b. The inertia forces may align themselves so as to op­

pose the applied force over some region of the cycle,
and by varying H, a more uniform stress throughout the
entire cycle may be obtained.
 49

The following should be noted in conjunction with the trade offs mentioned.

First, friction is a nonconservative effect; addition of friction automat­

ically increases the required input work. Second, if there is no friction

the inertia effects are conservative, the work will not change but the

reaction forces do change. Third, the inertia effects for the connecting

rod initially decrease when the connecting rod is increased from its mini­

mum length. This decrease is due to the rapid decrease in its acceleration.

Further increase in length increases the inertia effects because the con­

necting rod mass is increasing faster than the acceleration is decreasing.

To study the effects of the trade offs various cases of the slider-

crank are analyzed. The cases start from the simplest, no friction or in­

ertia effects, and proceed to the most complex case, complete friction and

inertia effects. For the simplest case, no friction or inertia effects,

the work is 0.949, and is the same for all mechanism configurations for

this case. This is the amount of work required to complete a compression

cycle. It was determined by evaluating the following integral:

- P)Ap dx = 0.949 (4-4)

This is the minimum work that any mechanism loaded by this compression

cycle can have. Since there is no friction or inertia effects the bear­

ings will approach an infinite diameter and therefore all the stress fac­

tors are zero.

4-2 Optimization Results for Special Cases

The results for the case when there are no inertia effects present are

given in Table 4-1 and Figure 4-2. For this data the coefficient of fric­

tion is 0.5. Figure 4-2 shows the trade off curves relating the stress