Vibrational Modeling
williar4@ohio.edu
people.ohio.edu/williar4
k
x(t)
x(t)
k m
x(t)
1. Introduction
1.3 Vectors. Cartesian Re-Im Representation (Phasors)
Here is an alternate vector representation.
P Pei
The phasor Pe i is a polar representation for vectors, where P is the length of vector P , e is the
natural logarithm base, i 1 is the imaginary operator, and is the angle of vector P . ei gives the
direction of the length P, according to Euler’s identity.
Phasor Re-Im representation of a vector is equivalent to Cartesian XY representation, where the real (Re)
axis is along X (or iˆ ) and the imaginary (Im) axis is along Y (or ĵ ).
P P cos
P P(cos i sin ) Re Pe
i
Im
P P sin
P P cos
P P(cos iˆ sin ˆj ) X
PY P sin
A strength of Cartesian Re-Im representation using phasors is in taking time derivatives of vectors – the
derivative of the exponential is easy ( d ds (e s ) e s ).
d 2 P d 2 Pei
dt 2 dt 2
d 2 P d i
( Pe iPei )
dt 2 dt
d 2 P i
Pe iPei iPei iPei i 2 P 2 ei
dt 2
d 2 P i
Pe 2iPei iPei P 2 ei
dt 2
d 2P d 2 Pcos
dt 2 dt 2 Psin
d 2 P d P cos Psin
dt 2 dt P sin Pcos
We obtain the same result, but the Re-Im phasor time differentiation is made in compact vector notation
along the way.
d dP (t ) i (t ) dei (t ) dei ( t )
product rule ( P (t )ei (t ) ) e P (t ) P(t )ei (t ) P (t )
dt dt dt dt
de i ( t ) de i ( t ) d (t )
chain rule ie i ( t ) (t )
dt d (t ) dt
d
( P(t )ei (t ) ) P (t )ei (t ) P(t )iei (t )(t )
dt
6
1.6 Matrices
Matrix an m x n array of numbers, where m is the number of rows and n is the number of columns.
a11 a12 a1n
a a22 a2 n
A
21
am1 am 2 amn
Matrices may be used to simplify and standardize the solution of n linear equations in n unknowns
(where m = n). Matrices are used in velocity, acceleration, and dynamics linear equations (matrices are
not used in position analysis which requires a non-linear solution).
Special Matrices
a11 a12 a13
square matrix (m = n = 3) A a21 a22 a23
a31 a32 a33
a11 0 0
diagonal matrix A 0 a22 0
0 0 a33
1 0 0
identity matrix I 3 0 1 0
0 0 1
x1
column vector (3x1 matrix) X x2
x
3
X x1 x3
T
row vector (1x3 matrix) x2
7
Matrix Addition add like terms and keep the results in place
a b e f a e b f
c d g
h c g d h
Matrix Multiplication with Scalar multiply each term and keep the results in place
a b ka kb
k
c d kc kd
Matrix Multiplication
C A B
In general, A B B A
C A B
(mxn) (mxp )( pxn)
That is, in a matrix multiplication product, the number of columns p in the left-hand matrix must equal
the number of rows p in the right-hand matrix. If this condition is not met, the matrix multiplication is
undefined and cannot be done.
The size of the resulting matrix [C] is from the number of rows m of the left-hand matrix and the
number of columns n of the right-hand matrix, m x n.
Multiplication proceeds by multiplying like terms and adding them, along the rows of the left-
hand matrix and down the columns of the right-hand matrix (use your index fingers from the left and
right hands).
g
a b c ag bh ci
C d e h
f dg eh fi
Example i
(2x1) (2x3)(3x1)
Note the inner indices (p = 3) must match, as stated above, and the dimension of the result is dictated by
the outer indices, i.e. m x n = 2x1.
8
Matrix Multiplication Examples
7 8
1 2 3
A 4 5 6 B 9 8
7 6
C A B
7 8
1 2 3
9 8
4 5 6 7 6
7 18 21 8 16 18 46 42
28 45 42 32 40 36 115 108
(2x2) (2x3)(3x2)
D B A
7 8
1 2 3
9 8
4 5 6
7 6
7 32 14 40 21 48 39 54 69
9 32 18 40 27 48 41 58 75
7 24 14 30 21 36 31 44 57
(3x3) (3x2)(2x3)
9
Matrix Inversion
Since we cannot divide by a matrix, we multiply by the matrix inverse instead. Given
C A B , solve for [B].
C A B
A C A A B
1 1
I B
B
B A C
1
A A A A I
1 1
where [I] is the identity matrix, the matrix 1 (ones on the diagonal and zeros everywhere else). To
calculate the matrix inverse use the following expression.
adjoint( A)
A
1
A
cofactor(A) a ij ( 1) i j M ij
minor minor Mij is the determinant of the submatrix with row i and
column j removed.
C A B
C B A B B
1 1
A I
A
A C B
1
In general the order of matrix multiplication and inversion is crucial and cannot be changed.
10
Matrix Determinant
The determinant of a square n x n matrix is a scalar. The matrix determinant is undefined for a
non-square matrix. The determinant of a square matrix A is denoted det(A) or A . The determinant
notation should not be confused with the absolute-value symbol. The MATLAB function for matrix
determinant is det(A).
a b
A c d
A ad bc
To calculate the determinant of 3 x 3 and larger square matrices, we can expand about any one
row or column, utilizing sub-matrix determinants. Each sub-determinant is formed by crossing out the
current row and its column and retaining the remaining terms as an n–1 x n–1 square matrix, each of
whose determinant must also be evaluated in the process. The pivot term (the entry in the cross-out row
and column) multiplies the sub-matrix determinants, and there is an alternating + / – / + / – etc. sign
pattern. Here is an explicit example for a 3 x 3 matrix, expanding about the first row (all other options
will yield identical results).
e f d f d e
A a b c
a b c h k g k g h
A d e f
g h k a(ek hf ) b(dk gf ) c(dh ge)
For a 3 x 3 matrix only, the determinant can alternatively be calculated as shown, by copying columns 1
and 2 outside the matrix, multiplying the downward diagonals with + signs and multiplying the upward
diagonals with – signs (clearly the result is the same as in the above formula).
a b c a b
A d e f d e
g h k g h
iˆ ˆj kˆ p p p1z p2 y
p1 y p1z p p1z p1x p1 y 1 y 2 z
P 1 P 2 p1x p1 y p1z iˆ ˆj 1x kˆ p1x p2 z p1z p2 x
p2 y p2 z p2 x p2 z p2 x p2 y
p2 x p2 y p2 z p1x p2 y p1 y p2 x
11
System of Linear Equations
We can solve n linear equations in n unknowns with the help of a matrix. Below is an example
for n = 3.
Where aij are the nine known numerical equation coefficients, xi are the three unknowns, and bi are the
three known right-hand-side terms. Using matrix multiplication backwards, this is written as
A x b .
where
x1
x x2 is the vector of unknowns to be solved and
x
3
b1
b b2 is the vector of known numerical right-hand-side terms.
b
3
There is a unique solution x A b only if [A] has full rank. If not, A 0 (the determinant of
1
coefficient matrix [A] is zero) and the inverse of matrix [A] is undefined (since it would require dividing
by zero; in this case the rank is not full, it is less than 3, which means not all rows/columns of [A] are
linearly independent). Gaussian Elimination is more robust and more computationally efficient than
matrix inversion to solve the problem A x b for {x}.
12
Matrix Example – solve linear equations
x1 2 x2 5 1 2 x1 5
6 x1 4 x2 14 6 4 x2 14
1 2 x1 5
A x b
6 4 x2 14
x A b
1
A 1 4 2 6 8
The determinant of [A] is non-zero so there is a unique solution.
1 4 2 1/ 2 1/ 4
A
1
6 1 3/ 4 1/ 8
A
1 0
A A A A I2
1 1
check
0 1
x1 1/ 2 1/ 4 5 1
answer.
x2 3 / 4 1/ 8 14 2
Check this solution by substituting the answer {x} into the original equations A x b and ensuring
the required original {b} results.
%-------------------------------
% Matrices.m - matrix examples
% Dr. Bob, ME 3011
%-------------------------------
clear; clc;
The first solution of the linear equations above uses the matrix inverse. To solve linear
equations, Gaussian Elimination is more efficient (more on this in the dynamics notes later) and more
robust numerically; Gaussian elimination implementation is given in the third to the last line of the
above m-file (with the back-slash).
Since the equations are linear, there is a unique solution (assuming the equations are linearly
independent, i.e. the matrix is not near a singularity) and so both solution methods will yield the same
answer.
14
Output of Matrices.m
A1 =
1 0 0
0 2 0
0 0 3
A2 =
1 0 0
0 1 0
0 0 1
Add =
6 8
10 12
MultSca =
10 20
30 40
Trans =
5 7
6 8
A7 =
46 42
115 108
A8 =
39 54 69
41 58 75
31 44 57
dA9 = -8
invA9 =
-0.5000 0.2500
0.7500 -0.1250
x = 1
2
Check = 5
14
xG = 1
2
15
2. Kinematics Analysis
2.1 Position Kinematics Analysis
2.1.1 Four-Bar Mechanism Position Analysis
t tan
2
i.e. the tangent of half of the unknown angle , we need to derive cos and sin as functions of
parameter t. This derivation requires the trigonometric sum of angles formulae.
cos cos
2 2
cos cos 2 sin 2
2 2
Multiplying by a ‘1’, i.e. cos 2 over itself yields
2
cos 2 sin 2
cos 2 2 cos 2 1 tan 2 cos 2
2 2 2 2
cos
2
The cosine squared term can be divided by another ‘1’, i.e. cos 2 sin 2 1 .
2 2
16
cos 2
2
cos 1 tan 2
2 cos 2 sin 2
2 2
Dividing top and bottom by cos 2 yields
2
1
cos 1 tan 2
2 1 tan 2
2
Remembering the earlier definition for t, this result is the first derivation we need, i.e.
1 t2
cos
1 t2
sin sin
2 2
sin sin cos cos sin 2sin cos
2 2 2 2 2 2
sin
sin 2 2 cos 2 2 tan cos2
2 2 2
cos
2
1
cos 2
2 1 tan 2
2
17
Substituting this term yields
1
sin 2 tan
2 1 tan 2
2
Remembering the earlier definition for t, this result is the second derivation we need, i.e.
2t
sin
1 t 2
The tangent half-angle substitution can also be derived using a graphical method as in the figure
below.
18
Alternate solution method
E cos F sin G 0
arises often in the position solutions for mechanisms and robots. It appeared in the 4 solution for the
four-bar mechanism in the ME 3011 NotesBook and was solved using the tangent half-angle
substitution.
Next we present an alternative and simpler solution to this equation. We make two simple
trigonometric substitutions based on the figure below.
E F
cos sin
E2 F 2 E2 F 2
E F G
cos sin
E F
2 2
E F
2 2
E2 F 2
G
cos cos sin sin
E2 F 2
Applying the sum-of-angles formula cos(a b) cos a cos b sin a sin b yields
G
cos( )
E2 F 2
19
G
1,2 cos1
E F
2 2
where
F
tan1
E
and the quadrant-specific inverse tangent function atan2 must be used in the above expression for .
There are two solutions for , indicated by the subscripts 1,2, since the inverse cosine function is
double-valued. Both solutions are correct. We expected these two solutions from the tangent-half-angle
substitution approach. They correspond to the open- and crossed-branch solutions (the engineer must
determine which is which) to the four-bar mechanism position analysis problem.
G G
1 1 or 1 1
E F
2 2
E F
2 2
If this condition is violated for the four-bar mechanism, this means that the given input angle 2 is
beyond its reachable limits (see Grashof’s Law).
20
2.1.1.3 Four-Bar Mechanism Solution Irregularities
For simplicity, let 1 = 0 (just rotate the entire four-bar mechanism model for zero ground link angle).
Case 1
When r1 r4 and r2 r3 , G E r12 r22 r22 r12 2 r12 2 r2 ( r1 r1 ) c 2 0 ALWAYS, regardless
of 2.
Example
Given r1 10, r2 6, r3 6, r4 10 ; this mechanism is ALWAYS singular. To fix this let
r1 10, r2 5.9999, r3 6.0001, r4 10 and MATLAB will be able to calculate the position
analysis reliably at every input angle.
Case 2
When r1 2 r3 and r4 2r2 , and furthermore 3r3 5r2 ,
This G E 0 occurs only when 2 90 . Case 2 is much less general than case 1.
Example
Given r1 10, r2 3, r3 5, r4 6 ; this mechanism is singular when 2 90 . To fix this ignore
2 90 or set your 2 array to avoid these values.
21
2.1.1.4 Grashof’s Law and Four-Bar Mechanism Joint Limits
Grashof’s Law
Grashof’s Law was presented in the ME 3011 NotesBook to determine the input and output link
rotatability in a four-bar mechanism. Applying Grashof’s Law we determine if the input and output
links are a crank (C) or a rocker (R). A crank enjoys full 360 degree rotation while a rocker has a
rotation that is a subset of this full rotation. This section presents more information on Grashof’s Law
and then the next subsection presents four-bar mechanism joint limits.
Grashof's condition states "For a four-bar mechanism, the sum of the shortest and longest link
lengths should not be greater than the sum of two remaining link lengths". With a given four-bar
mechanism, the Grashof Condition is satisfied if L S P Q where S and L are the lengths of the
shortest and longest links, and P and Q are the lengths of the other two intermediate-sized links. If the
Grashof condition is satisfied, at least one link will be fully rotatable, i.e. can rotate 360 degrees.
For a four-bar mechanism, the following inequalities must be satisfied to avoid locking of the
mechanism for all motion.
r2 r1 r3 r4
r4 r1 r2 r3
With reference to the figure below, these inequalities are derived from the fact that the sum of
two sides of a triangle must be greater than the third side, for triangles O4 A1B1 and O2 A2 B2 , respectively.
Note from our standard notation, r1 O2O4 , r2 O2 A , r3 AB , and r4 O4 B .
B B1
B2
A
A2 A1
O2 O4
22
L S P Q ( 10 6 8 7 )
102 62 (8 7) 2
2 L cos 1 cos 1 0.742 137.9
2(10)(6)
This method can also be used to find angular limits on link 4 when it is a rocker. In this case links 2 and
3 align.
102 7 2 (6 8) 2
cos 1 cos 1 0.336 109.6
2(10)(7)
4 L 180 70.4
In this example, the allowable input and output angle ranges are:
This example is shown graphically in the ME 3011 NotesBook, in the Grashof’s Law section (2. Non-
Grashof double rocker, first inversion).
Caution
The figure on the previous page does not apply in all joint limit cases. For Grashof
Mechanisms with a rocker input link, one link 2 limit occurs when links 3 and 4 fold upon each other
and the other link 2 limit occurs when links 3 and 4 stretch out in a straight line. See Example 4 (and
Example 3 for a similar situation with the output link 4 limits).
24
L S P Q ( 10 4 8 7 )
Since the S link is adjacent to the fixed link, we predict this Grashof Mechanism is a crank-rocker.
Therefore, there are no 2 joint limits.
102 42 (8 7) 2
2 L cos 1 cos 1.3625
1
2(10)(4)
There are limits on link 4 since it is a rocker. For 4min, links 2 and 3 are stretched in a straight line
(their absolute angles are identical).
102 72 (4 8)2
cos 1 cos 0.036 88.0
1
2(10)(7)
For 4max, links 2 and 3 are instead folded upon each other (their absolute angles are different by ).
102 7 2 (4 8) 2
cos1 cos 0.95 18.2
1
2(10)(7)
92.0 4 161.8
and 2 is not limited. This example is shown graphically in the ME 3011 NotesBook, in the Grashof’s
Law section (1a. Grashof crank-rocker).
25
L S P Q ( 11.18 3 8 7 )
This is the four-bar mechanism from Term Example 1 and it is a four-bar crank-rocker Grashof
Mechanism. There are no limits on 2 since link 2 is a crank.
120.1 4 172.5
and 2 is not limited. This example is NOT shown graphically in the ME 3011 NotesBook Grashof’s
Law section. However, these 4 limits are clearly seen in the F.R.O.M. plot for angle 4 in Term
Example 1 in the ME 3011 NotesBook.
26
L S P Q ( 10 4 8 7 )
so we predict this Grashof Mechanism is a double-rocker (S opposite fixed link). The 2 joint limits
are no longer symmetric about the ground link, as was the case in the Non-Grashof Mechanism double
rocker (Example 1). For 2min, links 3 and 4 are folded upon each other (their absolute angles are
identical).
102 82 (7 4) 2
2min cos 1 cos 0.969 14.4
1
2(10)(8)
For 2max, links 3 and 4 are instead stretched in a straight line (their absolute angles are different by as
in Example 1).
102 82 (7 4) 2
2max cos 1 cos 1 0.269 74.4
2(10)(8)
8 8
7 7
6 6
Y (m)
Y (m)
5 5
4 4
3 3
2 2
1 1
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
X (m) X (m)
This behavior reverses for the 4 joint limits. For 4min, links 2 and 3 are stretched in a straight line
(their absolute angles are identical).
102 72 (8 4)2
cos 1 cos 0.036 88.0
1
2(10)(7)
For 4max, links 2 and 3 are instead folded upon each other (their absolute angles are different by ).
102 7 2 (8 4) 2
cos 1
cos1 0.95 18.2
2(10)(7)
8 8
7 7
6 6
Y (m)
Y (m)
5 5
4 4
3 3
2 2
1 1
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
X (m) X (m)
200
150
100
50
st
3 1
0
nd
3 2
(deg)
st
4 1
-50
nd
4 2
-100
-150
-200
-250
10 20 30 40 50 60 70 80
2 (deg)
In this plot we can see the minimum and maximum values we just calculated for links 2 and 4.
Note at 2 min 14.4 , 3 138.6 and 3 221.4 are the same angle.
Again, this example is NOT shown graphically in the ME 3011 NotesBook Grashof’s Law section.
However, a similar case with the same dimensions, in different order, is shown in the ME 3011
NotesBook ( r1 7, r2 10, r3 4, r4 8 , 1d. Grashof double rocker).
Grashof’s Law only predicts the rotatability of the input and output links; it says nothing about
the rotatability of the coupler link 3 – in this case, what is the rotatability of the coupler link? (In this
case the coupler link S rotates fully, proving that the relative motion is the same amongst all four-bar
mechanism inversions, though the absolute motion with respect to the possible 4 ground links is very
different.)
R.L. Williams II and C.F. Reinholtz, 1987, “Mechanism Link Rotatability and Limit Position
Analysis Using Polynomial Discriminants”, Journal of Mechanisms, Transmissions, and Automation
in Design, Transactions of the ASME, 109(2): 178-182.
29
2.1.2 Slider-Crank Mechanism Position Analysis
Here are the same slider-crank mechanism position analysis XY component equations, rearranged
to isolate the 3 terms.
r3c3 x r2c2
r3s3 h r2 s2
We can square and add to eliminate 3, similar to the four-bar mechanism solution approach.
a 1
ax bx c 0
2
b 2 r2 c2
c r22 r32 h 2 2hr2 s2
There are two solutions for x, corresponding to the right and left branches.
3 atan2( h r2 s 2 , x1,2 r2 c 2 )
1,2
This alternate solution yields identical results as the earlier solution approach in the ME 3011
NotesBook for the right ( 3 , x1 ) and left ( 3 , x 2 ) branches.
1 2
30
Slider-Crank Mechanism Snapshot and F.R.O.M. MATLAB m-files
No sample m-files are given in the ME 3011 NotesBook for the slider-crank mechanism since
you can readily adapt the snapshot and F.R.O.M. m-files given for the four-bar mechanism previously.
However, below we include a partial m-file to show how to draw the slider and fixed piston
walls for the slider-crank mechanism graphics, since this was not required for the four-bar mechanism.
This establishes the rectangular corner coordinates for the slider link, centered at the origin of your
coordinate frame. It can be done once, outside the loop. Instead of typing numbers for Lp and Hp, I
scale them to a fraction of r2, for generality in different-sized slider-crank mechanisms. Note I only
included the four corner points – MATLAB patch (below) closes the rectangular figure, i.e. back to
the starting point.
Inside the loop (right after the plot command where links 2 and 3 are drawn to the screen)
where x(i) is the variable horizontal slider displacement and h is the constant vertical offset. These
position parameters shift the piston coordinates from the origin to the correct location in each loop. You
can use any piston color you like (I show green here, 'g').
Inside the loop (right after the plot command where links 2 and 3 are drawn to the screen)
line(Xpt,Ypt,'LineWidth',2); line(Xpb,Ypb,'LineWidth',2);
Set the piston wall width wall to allow a small clearance between the piston and the walls. Again, it
can be scaled to a small fraction of r2 for generality. The -1000 and 1000 coordinates used above are
to extend the piston wall lines off the screen to the left and to the right.
31
MATLAB subplot feature
Outside the F.R.O.M. loop you can do the subplot in this way:
Now, you can use the standard axis labels, linetypes, titles, axis limits, grid, etc., for each plot within a
subplot (repeat these formatting commands after each plot statement above to use similar formatting
for each). These options are not shown, for clarity.
As seen in the example syntax above, the integers need not be separated by spaces or commas.
However, I believe they may be so separated if you desire.
32
2.1.3 Inverted Slider-Crank Mechanism Position Analysis
2 4
Link 1 is the fixed ground link. Without loss of generality we may force the ground link to be
horizontal. If it is not so in the real world, merely rotate the entire inverted slider-crank mechanism so it
is horizontal. Both angles 2 and 4 are measured in a right-hand sense from the horizontal to the link.
Find r4 and 4
33
Step 3. Draw the Vector Diagram. Define all angles in a positive sense, measured with the right hand
from the right horizontal to the link vector (tail-to-head; your right-hand thumb is located at the vector
tail).
2 4
1
Step 4. Derive the Vector-Loop-Closure Equation. Starting at one point, add vectors tail-to-head until
you reach a second point. Write the VLCE by starting and ending at the same points, but choosing a
different path.
r 2 r1 r 4
Step 5. Write the XY Components for the Vector-Loop-Closure Equation. Separate the one vector
equation into its two X and Y scalar components.
r2c2 r1 r4c4
r2 s2 r4 s4
Step 6. Solve for the Unknowns from the XY equations. There are two coupled nonlinear equations in
the two unknowns r4, 4. Unlike the standard slider-crank mechanism, there is no decoupling of X and
Y. However, unlike the four-bar mechanism, there is only one unknown angle so the solution is easier
than the four-bar mechanism. First rewrite the above XY equations to isolate the unknowns on one side.
r4c4 r2c2 r1
r4 s4 r2 s2
r4 s4 rs
2 2
r4c4 r2c2 r1
4 atan2( r2 s2 , r2 c2 r1 )
Then square and add the XY equations to eliminate 4 and solve for r4.
r2 c2 r1 r2 s2
X) r4 Y) r4
c4 s4
Both of these r4 alternatives are valid; however, each is subject to a different artificial mathematical
singularity ( 4 90 and 4 0,180 , respectively), so only the former square-root formula should be
used for r4, which has no artificial singularity. The X algorithmic singularity 4 90 never occurs
unless r2 r1 , which is to be avoided (see below), but the Y algorithmic singularity occurs twice per full
range of motion.
Technically there are two solution sets – the one above and r4 r1 r2 2rr
1 2c2 , 4 .
2 2
However, the negative r4 is not practical and so only the one solution set (branch) exists, unlike most
planar mechanisms with two or more branches.
