Introduction
This laboratory investigation looks at the bending deflection of a circular frame under diagonal
tensile load. The finite element method is used to determine the deflection under the load
For a symmetric structure that is subject to a symmetric load, the frames deflection should also act
symmetrically for both vertical and horizontal directions. The finite element method can be used to
investigate the deflective behavior of the structure, but this process can be simplified by utilizing
axis of symmetry. Take a circle for example, which has two axis of symmetry; the vertical and the
horizontal, and can be simplified by observing the behavior of one quarter of the shape. As the load
is taken as symmetric, the simulation can be run for one quarter of the shape, and under one
quarter of the original load, and the deflection should be the same.
To determine the area inside a circle, a determinable shape like a square could be placed inside,
where the area of the shape gives an approximation to the area of the circle. As the number of
sides of the shape increases, the accuracy of the approximation increases as the space between the
sides of the internal shape and the curve decrease. To solve a circular frame problem, the loading
problem can be solved using a 2-Element idealization and a 3-element idealization, and then using
a relationship between the two to approximate the deflection value for the full curve.
RN =R E (1
R E=R3 +
K
)
N2
N 21
( R2R1 )
N 22N 21
Where RE is the exact value, RN corresponds to the value for N members. The second equation
outlines the relationship between the results at two different values for N, and then approximates
the RE value.
Experimental Arrangement
The experimental setup for this investigation consisted of the apparatus listed above, and allowed
the user to apply loads to a circular frame, which produced a deflection against the circular frame.
The deflection was measured by the digital gauge, so a table of load against deflection can be
constructed. Loads of 2kg were added from 0kg to 12kg, then back down to 0kg. From this, the
deflection against load can be plot on a graph and the constant /P can be obtained. This constant
gives us an experimental result for the behaviour for the circular frame.
To obtain a theoretical result to demonstrate the previously defined symmetrical relationship, the
finite element method can be used for a two beam and a three beam approximation. A stiffness
matrix [K] is produced using the beam properties, and is equated using the formula P=K, where
the ratio /P can be obtained by solving the matrix using gauss elimination.
For the beam, a value for EI was pre-determined as 111.3 Nm2, allowing the calculation for E and
other beam property values such as the second moment of area.
Results
0.005
f(x)
f(x) =
= 0x
0x
R
=
R = 1
1
0.004
0.003
Deflection /m
0.002
0.001
0.000
20
40
60
Load P / N
GRAPH 1 - P LOT
Up
Down
Avera
ge
OF
LOAD
D EFLECTION
/P
3.7403E05
3.70221E05
3.72125E05
FROM
Beam Description
Quantit Unit
y
s
Value
Width
m
0.0254
Height
m
0.00635
Ixx
m4
542.0E-12
0.0001612
Area
m2
9
Length
m
0.2334
E
Pa
205.4E+9
Pam
EI
4
111.3
Pam
EA
2
33.1E+6
G RAPH 1
80
100
120
140
EA/L
EI/L
EI/L2
EI/L3
Pam
Pam
3
Pam
2
Pam
141.89E+6
476.7884
2042.4723
8749.5693
FOR
2 B EAM
ANALYSIS
Beam Description
Quant Uni
ity
ts
Value
Width
m
0.0254
Height
m
0.00635
542.0EIxx
m4
12
0.000161
Area
m2
29
Length
m
0.1579
E
Pa
205.4E+9
Pam
EI
4
111.3
Pam
EA
2
33.1E+6
209.80E+
EA/L
Pam
6
Pam
EI/L
3
704.9675
Pam 4465.221
EI/L2
2
8
28282.44
EI/L3
Pam
61
TABLE 3 - B EAM
Graph
2
Beam
3
Beam
Rexp
Rexac
t
PROPERTIES FOR
/P
(m/N)
37.2E-6
33.9E-6
35.9E-6
37.6E-6
37.9E-6
TABLE 4 - R ESULTS
Graph
Rexp
FOR DELTA /P
/P
%
(m/N)
difference
37.21E6
-1.88%
37.56E-0.96%
BEAM ANALYSIS
Rexac
t
6
37.93E6
Discussion
From the experimental data, a plot of deflection against load was made, separating the values for
the loading and unloading. This is because during the load process, there might be some form of
deformation, which can be seen clearly for the last reading taken: 0kg -0.04mm, which shows an
element of deformation. This deformation affects the overall trend the data takes, so to reliably
obtain a relationship between the two, trend-lines were constructed for the two loading procedures,
and the average was taken of the two. The trend-line was set with a (x,y) intercept of 0. A value for
/P was obtained as 37.21E-06 N/m. The overall scatter for the data was low, so the experimental
approach towards obtaining these values were consistent in terms of reliability.