Full-rotation condition
For the inverted slider-crank mechanism to rotate fully, the fixed length of link 4, L4, must be
greater than the maximum value of the variable r4.
Slider Limits
The slider reaches its minimum and maximum displacements when 2 = 0 and , respectively.
Therefore, the slider limits are r1 r2 r4 r1 r2 . Thus, the fixed length L4 must be greater than r1 r2 .
In addition we require r1 r2 for full rotation.
35
Graphical Solution
The Inverted Slider-Crank mechanism position analysis may be solved graphically, by drawing
the mechanism, determining the mechanism closure, and measuring the unknowns. This is an excellent
method to validate your computer results at a given snapshot.
Draw the known ground link (points O2 and O4 separated by r1 at the fixed angle 1 = 0).
Draw the given input link length r2 at the given angle 2 (this defines point A).
Draw a line from O4 to point A.
Measure the unknown values of r4 and 4.
36
Inverted Slider-Crank Mechanism Position Analysis: Term Example 3
Given:
r1 0.20
r2 0.10
m
L4 0.32
1 0
4 (deg) r4 (m)
150.5 0.191
0.15
0.1
Y (m)
0.05
-0.05
-0.1
-0.1 -0.05 0 0.05 0.1 0.15 0.2
X (m)
Term Example 3 Position Snapshot
37
Full-Range-Of-Motion (F.R.O.M.) Analysis: Term Example 3
A more meaningful result from position analysis is to report the position analysis unknowns for
the entire range of mechanism motion. The subplots below gives r4 (m) and 4 (deg), for all
0 2 360 , for Term Example 3.
210
200
190
4 (deg)
180
170
160
150
0 50 100 150 200 250 300 350
0.3
0.25
r4 (m)
0.2
0.15
0.1
0 50 100 150 200 250 300 350
2 (deg)
Term Example 3 4 and r4
4 varies symmetrically about 180 , being 180 at 2 0,180, 360 . r4 varies like a negative cosine
function with minimum displacement r1 r2 0.1 at 2 0, 360 and maximum displacement r1 r2 0.3
. Since r1 is twice r2 in this example, whenever 2 60,300 , a perfect 30 60 90 triangle is
formed; the relative angle between links 2 and 4 is 90 which corresponds to the max and min of
4 150, 210 , respectively. At these special points, r4 ( 3 2)0.2 0.173 m.
There is another right triangle that shows up for 2 90 ; in these cases
r4 0.22 0.12 0.224 m and 4 153.4, 206.6 , respectively. Check all of these special values in the
F.R.O.M. plot results.
38
2.1.4 Multi-Loop Mechanism Position Analysis
Thus far we have presented position analysis for the single-loop four-bar, slider-crank, and
inverted slider-crank mechanisms. The position analysis for mechanisms of more than one loop is
handled using the same general procedures developed for the single loop mechanisms. A good rule of
thumb is to look for four-bar (or slider-crank) parts of the multi-loop mechanism as we already know
how to solve the complete position analyses for these.
This section presents position analysis for the two-loop Stephenson I six-bar mechanism shown
below as an example multi-loop mechanism. This is one of the five possible six-bar mechanisms shown
in the on-line Mechanisms Atlas.
We immediately see that the bottom loop of the Stephenson I six-bar mechanism is identical to
our standard four-bar mechanism model. Since we number the links the same as in the four-bar, and if
we define the angles identically, the position analysis solution is identical to the four-bar presented
earlier. With the complete position analysis of the bottom loop thus solved, we see that points C and D
can be easily calculated. Then the solution for the top loop is essentially another four-bar solution:
graphically, the circle of radius r5 about point C must intersect the circle of radius r6 about point D to
form point E (yielding two possible intersections in general). The analytical solution is very similar to
the standard four-bar position solution, as we will show.
For multi-loop mechanisms, the number of solution branches for position analysis increases
compared to the single-loop mechanisms. Most single-loop mechanisms mathematically have two
solution branches. For multi-loop mechanisms composed of multiple single-loop mechanisms, the
number of solution branches is 2n, where n is the number of mechanism loops. For the two-loop
Stephenson I six-bar mechanism, the number of solution branches for the position analysis problem is 4,
two from the standard four-bar part, and two for each of these branches from the upper loop.
39
Now let us solve the position analysis problem for the two-loop Stephenson I six-bar mechanism
using the formal position analysis steps presented earlier. Assume link 2 is the input link.
Step 1. Draw the Kinematic Diagram this is done in the figure above.
As usual, all angles are measured in a right-hand sense from the absolute horizontal to the link, as shown
in the kinematic diagram.
Given r1, 1, r2, r3, r4, r2a, r4a, r5, r6, 2, 4; plus 1-dof position input 2
Step 3. Draw the Vector Diagram. Define all angles in a positive sense, measured with the right hand
from the right horizontal to the link vector (tail-to-head with the right-hand thumb at the vector tail and
right-hand fingers towards the arrow in the vector diagram below).
40
Step 4. Derive the Vector-Loop-Closure Equations. One VLCE is required for each mechanism loop.
Start at one point, add vectors tail-to-head until you reach a second point. Write each vector equation by
starting and ending at the same points, but choosing a different path.
r 2 r 3 r1 r 4
r 2 a r 5 r1 r 4 a r 6
Note an alternative to the second vector loop equation is r 2b r 5 r 3 r 4b r 6 . See if you can identify
r 2b and r 4b , plus their angles 2b and 4b.
Step 5. Write the XY Components for each Vector-Loop-Closure Equation. Separate the two vector
equations into four XY scalar component equations.
r2c2 r3c3 rc
1 1 r4c4
r2 s2 r3s3 r1s1 r4 s4
2a 2 2
where
4a 4 4
Step 6. Solve for the Unknowns from the four XY Equations. The four coupled nonlinear equations in
the four unknowns 3, 4, 5, 6 can be solved in two stages, one for each mechanism loop.
Loop I. This solution is identical to the standard four-bar mechanism solution for 3, 4,
summarized here from earlier. From the first two XY scalar equations above, isolate 3 terms, square
and add both equations to obtain one equation in one unknown 4. This equation has the form
E cos4 F sin 4 G 0 , where terms E, F, and G are known functions of constants and the input angle
2. Solve this equation for two possible values of 4 using the tangent half-angle substitution. The two
4 values correspond to the open and crossed branches. Then return to the original XY scalar equations
with 3 terms isolated, divide the Y by the X equations, and solve for 3 using the atan2 function,
substituting the solved values for 4. One unique 3 will result for each of the two possible 4 values.
Loop II. The method is analogous to the Loop I solution above. Since 4 is now known, we also
know 4a 4 4 . From the second two XY scalar equations above, isolate 5 terms, square and add
both equations to obtain one equation in one unknown 6. This equation is of the form
E2 cos6 F2 sin 6 G2 0 , where terms E2, F2, and G2 are known functions of constants and the known
angles 2a 2 2 and 4a 4 4 . Solve this equation for two possible values of 6 using the tangent
41
half-angle substitution. The two 6 values correspond to open and crossed branches, for each of the
Loop I open and crossed branches. Then return to the original XY scalar equations with 5 terms
isolated, divide the Y by the X equations, and solve for 5 using the atan2 function, substituting the
solved values for 6. One unique 5 will result for each of the two possible 6 values from each 4.
Branches. There are two 5, 6 branches for each of the two 3, 4 branches, so there are four
overall mechanism branches for the two-loop Stephenson I six-bar mechanism.
Full-rotation condition
The range of motion of a multi-loop mechanism may be more limited than that of single-loop
mechanisms. One can perform a compound Grashof analysis when four-bars are the component
mechanisms. For the two-loop Stephenson I six-bar mechanism, the second loop may constrain the first
loop (e.g. it may change an expected crank motion of the input link to a rocker). This is an important
issue in design of multi-loop mechanisms if the input link must still rotate fully.
Graphical Solution
The two-loop Stephenson I six-bar mechanism position analysis may readily be solved
graphically, by drawing the mechanism, determining the mechanism closure, and measuring the
unknowns. This is an excellent method to validate your computer results at a given snapshot.
Loop II (this part is a modification of the standard four-bar graphical solution). Start with the end of the
procedure above, on the same drawing. For each solution branch above, perform the following steps.
Draw the link r2a at angle 2a 2 2 from point O2.
Draw the link r4a at angle 4a 4 4 from point O4.
Draw a circle of radius r5, centered at point C.
Draw a circle of radius r6, centered at point D.
These circles intersect in general in two places to yield two possible points E.
Connect the two branches and measure the unknown values 5 and 6.
4 1
3.5 0.9
0.8
3
0.7
2.5
0.6
Vy (m/s)
2
Y (m)
0.5
1.5
0.4
1
0.3
0.5 0.2
0 0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 -0.5 0 0.5
X (m) Vx (m/s)
4 1
3.5 0.9
0.8
3
0.7
2.5
0.6
Vy (m/s)
2
Y (m)
0.5
1.5
0.4
1 0.3
xL V
0.5 0.2
0 0.1
VP V0
0
0 0.5 1 1.5 2 2.5 3 3.5 4 -0.5 0 0.5
X (m) Vx (m/s)
1.8 0.4
V
0x
1.6 V
0.35 0y
P
0x V
P
0y
1.4
L
1.2
0.25
0.2
0.8
0.15
0.6
0.1
0.4
0.2 0.05
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (sec) time (sec)
0.5
0.5
0.4
0.4
0.3
0.3
Sliding and Tangential Velocities (m/s)
V 0.2
Px
0.2
V
Py
Point P Velocity (m/s)
0.1
0.1
0
0
-0.1
-0.1 V
sx
-0.2 V
sy
-0.2 V
tx
-0.3 V
ty
-0.3
-0.4
-0.4
-0.5
-0.5
Constant velocity terms V0 x V0 y V 0.2 0.1 0.3 0.4 lead to non-constant point P velocity
(due to the nonlinear position kinematics).
44
2.2.3 Four-Bar Mechanism Velocity Analysis
The four-bar mechanism velocity solution presented in the 3011 NotesBook used the simplifying
terms a – f. Here is the solution presented in the 3011 NotesBook:
a r3 s3
b r4 s4
ce bf af cd c r22 s2
3 4
ae bd ae bd d r3c3
e r4c4
f r22c2
Backsubstituting the terms a – f yields the following equivalent solutions, which simplify nicely using
the sum-of-angles formula sin( a b ) sin a cos b cos a sin b and better display the structure of the
solutions.
These solution forms clearly show the singularity condition derived for this problem in the 3011
NotesBook, i.e. the mechanism is singular and these solutions are both infinite when:
sin(4 3 ) 0 4 3 0 ,180 ,
Physically, this happens when links 3 and 4 are straight out or folded on top of each other, as explained
in the 3011 NotesBook (corresponding to joint limits on the input link 2).
45
2.2.5 Inverted Slider-Crank Mechanism Velocity Analysis
Again, link 2 (the crank) is the input and link 4 is the output. Remember r4 is a variable so
r4 0 in this problem.
Step 1. The inverted slider-crank mechanism Position Analysis must first be complete.
2 4
1
where i (i = 2,4) is the absolute angular velocity of link i. r4 is the slider velocity along link 4.
3 4 since the slider cannot rotate relative to link 4.
Here are the inverted slider-crank mechanism position equations from earlier.
r 2 r1 r 4
r2 c2 r1 r4 c4
r2 s2 r4 s4
These two linear equations in two unknowns can be written in matrix form.
c4 r4 s4 r4 r22 s2
s
4 r4 c4 4 r22 c2
r ( s c c s )
r4 1 r4 c4 r4 s4 r22 s2 2 2 2 4 2 4
r (s s c c )
4 r4 s4 c4 r22 c2 2 2 2 4 2 4
r4
r22 sin( 4 2 )
r4
r22 cos( 4 2 )
4
r4
The units are all correct in the solution above, m/s and rad/s, respectively.
47
Inverted slider-crank mechanism singularity condition
When does the solution fail? This is an inverted slider-crank mechanism singularity, when the
determinant of the coefficient matrix goes to zero. The result is dividing by zero, resulting in infinite
r4 , 4 .
c r4 s4
A s4 r4 c4
4
Physically, assuming r1 r2 as in the full rotation condition from the inverted slider-crank mechanism
position analysis, this is impossible, i.e. r4 never goes to zero.
This matrix determinant A r4 was used in the solution of the previous page.
Given
r1 0.20
2 70
r2 0.10
(m) 4 150.5 (deg and m)
L4 0.32
r4 0.191
1 0
Snapshot Analysis
Given this mechanism position analysis plus 2 25 rad/s, calculate r4 , 4 for this snapshot.
0.870 0.094 r4 2.349
0.493 0.166 0.855
4
r4 2.47
(m/s and rad/s)
4 2.18
Both are positive, so the slider link 3 is currently traveling up link 4 and link 4 is currently rotating in the
ccw direction, which makes sense from the physical problem.
48
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 3 continued
A more meaningful result from velocity analysis is to report the velocity analysis results for the
entire range of mechanism motion. The subplot below gives 4 (top, rad/s) and r4 (bottom, m/s), for all
0 2 360 , for Term Example 3. Since 2 is constant, we can plot the velocity results vs. 2 (since it
is related to time t via 2 2t ).
10
0
4 (rad/s)
-5
-10
-15
-20
-25
0 50 100 150 200 250 300 350
1
r4d (m/s)
-1
-2
As expected, 4 is zero at the max and min for 4 (at 2 60 ); also, 4 has a large range of nearly
constant positive velocity near the middle of motion – this can be seen in a MATLAB animation. r4 is
zero at the beginning, middle, and end of motion and is max at 2 60 .
49
2.2.6 Multi-Loop Mechanism Velocity Analysis
Thus far we have presented velocity analysis for the single-loop four-bar, slider-crank, and
inverted slider-crank mechanisms. The velocity analysis for mechanisms of more than one loop is
handled using the same general procedures developed for the single loop mechanisms.
This section presents velocity analysis for the two-loop Stephenson I six-bar mechanism shown
below as an example multi-loop mechanism. It follows the position analysis for the same mechanism
presented earlier.
The bottom loop of the Stephenson I six-bar mechanism is identical to the standard four-bar
mechanism model and so the velocity analysis solution is identical to the four-bar presented earlier.
With the complete velocity analysis of the bottom loop thus solved, the solution for the top loop is
essentially another four-bar velocity solution.
As in all velocity analysis, the velocity solution for multi-loop mechanisms is a linear analysis
yielding a unique solution (assuming the given mechanism position is not singular) for each position
solution branch considered. The position analysis must be complete prior to the velocity solution.
Now let us solve the velocity analysis problem for the two-loop Stephenson I six-bar mechanism
using the formal velocity analysis steps presented earlier. Again, assume link 2 is the input link.
50
Step 1. The Stephenson I six-bar mechanism Position Analysis must first be complete.
Given r1, 1, r2, r3, r4, r2a, r4a, r5, r6, 2, 4, and 2 we solved for 3, 4, 5, 6.
where i (i = 2,3,4,5,6) is the absolute angular velocity of link i. Triangular links 2 and 4 each have a
single angular velocity for the whole link. ri 0 for all links since all links are of fixed length (no
sliders).
Given the mechanism r1, 1, r2, r3, r4, r2a, r4a, r5, r6, 2, 4,
Vector equations
r 2 r 3 r1 r 4
r 2 a r 5 r1 r 4 a r 6
XY scalar equations
r2c2 r3c3 rc
1 1 r4c4
r2 s2 r3s3 r1s1 r4 s4
2 a 2 2
where
4 a 4 4
The first time derivatives of the Loop I position equations are identical to those for the standard
four-bar mechanism.
r22 s2 r33s3 r44 s4
r22c2 r33c3 r44c4
2a 2
where we have used since 2 and 4 are constant angles.
4a 4
52
Step 5. Solve the velocity equations for the unknowns 3, 4, 5, 6.
The two mechanism loops decouple so we find 3 and 4 from Loop I first and then use 4 to
find 5 and 6 from Loop II. The solutions are given below.
1
3 r3 s3 r4 s4 r22 s2
4 r3c3 r4 c4 r22 c2
1
5 r5 s5 r6 s6 r2 a2 s2 a r4 a4 s4 a
6 r5 c5 r6 c6 r2 a2 c2 a r4 a 4 c4 a
Remember, Gaussian elimination is more efficient and robust than the matrix inverse. Also, these
equations may be solved algebraically instead of using matrix methods to yield the same answers.
The velocity solution fails when the determinant of either coefficient matrix above goes to zero.
The result is dividing by zero, resulting in infinite angular velocities for the associated loop.
For the first loop, the singularity condition is identical to the singularity condition of the standard
four-bar mechanism, i.e. when links 3 and 4 either line up or fold upon each other, causing a link 2 joint
limit. For the second loop, the singularity condition is similar, occurring when links 5 and 6 either line
up or fold upon each other. These conditions also cause angle limit problems for the position analysis,
so the velocity singularities are known problems.
53
and t 0.1 sec was used; the initial and final snapshots, with their five-part acceleration diagrams, are
shown below.
4 1.2
3.5
1
3
2.5 0.8
Ay (m/s )
2
2
Y (m)
0.6
1.5
1 0.4
0.5
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 -0.8 -0.2 0.4
X (m) 2
A (m/s )
Initial Kinematic Diagram Initial Five-Part Acceleration Diagram
4 1.2
3.5
1
3
2.5 0.8
Ay (m/s )
2
2
Y (m)
0.6
1.5 xL
0.4
1 2 x V
x ( x L)
0.5
0.2
0 AP A
A0
0
0 0.5 1 1.5 2 2.5 3 3.5 4 -0.8 -0.2 0.4
X (m) 2
Ax (m/s )
0.9
2 0.8
V
0x
V0y
0.7
P
0x
P0y
1.5 0.6
L
0.5
1 0.4
0.3
0.5 0.2
0.1
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (sec) time (sec)
0.25
A
0x
0.4
A
0y
A
0.2
0.2
Acceleration Terms (m/s and rad/s )
2
0
0.15
2
APx
-0.2
A
Py
0.1
-0.4
0.05
-0.6
0 -0.8
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (sec) time (sec)
Constant acceleration terms A0 x A0 y A 0.2 0.05 0.15 0.1 lead to non-constant point P
acceleration (due to nonlinear position kinematics and centripetal acceleration).
55
2.3.3 Four-Bar Mechanism Acceleration Analysis
The four-bar mechanism acceleration solution presented in the 3011 NotesBook used the
simplifying terms a, b, C, d, e, F. Here is the solution presented in the 3011 NotesBook:
a r3 s3
b r4 s4
Ce bF aF Cd C r2 2 s2 r222 c2 r332 c3 r442 c4
3 4
ae bd ae bd d r3c3
e r4 c4
F r2 2 c2 r222 s2 r332 s3 r442 s4
Backsubstituting the terms a, b, C, d, e, F yields the following equivalent solutions, which simplify
nicely using the sum-of-angles formulae sin( a b ) sin a cos b cos a sin b and
cos( a b) cos a cos b sin a sin b and better display the structure of the solutions.
These solution forms clearly show the singularity condition derived for this problem in the 3011
NotesBook, i.e. the mechanism is singular and these solutions are both infinite when:
sin(4 3 ) 0 4 3 0 ,180 ,
Physically, this happens when links 3 and 4 are straight out or folded on top of each other, as explained
in the 3011 NotesBook (corresponding to joint limits on the input link 2).
56
2.3.4 Slider-Crank Mechanism Acceleration Analysis
Derivative/Integral Relationships
When one variable is the derivative of another, recall the relationships from calculus. For example:
dx(t )
x(t ) x(t ) x0 x (t )dt
dt
dx (t )
x(t )
x (t ) x0
x(t )dt
dt
0.2
x (m)
0.1
3
2
xd (m/s)
1
0
-1
-2
0 50 100 150 200 250 300 350
50
xdd (m/s )
2
-50
0 50 100 150 200 250 300 350
2 (deg)
The value of x at any point is the slope of the x curve at that point. The value of x at any point is
the integral of the x curve up to that point (the value of x at any point is the area under the x curve up
to that point plus the initial value x0). A similar relationship exists for x and x .
These graphs are plotted vs. 2, but the same type of relationships hold when plotted vs. time t
since 2 is constant. This is the Term Example 2 F.R.O.M. result.
Note these curves should be plotted vs. time t instead of 2 in order to see the true slope and area
values accurately.
57
2.3.5 Inverted Slider-Crank Mechanism Acceleration Analysis
Again, link 2 (the crank) is the input and link 4 is the output.
Step 1. The inverted slider-crank mechanism Position and Velocity Analyses must first be complete.
Given r1, 1 = 0, r2, and 2 we solved for r4 and 4; then given 2 we solved for r4 and 4.
2 4
1
where i (i = 2,4) is the absolute angular acceleration of link i. r4 is the slider acceleration along link 4.
3 4 since the slider cannot rotate relative to link 4.
r4 , 4 .
Step 5. Solve the acceleration equations for the unknowns
r4 1 r4c4 r4 s4 r2 2 s2 r222 c2 2r44 s4 r442 c4
4 r4 s4 c4 r2 2 c2 r222 s2 2r44c4 r442 s4
2
r2 2 sin( 4 2 ) r22 cos( 4 2 ) r44
r4 2
r cos( ) r 2sin( ) 2r
4 2 2 4 2 2 2 4 2 4 4
r4
A major amount of algebra and trigonometry is required to get the final analytical solution for
r4 , 4 above. Interestingly, the link 4 Coriolis term 2r44 cancelled in r4 while the link 4 centripetal
term r4 42 cancelled in 4. The units are all correct, m/s2 and rad/s2, respectively.
59
Inverted slider-crank mechanism singularity condition
The acceleration problem has the same coefficient matrix [A] as the velocity problem, so the
singularity condition is identical (see the singularity discussion in the inverted slider-crank mechanism
velocity section – the only singularity is when r4 goes to zero; this will never to occur if r1 r2 ).
Given
r1 0.20
2 70 2 25
r2 0.10
(m) 4 150.5 (deg and m)
r4 2.47 (rad/s and m/s)
L4 0.32
r4 0.191 4 2.18
1 0
Snapshot Analysis
Given this mechanism position and velocity analyses, plus 2 0 rad/s2, calculate
r4 , 4 for this
snapshot.
0.870 0.094
r4 16.873
0.493 0.166 48.957
4
r4 9.46
(m/s2 and rad/s2)
4 267.14
r4 is negative, so the slider link 3 is currently slowing down its positive velocity up link 4 and 4 is
positive, so the link 4 angular velocity is currently increasing in the ccw direction.
60
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 3 continued
A more meaningful result from acceleration analysis is to report the acceleration analysis results
for the entire range of mechanism motion. The subplot below gives 4 (top, rad/s2) and r4 (bottom,
m/s2), for all 0 2 360 , for Term Example 3. Again, since 2 is constant, we can plot the
acceleration results vs. 2 (since it is related to time t via 2 2t ).
1000
500
4 (rad/s2)
-500
-1000
0 50 100 150 200 250 300 350
100
r4dd (m/s )
2
50
-50
0 50 100 150 200 250 300 350
2 (deg)
As expected, 4 is zero at the beginning, middle, and end since the 4 curve flattens out at those
points. The maximum (and minimum) 4 values correspond to the greatest slopes for 4. r4 is
maximum (and minimum) at the beginning, middle, and end since the r4 curve is steepest at those
points; r4 is zero when the r4 curve is flat, i.e. 2 60 .
61
Derivative/Integral Relationships
When one variable is the derivative of another, recall the relationships from calculus (the
derivative is the slope of the above curve at each point; the integral is the area under the curve up to that
point, taking into account the initial value). For example:
d4 (t )
4 (t ) 4 (t ) 40 4 (t )dt
dt
d4 (t )
4 (t ) 4 (t ) 40 4 (t )dt
dt
200
4 (d e g )
180
160
10
0
4 (ra d /s )
-10
-20
1000
500
4 (ra d /s 2 )
-500
-1000
0 50 100 150 200 250 300 350
2 (deg)
62
dr4 (t )
r4 (t ) r4 (t ) r40 r4 (t )dt
dt
dr4 (t )
r4 (t )
r4 (t ) r40
r4 (t )dt
dt
0.25
r 4 (m )
0.2
0.15
0.1
0 50 100 150 200 250 300 350
2
1
r 4 d (m /s)
0
-1
-2
0 50 100 150 200 250 300 350
100
r 4 d d (m /s 2 )
50
-50
0 50 100 150 200 250 300 350
2 (deg)
Thus far we have presented acceleration analysis for the single-loop four-bar, slider-crank, and
inverted slider-crank mechanisms. The acceleration analysis for mechanisms of more than one loop is
handled using the same general procedures developed for the single loop mechanisms.
This section presents acceleration analysis for the two-loop Stephenson I six-bar mechanism
shown below as an example multi-loop mechanism. It follows the position and velocity analyses for the
same mechanism presented earlier.
The bottom loop of the Stephenson I six-bar mechanism is identical to the standard four-bar
mechanism model and so the acceleration analysis solution is identical to the four-bar presented earlier.
With the complete acceleration analysis of the bottom loop thus solved, the solution for the top loop is
essentially another four-bar acceleration solution.
As in all acceleration analysis, the acceleration solution for multi-loop mechanisms is a linear
analysis yielding a unique solution (assuming the given mechanism position is not singular) for each
solution branch considered. The position and velocity analyses must be complete prior to the
acceleration solution.
64
Now let us solve the acceleration analysis problem for the two-loop Stephenson I six-bar
mechanism using the formal acceleration analysis steps presented earlier. Again, assume link 2 is the
input link.
Step 1. The Stephenson I six-bar mechanism Position and Velocity Analyses must first be complete.
Given r1, 1, r2, r3, r4, r2a, r4a, r5, r6, 2, 4, 2, and 2, we solved for 3, 4, 5, 6, 3, 4, 5, 6.
where i (i = 2,3,4,5,6) is the absolute angular acceleration of link i. Triangular links 2 and 4 each have
a single angular acceleration for the whole link. ri 0 for all links since all links are of fixed length (no
sliders).
Given the mechanism r1, 1, r2, r3, r4, r2a, r4a, r5, r6, 2, 4,
the position analysis 2, 3, 4, 5, 6,
the velocity analysis 2, 3, 4, 5, 6,
1-dof acceleration input 2
2 a 2 2
where
4 a 4 4
The first time derivative of the Loop I velocity equations is identical to that for the standard four-
bar mechanism.
2a 2 2a 2
where we have used and since 2 and 4 are constant angles.
4a 4 4a 4
66
Step 5. Solve the acceleration equations for the unknowns 3, 4, 5, 6.
The two loops decouple so we find 3 and 4 from Loop I first and then use 4 to find 5 and 6
from Loop II. The solutions are given below.
1
3 r3 s3 r4 s4 r2 2 s2 r222 c2 r332 c3 r442 c4
4 r3c3 r4 c4 r2 2 c2 r222 s2 r332 s3 r442 s4
1
5 r5 s5 r6 s6 r2 a 2 s2 a r2 a22 c2 a r552 c5 r4 a 4 s4 a r4 a42 c4 a r662 c6
6 r5 c5 r6 c6 r2 a 2 c2 a r2 a22 s2 a r552 s5 r4 a 4 c4 a r4 a42 s4 a r662 s6
Remember, Gaussian elimination is more efficient and robust than the matrix inverse. Also, these
equations may easily be solved algebraically instead of using matrix methods.
The acceleration solution fails when the determinant of either coefficient matrix above goes to
zero. The result is dividing by zero, resulting in infinite angular accelerations for the associated loop.
Note the two coefficients matrices in the acceleration solutions are identical to those for the velocity
solutions. Therefore, the acceleration singularity conditions are identical to the velocity singularity
conditions.
For the first loop, the singularity condition is identical to the singularity condition of the standard
four-bar mechanism, i.e. when links 3 and 4 either line up or fold upon each other, causing a link 2 joint
limit. For the second loop, the singularity condition is similar, occurring when 5 and 6 either line up or
fold upon each other. These conditions also cause problems for the velocity and position analyses, so
the acceleration singularities are known problems.
67
A
C
Here are the kinematics equations (we already presented the point C kinematics equations in the ME
3011 NotesBook).
cx ax rCA cos( 3 3C ) d x bx rDB cos(3 3 D )
C D
c y a y rCA sin(3 3C ) d y by rDB sin( 3 3 D )
In this simple straight-line case, use 3C 180 and 3 D 0 . Here is partial MATLAB code for link 3
extensions animation.
x2 = [0 ax(i)]; % coordinates of link 2
y2 = [0 ay(i)];
x3 = [cx(i) dx(i)]; % coordinates of link 3
y3 = [cy(i) dy(i)];
x4 = [r1x bx(i)]; % coordinates of link 4
y4 = [r1y by(i)];
figure; plot(x2,y2,'r',x3,y3,'g',x4,y4,'b');
68
Four-bar mechanism link 4 extensions
C
O4
In this simple straight-line case, use 4C 0 and 4 D 180 . bx and by were given above and O 4 is:
r cos 1
O4 1
r1 sin 1
Of course, one can use combinations of these graphics approaches as necessary. Also, use
patch for drawing solid polygonal links rather than straight lines. You can also use patch for
drawing solid circles. This method is also extendable to non-straight-line links by using the appropriate
angles.
69
Jerk can be important for kinematic motion analysis in general. Position, velocity, and
acceleration analyses must be completed first. Jerk is the first time derivative of the acceleration, the
second time derivative of the velocity, and the third time derivative of the position. Like all the terms
preceding it, jerk is also a vector quantity. There are both translational and rotational jerk terms.
d A(t ) d 2 V (t ) d 3 P (t ) m
J (t ) SI units:
dt dt 2 dt 3 s3
d (t ) d 2 (t ) d 3 (t ) rad
(t ) SI units:
dt dt 2 dt 3 s3
2.5.1 Jerk Analysis Introduction
In this section we will derive the n-part jerk formula, showing the most general jerk terms
possible for planar devices.
This is a four-dof system consisting of a translating/rotating rigid rod with a slider. The same
system was used for the two-part position, three-part velocity, and five-part accelerations formula
derivations in the ME 3011 NotesBook. Find the total jerk of the point P, which is on the slider.
70
Start with the five-part acceleration formula from before and take another time derivate of the XY
components. What is n? (Hint – clearly n must be greater than 5.)
Recall the two-part position, three-part velocity, and five-part acceleration formula results below.
P P PO L
P L cos
OX
POY L sin
V P V O V L
V V cos L sin
OX
VOY V sin L cos
AP AO A 2 V L L
AOX A cos 2V sin L sin L 2 cos
AOY A sin 2V cos L cos L sin
2
The angle , angular velocity , angular acceleration , and angular jerk are all changing with respect
to time (only the planar case is this simple; the spatial rotation case is more complicated).
d d 2 d 3
2 3
dt dt dt
The rod length L, sliding velocity V, sliding acceleration A, and sliding jerk J are all changing with
respect to time.
d A d 2V d 3 L
J 2 3
dt dt dt
Here are the same relationships, using the dot notation to indicate time differentiation.
d d 2 d 3 d L d 2 L d 3 L
2 3
L 2 3
dt dt dt dt dt dt
71
Product and Chain Rules of Differentiation
Again, we’ll need to use the product and chain rules repeatedly in jerk analysis derivations.
Product rule
d dx(t ) dy(t )
( x(t ) y(t )) y(t ) x(t )
dt dt dt
Chain rule
d df ( x(t )) dx(t )
( f ( x(t )))
dt dx dt
Examples
d d
(2V sin ) 2 A sin 2V sin 2V (sin )
dt dt
d (sin ) d
2 A sin 2V sin 2V
d dt
2 A sin 2V sin 2V cos
2
d d d
( L 2 cos ) V 2 cos L ( 2 ) cos L 2 (cos )
dt dt dt
d ( ) d
2
d (cos ) d
V 2 cos L cos L 2
d dt d dt
V cos 2 L cos L sin
2 3
Many terms will combine (like in the Coriolis acceleration case). Check the resulting units of all
components to check your results. The full n-part jerk derivation is left to the interested student.
72
2.5.2 Mechanism Jerk Analysis
Given the mechanism, complete position, velocity, and acceleration analyses, and one-dof of jerk
input, calculate the jerk unknowns.
For a given branch of a known mechanism, this will yield linear equations so a matrix-vector
approach may be used to obtain the unique solution (assuming no singularity).
Since jerk kinematics is not required for dynamics analysis or machine design (generally), this
solution is beyond the scope of the class.
For our jerk needs, we may use the time differentiation approach of the previous subsection.
However, for the x and x slider-crank mechanism results, this symmetry is not immediately
evident. This special symmetry is revealed when first cutting the plots at 2 180 , and then performing
the horizontal and/or vertical flipping.
77
r1 11.18 r1 0.284
r2 3 r2 0.076
Given: in m
r3 8 r3 0.203
r4 7 r4 0.178
and 1 10.3 (Ground link is 11" over and 2" up). Also given rCA 5 in (0.127 m) and 3 36.9 for
the coupler link point of interest.
Given this mechanism and 2 30 , calculate 3 ,4 , , and PC for both branches. Results:
E 0.076
F 0.005
G 0.036
branch t 3 4 PC
These two branch solutions are demonstrated in the figures below. The length units are m. Note
is identical for both branches due to the conventions presented earlier.
78
0.25
0.2
0.15
Y (m) 0.1
0.05
-0.05
-0.1
0.25
0.2
0.15
0.1
Y (m)
0.05
-0.05
-0.1
A more meaningful result from position analysis is to solve and plot the position analysis
unknowns for the entire range of mechanism motion.
The plot below gives 3 (red), 4 (green), and (blue), all deg, for all 0 2 360 , for Term
Example 1, the open branch only.
180
160
140
120
3
i (deg)
100 4
80
60
40
20
0 50 100 150 200 250 300 350
2 (deg)
The plot below gives the initial (and final) animation position, for 2 0,360 . It also gives the
coupler curve for the open branch, plotting PCY vs. PCX in green circles.
0.25
0.2
0.15
0.1
Y (m)
0.05
-0.05
-0.1
P C Coupler Curve
Given r1 = 0.284, r2 = 0.076, r3 = 0.203, r4 = 0.178, rCA = 0.127 m, and 1 10.3 , 2 30 ,
3 53.8 , 4 121.7 , 3 36.9 . This is the open branch of the Term Example 1 four-bar mechanism.
Snapshot Analysis
Given this mechanism position analysis plus 2 20 rad/s (positive, which indicates ccw),
calculate 3 , 4 , and V C for this instant in motion (snapshot).
Both 3 and 4 are negative, so they are in the cw direction for this snapshot. These results are the
absolute angular velocities of links 3 and 4 with respect to the ground link. The associated coupler point
translational velocity vector for this snapshot is:
0.265
VC (m/s)
1.330
82
Full-Range-Of-Motion (F.R.O.M.) Analysis: Term Example 1 continued
A more meaningful result from velocity analysis is to solve and plot the velocity analysis
unknowns for the entire range of mechanism motion.
The plot below gives 3 (red) and 4 (green) (rad/s) for all 0 2 360 , for Term Example 1,
for the open branch only. For all of Term Example 1, assume the 2 given above is constant. Since 2
is constant, we can plot the velocity results vs. 2 (since it is related to time t via 2 2t ).
10
2
i (rad/s)
-2
3
-4
4
-6
-8
-10
0 50 100 150 200 250 300 350
2 (deg)
3 and 4
83
The plot below gives the absolute translational coupler point C velocity for all 0 2 360 , for
Term Example 1, for the open branch only.
1.5
1
VCX
VCY
0.5
VC (m/s)
-0.5
-1
-1.5
0 50 100 150 200 250 300 350
2 (deg)
VC
Given r1 = 0.284, r2 = 0.076, r3 = 0.203, r4 = 0.178, rCA = 0.127 m, and 1 10.3 , 2 30 ,
3 53.8 , 4 121.7 , 3 36.9 ; 2 20 (constant), 3 8.073 , 4 3.729 rad/s. This is the open
branch of the position and velocity example (Term Example 1).
Snapshot Analysis
Given this mechanism position and velocity analysis, plus 2 0 rad/s2, calculate 3 , 4 for
this instant in motion (snapshot).
3 7.994
4 243.018
Both are positive, so they are ccw in direction. These results are the absolute angular accelerations of
links 3 and 4 with respect to the ground link. The coupler point translational acceleration vector is:
27.230
AC m/s
2
23.490
85
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 1 continued
A more meaningful result from acceleration analysis is to solve and plot the acceleration analysis
unknowns for the entire range of mechanism motion.
The plot below gives 3 (red) and 4 (green), (rad/s2), for all 0 2 360 , for Term Example
1, open branch. In the Term Example 1 velocity section it was assumed that the given 2 is constant,
which means that the given 2 is always zero. Since 2 is constant, we can plot the acceleration results
vs. 2 (since it is related to time t via 2 2t ).
250
200
3
150
4
100
50
i (rad/s )
2
-50
-100
-150
-200
-250
0 50 100 150 200 250 300 350
2 (deg)
3 and 4
86
The plot below gives the translational coupler point acceleration for all 0 2 360 , for Term
Example 1, the open branch only.
30
20
10
AC (m/s )
0
2
-10
ACX
ACY
-20
-30
AC
When one variable is the derivative of another, recall the relationships from calculus. e.g.:
d (t )
4 (t ) 4 4 (t ) 40 4 (t )dt
dt
d 4 (t ) 4 (t ) 40 4 (t )dt
4 (t )
dt
3
4 (rad)
2.5
2
0 50 100 150 200 250 300 350
10
5
4 (rad/s)
-5
-10
0 50 100 150 200 250 300 350
300
200
4 (rad/s )
2
100
0
-100
-200
0 50 100 150 200 250 300 350
2 (deg)
The value of 4 at any point is the slope of the 4 curve at that point. The value of 4 at any point
is the integral of the 4 curve up to that point (the value of 4 at any point is the area under the 4 curve
up to that point plus the initial value 40). A similar relationship exists for 4 and 4.
These graphs are plotted vs. 2, but the same type of relationships hold when plotted vs. time t
since 2 is constant. This is the Term Example 1 F.R.O.M. result, but 4 was changed from deg to rad
for better comparison.
These curves should be plotted vs. time t instead of 2 in order to see the true slope and area
values accurately.
88
2.7.2 Term Example 2: Slider-Crank Mechanism
Given:
r2 4 r2 0.102
r3 8 in r3 0.203 m
h3 h 0.076
Given this mechanism and 2 30 , calculate x and 3 for both branches. Results:
These two branch solutions are demonstrated in the figures below (length units are m).
0.2 0.2
0.1 0.1
Y (m)
Y (m)
0 0
-0.1 -0.1
-0.2 -0.2
A more meaningful result from position analysis is to solve and plot the position analysis
unknowns for the entire range of mechanism motion.
The top plot gives 3 (deg) and the bottom plot gives x (m), for all 0 2 360 , for Term
Example 2, the right branch only.
70
60
50
40
3 (deg)
30
20
10
0
-10
0 50 100 150 200 250 300 350
0.3
0.25
0.2
x (m)
0.15
0.1
0.05
0 50 100 150 200 250 300 350
2 (deg)
3 and x
90
The plot below gives the initial (and final) animation position, for 2 0,360 . It also gives the
coupler curve to scale for the right branch, plotting PCY vs. PCX in green. In this case the coupler point is
taken to be the midpoint of coupler link 3.
0.2
0.15
0.1
0.05
Y (m)
-0.05
-0.1
-0.15
-0.2
-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
X (m)
P C Coupler Curve
For Term Example 2, the slider translation limits are 0.067 x 0.295 , as seen in the x plot
above, calculated from the equations in the 3011 NotesBook.
91
Slider-crank mechanism velocity example – Term Example 2 continued
Given r2 = 0.102, r3 = 0.203, h = 0.076 m, and 2 30 , 3 7.2 , x = 0.290 m. This is the right
branch of the slider-crank position example of Term Example 2.
Given this mechanism position analysis plus 2 15 rad/s (+ so ccw), calculate 3 , x for this
instant in time (snapshot).
3 6.577
x 0.601
These results are the absolute rotational and translational velocities of links 3 and 4 with respect to the
fixed ground link. Both are negative, so the coupler link 3 is currently rotating in the clockwise direction
and the slider 4 is currently translating to the left.
92
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 2 continued
A more meaningful result from velocity analysis is to solve and plot the velocity analysis
unknowns for the entire range of mechanism motion. The subplot arrangement below gives 3 (top,
rad/s) and x (bottom, m/s), for all 0 2 360 , for Term Example 2, the right branch only. For all of
Term Example 2, assume the 2 given above is constant. Since 2 is constant, we can plot the velocity
results vs. 2 (since 2 changes linearly, as it is related to time t via 2 2t ).
10
5
3 (rad/s)
-5
-10
0 50 100 150 200 250 300 350
1
xd (m/s)
-1
-2
0 50 100 150 200 250 300 350
2 (deg)
Given this mechanism position and velocity analysis plus 2 0 rad/s2, calculate 3 ,
x for this
snapshot in time.
0.025 1 3 28.590
0.202 0
x 12.557
3 62.329
x 30.148
These results are the absolute angular and linear accelerations of links 3 and 4 with respect to the fixed
ground link.
94
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 2 continued
A more meaningful result from acceleration analysis is to solve and plot the acceleration analysis
unknowns for the entire range of slider-crank mechanism motion. The top plot gives 3 (rad/s2) and the
bottom plot gives x (m/s2), for all 0 2 360 , for Term Example 2, for the right branch only. In the
Term Example 2 velocity section it was assumed that the given 2 is constant, which means that the
given 2 is always zero. Since 2 is constant, we can plot the velocity results vs. 2 (since it is related to
time t via 2 2t ).
150
100
50
0
3 (rad/s )
2
-50
-100
-150
-200
-250
0 50 100 150 200 250 300 350
50
xdd (m/s )
2
-50
0 50 100 150 200 250 300 350
2 (deg)
3. Dynamics Analysis
3.1 Dynamics Introduction
D’Alembert’s Principle
We can convert dynamics problems into statics problems by the inclusion of a vector inertial
force F 0 mAG and a vector inertial moment M 0 I G . Centrifugal force mr 2 , directed away
from the center of rotation, is an example of an inertial force vector. It’s not really a force but a felt
effect of an inertia in acceleration. Using D’Alembert’s principle, the right-hand side of the translational
and rotational dynamics equations is subtracted to the other side of the equation. Then the forces and
moments balance to zero as in statics, when the inertial forces are included in the FBD.
We won’t use this method, but it is mentioned for completeness. We would instead to prefer to
consider statics problems as a subset of dynamics problems, with zero accelerations.
R mAG 0
R FO 0
T r R IG 0
T rRMO 0
xkcd.com
96
Newton’s Second Law requires m and CG, Euler’s Rotational Dynamics Equation requires CG and IGZ.
m PCG IGZ
translational mass center of gravity
rotational center of gravity mass moment of inertia
Mass
In Newton’s Second Law F mA G
, the mass m is the proportionality constant. Mass is
measure of translational inertia – a resistance to change in motion according to Newton’s First Law.
1 2
Mass is also a measure of the storage of translational kinetic energy KET mv and the units are kg.
2
Mass calculation
System of particles General rigid body
N
m mi m dm
i 1 body
b/2 h/2
m tdxdy tbh m V (an obvious result)
b / 2 h / 2
97
Center of Gravity (CG, G)
The CG is the point at which a body is balanced with respect to gravity. It is also the point at
which the body weight acts. The CG is also called the mass center, center of mass, and centroid. It is a
vector quantity and the units are length units, m.
CG calculation
System of particles
X CG X
m x i i
P CG
m r i i
Cartesian components
m i
m i
YCG Y
m y i i
m i
dm
Rectangular rigid body
Using an XY coordinate frame at the geometric center, the CG is calculated below.
xdm ydm
X x
Y
y
dm dm
m m y
xdV ydV
x
b/2
m
h/2
m
b / 2
xthdx ytbdy
h / 2
th b/2
tb
h/2
m b / 2
xdx
m h / 2
ydy
b/2
th x 2
h/2
tb y 2
m 2 b / 2 m 2 h / 2
th b 2
b
2
tb h 2
h2
0 0
2m 4 4 2m 4 4
X 0
P CG
Y 0
For a homogeneous, regular geometric body of uniform thickness, the CG is the geometric center.
98
Mass Moment of Inertia IG
This is not the same as area moment of inertia (IA) for beam bending, which is recalled below.
I Ax y 2 dA I Ay x 2 dA units: I A m 4
y x
What is the mass moment of inertia, a scalar, vector, matrix, or something else? Answer – it is a tensor.
System of particles
I axis mi ri 2
where ri is the scalar perpendicular distance from the axis to the ith particle. With squaring, all terms
will be positive, so there can be no canceling like for the CG. If the first moment (CG) is balanced, the
second moment (IGZ) terms do not cancel since the squared terms are all positive.
I XX I XY I XZ
I axis I XY IYY IYZ
I XZ IYZ I ZZ
What is the only term that matters for XY planar motion? Answer – IZZ.
b/ 2 h /2
I GZ
body
( x 2 y 2 )dm
b /2 h /2
( x 2 y 2 ) tdxdy
b /2 y3
h /2
t x2 y dx
b /2 3
h /2
2 h h 1 h3 h3
b /2
t x dx
2 2 3 8
b /2 8
b /2 h3 hx 3 h3 x b /2
t hx dx t
2
b /2
12 3 12 b / 2
h b3 b3 h3 b b
t
3 8 8 12 2 2
b3 h bh3 tbh 2
t (b h 2 )
12 12 12
m 2 2
IGZ (b h ) (using m V tbh )
12
units: mass times distance squared, kg-m2
How do we find mass, center of gravity, and mass moment of inertia in the real world?
From tables – for example, see the three tables at the end of this section.
CAD packages (such as SolidEdge and AutoCAD) calculate m, CG, and IGZ automatically for
each link drawn, once material is associated with the 3D model.
100
Parallel Axis Theorem
The mass moment of inertia through the CG is related to mass moments of inertia of parallel
axes through different points as follows.
I ZZO I ZZG md 2
where d is the scalar distance separating the axis of interest O from the axis through the CG. Notice IZZG
is as small as possible. Any IZZO must be greater, due to the term md2 which is always positive.