For the 2 beam analysis, a matrix of K was constructed in excel, using beam properties in the form
of the values; A, B, F, G, H, P, Q, which are all ratios between the angle of the member and the
elastic modulus E of the beam. A resultant matrix was created by combining the individual values
for beam (a) and beam (b) together, which produced a 9x9 matrix. As points 1 and 3 were fixed;
x1, y3, 1 and 3 were cancelled out as they were zero, reducing the matrix to a 5x5. The
standard gauss elimination was used and the elimination took 10 steps to complete. The initial and
final steps are shown in the appendix, but the other 8 steps were omitted from the report and can
be observed in the excel file due to the impact of page length. A value for x3 was obtained, and
used to obtain a theoretical value for /P, calculated as 33.9E-06 m/N. For the two beam analysis,
the load was applied to node 3 in the negative x direction (load of -P).
Following this, a 3 beam analysis was also used in the same method as for the two beam, but
produced a 12x12 K matrix, which was then reduced to an 8x8 matrix, due to the fixed values x1,
y4, 1 and 4. A Load of P was applied to node 4 in the x direction, and the value for x4 was
obtained, and then used to calculate a value for /P: 35.9E-06 m/M. To solve this larger matrix, the
Gaussian elimination method had to be done column by column, not row by row. There are two
worked answers for /P, but one is crossed out. The Gaussian elimination for the three beam
analysis consists of 23 steps. The other steps were omitted from the report and can be observed in
the excel file due to the impact of page length.
4
Rexp ( E06 )=35.94+ (35.9433.9 ) =37.6E-06 m/N
5
The above theoretical values equate using the above formula to give a value for R E. An additional
method to calculate RE was given as:
3
R Exact
r 2
=37.9E-06 m/ N
EI 4
In comparison with the values, a percentage difference table between the three values was made,
with RExact being used as the baseline value. The experimental data which used a graph to determine
the deflection constant was fairly accurate, and had a -1.88% percentage difference. 5% is the
threshold for whether a set of data can be considered reliable, and the values obtained were well
within the limit, showing that the graphical/experimental method can be a reliable approach
towards solving for the stiffness of a circular beam.
The matrix based FEA gave a much closer value with a -0.96% difference, and deviated by a
fraction of a mm/N, showing that the FEA using the two and three element comparison gives a good
approximation to the actual deflection property of the circular beam. It should also be noted that
the FEA method was more accurate than the graphical solution, but at the extra cost of calculation
time. The graphical approach took seconds to measure and calculate, while the FEA requires
multiple set-up matrices, which is more time-consuming, suggesting that for small-scale projects,
where % difference of 2% is not significant, the graphical approach may be better for resource
management, although a computer program can be written to optimise the calculation process and
automate it.
In terms of the recorded datas accuracy, the loads were all assumed to be equal to 2kg, and no
error was taken into account for the internal deviation the masses may have had, i.e. the mass
could have been 1.9kg, and although the individual error is small, those individual errors add up
and propagate through. The displacement was also only measured to 2dp, and a more precise and
reliable reading could be taken by apparatus with a higher degree of precision. As the output for
this investigation were values approximately equal to 40x10-6, small errors hold a greater value and
have a greater significance. An error of 2N is nothing to a 400,000 N load, but 2N to a 20N load is a
larger percentage.
Appendix
Recorded Data
Load
Displacement
k
g
0
2
4
6
8
10
12
10
8
6
4
2
0
mm
0.00
0.76
1.45
2.21
2.95
3.67
4.39
3.67
2.94
2.10
1.32
0.71
-0.04
0
19.62
39.24
58.86
78.48
98.1
117.72
98.1
78.48
58.86
39.24
19.62
0
TABLE 5 -
m
0.00000
0.00076
0.00145
0.00221
0.00295
0.00367
0.00439
0.00367
0.00294
0.00210
0.00132
0.00071
-0.00004
Member
Length (m)
Angle
(global)
Cos
Sin
b
0.2334
0.2334
22.5
0.9239
0.3827
67.5
0.3827
0.9239
TABLE 6 - L ENGTH
ANGLE
COS S IN
TABLE FOR
Stiffness Coefficients
a
b
A
1907.15
1907.15
B
953.58
953.58
F 121128207.20
20869295.35
ELEMENTS
G
H
P
Q
50129455.92
-4689.72
20869295.35
11321.99
50129455.92
-11321.99
121128207.20
4689.72
FOR
ELEMENTS
K11a
K12a
121128207
.20
50129455.