I ZZO I ZZG md 2
m 2 b2 h2
(b h 2 ) m
12 4 4
b2 b2 h2 h2
m
12 4 12 4
b2 h2
m
3 3
m
(b 2 h 2 )
3
To combine any number of bodies n of known material, shape, size, and location into one rigid
body, use the following equations for mass, center of gravity, and mass moment of inertia.
n
mT m1 m2 mn mi
i 1
n
m1 x1 m2 x2 mn xn mi xi
m
i 1
X T mT
P CGT n
T
YT m1 y1 m2 y2 mn yn m y
i i
mT i 1
mT
n
I GZT I GZ 1 m1d12 I GZ 2 m2 d 22 I GZn mn d n2 I GZi mi di2
i 1
The subscript ‘T’ indicates total (or combined) and di is the distance between the combined CG ( P CG ) T
and the CG of body i ( PCG i xi yi ). These equations are obtained from the mass, CG, and IGZ
T
equations for particles, where now each particle is instead a rigid body.
101
Example 1. Two rectangles joined as shown below
1.5
Y (m)
0.5
0
0 0.5 1 1.5 2
X (m)
1.5
Y (m)
0.5
0
0 0.5 1 1.5 2
X (m)
Example 2. Two Equal Steel Rectangles
(again, b1 is vertical and b2 is horizontal, with b1 = b2 = b here)
103
Tables of Mass, Center of Gravity, and Mass Moment of Inertia
The tables below present the mass (kg), center of gravity (m), and mass moment of inertia (kg-
m2) for some common planar link shapes.
Note that mass moment of inertia (kg-m2) is not the same as area moment of inertia for beam
bending (m4). The former represents rotational inertia while the latter is a measure of resistance to beam
bending.
We assume that all link materials are homogeneous and uniformly distributed, with mass density
(kg/m3), all links have regular geometry, and all links have a constant thickness t in the Z direction.
The general equations for mass, center of gravity, and mass moment of inertia are given below
for general rigid bodies.
m
body
dm
rdm
P CG body
dm
I ZZ
body
( x 2 y 2 )dm
All of these terms require double integrals over the rigid body in the XY plane. Both center of
gravity and mass moment of inertia depend on the origin of the chosen XYZ Cartesian coordinate
system.
For general 3D rigid bodies, mass moment of inertia is a 3x3 tensor. For planar mechanism
dynamics we only need one scalar term out of these 9 terms, IZZ.
In the drawings below, the planar reference Cartesian coordinate system is shown, with origin O,
and the standard symbol is used for center of gravity, denoted as point G. The mass moment of inertia
about axis O is related to the mass moment of inertia about axis G is via the parallel axis theorem, where
d is the scalar distance (vector length) between axes G and O in the XY plane:
I OZ I GZ md 2
104
Mass Properties for Planar Links
Center of
Mass Moment of Inertia
Name Model Mass (kg) Gravity
(kg-m2)
(m)
Y mR 2
I GZ
R 2
cylinder
R 5mR 2
R2t I OZ
2
R
O
X
Y m ( Ro2 Ri2 )
Ro I GZ
Ri R0 2
hollow m (5 Ro2 Ri2 )
cylinder Ro2 Ri2 t R0 I OZ
2
O
X
Y
R R I GZ mR 2
thin ring
m R I OZ 3mR 2
O
X
Y a a b I GZ
m ( a 2 ab b 2 h 2 )
3 18
triangle bht
h h m ( a ab b 2 h 2 )
2
I OZ
2 3 6
O b
X
The formula for the mass moment inertia of a triangle was derived via double integral over the
body by Ohio University Ph.D. student Elvedin Kljuno – it could not be found in any sophomore-level
dynamics book, or in any other textbook, nor in any Internet search.
106
Mass Properties for Cylindrical Links
Now, the previous nine shapes for which the mass properties were summarized are all planar
shapes, with a constant thickness t in the Z direction (except for the point mass, point mass on massless
rod, and slender rod, whose Z dimensions are unimportant). Here we give one more shape, a cylinder
that is 3D but useful for planar slider-crank mechanisms and other planar mechanisms with a prismatic
joint and sliding cylindrical piston. The cylinder given on the previous page was arranged with the
circle in the XY plane. Now we need to rotate this so the rectangular projection of the cylinder is the XY
plane. The mass moment of inertia is quite different from that of the rectangular parallelepiped, due to
the effect of the radius R in this case, as opposed to the constant thickness t in the rectangular
parallelepiped case.
Not all pistons are solid, so we also include a similar model for the hollow piston cylinder, with
outer radius Ro and inner radius Ri.
ME 3011 uses SI units exclusively. However, many of you perform your capstone project work
using English units, which is fine, since we live in the U.S. In the 1970s the U.S. government mandated
a change to the SI system – this failed spectacularly (why?).
One big benefit of the SI system is seen in the units for Newton’s Second Law, F = ma. Using
standard SI units, this equation uses all ones (1s):
Sadly the English units DO NOT behave with ones in Newton’s Second Law:
Further, the English system has another confusion which does not exist for the SI system. The
same unit, pound (lb), applies both to force (lbf) and mass (lbm), depending on the context. Please
always use the correct subscript for clarity. Happily, a mass of 1 lbm does weigh 1 lbf at standard gravity
(g = 32.2 ft/s2 or 386.1 in/s2).
Now we present the standard English mass units; there are two, depending on if you use feet or
inches for the length unit.
WTF?!? slug? blob? I promise you I am not making this up. A slug is a rather large mass,
equivalent to 32.2 lbm (14.6 kg). A blob is even larger, equivalent to 12 slugs, 386.1 lbm (175.1 kg).
In conclusion, do not use lbm in dynamics equations for your project. Instead use slugs if you are
using feet or blobs if you are using inches. If you have estimated your masses in lbm, simply divide by
32.2 to get slugs, or divide by 386.1 to get blobs.
slug ft lb f s 2
1 lb f 1 so 1 slug 1
s2 ft
blob in lb f s 2
1 lb f 1 so 1 blob 1
s2 in
108
r R cos( 2 2 ) r R cos( 4 4 )
r 12 12 X G 2 r 14 14 X G 4
r12Y RG 2 sin( 2 2 ) r14Y RG 4 sin( 4 4 )
r r cos 2 r r cos 4
r 32 r12 r 2 12 X 2 r 34 r14 r 4 14 X 4
r12Y r2 sin 2 r14Y r4 sin 4
F F cos E 4
F E4 E4X E4
FE 4Y FE 4 sin E 4
r rE 4 cos rE 4
r E4 E4X
rE 4Y rE 4 sin rE 4
M E 4 M E 4kˆ
109
Four-bar mechanism inverse dynamics matrix equation
1 0 1 0 0 0 0 0 0 F21 X m2 AG 2 X
1 m2 ( AG 2Y g )
0 0 1 0 0 0 0 0 F21Y
r12Y r12 X r32Y r32 X 0 0 0 0 1 F32 X IG 2 Z 2
0 0 1 0 1 0 0 0 0 F32Y m3 AG 3 X FE 3 X
0
0 0 1 0 1 0 0 0 F43 X m3 ( AG 3Y g ) FE 3Y
0 0 r23Y r23 X r43Y r43 X 0 0 0 F43Y I G 3 Z 3 rE 3 X FE 3Y rE 3Y FE 3 X M E 3
0 0 0 0 1 0 1 0 0 F14 X m4 AG 4 X FE 4 X
0 0 0 0 0 1 0 1 0 F14Y m4 ( AG 4Y g ) FE 4Y
0 r34 X r14Y 0 2 I G 4 Z 4 rE 4 X FE 4Y rE 4Y FE 4 X M E 4
0 0 0 r34Y r14 X
A v b
1 0 1 0 0 0 F32 X m3 AG 3 X FE 3 x
0 1 0 1 0 0 F m3 ( AG 3Y g ) FE 3 y
32Y
r23Y r23 X r43Y r43 X 0 0 F43 X I G 3Z 3 rE 3 x FE 3 y rE 3 y FE 3 x M E 3
0 0 1 0 1 0 F43Y m4 AG 4 X FE 4 x
0 0 0 1 0 1 F14 X m4 ( AG 4Y g ) FE 4 y
0 0 r34Y r34 X r14Y r14 X F14Y I G 4 Z 4 rE 4 x FE 4 y rE 4 y FE 4 x M E 4
Solve for six unknowns v34 A34 b34 and then use F32X and F32Y in the following 3x3 set of linear
1
equations, from the link 2 FBD, very similar to the single rotating link.
1 0 0 F21 X m2 AG 2 X F32 X
0
1 0 F21Y m2 ( AG 2Y g ) F32Y
r12Y r12 X 1 2 I G 2 Z 2 r32Y F32 X r32 X F32Y
1
R.L. Williams II, 2009, “Partial Decoupling of the Matrix Method for Planar Mechanisms Inverse Dynamics”, CD
Proceedings of the ASME International Design Technical Conferences, 33rd Mechanisms and Robotics Conference, Paper #
DETC2009-87054, San Diego CA, August 30-September 2.
110
We do not need a matrix solution here since the X and Y force equations are decoupled. The solution is:
3n3
Matrix inversion requires approximately and Gaussian elimination requires approximately
log n
(n2 1)n 2
n multiplications/divisions2.
3
There is a substantial 65% reduction in computational cost for Gaussian elimination with the
6x6 plus decoupled link 2 method. Also, the numerical accuracy may also improve with this method
since we needn’t do unnecessary calculations with the three 6x1 columns of zeros.
2
E.D. Nering, 1974, Elementary Linear Algebra, W.B. Saunders Company, Philadelphia: 38-39.
111
1 0 1 0 0 0 0 0 F21 X m2 AG 2 X
0 1 0 1 0 0
0 0 F21Y m2 ( AG 2Y g )
r12Y r12 X r32Y r32 X 0 0 0 1 F32 X IG 2 Z 2
0 0 1 0 1 0 0 0 F32Y m3 AG 3 X FE 3 X
0
0 0 1 0 1 0 0 F43 X m3 ( AG 3Y g ) FE 3Y
0 0 r23Y r23 X r43Y r43 X 0 0 F43Y I G 3 Z 3 rE 3 X FE 3Y rE 3Y FE 3 X M E 3
0
0 0 0 1 0 0 F14Y m4 AG 4 X FE 4 X
0 1
1 0 2 m4 g FE 4Y
0 0 0 0
A v b
Like the four-bar mechanism, it is possible to partially decouple the solution to this problem1. If
we consider the FBDs of only links 3 and 4 first, this is 5 equations in 5 unknowns – this is verified by
looking at the original 8x8 matrix and noting three 5x1 columns of zeros (1, 2, 8) in rows 4 through 8.
Here is a more efficient solution. The reduced 5x5 set of equations for links 3 and 4 are given below.
1 0 1 0 0 F32 X m3 AG 3 X FE 3 x
0 m3 ( AG 3Y g ) FE 3 y
1 0 1 0 F32Y
r23Y r23 X r43Y r43 X 0 F43 X I G 3 Z 3 rE 3 x FE 3 y rE 3 y FE 3 x M E 3
0 0 1 0 F43Y m4 AG 4 X FE 4 x
0 0 0 1 1 F14Y m4 g FE 4 y
1 0 0 F21 X m2 AG 2 X F32 X
0
1 0 F21Y m2 ( AG 2Y g ) F32Y
r12Y r12 X 1 2 I G 2 Z 2 r32Y F32 X r32 X F32Y
112
We do not need a matrix solution here since the X and Y force equations are decoupled. The solution is
identical to that of the four-bar mechanism.
3n3
Matrix inversion requires approximately and Gaussian elimination requires approximately
log n
(n2 1)n 2
n multiplications/divisions (Nering, 1974).
3
There is a substantial 69% reduction in computational cost for Gaussian elimination with the
5x5 plus decoupled link 2 method. Also, the numerical accuracy may also improve with this method
since we needn’t do unnecessary calculations with the three 5x1 columns of zeros.
113
Step 1. The inverted slider-crank Position, Velocity, and Acceleration Analyses must be complete.
Given r1, 1 = 0, r2
2, r4, 4
2, r4 , 4
2, r4 , 4
assume zero external forces and moments
Link 2
F 2 F 42 F 21 W 2 m 2 A G 2
Link 4
F 4 F 14 F 42 W 4 m 4 A G 4
Link 2
M G2z 2 r 42 F 42 r 12 F 21 I G 2 Z 2
Link 4
M G4z r 14 F 14 r 24 F 42 I G 4 Z 4
Count the number of unknowns and the number of equations: 6 scalar equations and 7 scalar unknowns.
We need an additional equation; let us assume zero friction between links 2 and 4. Therefore, F42 is
always perpendicular to link 4 and there is only one unknown from this vector, the magnitude F42.
F F cos( 4 2)
F 42 42 X 42
F42Y F42 sin( 4 2)
115
Step 5. Derive the XYZ scalar dynamics equations from the vector dynamics equations.
Link 2
F42 X F21 X m2 AG 2 X
F42Y F21Y m2 ( AG 2Y g )
Link 4
F14 X F42 X m4 AG 4 X
F14Y F42Y m4 AG 4Y g
Express these scalar equations in matrix/vector form. The simplified inverted slider-crank mechanism
inverse dynamics matrix equation is given below.
1 0 c 0 0 0 F21 X m2 AG 2 X
0
1 s 0 0 0 F21Y m2 ( AG 2Y g )
r12Y r12 X r42 X s r42Y c 0 0 1 F42 I G 2 Z 2
0 0 c 1 0 0 F14 X m4 AG 4 X
0 0 s 0 1 0 F14Y m4 ( AG 4Y g )
0 0 r24 X s r24Y c r14Y r14 X 0 2 I G 4 Z 4
c cos( 4 2)
where:
s sin( 4 2)
A v b
116
Step 6. Solve for the unknowns
The coefficient matrix [A] is dependent on the mechanism geometry (i.e. the angles from the
position kinematics solution). The right-hand-side vector {b} is dependent on inertial terms and gravity.
v A b
1
Matrix/vector solution
The solution to the internal forces and input torque are contained in the components of v. To save these
values for later plotting, use the following MATLAB code, inside the i loop.
f21x(i) = v(1);
f21y(i) = v(2);
tau2(i) = v(6);
Like the four-bar and slider-crank mechanisms, it is possible to partially decouple the solution to
this problem. If we consider the FBDs of only 4 first, this is 3 equations in 3 unknowns – this is verified
by looking at the original 6x6 matrix and noting three 3x1 columns of zeros (1, 2, 6) in rows 4 through
6. Here is a more efficient solution. The reduced 3x3 set of equations for link 4 is given below.
c 1 0 F42 m4 AG 4 X
s 0 1 F14 X m4 ( AG 4Y g )
r24 X s r24Y c r14Y r14 X F14Y I G 4 Z 4
A4 v4 b4
Solve for three unknowns v4 A4 b4 and then use F42 in the following 3x3 set of linear equations,
1
1 0 0 F21 X m2 AG 2 X cF42
0
1 0 F21Y m2 ( AG 2Y g ) sF42
r12Y r12 X 1 2 I G 2 Z 2 (r42 X s r42Y c) F42
We do not need a matrix solution here since the X and Y force equations are decoupled. The solution is
identical to that of the four-bar mechanism.
F21 X m2 AG 2 X cF42
F21Y m2 ( AG 2Y g ) sF42
2 I G 2 Z 2 r12Y F21 X r12 X F21Y (r42 X s r42Y c) F42
117
3n3
Matrix inversion requires approximately and Gaussian elimination requires approximately
log n
(n2 1)n 2
n multiplications/divisions (Nering, 1974).
3
Number of Multiplications/Divisions
Method Inversion Gaussian Reduction
6x6 833 106 87%
3x3 plus decoupled link 2 177 24 86%
Reduction in cost 79% 77%
There is a substantial 77% reduction in computational cost for Gaussian elimination with the 3x3 plus
decoupled link 2 method. Also, the numerical accuracy may also improve with this method since we
needn’t do unnecessary calculations with the three 3x1 columns of zeros.
After the basic inverse dynamics problem is solved, we can calculate the vector shaking force
and vector shaking moment, which is the force/moment reaction on the ground link due to the motion,
inertia, weight, and external loads (which we assumed to be zero in this problem). The shaking force
and moment for the inverted slider-crank mechanism is identical to the four-bar in notation and terms.
Shaking force
F F14 X
FS F21 F41 F21 F14 21 X
F21Y F14Y
Shaking moment
M S 2 r21 F21 r41 F14
Given
r1 0.20
2 70
r2 0.10
(m) 4 150.5 (deg and m)
L4 0.32
r4 0.19
1 0
2 25 2 0
r4 2.47 (rad/s and m/s) r4 9.46 (rad/s2 and m/s2)
4 2.18 4 267.14
In this problem the external forces and moments are zero for both links 2 and 4. In inverse
dynamics we ignore the slider link mass and inertia. The mechanism links 2 and 4 are uniform,
homogeneous rectangular solids made of steel ( = 7850 kg/m3) with a constant thickness of 2 cm and
link widths of 3 cm. The CGs are in the geometric center of each link. This yields the following fixed
dynamics parameters.
m2 = 0.47 and m4 = 1.51 (kg) IGZ2 = 0.0004 and IGZ4 = 0.013 (kg-m2)
Snapshot Analysis
Given the previous mechanism position, velocity, and acceleration analyses, solve the inverse
dynamics problem for this snapshot ( 2 70 ). The matrix-vector equation to solve is given below.
F 35.81
The associated vector shaking force and moment are FS SX (N) M S 8.53kˆ (Nm)
SY
F 51.03
119
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 3 continued
A more meaningful result is to solve and plot the inverse dynamics analysis results for the entire
range of mechanism motion. The plots below give the inverse dynamics results for all 0 2 360 ,
for Term Example 3. Since 2 is constant, we can plot the velocity results vs. 2 (since it is related to
time t via 2 2t ).
30
2
20 AVG
RMS
10
2 (Nm)
-10
-20
Input Torque 2
120
150
FSX
100
FSY
50
0
FS (N)
-50
-100
-150
-200
Shaking Force
20
15
10
5
MS (Nm)
-5
-10
-15
-20
Shaking Moment
121
For example, here are the Watt II six-bar mechanism FBDs, ignoring external forces and moments.
123
The Watt II six-bar mechanism inverse dynamics 15x15 matrix-vector equation is given below.
1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 F21 X m2 AG 2 X
0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 F21Y m2 ( AG 2Y g )
r12Y r12 X r32Y r32 X 0 0 0 0 0 0 0 0 0 0 1 F32 X I G 2 Z 2
0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 F32Y m3 AG 3 X
0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 F43 X m3 ( AG 3Y g )
0 0 r23Y r23 X r43Y r43 X 0 0 0 0 0 0 0 0 0 F43Y I G 3 Z 3
0 0 0 0 1 0 1 0 1 0 0 0 0 0 0 F14 X m4 AG 4 X
0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 F14Y m4 ( AG 4Y g )
0 0 0 0 r34Y r34 X r14Y r14 X r54Y r54 X 0 0 0 0 0 F54 X I G 4 Z 4
0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 F54Y m5 AG 5 X
0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 F65 X m5 ( AG 5Y g )
0 0 0 0 0 0 0 0 r45Y r45 X r65Y r65 X 0 0 0 F65Y I G 5 Z 5
0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 F16 X m6 AG 6 X
0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 F16Y m6 ( AG 6Y g )
0 0 0 0 0 0 0 0 0 0 r56Y r56 X r16Y r16 X 0 2 I G 6 Z 6
A v b
124
Step 6. Solve for the unknowns (continued)
Like the four-bar mechanism, it is possible to partially decouple the solution to this problem. If
we consider the FBDs of only links 5 and 6 first, this is 6 equations in 6 unknowns – this is verified by
looking at the original 15x15 matrix and noting nine 6x1 columns of zeros (1, 2, 3, 4, 5, 6, 7, 8, 15) in
the six rows 10 through 15. Here is a more efficient solution. The reduced 6x6 set of equations for links
5 and 6 is given below.
1 0 1 0 0 0 F54 X m5 AG 5 X
0 1 0 1 0 0 F54Y m5 ( AG 5Y g )
r45Y r45 X r65Y r65 X 0 0 F65 X I G 5 Z 5
0 0 1 0 1 0 F65Y m6 AG 6 X
0 0 0 1 0 1 F16 X m6 ( AG 6Y g )
0 0 r56Y r56 X r16Y r16 X F16Y I G 6 Z 6
We do not need a matrix solution here since the X and Y force equations are decoupled. The solution is
identical to that of the four-bar mechanism.
F21 X m2 AG 2 X F32 X
F21Y m2 ( AG 2Y g ) F32Y
2 I G 2 Z 2 r32Y F32 X r32 X F32Y r12Y F21X r12 X F21Y
125
3n3
Matrix inversion requires approximately and Gaussian elimination requires approximately
log n
(n2 1)n 2
n multiplications/divisions (Nering, 1974).
3
Number of Multiplications/Divisions
There is an astonishing 86% reduction in computational cost for Gaussian elimination with the
methods using 6x6 inversion twice, plus the decoupled link 2 solution. Also, the numerical accuracy
may also improve with this method since we needn’t do unnecessary calculations with the sixteen 6x1
columns of zeros.
Any mechanism with a dyad of binary links may be decoupled in this manner. Thus, the method
is similar and the computational complexity identical for the Stephenson I, Stephenson III, Watt I, and
Watt II six-bar mechanisms.
The Stephenson II six-bar mechanism does not include a dyad of binary links and so it cannot
be solved like the other 4 six-bar mechanisms (first links 5 and 6, then links 3 and 4 with one unknown
vector force from 5 and 6, then link 2 independently). But links 3, 4, 5, and 6 can be solved first
independently of link 2: a 12x12 solution followed by the standard link 2 solution. The computational
savings is not as impressive as in the former six-bar cases.
There is a 46% reduction in computational cost for Gaussian elimination with the 12x12 plus
decoupled link 2 method. Also, the numerical accuracy may also improve with this method since we
needn’t do unnecessary calculations with the three 12x1 columns of zeros.
126
1) Static Balance Moments about the rotating shaft must be balanced statically.
Inertial forces are not actual forces but are effects of acceleration, e.g. centripetal force. Assuming
constant input angular velocity , the inertial force is directed outward, opposite to the centripetal
acceleration directed inward.
For dynamic balance, we again add a balance mass. The dynamic balance condition is:
F I F IB 0
The original and balance inertial forces must be equal in magnitude and opposite in direction. This
vector balance condition is equivalent to the two scalar equations below.
Now let us include a system of idealized point masses attached to the same rotating shaft.
1) Static Balance
The rotating shaft will be balanced statically if the system CG lies on the axis of rotation.
X 0
P CG
Y 0
X CG X
m x i i
YCG Y
m y i i
m i m i
Let us consider a system of four point masses. There are two equations to satisfy:
4 4
mi ri cos i 0
i 1
m r sin
i 1
i i i 0
4
where xi ri cos i , yi ri sin i , and the m
i 1
i term in the denominator cancels out.
If we fix each miri, these are 2 equations in the four unknowns i. Therefore, arbitrarily fix 1,2
and solve for 3,4. This is equivalent to solving the four-bar linkage position problem.
1 1 m2 r2 c2 m3r3c3 m4 r4 c4 0
m1rc r2c2 r3c3 rc
1 1 r4c4
2) Dynamic Balance
For dynamic balance, we must consider the front view in addition to the previously-shown side
view.
If the following conditions are satisfied assuming constant , we will have dynamic balance.
FX 0 MX 0
F Y 0 M Y 0
F Z 0 M Z 0
4 4
F m r
Y
i 1
i i
2
sin i mi ri sin i 0
i 1
F Z 00
The F X
and F Y
equations are already satisfied by the static balancing, because 2 divides out.
The F Z
equation yields nothing because all inertial forces are in the XY plane.
129
M X mi ri 2 sin i Li 0
i 1
4
M Y mi ri 2 cos i Li 0
i 1
M Z 00
The M X
and M Y
equations must be solved to satisfy dynamic balancing. The M Z
equation
yields nothing because all inertial forces pass through the axis of rotation.
We fixed each miri, we previously determined i, and the 2 term divides out. Therefore, we
have two equations in the four unknowns Li; arbitrarily fix L1, L2 and solve for L3, L4. The result is two
linear equations in the two unknowns.
Solve for L3, L4 and the system will be balanced statically and dynamically.
130
At 150 , given this link, motion, and external force, calculate F12 X , F12Y , and F S , M S .
8510
F S F 21 F 12 N M S 66.5kˆ Nm
4980
A more meaningful result from inverse dynamics analysis is to solve and plot the dynamics
unknowns for the entire range of mechanism motion. For the same example as the snapshot we specify
that the given is constant. Prior to solving the inverse dynamics problem, the CG translational
acceleration results for all 0 360 are given in the plot below. The X components are red and the
Y green. Is the static link weight (mg) significant in this problem?
131
-5000
0 50 100 150 200 250 300 350
(deg)
0.5
FSX and FSY (N)
-0.5
-1
0 50 100 150 200 250 300 350
(deg)
The plot above gives the Shaking Force F S for all 0 360 . The X component is red and the Y
green. The Shaking Moment M S is simply the negative of the driving torque plot shown next.
132
The plot below gives the required driving torque (Nm, red) for all 0 360 , assuming
the given is constant, for the same example as the snapshot. This shows the torque that must be
supplied by an external DC servomotor to cause the specified motion. Also plotted is the average torque
(green) AVG = 0 and the root-mean-square (RMS) torque value (blue) RMS = 106.1 Nm.
100
50
(Nm)
-50
-100
-150
0 50 100 150 200 250 300 350
(deg)
Single Rotating Link Torque Results
Here are the calculations for the average and root-mean-square torques.
0 1 2 k 02 12 22 k2
AVG RMS
k 1 k 1
where k+1 is the total number of elements in the array (since the counting index k starts at zero).
MATLAB can be used to calculate and plot the average and root-mean-square torques on the plot of
for easy comparison.
The coupler link 3 is a rectangle of dimensions 0.203 x 0.152 (m). The triangle tip we have been
using all along in Term Example 1 (previously called point C) is the CG of the rectangular link shown
below for inverse dynamics.
0.3
0.25
0.2
0.15
Y (m)
0.1
0.05
-0.05
-0.1
-0.1 0 0.1 0.2 0.3
X (m)
134
Four-bar mechanism inverse dynamics example – Term Example 1 continued
This is the mechanism from Term Example 1 (open branch), with kinematics solutions as
presented before. Given r1 = 0.284, r2 = 0.076, r3 = 0.203, r4 = 0.178, 1 10.3 , 2 30 , 3 53.8 ,
4 121.7 , RG2 = 0.038, RG3 = 0.127, RG4 = 0.089 (m), 2 0 , 3 36.9 , 4 0 , 2 20 , 3 8.09 ,
4 3.73 (rad/s), 2 0 , 3 8.65 , and 4 244.4 (rad/s2).
All moving links are wood, with mass density 830.4 kg/m3. Links 2 and 4 have rectangular
dimensions ri x 0.019 x 0.013 thick (m; i=2,4); link 3 has rectangular dimensions 0.203 x 0.152 x 0.013
thick (m), as shown on the previous page. The calculated mass and inertia parameters are m2 0.015 ,
m3 0.327 , m4 0.036 (kg) and I G 2 Z 7.9 106 , I G 3 Z 1.8 103 , I G 4 Z 9.5 105 (kgm2). All
external forces and moments are zero but gravity, g = 9.81 m/s2, is included.
1 0 1 0 0 0 0 0 0 F21x 0.202
0
1 0 1 0 0 0 0 0 F21 y 0.034
0.019 0.033 0.019 0.033 0 0 0 0 1 F32 x 0
0 0 1 0 1 0 0 0 0 F32 y 8.955
0
0 0 1 0 1 0 0 0 F43 x 4.497
0 0 0.127 0.002 0.037 0.122 0 0 0 F43 y 0.015
0 0 0 0 1 0 1 0 0 F14 x 0.638
0 0 0 0 0 1 0 1 0 F14 y 0.095
0 0 2 0.023
0 0 0 0.076 0.047 0.076 0.047
The solution is accomplished by Gaussian elimination, or v A b , or by the reduced 6x6
1
plus decoupled link 2 method (see the on-line ME 3011 NotesBook Supplement). All methods yield the
same results. Snapshot answers:
F21x 6.20
F 10.08
21 y
F32 x 5.99
F32 y 10.11 9.80
v F43 x 2.96 (N, Nm) FS (N) M S 1.68kˆ (Nm)
F 5.61 4.56
43 y
F14 x 3.60
F
14 y 5.52
2 0.43
135
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 1 continued
A more meaningful result from inverse dynamics analysis is to solve and plot the dynamics
unknowns for the entire range of four-bar mechanism motion. Prior to solving the inverse dynamics
problem, the plot below shows the CG translational acceleration results for link 3 for all 0 2 360 .
The X component is red and the Y component is green.
30
20
10
AG3 (m/s )
2
-10
AG3X
AG3Y
-20
-30
Term Example 1, AG 3
136
0.8
0.6
0.4
0.2
2 (Nm)
-0.2
-0.4
The plot above gives the required driving torque 2 (Nm) for all 0 2 360 , for the Term
Example 1 mechanism, assuming the given 2 20 rad/s is constant. This plot shows the torque (red)
that must be supplied in all configurations by an external DC servomotor to cause the specified motion.
Also plotted is the average torque (green) 2 AVG 0.003 and the root-mean-square torque value (blue)
2 RMS 0.354 Nm. The root-mean-square (RMS) torque is more meaningful than the average torque
since its terms do not cancel each other (k+1 is the number of elements in the 2 array).
22 22 22 22
2 RMS 0 1 2 k
k 1
137
The plot below gives the shaking force F S (N) results for all 0 2 360 . The X component
is red and the Y component is green.
10
FSX
FSY
0
FS (N)
-5
-10
The plot below gives the shaking moment M S (Nm) results for all 0 2 360 . There is only
the Z component since a planar moment is a k̂ vector.
1.5
0.5
MS (Nm)
-0.5
-1
-1.5
-2
The Term Example 2 slider-crank mechanism is shown below at the starting (or ending) position,
with zero (or 360 ) input angle 2.
0.2
0.15
0.1
0.05
Y (m)
0
-0.05
-0.1
-0.15
-0.2
-0.25
-0.1 0 0.1 0.2 0.3
X (m)
140
Slider-crank mechanism inverse dynamics example – Term Example 2 continued
This is the mechanism from Term Example 2 (right branch), with kinematics solutions as
presented before. Given r2 = 0.102, r3 = 0.203, h = 0.076 m, 2 30 , 3 7.1 , x = 0.29 m, 2 15
rad/s (constant), 3 6.58 rad/s, x 0.60 m/s, 2 0 , 3 62.33 rad/s2, and
x 30.15 m/s .
2
All moving links are wood, with mass density 830.4 kg/m3. Links 2 and 3 have rectangular
dimensions ri x 0.019 x 0.013 thick (m; i=2,3); link 4 has rectangular dimensions 0.076 x 0.019 x 0.013
thick (m), as shown on the previous page. The calculated inertia parameters are m2 = 0.020, m3 = 0.041,
m4 = 0.015 (kg) and IG2Z = 1.819e-005, IG3Z = 1.418e-004 (kgm2). The CGs all lie at their respective link
centers. There is a constant external force of 1 N acting at the center of the piston end, directed
horizontally to the left; gravity is included but all other external forces and moments are zero. We
assume the coefficient of kinetic friction between the piston and the fixed wall is = 0.2.
At 2 30 , given this mechanism and motion, calculate the four vector internal joint forces, the
driving torque 2 , and the shaking force and moment F S , M S for this instant (snapshot).
1 0 1 0 0 0 0 0 F21 X 0.202
0
1 0 1 0 0 0 0 F21Y 0.084
-0.025 0.044 -0.025 0.044 0 0 0 1 F32 X 0
0 0 1 0 1 0 0 0 F32Y 1.018
0
0 0 1 0 1 0 0 F43 X 0.168
0 0 -0.013 0.101 -0.013 0.101 0 0 F43Y 0.009
0
0 0 0 1 0 0.2 0 F14Y 0.540
0 0 0 0 0 1 1 0 2 0.150
The solution is accomplished by Gaussian elimination, or v A b , or by the reduced 5x5 plus
1
decoupled link 2 method (see the on-line ME 3011 NotesBook Supplement). All methods yield the
same results. Snapshot answers:
F21 X 0.736
F 0.121
21Y
F32 X 0.534
F32Y 0.037 0.680
v (N, Nm) FS (N) M S 0.116kˆ (Nm)
F43 X 0.484 0.401
F43Y 0.131
F14Y 0.281
0.039
2
141
Full-Range-Of-Motion (F.R.O.M.) Analysis – Term Example 2 continued
A more meaningful result from inverse dynamics analysis is to solve and plot the dynamics
unknowns for the entire range of mechanism motion. Prior to solving the inverse dynamics problem, the
plot below gives the CG translational acceleration results for link 3 for all 0 2 360 . Here CG3 is
taken as the midpoint of link 3. The X component is red and the Y component is green.
25
20
15
10
5
AG3 (m/s )
0
2
-5
AG3X
-10
AG3Y
-15
-20
-25
-30
0 50 100 150 200 250 300 350
2 (deg)
0.2
0.15
0.1
2 (Nm)
0.05
-0.05
-0.1
The plot above gives the required driving torque 2 (Nm) for all 0 2 360 , for the Term
Example 2 slider-crank mechanism, right branch only, assuming the given 2 15 rad/s is constant.
This plot shows the torque (red) that must be supplied in all configurations by an external DC
servomotor to cause the specified motion. Also plotted is the average torque (green) 2 AVG 0.004
and the root-mean-square torque value (blue) 2 RMS 0.099 Nm. The root-mean-square (RMS) torque is
more meaningful than the average torque since its terms do not cancel each other (k+1 is the number of
elements in the 2 array).
22 22 22 22
2 RMS 0 1 2 k
k 1
143
The plot below gives the shaking force F S (N) results for all 0 2 360 . The X component
is red and the Y component is green.
1.5
FSX
0.5
FSY
-0.5
FS (N)
-1
-1.5
-2
-2.5
-3
0 50 100 150 200 250 300 350
2 (deg)
The plot below gives the shaking moment M S (Nm) result for all 0 2 360 . There is only
the Z component since a planar moment is a k̂ vector.
0.15
0.1
0.05
MS (Nm)
-0.05
-0.1
Gears are used to transfer motion between rotating shafts in machinery, mechanisms, robots,
vehicles, toys, and other electromechanical systems. Gears cause changes in angular velocity, torque,
and direction. Gears are used in various applications, from can openers to aircraft carriers. Belt and
chain drives are related to gear mechanisms.
Gear Classification
“The harmonic gear allows high reduction ratios with concentric shafts and with very low
backlash and vibration. It is based on a very simple construction utilizing metal’s elasto-mechanical
property.”
“Harmonic drive transmissions are noted for their ability to reduce backlash in a motion control
system. How they work is through the use of a thin-walled flexible cup with external splines on its lip,
placed inside a circular thick-walled rigid ring machined with internal splines. The external flexible
spline has two fewer teeth than the internal circular spline. An elliptical cam enclosed in an antifriction
ball bearing assembly is mounted inside the flexible cup and forces the flexible cup splines to push
deeply into the rigid ring at two opposite points while rotating. The two contact points rotate at a speed
governed by the difference in the number of teeth on the two splines. This method basically preloads
the teeth, which reduces backlash.”
roymech.co.uk
roymech.co.uk
150
The wave generator is attached to the input shaft, the flexible spline is attached to the output
shaft, and the circular spline is fixed.
roymech.co.uk
For harmonic gearing, the gear ratio is also calculated from the numbers of teeth in each gear.
IN WG N FS
n
OUT FS N FS N CS
where WG stands for wave generator, FS stands for flexible spline, and CS stands for circular spline.
For example, if NFS = 200 and NCS = 202, the gear ratio is
IN N FS 200
n 100
OUT N FS N CS 200 202
which means that the output shaft rotates 100 times slower than the input shaft, but the output shaft
carries 100 times more torque than the input shaft. Therefore, this example would be good for the robot
joint case, i.e. reducing speed and increasing torque, with n >> 1. The negative sign indicates the
angular velocity and torque of the output shaft are in the opposite direction of the angular velocity and
torque of the input shaft.
151
4.1.2 Gear Ratio
2 1
If n 1
2 1
The output has reduced speed and increased torque.
2 1
If n 1
2 1
The output has increased speed and reduced torque.
This is the bicycle transmission case, except for some granny gears where n can be as
high as 1.5.
2 1
If n 1
2 1
This case is called an idler, where the output speed and torque are unchanged, but the
direction reverses (for external spur gears)
152
Gear ratio examples – Bicycle Transmissions
N OUT N R F R
Gear Ratio: n
N IN N F R F
The Cannondale mountain bike has a traditional front/rear derailleur transmission with three
gears in the front and nine gears in the rear, for a total combination of 27 gears. The standard BikeE
recumbent bike has a traditional derailleur transmission with seven gears only in the rear. Instead of a
traditional derailleur transmission in the front with three gears, the BikeE has an internal hub planetary
gear arrangement with three selectable ratios of 1.2913 : 1 (high), 1 : 1 (medium), and 0.7 : 1 (low). The
BikeE then has a single chain ring in front to drive the rear chain rings.
The standard BikeE original front single chain ring had 46 teeth – I changed this to a smaller
front chain ring of 34 teeth for more granny gear in order to climb Mulligan Hill; the cost of this is loss
of high gear for the bike path. I designed the lowest gear to mimic the lowest Cannondale gear ratio
since I knew that granny gear climbed well. The Cannondale mountain bike has a standard wheel size
of 26” diameter and the BikeE has a rear wheel size of 20” diameter. We must consider this difference
in wheel sizes to compute the overall effective BikeE gear ratios. The table below reflects this
calculation.
153
BikeE standard recumbent bike
front teeth
34 34 34
Ni 1.2913 : 1 1:1 0.7 : 1
11 0.33 0.42 0.60
13 0.38 0.50 0.71
15 0.44 0.57 0.82
rear teeth 18 0.53 0.69 0.98
21 0.62 0.80 1.15
24 0.71 0.92 1.31
28 0.83 1.07 1.53
I was able to obtain a used deluxe BikeE recumbent bike from noted luthier Dan Erlewine. I
decided to keep the front chain ring of 46 teeth (numbers of teeth in the rear chain ring and the planetary
gear ratios in the rear hub are identical between the standard and deluxe BikeE models). This means my
new deluxe BikeE doesn’t climb as well as my modified standard BikeE, but it flies much faster on the
bike path in high gear than the standard BikeE! Again, the difference in wheel diameter is taken into
account in the table below.
We see that the standard BikeE that was designed to equal the granny gear of the Cannondale (it
was exceeded, 1.53 vs. 1.45). However, the mountain bike still climbs better in granny gear, since your
legs are positioned above the pedals in the mountain bike case, and your legs are positioned straight out
in front of you in the recumbent bike case.
Unlike the robot joint example, bicycle gearing generally has n < 1 and so the transmission
increases angular velocity
decreases torque
by the gear ratio n. The exception is the granny gears with n > 1.
154
4.1.3 Gear Trains
To obtain a higher gear ratio than practical with a single pair of standard involute spur gears, one
can mate any number of spur gears in a gearbox, or gear train. The leftmost gear is the driving gear and
the rightmost is the output gear. All intermediate gears are first the driven gear and then the driving gear
as we proceed from left to right. Let us calculate the overall gear ratio nGT.
IN
nGT
OUT
Example
We can find the overall gear ratio by canceling neighboring intermediate angular velocities.
Each term in the above product may be replaced by its known number of teeth ratio.
We could have done the same with pitch radii instead of number of teeth because they are in direct
proportion.
So, the intermediate gears are idlers. Their number of teeth effect cancels out, but they do change
direction. We should have included the +/– signs, by inspection. For gear trains composed of
externally-meshing spur gear pairs:
odd number of gears the output is in the same direction as the input
even number of gears the output is in the opposite direction as the input
155
Improved gear train
That gear train concept did not work. Now let us mate any number of spur gears, where the
driving and driven gears are distinct, because each pair is rigidly attached to the same shaft. Again, let
us calculate the overall gear ratio.
IN
nGT
OUT
Example
But now the gears rigidly attached to the same shaft have the same angular velocity ratio, so
Rolling Cylinders
Mating spur gears are based on two pitch circles rolling without slip. These are fictitious circles,
i.e. you cannot look on a gear to see them. The actual gear teeth both roll and slide with respect to each
other (via the two-dof gear joint).
The angular velocity between the gears of a gearset must remain constant throughout the mesh.
From our study of linkage velocity, we know this is no easy feat. Velocity ratios in a linkage
vary wildly over the range of motion.
OUT r 1
Velocity Ratio VR IN
IN rOUT n
IN r
Torque Ratio TR OUT n
OUT rIN
The velocity ratio is the inverse of the gear ratio n and the torque ratio is the same as the gear
ratio n defined previously. The torque ratio is also called Mechanical Advantage (MA).
Involute Function
Standard spur gears have an involute tooth shape. If the gears’ center distance is not perfect
(tolerances, thermal expansion, wear – in design the center distance is increased slightly by the engineer
to allow for these effects; this is called clearance), the angular velocity ratio will still be constant to
satisfy the Fundamental Law of Gearing.
157
The involute of a circle is a curve generated by unwrapping a taut sting from the circumference of a so-
called base circle, always keeping it tangent to the circle. The figure below has th = [0:5:60]*DR
and rb = 1 m. The red circle is the gear base circle, the blue lines are the taut tangent construction lines,
and the green curve is the involute function.
0.5
Y (m)
-0.5
-1
In polar coordinates (, r), the parametric equations for the involute of a circle are given below.
tan t t
r rb sec t
where t is the independent parameter and rb is the base circle radius. In Cartesian coordinates (xI, yI):
xI r cos rb sec t cos(tan t t )
yI r sin rb sec t sin(tan t t )
158
Involute Function Example
The plot below shows the involute function for a circle of rb = 1 m. The parameter range is t =
[0:1:80]*DR; near the circle the involute points are very close to each other and farther away the
step size increases dramatically.
The involute function is symmetric (try t = [0:5:360]*DR) but watch out for those
intermediate steps). Gear teeth only require the involute near the base circle, with two symmetric sides.
1
Y (m)
-1
-2
-3
-4
-6 -5 -4 -3 -2 -1 0 1 2
X (m)
Norton (2008)
Pitch Circle A fictitious circle (you cannot see it on a spur gear) with theoretical pure
rolling in contact between two cylinders of the pinion and gear.
Pitch Point The contact point between the two pitch circles.
Pressure Angle The angle between the common normal (also called axis of transmission)
of the two meshing teeth and the velocity of the pitch point (the tangent to
both pitch circles). The point of contact slides along this line. A similar
angle is defined for cams and followers.
The relationship among the base circle radius rb, pitch circle radius rp, pressure angle is
rb rp cos
3
R. Norton, 2008, Design of Machinery: An Introduction to the Synthesis and Analysis of Mechanisms and Machines, McGraw-Hill.
160
Norton (2008)
The length of contact is measured along the axis of transmission. The beginning of contact is
when the tip of the driven gear tooth intersects the axis of transmission. The end of contact is when the
tip of driving gear tooth intersects the axis of transmission, as shown in the figure above. Only one or
two teeth are in contact at any one time for standard spur gears.
For harmonic gearing, many teeth are in contact at any one time, which provides a higher gear
ratio in a smaller package.
161
Norton (2008)
Increasing the center distance increases the pressure angle and increases the pitch circle
radii, but doesn’t change the base circles (obviously – the gears are made based on their own constant
rb). Thanks to the involute tooth shape, increasing the center distance does not affect the angular
velocity ratio. This is why the involute function is so widely used in spur gears.
The relationship from before still applies with an increase in center distance.
rb rp cos
Again, rb is fixed, and rp and both increase – the cosine function maintains the constant rb.
162
Spur Gear Standardization
N
diametral pitch pd
d
Where N is the number of teeth and d is the pitch diameter, both for each gear.
d 1
module m
N pd
Module is the SI version of diametral pitch (it is the inverse). SI gears are not
interchangeable with English system gears because of different tooth proportion
standards.
d
circular pitch pc
N
Circular pitch is the circumferential distance (arc length) between teeth along the pitch
circle of a spur gear.
163
pc 2t b
164
Backlash
Backlash B is the distance between mating teeth measured along the pitch circle circumference.
Backlash can be thought of as circumferential clearance. All real-world gears must have some backlash
in order to still function despite real-world problems of manufacturing tolerances, thermal expansion,
wear, etc. However, one must minimize backlash for smooth operation. For example, robot joints must
be driven both directions. Upon changing direction, nothing happens until the backlash is passed, and
then an impact occurs, which is bad for gear dynamics. This is a non-linear effect in robotics. On earth
gravity tends to load the backlash for predictable effects. In space however, the backlash is less
predictable.
165
4.1.5 Planetary Gear Trains
Planetary gear trains are also called epicyclic gear trains. The sun gear rotates about a fixed axis.
Each planet gear rotates about its own axis and also orbits the sun gear. This happens with direct
meshing of teeth, unlike celestial planetary motion. The arm link (which is a rigid body with no teeth)
carries the planet(s) around the sun. The arm has a revolute joint to the sun gear on one side and another
revolute joint to the planet gear(s) on the other side. Planetary gear trains have two-dof, so two inputs
must be given to control the mechanism. For instance, one can drive the sun gear and the arm link with
independent external motors. Alternatively, the sun gear may be fixed and the arm link driven. Using
planetary gear trains, one can obtain a higher gear ratio in a smaller package, compared to non-planetary
gear trains.
We present the tabular method below to determine the gear ratio for various planetary gear trains.
166
Planetary Gear Train Applications
Planetary gear train applications include airplane propellers, some automotive transmissions and
differentials, machine tools, hoists, and a hub-enclosed multi-speed bicycle transmission.
appauto.files.wordpress.com
167
This is a commercial planetary gear set with 8 possible ratios (4:1, 5:1, 16:1, 20:1, 25:1, 80:1,
100:1, 400:1).
servocity.com
168
Table Method Analysis
First, consider a simplified planetary gear system with S 0 . With the sun gear as the fixed
link, we have a 1-dof system. Given the input A, calculate the absolute value of P. The Table
Method is based on the following relative velocity equation.
G A G / A
This is a vector equation, but since all rotations are about the Z axis, we just use the magnitudes and +
for CCW and – for CW, according to the right-hand-rule. This equation is written for each gear in the
system (replace G with the appropriate index). A stands for arm. G , A are the absolute angular
velocities of a gear and the arm link. G / A is the relative velocity of a gear with respect to the moving
arm link. Construct a table as below; each row is the relative equation written for a different gear.
G G = A + G/A
S
P
G G = A + G/A
S 0 A
P A
G G = A + G/A
S 0 A –A
P A
Now we can fill in down the relative column, using a simple gear ratio (relative to the arm).
P / A N N N
S P / A S S / A S A
S / A NP NP NP
G G = A + G/A
S 0 A –A
P A NS
A
NP
169
The last row must add up according to the relative equation, to finish the table.
G G = A + G/A
S 0 A –A
P NS A NS
1 A A
NP NP
N
Therefore, the absolute angular velocity of the planetary gear is P 1 S A . Since the sign is
NP
positive, it has the same direction as A (CCW).
IN 1 NP
n A
OUT P 1 N S NP NS
NP
Example
10
P 1 100 125 rpm, CCW
40
NP 40
n 0.8
N P N S 40 10
A A 100
Check n P 125
P n 0.8
So we see that with S 0 , the gear ratio is not higher than the conventional gear train.
N P 40
n .
N S 10
3) Fill down the right column using a simple gear train, relative to arm A.
P / A N N N N
S P / A S S / A S (S A ) S ( A S )
S / A NP NP NP NP
G G = A + G/A
S * S * A S A
P NS * A NS
NS (A S )
1 A S NP
NP NP
Examples S 0
40 40
P 1 100 (100) 500 400 900 rpm, CCW That’s a big ratio.
10 10
40 40
P 1 100 (125) 500 500 0
10
10
In Example 3, the Sun and Arm rotational velocities cancel so the absolute angular velocity of the Planet
is zero.
171
Table Method Analysis
Now let us consider an even more general system with S 0 , also adding an internal-teeth ring
gear.
3) Fill down the right column using a simple gear train, relative to arm A.
R / A N N N N N
P R / A P P / A P S ( A S ) S ( A S )
P / A NR NR NR NP NR
G G = A + G/A
S * S * A S A
P NS * A NS
NS (A S )
1 A S NP
NP NP
R NS * A NS
NS (A S )
1 A S NR
NR NR
172
10 10
R 1 100 (100) 110 10 120 rpm, CCW
100 100
NS NS
P 1 A S
NP NP
40 40
P 1 (200) (100) 600 200 400 rpm, CW
20 20
NS NS
R 1 A S
NR NR
40 40
R 1 (200) (100) 300 50 250 rpm, CW
80 80
173
4.2 Cams
Applications
Compared to linkages, it is easier to design desired motion with cams, but it is more expensive and
difficult to produce. Also, the cam contact and wear properties are worse than for linkages.
Cam Classification
a) disk cam with knife-edge b) disk cam with radial roller c) disk cam with offset radial
follower follower roller follower
d) disk cam with oscillating e) disk cam with radial flat-faced f) disk cam with oscillating flat-
roller follower follower faced follower
4
H.H. Mabie and C.F. Reinholtz, 1987, Mechanisms and Dynamics of Machinery, Wiley.
174
Disk cams with followers
Norton (2008)
Norton (2008)
175
Norton (2008)
Two cam/follower systems are shown above with equivalent four-bar and slider-crank
mechanism models.
Caution
These are instantaneous equivalents only, i.e. the virtual link lengths for both the four-bar and
slider-crank models change with cam mechanism configuration.
Function Generation
In function generation, the output parameter is a continuous function of the input parameter in a
mechanism. With linkages, we can only satisfy a function exactly at a finite number of points: 3, 4, or 5,
usually. For example, for a four-bar linkage with 4 f 2 , this function is only exact at a few points.
With a cam and follower mechanism, however, we can satisfy function generation at infinite
points. is the cam input angle and the output is S for a reciprocating (translating) follower and the
output is for an oscillating (rotating) follower.
S f ( ) f ( )
176
4.2.2 Cam Motion Profiles
Up to this point, we have been mostly concerned with mechanism analysis: given a mechanism
design and its input parameters, determine the position, velocity, acceleration, and dynamics behavior.
With cams we must consider mechanism synthesis for the first time: given the motion requirements
(follower motion and timing with the input cam angle), design the cam. The first step is to determine a
smooth cam follower motion profile. In general a cam follower has 4 motion zones (rise, dwell, fall,
dwell), as shown below.
When the motion transitions between different motion functions, we must ensure smooth motion.
The cam function must be continuous through the first and second derivatives of displacement
across the entire motion interval.
Which means:
Position, velocity, and acceleration must be continuous for the entire 360 of cam rotation. The
jerk function (the derivative of the acceleration) must be finite, but need not be continuous.
If the Fundamental law of Cam Design is satisfied, the resulting dynamic performance will be
acceptable for high-speed cam/follower operation. If not, there will be performance degradation due to
noise, vibrations, high wear, etc. There is a cyclical impulse hammering at each point in the cam cycle
when acceleration is not continuous (even worse if position or velocity is not continuous).
177
S V A J Diagrams
In cam synthesis (design), we are only given the total motion range and perhaps some timing
requirements. It is the engineer’s job to determine the position curves and to match the velocity and
acceleration across junctions. Position is automatically matched by shifting the dependent function
axes. Draw S V A J diagrams vs. time to graphically see if the Fundamental Law of Cam Design is
satisfied for candidate curves. We can plot vs. time or vs. input cam angle (assuming constant angular
velocity, t ).
The slope of a function is the value of its derivative at a point in time (or ). Therefore, for
continuous velocity and acceleration curves, the slopes of the position and velocity curves must match
across all junctions. The slope of the acceleration can be discontinuous (leading to finite jumps in jerk),
but the acceleration itself must be continuous.
Cam motion curves are very much like the input link motion curves discussed earlier, for input
links that start and stop at zero velocity and acceleration. In fact, I adapted the input motion curves from
cam motion curve design.
Define each separate function so the value is zero at the initial angle, which is zero. Then to put the
whole cam motion profile together, just shift the and S axes.
rise – dwell portion. Specify parabolic (constant acceleration) to straight line (constant velocity) rise,
followed by a dwell.
1
S: f1 (1 ) A012 f 2 ( 2 ) V0 2 f 3 ( 3 ) 0
2
V: v1 (1 ) A01 v2 ( 2 ) V0 v3 ( 3 ) 0
A: a1 (1 ) A0 a2 ( 2 ) 0 a3 ( 3 ) 0
J: j1 (1 ) 0 j2 ( 2 ) 0 j3 ( 3 ) 0
Match V at junction B
v1 (1 1 ) v2 ( 2 0) A0 1 V0 so V0 A0 1
A0 0 is impossible, or else the parabola is degenerate, which we cannot allow. This case violates the
Fundamental Law of Cam Design since the acceleration function cannot be made to be continuous at
junction B. Therefore, this cam motion profile example cannot be used for cam design.
We have a bigger problem at junction C, between functions 2 and 3: the velocity function cannot
be made to be continuous at junction C. Discontinuous velocity is one level worse than discontinuous
acceleration; either renders the resulting cam motion profile unacceptable.
Example 1 Plots
179
Cam Follower Motion Profile Examples Example 2
Let us fix the rise portion only, at junction B. Then the problem at junction C can be fixed using
symmetry. We specify a half-cycloidal function (sinusoidal in cam angle) to a straight line (constant
velocity) rise.
half-cycloidal function constant velocity function
1
S f1 (1 ) L1 1 sin 1 f 2 ( 2 ) V0 2
1 1
L
V v1 (1 ) 1 1 cos 1 v2 ( 2 ) V0
1 1
L
A a1 (1 ) 21 sin 1 a2 ( 2 ) 0
1 1
2L
J j1 (1 ) 3 1 cos 1 j2 ( 2 ) 0
1 1
0
0 20 40 60 80
0.05
V (m/deg)
0 -4
0x 10 20 40 60 80
J (m/deg ) A (m/deg2)
0 -5
0x 10 20 40 60 80
3
2
0
-2
0 20 40 60 80
(deg)
Half-Cycloidal Rise (to connect with constant velocity)
180
Cam Follower Motion Profile Examples Example 3
We now specify a full-cycloidal function (sinusoidal in cam angle). This will rise all the way to
meet a dwell smoothly; it satisfies the Fundamental Law of Cam Design. This is the same function
used in term project input link motion specification earlier, when starting at stopping at rest.
1 21
S f1 (1 ) L1 1 sin f 2 ( 2 ) 0
1 2 1
L1 21
V 1 cos v1 (1 ) v2 ( 2 ) 0
1 1
2 L1 21
A a1 (1 ) 2
sin a2 ( 2 ) 0
1 1
4 2 L1 21
J j1 (1 ) cos j2 ( 2 ) 0
1
3
1
The full-cycloidal function plots are shown below, rising through the displacement to connect with a
dwell.
Full Cycloid
2
S (m)
0
0 20 40 60 80
0.05
V (m/deg)
0 -3
0x 10 20 40 60 80
J (m/deg ) A (m/deg )
2
2
-2 -4
0x 10 20 40 60 80
2
3
-2
0 20 40 60 80
(deg)
Full-Cycloidal Rise (to connect with a dwell)
Note that the derivatives above are with respect to (deg). To find the time derivatives, use the chain
rule (e.g., for velocity, multiplying by d (t ) dt (t ) , a constant). However, for a constant , it is
customary to use the derivatives with respect to for cam design. The full-cycloidal function matches
the ensuing dwell: the displacement functions are made to match, and the velocity and acceleration are
zero at the end of the full-cycloidal function and the start of the ensuing dwell. The jerk does not match,
but the discontinuity in jerk is finite, which satisfies the Fundamental Law of Cam Design.
181
4.2.3 Analytical Cam Synthesis
Assume a valid cam motion profile has been designed according to the Fundamental Law of
Cam Design; i.e. we now have continuous S, V, A curves. Given the motion profile found by the
engineer, now we must determine the cam contour.
Is it as simple as polar-plotting S = f() vs. cam angle ? No – that approach would not account
for the face width of the cam follower, i.e. the contact points are not along radial lines in general. We
will use kinematic inversion to simplify the synthesis process.
DCRFFF Figure
As seen in the figure, the radius R out to the flat-faced follower (not to the point of contact x, y) is:
R C f ( )
where C is the minimum cam radius, a design variable, and S = f() is the given cam motion profile.
The radius R and the flat-faced follower length L can be related to the contact point x, y and the cam
angle through geometry.
182
R x cos y sin
L x sin y cos
Notice that
dR
x sin y cos L
d
d df
L (C f ( ))
d d
To calculate the follower flat-face length, double the maximum of L from above. It is doubled because
by symmetry the contact point will change to the other side at 180 .
This is sufficient to manufacture the cam since it is machined with , R, L coordinates. If we want to
know the cam contour in Cartesian coordinates, we must solve the relationships for x, y. In matrix form:
This special coefficient matrix [A] is orthonormal, which means both columns and rows are
perpendicular to each other and both columns and rows are unit vectors. One unique property of
orthonormal matrices is A A . The Cartesian cam contour solution is thus:
1 T
df
x (C f ) cos sin
d
df
y (C f ) sin cos
d
183
Minimum Cam radius to Avoid Cusps
A cusp is when the cam becomes pointed or undercut. Clearly, this must be avoided for good
cam motion. The cusp condition is that for a finite , there is no change in x, y.
dx dy
0 will cause a cusp.
d d
dx df df d2 f
(C f ) sin cos cos 2 sin
d d d d
dy df df d2 f
(C f ) cos sin sin 2 cos
d d d d
dx d2 f
C f 2 sin
d d
dy d2 f
C f 2 cos
d d
dx dy
0 occurs simultaneously only when
d d
d2 f
C f ( ) 0
d 2
Therefore, to avoid cusps on the entire cam contour, we must ensure that
d2 f
C f ( ) 2 0
d
Note that C is always positive and f() starts and ends at zero and never goes negative. So the sum of
these positive terms and the sometimes-negative second derivative of the cam motion profile must
always be greater than zero to avoid cusps or undercutting in the practical cam you are designing.
184
Disk Cam with Radial Flat-Faced Follower Design Example
Specify a full-cycloidal rise with a total lift of 50 mm, followed by a high dwell, a symmetric
full-cycloidal return with a total fall of 50 mm, and then a low dwell. Each of these four motion steps
occurs for 90 of cam shaft rotation.
The full-cycloidal rise and fall cam motion profile associated with this specification is shown
below. Clearly, this satisfies the Fundamental Law of Cam Design because the position, velocity, and
acceleration curves are continuous. The jerk is not continuous, but it remains finite over all cam angles.
50
40
30
S
20
10
0
0 50 100 150 200 250 300 350
50
0
V
-50
0 50 100 150 200 250 300 350
100
50
0
A
-50
-100
0 50 100 150 200 250 300 350
500
0
J
-500
0 50 100 150 200 250 300 350
185
Choosing a minimum cam radius of C = 100 mm, the resulting cam contour is shown below.
150
100
50
Y (mm)
-50
-100
-150 -100 -50 0 50 100
X (mm)
Cam Cartesian Contour
186
Let us check the cusp avoidance plot. To avoid cusps in this cam, we require that
d2 f
C S ( ) A( ) C f ( ) 0
d 2
As seen in the plot below, this inequality is satisfied for the entire range of motion, so this cam design is
acceptable with respect to avoiding cusps and undercutting.
250
200
150
C+S+A
100
50
0
0 50 100 150 200 250 300 350
200
150
100
50
Y (m)
-50
-100
-150
-200
-200 -150 -100 -50 0 50 100 150 200
X (m)
degrees of freedom dof – the number of independent parameters required to fully specify the
location of a device. The number of different ways something can move.
free-body diagram FBD, a diagram drawn out of context for each separate mass or inertia
with all external and internal forces and moments shown to give context.
cyclical frequency f, number of cycles per second, the inverse of time period, 1 / T, Hz.
circular frequency 2 f , scalar measure of rotational rate, i.e. the magnitude of the
angular velocity vector, rad/sec.
natural frequency The rate at which an unforced, undamped vibratory system tends to
vibrate, either cyclical frequency fn, or circular frequency n.
mass moment of inertia Idealized mechanical element that provides the rotational inertia in a
vibrational system.
189
damping A linear model for energy loss due to friction and other energy dissipaters,
for either translational or rotational motion. Damping is provided by a
virtual viscous dashpot (an automotive shock system is a real example).
phase angle Angle in a single sinusoidal function that indicates the amount of lead or
lag of a cosine or sine wave relative to the standard cosine or sine wave.
differential equation An equation in which the unknown appears with its time derivatives.
initial conditions Given values at time t = 0 for the output and its time derivatives. There
must be as many initial conditions as the order of the differential equation.
homogeneous solution Transient solution due to initial conditions and the external input forcing
function, yielding zero in the ODE.
particular solution Steady-state solution due to the external input forcing function. There is
also a transient solution due to the external input forcing function.
driving frequency The rate at which an external input forces a vibratory system, either
cyclical f, or circular .
damped frequency The rate at which an unforced, damped vibratory system tends to vibrate,
either cyclical fd, or circular d.
logarithmic decrement The natural logarithm of the ratio of two consecutive points, separated by
the damped time period Td, on a unforced underdamped vibratory wave.
beat frequency Vibrational phenomenon wherein two frequencies in a dynamic system are
close in value, causing a vibration within a vibrational envelope.
resonance This condition occurs when the driving frequency is the same as, or
close to, the system’s natural frequency n. In this case the resulting
amplitude of the steady-state vibration becomes very large.
190
frequency response The plots of system amplitude and phase angle vs. independent variable
input frequency.
critical shaft speed The angular shaft velocity that excites the natural frequency of a rotating
dynamic system. When driven at the critical speed, resonance occurs, a
condition generally to be avoided at all costs.
transmissibility The ratio of the maximum force transmitted to the base over the maximum
input force magnitude in a vibratory dynamic system.
vibration mode Characteristic shape at which a dynamic system will vibrate. Multi-dof
systems have multiple modes that sum to yield the total time response.
A linear system is one in which all governing equations (differential, algebraic) are linear.
Linear systems satisfy the principles of linear superposition and homogeneity. Let u(t) be the input
and y(t) be the output. Further, u (t ) y ( t ) indicates a system yielding output y(t) given input u(t).
1) linear superposition
if u1 (t ) y1 (t ) and u2 (t ) y2 (t ) then u1 (t ) u2 (t ) y1 (t ) y2 (t )
2) homogeneity
if u (t ) y (t ) then u (t ) y (t )
Helical Coiled Spring. The jagged springs indicated so far represent physical translational
springs coiled from an elastic wire into a helical shape about a central axis. For this physical spring,
from Strength of Materials, the linear spring constant k for small displacements is given below.
f(t) f(t)
D
Gd 4
k
8nD3
where
G material shear modulus N/m2
d wire diameter m
n number of spring coils unitless
D diameter of spring coiling m
This type of coiled physical spring is common; however, there are many mechanical systems
without a clear spring such as above. Instead, their elastic properties are provided by distributed
material in the system. Here we present several such distributed springs that are modeled as linear
springs for small displacements, as follows.
For the following 4 translational springs, the following method is used. For each given physical
situation, we find the displacement under an applied force F, from Strength of Materials. Then the
spring stiffness constant k is found from Hooke’s Law F = k: k = F / . Below we assume that the force
is applied and the displacement measured at the end of the cantilevered beam and in the center of the
other beams.
192
Cylindrical Rod Spring. An elastic rod has an equivalent spring constant along the longitudinal
axis of the rod. From Strength of Materials, the linear spring constant k for small displacements is given
below.
r
f(t) f(t) f(t) f(t)
r1
r2
L L
EA
k
L
where
E material Young’s modulus N/m2
A rod cross-sectional area, r 2 m2
r radius of rod m
L length of rod m
So the cylindrical rod spring constant for a constant circular cross section becomes
E r2
k
L
If the elastic rod is tapered with different radii r1 and r2 on opposite ends of the rod, the equivalent
spring constant is
E rr
k 1 2
L
Cantilever Beam Spring. A rectangular cantilevered beam has an equivalent spring constant
for small vertical displacements. From Strength of Materials, the linear spring constant k for small
displacements is:
h m
L x(t)
3EI
k
L3
where
E material Young’s modulus N/m2
I beam area moment of inertia, bh 3 12 m4
b beam width m
h beam height m
L length of beam m
h m
L
x(t)
48EI
k
L3
where
E material Young’s modulus N/m2
b beam width m
h beam height m
L length of beam m
This doubly-simply-supported beam is 16 times stiffer in vertical displacement than the cantilever beam.
Doubly-Fixed Beam Spring. A rectangular beam rigidly fixed at both ends has an equivalent
spring constant for small vertical displacements. From Strength of Materials, the linear spring constant
k for small displacements is given below.
h m
L
x(t)
192EI
k
L3
where
E material Young’s modulus N/m2
b beam width m
h beam height m
L length of beam m
This doubly-fixed beam is 64 times stiffer in vertical displacement than the cantilever beam, and 4 times
stiffer than the simply-supported beam presented earlier.
194
Air Spring. An enclosed volume of air has an equivalent spring constant for small
displacements. From Fluid Mechanics for compressible air, the linear spring constant k for small
displacements is given below.
f(t)
PA2
k
V
where
ratio of specific heats unitless
P enclosed air pressure N/m2
A enclosed air cross-sectional area m2
V enclosed air volume m3
These six spring constant examples are all for translational springs, so their spring constant k
units should all be N/m. The material Young’s Modulus E and Shear Modulus G for steel are:
Let us now check the units for the previous six spring constants.
We have also used torsional springs in our mechanical systems models. Torsional springs can be
constructed of a coiled elastic material, spiraling in one plane. An elastic cylindrical rod also provides a
torsional spring effect when torque about the longitudinal axis of the rod. Additionally, a fixed-simply-
supported beam also provides a torsional spring effect.
195
Torsional Rod Spring. An elastic rod has an equivalent spring constant about the longitudinal
axis of the rod. From Strength of Materials, the torsional spring constant kR for small angular
displacements is given below.
r
(t) (t)
G r4
kR
2L
where
G material shear modulus N/m2
r radius of rod m
L length of rod m
G (ro4 ri4 )
kR
2L
where ro and ri are the outer and inner shaft radii, respectively.
Fixed-Simply-Supported Torsional Beam Spring. A rectangular beam rigidly fixed at one end
and simply supported on the other has an equivalent torsional spring constant for small angular
displacements. From Strength of Materials, the torsional spring constant kR for small displacements is
given below.
(t) m h
L
4EI
kR
L
where
E material Young’s modulus N/m2
b beam width m
h beam height or thickness m
L length of beam m
G r4 Nm4 Nm Nm Ebh3 N m4 Nm Nm
= =
2L m2m 1 rad 3L m2 m 1 rad
Recall that rad doesn’t count as a unit so we are free to include it above, yielding the correct SI units N-
m/rad for torsional spring constant.
Non-massless Spring. For translational mechanical systems in which the distributed mass of the
spring is significant and cannot be ignored, we use the following formula for the lumped system mass.
mS
mE m kg
3
Where m is the modeled point mass, mS is the spring mass, and mE is the calculated equivalent point
mass to include in the model with a massless spring. The spring constant k does not change and is
calculated as shown earlier in this subsection.
For the cantilever and simply-supported beam spring models, the equivalent mass is given below
(left and right below, respectively, and mB is the distributed beam mass, all kg).
mB
mE 0.23mB m mE m
2
197
Calculation of Damping Constants
In general the damping coefficient for a physical system is harder to identify and calculate than
the mass and spring constants. This subsection presents 4 equivalent damping coefficient cases. Many
physical systems have no identifiable damping dashpot, but a viscous damping coefficient c is included
to account for friction losses in the system, using a simple linear model (the law for linear dashpots
where the friction force is proportional to the velocity).
Also see the later section on logarithmic decrement, where the damping coefficient c can be
estimated from experimental observations.
Dashpot. The piston-based dampers indicated so far represent physical dashpots with a piston
moving (translating in a reciprocating fashion) in a viscous fluid to remove energy from the system,
such as in an automotive shock absorber. For this physical dashpot, from Fluid Mechanics, the linear
viscous damping coefficient c for small velocities is given below.
f(t)
h w
6 r 3h w
c 1
w3 r
where
fluid viscosity N-s/m2
r piston radius m
h piston height m
w wall thickness ro – r m
198
Coulomb damping. Coulomb damping arises from the so-called dry friction of sliding. This is
a nonlinear effect that can be approximated by the following equivalent linear viscous damping
coefficient ceq.
v(t)
f f
4ff 4 N
ceq
B B
where
ff friction force N N
coefficient of friction unitless
N normal force N
B amplitude of vibration m
frequency of vibration rad/s
Flat plates. When two flat plates move relative to each other, separated by a layer of viscous
fluid, damping arises. From Fluid Mechanics, the linear viscous damping coefficient c for small
velocities is given below.
f(t)
A
ceq
w
where
fluid viscosity N-s/m2
A area of smaller flat plate m2
w viscous fluid layer thickness m
These three damping examples are all for translational dampers, so their viscous damping
coefficient c units should all be N-s/m. The SI units for fluid viscosity are N-s/m2 and the Coulomb
friction coefficient is unitless. Let us now check the units for the previous three damping coefficients.
6r 3h w Ns m3m m Ns 4 N N Ns A Ns m2 Ns
1
w3 r m2 m3 m m B m rad m w 2
m m m
s
(t)
h w
h0
h r
cR 2 r 3
w 8h0
where
fluid viscosity N-s/m2
r piston radius m
h piston height m
w wall thickness ro – r m
h0 height from base to rotating piston m
200
Journal bearing. A journal bearing also has a fluid providing rotational damping. From Fluid
Mechanics, the linear rotational viscous damping coefficient cR for small angular velocities is given
below.
r
2 wr 3
cR
where
fluid viscosity N-s/m2
w base thickness m
r bearing radius m
journal bearing fluid clearance m
Let us now check the units for these torsional damping coefficients.
Recall that rad doesn’t count as a unit so we are free to include it above, yielding the correct SI units N-
m-s/rad for torsional damping coefficient.
201
Equivalent Springs
x(t)
k1 k2
m
From the FBD (not shown), again there is a common displacement x(t). Assuming a small
positive displacement x(t), both spring restoring forces are negative (to the left). The equivalent spring
derivation is now presented.
f f1 f 2
f k1 x k2 x (k1 k2 ) x
Therefore, the equivalent spring constant for parallel springs case is again:
f
kpar k1 k2
x
This result is identical to the original parallel springs case presented in the ME 3011
NotesBook.
k1k2 k3
kser3
kser2
k1k2 k1k2 k1k3 k2 k3 kser4
k1k2k3k4
k1 k2 k1k2k3 k1k2 k4 k1k3k4 k2 k3k4
Actually, the original general formula below yields the same answers, with less computation:
1
kser n
1
k
i 1 i
202
6.2.3 Pendulum System Dynamics Model
Linearized ODE models can be derived that are very similar to the simple pendulum in the ME
3011 NotesBook, for the two pendulum models shown below. The figure on the left shows a pendulum
consisting of a slender rod of length L, with distributed mass m. The figure on the right shows a
pendulum consisting of a slender rod of length L, with distributed mass m1, plus an end-mounted point
mass of m2.
O O
g g
m m1 L
L
(t) (t)
m2
Two Additional Pendulum System Diagrams
Only the mass moment of inertia IOZ changes; for the cases on the left and right above, respectively, the
formulae are given below.
mL2 m L2
I OZ I OZ 1 m2 L2
3 3
Where, for both slender rod cases, in the sum of moments we used the fact that the slender rod weight
(mg and m1g, respectively), acts at the slender rod CG, at a radius of L/2 from point O.
It is possible to account for friction in the simple pendulum system by including a rotational
viscous damping coefficient at the rotational bearing.
cR g
(t ) 2
(t ) (t ) 0
mL L
And while not common, it is also possible to include an input forcing function, input torque (t)
at the bearing.
203
cR g 1
(t) 2
(t) (t) 2 (t)
mL L mL
Finally, if we model a horizontal pendulum, the restoring torque of gravity disappears and we
have the dynamic model of a gravity-neutral single rotating link.
cR 1
(t) 2
(t) 2 (t)
mL mL
204
6.2.4 Uniform Circular Motion
Uniform circular motion (UCM), wherein a point P at the end of a fixed-length link of length r
rotating in the XY plane about a fixed point O with a constant angular velocity (see figure below)
generates Simple Harmonic Motion, like an unforced, undamped m-k translational mechanical system.
Y P
r
(t)
X
O
Uniform Circular Motion Diagram
In this UCM case no differential equations or solutions are required. Simultaneously simple
harmonic motion (SHM) is generated in the X with a pure cosine wave (with no phase angle) and in the
Y with a pure sine wave (with no phase angle); see the figure below, generated by UCM.m. Since
angular velocity is constant, the angle linearly increases with time, (t) t . With 0 = 0 at t = 0, this
system will generate SHM in the X and Y directions indefinitely, repeating every 360 deg or 2 rad.
Derive the vibrational equation of motion for each given mechanical system below.
1. m-c-k translational systems with parallel and series springs and dampers
The ME 3011 NotesBook presented the derivation of equivalent translational springs for the
parallel springs and series springs cases. Translational dashpots also follow the same formulas for their
respective parallel and series cases.
x(t)
k1 k2
c1 m f(t)
c2
1-dof Translational Mechanical System with Parallel Springs and Dashpots
x(t)
k1 k2
c1 c2 m f(t)
cc kk
mx(t ) 1 2 x (t ) 1 2 x(t ) f (t )
c1 c2 k1 k2
206
2. J-cR-kR rotational systems with parallel and series springs and dampers
The same type of second-order models result for the rotational systems as in the translational
systems presented on the last page, with regard to parallel and series arrangements of rotational springs
and dashpots.
J (t ) (c R1 c R 2 ) (t ) ( k R1 k R 2 ) (t ) (t )
c c k k
J(t ) R1 R 2 (t ) R1 R 2 (t ) (t )
cR1 cR 2 k R1 k R 2
207
Compound m-c-k translational systems with parallel and series springs and dampers
Certain real-world systems may require more complicated models with combinations of springs
and dampers in parallel and in series. There are infinite possible combinations of such models, dictated
by the real-world component arrangement. The next two sections present two such possibilities.
The figure below shows two sets of parallel springs arranged in series. For clarity, the dashpots
are not shown but they have the same arrangement.
x(t)
k1 k2
m f(t)
k3 k4
x(t)
ka kb
m f(t)
Using the parallel- and then serial-springs equivalent spring formulas derived in the ME 3011
NotesBook:
k a k1 k3 k a kb k k k k k k k k
keq 1 2 2 3 1 4 3 4
kb k 2 k 4 k a kb k1 k2 k3 k4
mx(t ) ceq x (t ) k eq x (t ) f (t )
This example is for a translational system, but equivalent results exist for an analogous rotational
system.
208
4. Series springs and dampers in parallel
The figure below shows two sets of series springs arranged in parallel. For clarity, the dashpots
are not shown but they have the same arrangement.
x(t)
k1 k2
m f(t)
k3 k4
x(t)
ka
m f(t)
kb
Using the series- and then parallel-springs equivalent spring formulas derived in the ME 3011
NotesBook:
k1k 2
ka
k1 k 2 k1k2 kk k k k k k k k k k k 2 k3 k 4
keq ka kb 3 4 1 2 3 1 2 4 1 3 4
k3 k 4 k1 k2 k3 k4 k1k3 k2 k4 k1k 4 k 2 k3
kb
k3 k 4
mx(t ) ceq x (t ) k eq x (t ) f (t )
This example is for a translational system, but equivalent results exist for an analogous rotational
system.
209
5. Electric motor with vibration isolation FBD
(t)
J
k k
Often a vibration isolation system will be designed to minimize the transference of shaking
forces and shaking moments from rotating machinery to the ground. The rotating electric motor shown
above, whose moment of inertia about the motor shaft axis is J, is supported by four identical springs (2
shown, 2 symmetric into the page on the back side of the motor) of spring constant k. We measure
angular displacement (t) according to the right-hand-rule (positive CCW) such that (t) is zero when
the system is horizontal. First, draw the FBD.
Using Euler’s Rotational Dynamics Law for the motor shaft, in words we can say that the
excitation torque caused by the inertia J and angular acceleration (t ) of the motor shaft must be
balanced by the restoring torque due to the four translational springs through the distance a. Assuming
positive angular acceleration, the two left translational springs are in compression and the two right
translational spring are in tension in applying the restoring torques.
M J (t )
4 kax (t ) J (t )
Where x(t) is the total displacement of each spring. Now substitute x(t ) a (t ) (approximation
for small displacement) into the above equation to yield the final 1-dof model for this system.
J(t ) 4ka 2 (t ) 0
210
6. Sprung mass/cylinder FBDs
r (t)
m
k
m
x(t)
In this dynamic system, m1 is the point mass suspended by cable about a cylindrical pulley of
mass m2 and radius r. The other end of the cable is attached to a linear translational spring with
constant k. We measure displacement x(t) (positive downward) from the neutral position of the spring.
We also measure angular displacement (t) with the right-hand-rule (positive CCW) such that it is zero
when x(t) is zero. First, draw the FBDs.
m2 r 2
Tr kr ( r (t )) (t )
2
We can combine the above two equations to eliminate the cable tension T.
T m1 g m1r(t )
m2 r 2
m1 gr m1r 2(t ) kr ( r (t )) (t )
2
At static equilibrium we have m1 gr kr , so the final 1-dof model for this system is
m2 r 2
m1r 2(t ) kr 2 (t ) (t )
2
m2
m1 (t ) k (t ) 0
2
211
7. Doubly-sprung torsional rod FBD
k
y(t)
kR (t)
L
k
In this dynamic system, m is the distributed mass of the slender rigid rod of length L. The rod’s
left end is connected to ground by a torsional spring kR and the rod’s right end is supported by two
identical linear translational springs k as shown. We measure angular displacement (t) in a right-
handed sense from the horizontal rod position. To linearize we will make a small angle assumption; we
further ignore the small horizontal motion and gravity loading of the springs. First, draw the FBD.
The mass moment of inertia of the slender rod about its end is J mL2 3 . Also, y(t ) L sin (t ) ,
leading to:
mL2
(t ) kR (t ) 2kL2 sin (t ) 0
3
Applying the small angle assumption, we have sin (t ) (t ) and the final 1-dof model for this system
is
mL2
(t ) kR 2kL2 (t ) 0
3
212
8. Sprung rolling cylinder FBD
x(t)
k r (t)
In this dynamic system, a cylinder of mass m and radius r rolls without slipping on a horizontal
surface. The cylinder center is attached to the ground via a linear translational spring with constant k.
We measure displacement x(t) (positive to the right) from the neutral position of the spring. We also
measure angular displacement (t) with the right-hand-rule (positive CCW) such that it is zero when x(t)
is zero. First, draw the FBD.
Using Newton’s Second Law for the cylinder (Ff is the force due to friction):
F X Ff kx(t ) mx(t )
M Z Ff r J(t )
We can combine the above two equations to eliminate the friction force Ff:
F f mx(t ) kx (t )
mr 2
x(t )
(mx(t ) kx(t ))r
2 r
3
mx(t ) kx(t ) 0
2
213
9. Doubly-sprung rotating cylinder FBD
r (t)
m
k k
In this dynamic system, a rotating cylindrical pulley of mass m and radius r is suspended from
the ground link as shown. A massless inextensible cable is passed over the pulley and connected to two
grounded linear translational springs of equal spring constant k. At static equilibrium both springs are
pre-tensioned by a vertical displacement . We measure angular displacement (t) with the right-hand-
rule (positive CCW). First, draw the FBD.
M Z T1 r T2 r J (t )
Where the left cable tension is T1 k ( r (t )) and the right cable tension is T2 k ( r (t )) .
Also, the mass moment of inertia of the cylindrical pulley is J mr 2 2 .
mr 2
kr ( r (t )) kr ( r (t )) (t )
2
m(t ) 4k (t ) 0
214
10. Sprung pendulum FBD
L0
k
L
(t)
In this dynamic system, a standard point-mass m pendulum is tied to the ground through a linear
translational spring with constant k, connected a distance L0 along the length L of the massless pendulum
rod. We measure angular displacement (t) with the right-hand-rule (positive CCW) such that it is zero
when the pendulum is vertical (down). First, draw the FBD.
Where kasin (t) is the amount the spring stretches. The mass moment of inertia of the point-
mass pendulum about the point of rotation is J mL2 .
mL2(t ) (mgL ka 2 cos (t ))sin (t ) 0
This is a nonlinear model; in addition to the sine of the regular pendulum, now we also have a cosine
term of the unknown angular displacement (t). To linearize let us assume small angle motion.
We are implicitly ignoring the vertical motion of the spring. The final 1-dof model for this system is
mL2(t ) (mgL ka 2 ) (t ) 0
215
11. Doubly-sprung rolling cylinder FBD
x(t)
k k
a r (t)
In this dynamic system, a cylinder of mass m and radius r rolls without slipping on a horizontal
surface. The cylinder is attached off-center (distance a from the center to the springs connection point)
to the ground via two identical linear translational springs with constant k. We measure displacement
x(t) (positive to the left) from the equilibrium position of the springs. At static equilibrium both springs
are pre-tensioned by an equal horizontal displacement . We also measure angular displacement (t)
with the right-hand-rule (positive CCW) such that it is zero when the springs are in their equilibrium
position. First, draw the FBD.
Using Newton’s Second Law for the cylinder (Ff is the force due to friction):
M Z F f r ka ( ( r a ) (t )) ka ( ( r a ) ( t )) J ( t )
Where the left spring force tension is k( (r a) (t)) and the right cable tension is
k( (r a) (t)) . Also, the mass moment of inertia of the cylinder is J mr 2 2 . We can combine the
above two equations to eliminate the friction force Ff:
Ff mr(t ) 2k (r a) (t )
2
mr
(mr(t ) 2k (r a) (t ))r 2ka(r a) (t ) (t )
2
3 2
mr (t ) 2k (r a)2 (t ) 0
2
216
12. Hung-sprung mass/cylinder FBDs
r (t)
x (t) m
m
x(t)
In this dynamic system, m1 is the point mass suspended by an inextensible cable about a
cylindrical pulley of mass m2 and radius r. The other end of the cable is rigidly attached to the ground.
The pulley is attached to ground via a linear translational spring with constant k. We measure
displacements x1(t) and x2(t) (both positive downward) from the neutral position of the spring. We also
measure angular displacement (t) with the right-hand-rule (positive CCW) such that it is zero when x(t)
is zero. First, draw the FBDs.
F X m1 g T1 m1
x1 (t )
FX T1 T2 m2 g k ( x2 (t )) m2
x2 (t )
Where is the static deflection of the spring due to gravity. Using Euler’s Rotational Dynamics Law for
the cylinder:
M Z T1 r T2 r J (t )
The mass moment of inertia of the cylindrical pulley is J m2 r 2 2 . Substituting the mass moment of
inertia and eliminating the cable tensions T1 and T2 amongst these three equations yields
217
T1 m1 g m1
x1 (t )
x2 (t ) m2 g k ( x2 (t )) T1
T2 m2
m2 r
2(m1 g m1 x2 (t ) m2 g k ( x2 (t ))
x1 (t )) m2 (t )
2
T1 T2 m2 g k
(2m1 m2 ) g k
m2r
(t ) 2m1x1 (t ) m2 x2 (t ) kx2 (t ) 0
2
3
4m1 m2
x1 (t ) kx1 (t ) 0
2
Interestingly, the same exact equation form results, regardless of whether the model variable is x1(t), as
above, or x2(t), or (t).
218
13. Horizontally-sprung mass/cylinder FBDs
a r (t)
m
m
x(t)
In this dynamic system, m1 is the point mass suspended by cable about a cylindrical pulley of
mass m2 and radius r. The pulley is grounded via a linear translational spring with constant k, attached
distance a from the rotational center of the pulley. We measure displacement x(t) (positive downward)
from the neutral position of the spring. We also measure angular displacement (t) with the right-hand-
rule (positive CCW) such that it is zero when x(t) is zero. First, draw the FBDs.
T m1 g m1
x(t )
m2 r 2
x(t ))r ka ( a (t ))
(m1 g m1 (t )
2
2
219
14. Cylinder rolling in a big cylinder FBD
R 2(t)
m
1(t)
In this dynamic system, a cylinder of mass m and radius r rolls without slipping in a concave
cylindrical surface of radius R as shown. 1(t) is the angular displacement of the cylinder and 2(t) is the
angular displacement of the cylinder CG from the vertical. Both angles are measured with the right-
hand-rule (positive CCW) such that they are zero when the cylinder is at static equilibrium at the bottom
of the well. We define the moving X axis such that it is perpendicular to the radial direction, pointing to
the left. First, draw the FBD.
Ff mg sin 2 (t ) mx(t )
mr
mg sin 2 (t ) mx(t ) (1 (t ) 2 (t ))
2
Note that from kinematics we know x (t ) ( R r )2 (t ) and 1 (t ) R 2 (t ) r . Substituting the
kinematic equations yields
3( R r )
2 (t ) g sin 2 (t ) 0
2
We apply a small angle assumption such that sin 2 (t ) 2 (t ) . The final linearized 1-dof model for this
system is
3( R r )
2 (t ) g2 (t ) 0
2
220
15. Doubly-sprung pulleys/mass FBDs
k
x(t)
x(t)
k m
x(t)
In this dynamic system, m is the point mass suspended by an inextensible cable about two
pulleys of negligible mass as shown. Each pulley is connected to the ground via linear translational
springs with constants k1 and k2, respectively. We measure displacements x(t) (positive downward),
x1(t), and x2(t) as shown.
m2
(t)
m1
In this dynamic system, a standard point-mass pendulum is rigidly connected to a cylinder that is
free to rotate without slipping on a horizontal surface. M1 is the point mass and the cylinder has mass m2
and radius r. The length of the massless pendulum rod is L. We measure angular displacement (t) with
the right-hand-rule (positive CCW) such that it is zero when the pendulum is vertical (down). First,
draw the FBD.
At any instant of the motion, the rotation is about (moving) point C, the instantaneous center of
zero velocity. Therefore let us take point C as the moment center for deriving the dynamic model
M CZ JC(t ) . The mass moment of inertia of the compound rigid link, using the parallel axis
theorem, is:
m2 r 2 3
J C m1R
2
m2 r 2 m1R2 m2 r 2
2 2
2
3 2
m1 ( L 2 Lr cos (t ) r ) m2 r (t ) m1 gL sin (t ) 0
2 2
2
This is a nonlinear model; in addition to the sine of the regular pendulum, now we also have a cosine
term of the unknown angular displacement (t). To linearize let us assume small angle motion.
sin (t) (t) cos(t) 1
2
222
17. Rigid-body pendula
Earlier we derived the linearized dynamic model for a simple pendulum consisting of a massless
rigid rod of length L and a point mass m.
g
(t ) (t ) 0
L
We also derived the dynamic model for a slender-rod pendulum with distributed mass and also
for a slender-rod pendulum with distributed mass plus point mass. Here we present the dynamic models
for pendula of various other rigid-body shapes, including for the general case. The pendulum mass is m,
O is the fixed point of rotation, G is the pendulum center of mass, and rG is the distance between these
two points. As before, we measure angular displacement (t) with the right-hand-rule (positive CCW)
such that it is zero when the pendulum is vertical (down). Draw the FBD for the general rigid-body
pendulum below.
Note that if we define a virtual point mass pendulum length of LEQ J O mrG , we could use the simple
point-mass pendulum model for any rigid body pendulum. Now we can apply these results to specific
rigid-body pendula below.
g g
g
(t)
g L s
(t)
m m
m r m
(t) r (t)
h s
Semi-circular Circular Rectangular Square
rG
4r mr 2 m( L2 h 2 ) ms 2
JG JG JG
3 2 12 6
mr 2 3mr 2 m(4 L2 h 2 ) 5ms 2
JO JO JO JO
2 2 12 12
8g 2g 6 gL 6g
(t ) (t ) 0 (t ) (t ) 0 (t ) (t ) 0 (t ) (t ) 0
3 r 3r 4L2 h2 5s
223
18. Beams with a point mass
Earlier we derived the equivalent spring constants kEQ for three massless beams with point
masses attached as shown, the cantilevered beam, the simply-supported beam, and the rigidly-clamped
beam. For all three cases the general model is mx(t ) k EQ x (t ) 0 .
m h m
h
L x(t) L
x(t)
Cantilevered beam Simply-supported beam
Ebh3 4Ebh3
mx(t ) x(t ) 0 mx(t ) x(t ) 0
4 L3 L3
h m
L
x(t)
Doubly-fixed beam
16Ebh3
mx(t )
x(t ) 0
L3
Machines with rotating parts are common in industry. The figure below represents a simplified
idealized machine with an offset mass m1 that is unbalanced in rotation. The total mass of the machine
is m2, supported by equivalent spring and damping coefficients k and c, respectively, to the ground
(there may be four machine mounts, one on each corner of the machine, in parallel, so their individual ki
and ci coefficients add up to yield the overall k and c). The machine is constrained to move in the
vertical direction only. We measure displacement x(t) in the vertical (up) direction, from the gravity-
loaded spring location.
x(t)
m1
e
t
m2
k c
The model is the same we have derived earlier for the standard m-c-k translational mechanical
system.
x(t ) cx (t ) kx (t ) f (t )
m2
Where now the forcing function f(t) is supplied by the rotating imbalance, mass m1 at an offset of e, as
shown. Assume the rotation is at a constant angular velocity of . Then the centripetal acceleration of
the rotating mass is e2 (directed inward) and thus the inertial force of the rotating mass is m1e 2
(directed outward). The angle of rotation is t and so the vertical component of the rotating imbalance
is:
f (t ) m1e 2 sin t
u(t) x(t)
k
m
c
In this case, the base is not fixed, but harmonically-moved via base displacement u(t) in units of
meters. We measure displacement x(t) in the same manner (positive to the right, from the neutral
position of the spring, this time with u(t) = 0). There is no input forcing function f(t). The moving base
portion is massless.
The model is similar to what we have previously derived. Assume that displacements x(t) and
u(t) are small and positive at this instant in time, with x(t) u(t) . We know that the model we derive
based on these assumptions will hold good for all small displacement motions, as both x(t) and u(t)
vibrate at different phases in different directions. With this assumption, the spring is in tension, and the
damper behaves similarly. First, draw the FBD. Using Newton’s Second Law for mass m:
FX k ( x(t ) u(t )) c( x(t ) u(t )) mx(t )
mx(t ) cx (t ) kx(t ) cu (t ) ku (t )
We can genericize this model similar to previous derivations by using the dimensionless
c k
damping ratio and natural frequency n as follows.
2 km m
x(t ) 2n x (t ) n2 x(t ) 2nu (t ) n2u (t )
It may be convenient in some cases to use a new variable in the above model, the relative
difference in displacements, z(t) x(t) u(t) . Then we have
mz(t) cz(t) kz(t) mu(t)
Where now mu(t) plays the role of input forcing function. If we specify a sinusoidal input
displacement u(t ) Asin t then the model becomes
mz(t ) cz (t ) kz (t ) mA 2 sin t
Note that this version of the model is identical in form to the forced, damped m-c-k system where the
force amplitude F has been replaced by mA2 .
226
21. Horizontal sprung pendulum FBD
(t)
a b x(t)
m
In the figure above, a horizontal simple pendulum has a point mass on the end of a massless rigid
rod of length a. The rod extends by length b to the other side to connect to the ground through a linear
translational spring with constant k. We measure angular displacement (t) with the right-hand-rule
(positive CCW) such that it is zero when the pendulum is horizontal. We further measure displacement
variable x(t) as the extension/compression of the spring as shown. First, draw the FBD.
Where x(t) is the total displacement of the spring ( is the static deflection of the spring in
tension due to gravity). Also, the mass moment of inertia of the pendulum about the axis of rotation is
J ma2 .
This is a nonlinear model, due to the cosine term of the unknown angular displacement (t). To
linearize let us assume small angle motion cos (t ) 1 .
At static equilibrium we have mga kb , so the simplified model for this system is
ma 2(t ) bkx (t ) 0
For a small displacement x(t) the approximation x(t ) b sin (t ) holds good. Once again assuming
small angle motion, to linearize we use sin (t ) (t ) , so the final 1-dof model for this system is
ma 2(t ) kb 2 (t ) 0
227
22. Sprung cylinder FBD
(t)
r
x(t) y(t)
In this dynamic system, a cylinder of mass m and radius r is suspended via a cable on one side
and connected to ground via a linear translational spring with constant k on the other side. We measure
displacement x(t) (positive downward) from the neutral position of the spring and y(t) is the
displacement of the cylinder center. We also measure angular displacement (t) with the right-hand-rule
(positive CCW) such that it is zero when x(t) is zero. First, draw the FBD. Using Newton’s Second
Law for the cylinder:
FX mg k ( x(t )) T (t ) my(t )
Where is the static deflection of the spring due to gravity. Using Euler’s Rotational Dynamics Law for
the cylinder:
M Z T (t ) r k ( x (t )) r J (t )
The mass moment of inertia of the cylindrical pulley is J mr 2 2 . Substituting the mass moment of
inertia and eliminating the cable tension T(t) between these two equations yields
T (t ) mg k ( x(t )) my(t )
mr
mg 2k ( x(t )) my(t ) (t )
2
Note that from kinematics we know x(t) 2y(t) and y(t) r(t) . Substituting the kinematic equations
yields the final 1-dof model for this system.
3
m (t ) 4k (t ) 0
2
228
229
23. Vertical sprung pendulum with damping FBD
k xb(t)
c
a
(t)
In the figure above, a simple pendulum rotating about point O has a point mass on the end of a
massless rigid rod of length L. The rod extends by length b to the other side of point O to connect to the
ground through a horizontal linear translational spring with constant k. We measure angular
displacement (t) with the right-hand-rule (positive CCW) such that it is zero when the pendulum is
vertical. We further measure displacement variable xb (t ) b (t ) as the compression/extension of the
spring as shown. A viscous dashpot is connected by length a down along the rod to provide damping,
with velocity variable x a (t ) a(t ) . First, draw the FBD.
The mass moment of inertia of the pendulum about the axis of rotation is J mL2 .
This is a nonlinear model; in addition to the sine of the regular pendulum, now we also have a cosine
term of the unknown angular displacement (t). To linearize let us assume small angle motion.
L1 L2
m
L
T T
x(t)
L1 L2
In this system a point mass m is suspended between two strings of tension T. Assume that for
small motions the tension can be considered to be constant. Also assume there is only vertical
displacement, i.e. ignore horizontal motion of the mass. We measure displacement x(t) positive
vertically (up).
F X f mx(t )
Assuming small displacement we also have small angular motions. For small angular motions
sin (t) tan (t) and then we have
1 1
mx(t ) T ( ) x(t ) 0
L1 L L1
L
mx(t ) T ( ) x(t ) 0
L1 ( L L1 )
231
25. Manometer
x(t)
x(t)
A manometer is used for measuring pressure. The manometer shown has a uniform cross-
sectional area A, a liquid column of length L, and a liquid density of . We measure displacement x(t)
positive vertically as shown, relative to atmospheric pressure, where the two columns have equal height.
F X w(t ) mx(t )
Where w(t) is the weight acting due to the difference in height in the manometer. Therefore:
Since the total liquid mass m in the manometer is AL , the final 1-dof manometer dynamics
model is
mx(t ) 2 gAx(t ) 0
ALx(t ) 2 gAx(t ) 0
Lx(t ) 2 gx(t ) 0
232
26. Cylinder in water bath
x(t)
A solid buoyant cylinder is immersed in a water bath. Derive the vibratory dynamic equation of
motion assuming purely vertical motion with no rotations. The cylinder has a height h and radius r and
thus a mass of m V r 2 h , where is the cylinder material density. Further let W be the water
density. Displacement x(t) is the amount of the cylinder height under water at any instant, as shown.
From Archimedes’ principle, the time-varying restoring force fR due to the water is the weight of water
displaced.
f R (t ) W VW g W g r 2 x (t )
mx(t ) kx(t ) 0
r 2 hx(t ) W g r 2 x(t ) 0
hx(t ) W gx(t ) 0
233
27. Human skeletal muscle model
This section presents a common single skeletal muscle dynamics model, including the figure,
equations, state-space form, and simulations. This is a vibrational model, included for completeness, to
show one type of engineering mechanics modeling for muscles. In this class we will generally treat the
effect of muscles more simply, i.e. as simple cables in tension with no mass, spring constants, or
dynamics of their own.
The model in this section was suggested by Dr. Scott Hooper (OU Biological Sciences) and the
equations were derived and the simulations performed by Elvedin Kljuno (OU Fulbright Scholar from
Bosnia). Dr. Hooper uses this model to study neural muscular control in the stomachs of lobsters, but he
says it can be scaled to adequately model skeletal muscle in many animals, including humans.
The skeletal muscle vibrational dynamics model is shown below. The lumped mass m represents
the load the muscle is lifting vs. gravity g. The muscle itself is represented by linear elastic spring
stiffness k1, in parallel with linear elastic spring stiffness k2 that is in series with linear dissipative
dashpot b. The absolute displacement of the dashpot end and the muscle end are measured by
coordinates x and y, respectively. Lengths L1R and L2R (not shown) are the resting lengths (not stretched
by gravity) of springs 1 and 2, respectively. In vibrations it is more common to measure change in
displacements relative to the gravity-stretched position, but biologists need to include the absolute spring
lengths as well. The actuator FA represents the contractile element of the muscle and Fm is the force
generated by the muscle. Note: this derivation assumes zero pennation angle, i.e. zero angle between
the muscle fibers and the tendons.
k1 ( y L1R ) F1 (1)
From the right figure above, the equations for the right spring and dashpot are:
k2 ( y x L2R ) F2
(2, 3)
bx F2 FA
From the FBD shown below, the dynamics equation for the mass is obtained using Newton’s Second
law:
F vert ma vert
(4)
F1 F2 Fm mg my
235
Substituting (1) and (2) into (4), and also substituting (2) into (3) yields the following coupled
second/first order linear vibrational dynamics model for skeletal muscle.
1 k 1 F
y
(k1 k2 ) y 2 x (k1 L1R k2 L2R ) g m
m m m m
(7, 8)
k k F k
x 2 x 2 y A 2 L2R
b b b b
As mentioned above, (7, 8) can be simplified by choosing more adequate coordinates that will
measure the deviation from the static position only. The form shown above is formed intentionally for
biological systems, using absolute coordinates and initial spring lengths.
The system of differential equations (7, 8) can be solved for coordinates x and y as functions of
time given time functions for FA and Fm. In the case of a simple form of FA and Fm the solution could
be found using reduction and decoupling by a formation of a single third-order linear ordinary
differential equation.
k2 (k k ) kk Fm (t ) k2
y (t )
y(t ) 1 2 y (t ) 1 2 y(t )
( FA (t ) Fm (t ) mg k1L1R ) (9)
b m mb m mb
Equation (9) can be solved for y(t) which is then substituted into (8) to directly solve for x(t).
236
Common Model Features
Considering all vibrational models we have derived, including the many cases in this section,
some general comments can be made about all models.
1. When gathering the unknown and its time derivatives on one side of the ODE, there will
never be a change in sign amongst these terms; generally they will all be positive.
2. The effect of gravity always cancels out. That is, any mg static weight terms will load the
spring with a static deflection and these terms always subtract out of the final ODE.
3. As always, the resulting units must be consistent. This is a good check to perform on your
resulting model.
237
Many electrical circuits can exhibit electrical vibrations, i.e. oscillation of the electrical output
variable of interest, which is why their modeling is included here.
Kirchoff’s Laws
Instead of Newton’s and Euler’s dynamics laws for mechanical systems, we use Kirchoff’s Laws
to derive the models for electrical circuits.
Kirchoff’s Current Law (KCL) states that the sum of currents flowing into any circuit node is
zero.
Kirchoff’s Voltage Law (KVL) states that the sum of voltages around any circuit loop is zero.
n m
i (t) 0
j 1
j v (t) 0
k 1
j
In some electrical circuit modeling you may encounter two more variable types, electrical charge
and electrical flux. As shown in Dr. Bob’s on-line Atlas of Models and Transfer Functions,
www.ohio.edu/people/williar4/html/PDF/ModelTFAtlas.pdf
electrical charge q(t) is the time integral of current i(t) and electrical flux (t) is the time integral of
voltage v(t).
q (t ) i (t ) dt (t ) v (t ) dt
When early electricity experimenters and early electrical engineer pioneers established the
convention for the flow of electrons composing current, they got it backwards. This error persists to this
day, because it really doesn’t matter in the equations.
xkcd.com
239
Current-Driven Parallel R-L-C Electrical Circuit Model
The system diagram for the current-driven parallel R-L-C electrical circuit is shown below. The
input is current i(t), and the output is voltage v(t). The voltage across the current generator, the resistor
R, the inductor L, and the capacitor C are all identical, v(t).
+
i(t) iR(t) R iL(t) L iC (t) C v(t)
-
Using Kirchoff’s Current Law (the sum of currents into any circuit node is zero), we have:
i(t ) iR (t ) iL (t ) iC (t ) 0
iC (t ) iR (t ) iL (t ) i (t )
Substituting the current through each element as a function of voltage v(t) from the previous table, we
obtain the model for the current-driven parallel R-L-C circuit.
dv(t ) 1 1
C v(t ) v(t )dt i(t )
dt R L
dv(t ) 1 1
C v(t ) v(t )dt i(t )
dt R L
Compare the mechanical systems ODEs to the parallel R-L-C circuit model (repeated above).
Translational Rotational
Energy Role Electrical
Mechanical Mechanical
dv(t ) d (t ) dv(t )
store energy f S (t ) m S (t ) J iS (t ) C
dt dt dt
1
dissipate energy f D (t ) cv(t ) D (t ) cR (t ) iD (t ) v(t )
R
1
fO (t ) k v(t )dt O (t ) kR (t )dt
L
oscillate energy iO (t ) v(t )dt
Force-Current Analogy
inertia m J C
damping c cR 1/R
stiffness k kR 1/L
Caution
The Force-Voltage Analogy, presented next, based on the voltage-driven series R-L-C circuit
model, is different.
241
242
Voltage-Driven Series R-L-C Electrical Circuit Model
The system diagram for the voltage-driven series R-L-C electrical circuit is shown below. The
input is voltage v(t), and the output is current i(t). The current i(t) flowing through all elements is the
same. The input voltage source v(t) increases as shown negative to positive; the voltage drops across
each of the circuit elements, positive to negative in each case.
+
v(t) i(t) L
-
Using Kirchoff’s Voltage Law (the sum of voltages around the circuit loop is zero), we have:
v(t ) vR (t ) vL (t ) vC (t ) 0
vL (t ) vR (t ) vC (t ) v(t )
Substituting the voltage drop across each element as a function of current i(t) from the first table in this
section, we obtain the model for the voltage-driven series R-L-C circuit.
di(t ) 1
L Ri(t ) i(t )dt v(t )
dt C
We can make a different analogy between the mechanical and electrical modeling worlds by
comparing the translational (and rotational) mechanical system with the voltage-driven series R-L-C
circuit model. This is called the force-voltage analogy.
We again use the translational and rotational mechanical system models, rewritten in terms of
velocity v(t) instead of displacement x(t) and rewritten in terms of angular velocity (t) instead of
angular displacement (t).
di(t ) 1
L Ri(t ) i(t )dt v(t )
dt C
Force-Voltage Analogy
inertia m J L
damping c cR R
stiffness k kR 1/C
244
Subset Electrical Circuit Models
This subsection presents subset models derived from the standard current-driven parallel R-L-C
electrical circuit and the voltage-driven series R-L-C electrical circuit models.
dv(t ) 1 1 di(t ) 1
C v(t ) v(t )dt i(t ) L Ri(t ) i(t )dt v(t )
dt R L dt C
1 1 di (t )
v(t ) v(t )dt i(t ) L Ri (t ) v(t )
R L dt
dv(t ) 1 1
C
C v(t ) i(t ) Ri(t ) i(t )dt v(t )
dt R
dv(t ) 1 di(t ) 1
v(t )dt i(t )
dt C
C L i(t )dt v(t )
dt L
245
Parallel R-L-C models based on flux (t) and series R-L-C models based on charge q(t)
Recall that, in an electrical circuit, electrical flux (t) is the time integral of voltage v(t) and
electrical charge q(t) is the time integral of current i(t). The subset models for the current-driven parallel
R-L-C electrical circuit and voltage-driven series R-L-C electrical circuit can thus be rewritten as
follows, using flux (t) as the output variable in place of voltage v(t) and using charge q(t) as the output
variable in place of current i(t), respectively. Here are the flux/voltage and charge/current substitution
relationships.
(t ) v(t )dt q (t ) i (t ) dt
(t ) v(t ) q (t ) i (t )
dv(t ) di (t )
(t ) q(t )
dt dt
1 1 1
C(t ) (t ) (t ) i(t ) Lq(t ) Rq (t ) q(t ) v(t )
R L C
1 1
(t ) (t ) i(t ) Lq(t) Rq(t) v(t)
R L
1 1
C(t ) (t ) i(t ) Rq (t ) q(t ) v(t )
R C
1 1
C(t ) (t ) i(t ) Lq(t ) q (t ) v(t )
L C
In the Force-Current Analogy the new output variable would be flux (t) instead of voltage v(t)
and in the Force-Voltage Analogy the new output variable would be charge q(t) instead of current i(t).
The rest of these analogies are unchanged.
246
y1(t) y2(t)
u1(t) u2(t)
k1 k2
m1 m2
c1 c2
u1(t) u2(t)
y1 (t ) c1 c2 y1 (t ) k1 k2 y1 (t ) c2 y 2 (t ) k2 y2 (t ) u1 (t )
m1
y2 (t ) c2 y 2 (t ) k2 y2 (t ) c2 y1 (t ) k2 y1 (t ) u2 (t )
m2
m1 y1 t c1 c2
0 c2 y1 t k1 k 2 k 2 y1 t u1 t
0
y2 t c2
m2 c2 y 2 t k 2
k 2 y2 t u2 t
247
2. Automotive Shock Model
This is a ¼ car suspension model. M2 is a point mass representing one-fourth of the car mass
(excluding wheels/tires). Spring constant k2 and viscous dashpot coefficient c2 represent the shock
spring and damping fluid friction dissipation elements. Point mass m1 represents one wheel/tire mass
and spring constant k1 and viscous dashpot coefficient c1 represent the spring effect and energy
dissipation quality of the tire, respectively. The input is the road displacement z(t), and there are two
displacement outputs y1(t) and y2(t). The output y2(t) relates to passenger comfort. Since this is a
passive system, there are no force inputs.
y2(t)
m2
k2 c2
y1(t)
m1
z(t) k1 c1
y1 (t ) c1 c2 y1 (t ) k1 k2 y1 (t ) c2 y 2 (t ) k2 y2 (t ) c1 z (t ) k1 z (t )
m1
y2 (t ) c2 y 2 (t ) k2 y2 (t ) c2 y1 (t ) k2 y1 (t ) 0
m2
m1 y1 t c1 c2
0 c2 y1 t k1 k 2 k 2 y1 t c1 z (t ) k1 z (t )
0 y2 t c2
m2 c2 y 2 t k 2
k 2 y2 t 0
248
3. Another Two-Mass Vibrational System
y1(t) y2(t)
f(t)
k1 k2 c
m1 m2
y1 (t ) k1 y1 (t ) k2[ y2 (t ) y1 (t )] f (t )
m1
y2 (t ) cy2 (t ) k2[ y2 (t ) y1 (t )] 0
m2
m1 0 y1 t 0 0 y1 t k1 k 2 k 2 y1 t f t
0 m2
y2 t 0 c y 2 t k 2
k 2 y2 t 0
249
4. Inverted Pendulum
m2
Y
g (t)
L
X
f(t)
m1
w(t)
m1 m2 m2 L w
(t ) 0 0 w(t ) f (t )
m L
2 m2 L (t ) 0 m2 gL (t ) 0
2
250
5. Ball and Beam
)
p(t
(t)
(t)
Jb
p (t ) m g sin (t ) m p (t ) (t ) 2 0
r 2 m
m p (t ) J J b (t ) 2 m p (t ) p (t ) (t ) m g p (t ) cos (t ) (t )
2
Jb
r 2 m p (t ) m g (t ) 0
m p (t ) 2 J J b (t ) m g p (t ) (t )
Jb r 2 m 0
p (t ) 0 mg p (t ) 0
0 m p (t ) J J b (t ) mg
2
0 (t ) (t )
251
6. Proof-Mass Actuator
q(t)
M
n(t)
k f(t)
e
J
(t) m
( M m)q(t ) kq (t ) me(t ) 0
( J me 2 )(t ) meq(t ) n(t )
M m me q(t ) k 0 q(t ) 0
me 2 (t ) n(t )
J me (t ) 0 0
252
7. Two Masses suspended via Tensed Strings
m1 m2
L L L
T T T
x1(t) T
x2(t)
L L L
2T T
x1 (t )
m1 x1 (t ) x2 (t ) 0
L L
2T T
x2 (t )
m2 x2 (t ) x1 (t ) 0
L L
m1 x1 t 2T L T L x1 t
0 0
0
x2 t T L 2T L x2 t
m2 0
253
8. Double Pendulum
L1
g
1 (t)
m1
L2
2 (t)
m2
If m1 = m2 and L1 = L2
2 1 1 (t ) 2 g L 0 1 (t ) 0
1 1 0
g L 2 (t ) 0
2 (t )
254
9. Sprung Double Pendulum
L0 L0
k
L L
1 (t) 2 (t)
m m
If m1 = m2 and L1 = L2
g k L0 2 k L0
2
1 0 1 (t ) L m L m L 1 (t ) 0
0 1 2
2 (t ) k L0 g k L0 2 (t ) 0
2
m L L m L
255
10. Doubly-sprung mass/cylinder
r (t)
m
k
k
x(t)
x (t ) k1 x (t ) k1r (t ) 0
m1
m2 r 2
(t ) ( k1 k 2 ) r 2 (t ) k1rx (t ) 0
2
k
r (t)
m k
r (t)
m k
m1r12
1 (t ) (k1 k2 )r121 (t ) k2 r1r2 2 (t ) 0
2
m2 r22
2 (t ) (k2 k3 )r22 2 (t ) k2 r1r21 (t ) 0
2
m1r12
0
1 (t ) ( k1 k2 ) r12 k2 r1r2 1 (t ) 0
2
m2 r22 2 (t ) k2 r1r2 ( k2 k3 ) r22 2 (t ) 0
0 2
4k 2k
m
1 0
1 (t ) m 1 (t ) 0
0 1
2 (t ) 2k 4k 2 (t ) 0
m m
257
12. Sprung Cart Pendulum
x(t)
k k
m1
(t)
m2
m1 m2 m2 L
x(t ) 2k 0 x(t ) 0
mL 2
2 m2 L (t ) 0 m2 Lg (t ) 0
258
13. Translating rotating rigid body
This is a 2-dof system, still constrained to vibrate in the vertical direction, with small-angle
rotational vibrations as well.
m, J
(t)
x(t)
L1 L2
k1 k2
x(t ) k1 k2
m 0 k1 L1 k2 L2 x(t ) 0
0 J (t ) k L k L
k1 L12 k2 L22 (t ) 0
1 1 2 2
259
14. Doubly sprung cylinder mass
x2(t)
x1(t)
k2 r (t)
k1
m2
m1
m2 m
(m1 x1 (t ) 2
) x2 (t ) (k1 k2 ) x1 (t ) k2 x2 (t ) 0
2 2
m 3m
2 x1 (t ) 2 x2 (t ) k2 x1 (t ) k2 x2 (t ) 0
2 2
m2 m2
m1 2
2
x1 (t ) k1 k2 k2 x1 (t ) 0
m2 x2 (t ) k2
3m2 k2 x2 (t ) 0
2 2
260
15. Cylinder rolling in a sprung big cylinder
r1 (t)
r2
m2
x(t) m1
( m1 m2 )
x (t ) kx (t ) 0
3
(r1 r2 )(t ) g (t ) 0
2
261
16. Dual torsional disks coupled by a spring
a r2 2(t) a r1 1(t)
m2 k R2 m1 k R1
m1r12
1 (t ) (k R1 ka 2 )1 (t ) ka 2 2 (t ) 0
2
m2 r22
2 (t ) (k R 2 ka 2 ) 2 (t ) ka 21 (t ) 0
2
m1r12
0
1 (t ) k R1 ka 2 ka 2 1 (t ) 0
2
m2 r22 2 (t ) ka k R 2 ka 2 2 (t ) 0
2
0 2
262
17. Dual point masses
f(t)
x2(t) x1(t)
m2 m1
L L
k2 c k1
x1 (t ) (k1 k2 ) x1 (t ) 2k2 x2 (t ) f (t )
m1
x2 (t ) cx2 (t ) 4k2 x2 (t ) 2k2 x1 (t ) 0
m2
x1 (t ) 0 0 x1 (t ) k1 k2
m1 0 2k2 x1 (t ) f (t )
0 m
2 x2 (t ) 0 c x2 (t ) 2k2 4k2 x2 (t ) 0
263
18. Rotating sprung point masses
k
(t)
m m
L L
(t) k
m
x(t)
m1 m2 m3
c1 c2 c3 c4
m1 0 0 y1 (t ) c1 c2 c2 0 y1 (t ) k1 k 2 k2 0 y1 (t ) u1 (t )
0 m2 0 y2 (t ) c2 c2 c3 c3 y 2 (t ) k 2 k 2 k3 k3 y2 (t ) u2 (t )
0 0 m3
y3 (t ) 0 c3 c3 c4 y 3 (t ) 0 k3 k3 k 4 y3 (t ) u3 (t )
265
20. Three Masses suspended via Tensed Strings
m1 m2 m3
L L L L
T T T T
T T
x2(t)
x1(t) x3(t)
L L L L
2T T
x1 (t )
m1 x1 (t ) x2 (t ) 0
L L
2T T T
x2 (t )
m2 x2 (t ) x1 (t ) x3 (t ) 0
L L L
2T T
x3 (t )
m3 x3 (t ) x2 (t ) 0
L L
m1 0 0 x1 t 2T L T L 0 x1 t 0
0
m2 0 x2 t T L 2T L T L x2 t 0
0 0 m3
x3 t 0 T L 2T L x3 t 0
266
21. Triple Pendulum
L1
g
1 (t)
m1
L2
2 (t)
m2
L3
3 (t)
m3
Let m1 = m2 = m3 and L1 = L2 = L3
3g
0 0
3 2 1 1 (t ) L 1 (t ) 0
2 2 1 (t ) 0 2g
0 2 (t ) 0
2 L
1 1 1 3 (t ) (t ) 0
0 g 3
0
L
267
22. Sprung Triple Pendulum
L0 L0 L0
k k
L L L
m m m
where L0 = L / 2.
kL2 kL2
mL21 (t ) (mgL )1 (t ) 2 (t ) 0
4 4
kL2 kL2 kL2
mL22 (t ) ( mgL ) 2 (t ) 1 (t ) 3 (t ) 0
2 4 4
2 kL2 kL2
mL 3 (t ) (mgL ) 3 (t ) 2 (t ) 0
4 4
g k k
0
1 0 0 1 (t ) L 4m 4m
1 (t ) 0
0 1 0 (t ) k g k
k
2 (t ) 0
2 4m L 2m
4m
0 0 1 3 (t ) (t ) 0
0 k g k 3
4m L 4m
268
23. Sprung Cylinder with Point Masses
k
m
r
r/3 (t)
x(t)
k
m
m
x(t)
x(t)
Considering all multi-dof vibrational models we have derived, some general comments can be
made about the models.
1. When gathering the unknown i and its time derivatives on one side of the ODE i, there will
never be a change in sign amongst these terms; generally they will all be positive. Generally
the unknown j appears on the ODE i with a negative sign.
2. The effect of gravity always cancels out. That is, any mg static weight terms will load the
springs with static deflections and these terms always subtract out of the final ODE.
3. As always, the resulting units must be consistent. This is a good check to perform on your
resulting model.
a. The matrix multiplying the vector of second-order derivatives is called the mass
matrix. It is square, symmetric, and positive-definite.
b. The matrix multiplying the vector of first-order derivatives is called the damping
matrix. It is square and symmetric.
c. The matrix multiplying the vector of zeroth-order unknowns is called the stiffness
matrix. It is square and symmetric.