92
-4689.72
50129455.
92
-4689.72
20869295.
35
11321.99
11321.99
1907.15
121128207.
20
50129455.9
2
4689.72
K21a
121128207
.20
50129455.
92
-4689.72
4689.72
11321.9
9
953.58
K22a
50129455.
92
20869295.
35
11321.99
4689.72
121128207.
20
50129455.9
2
-11321.99
953.58
50129455.9
2
4689.72
20869295.3
5
-11321.99
K11b
4689.72
11321.9
9
1907.15
K12b
20869295.
35
50129455.
92
-11321.99
50129455.
92
-11321.99
121128207
.20
4689.72
4689.72
1907.15
20869295.3
5
50129455.9
2
11321.99
K21b
20869295.
35
50129455.
92
-11321.99
50129455.9
2
20869295.3
5
-11321.99
50129455.9
2
121128207.
20
-4689.72
11321.9
9
4689.72
953.58
K22b
50129455.
92
121128207
.20
4689.72
11321.99
20869295.3
5
50129455.9
2
11321.9
9
-4689.72
953.58
50129455.9
2
11321.99
121128207.
20
-4689.72
4689.72
1907.15
FOR
ELEMENT
MATRIX
20869295.
35
50129455.
92
20869295.
35
11321.99
0.00
TABLE 10 - K
50129455.
92
20869295.
35
14199750
2.55
10025891
1.85
10025891
1.85
14199750
2.55
-6632.27
20869295.
35
-6632.27
50129455.
92
MATRIX BEFORE
11321.
99
6632.2
7
6632.2
7
3814.3
1
11321.
99
20869295
.35
20869295.3
5
0.00
21583166.1
2
50129455.9
2
0.00
0.00
4696617.10
20563.9
2
43072.4
1
0.00
0.00
0.00
0.00
0.00
0.00
3393.56
0.00
MATRIX AFTER
Member
Length
(m)
Angle
(global)
Cos
Sin
dx2
0
= 0
11321.99
dy2
thet
a2
20869295.
35
dx3
11321.9
9
0.00
20869295.3
5
1658049.30 X
15999.87
29481.34
0.1579
0.1579
0.1579
15.0
0.9659
0.2588
45.0
0.7071
0.7071
75.0
0.2588
0.9659
TABLE FOR
Stiffness Coefficients
b
c
ELEMENT
dy1
G AUSSIAN ELIMINATION 2 ND
50129455.9
2
TABLE 11 - K
0.00
20869295.
35
50129455.
92 X
ELEMENTS
dy1
dx2
dy2
theta
2
dx3
= 0
0
P
A
B
F
G
H
P
Q
2819.87
1409.94
195767632.
43
52364839.9
1
-6934.11
14370505.9
2
25878.44
2819.87
1409.94
105069069.
17
104729679.
82
-18944.33
105069069.
17
18944.33
TABLE 13 - S TIFFNESS
COEFFICIENTS FOR
K11a
ELEMENTS
K12a
19576763
2.43
52364839.
91
52364839.
91
14370505.
92
-6934.11
25878.44
6934.1
1
25878.
44
2819.8
7
K21a
19576763
2.43
52364839.
91
52364839.
91
14370505.
92
6934.11
-25878.44
6934.1
1
25878.
44
1409.9
4
K22a
19576763
2.43
52364839.
91
52364839.
91
14370505.
92
-6934.11
25878.44
6934.1
1
25878.
44
1409.9
4
10506906
9.17
10472967
9.82
18944.
33
10472967
9.82
10506906
9.17
-18944.33
18944.33
K11b
19576763
2.43
52364839.
91
52364839.
91
14370505.
92
6934.11
-25878.44
6934.1
1
25878.
44
2819.8
7
K12b
18944.
33
2819.8
7
K21b
10506906
9.17
10472967
9.82
2819.87
1409.94
14370505.9
2
52364839.9
1
-25878.44
195767632.
43
6934.11
10506906
9.17
10472967
9.82
10472967
9.82
10506906
9.17
18944.33
-18944.33
18944.
33
18944.
33
1409.9
4
K22b
10472967
9.82
10506906
9.17
18944.
33
18944.
33
10506906
9.17
10472967
9.82
10472967
9.82
10506906
9.17
18944.
33
18944.
33
-18944.33
18944.33
1409.9
4
18944.33
K11c
-18944.33
2819.8
7
K12c
14370505.
92
52364839.
91
52364839.
91
19576763
2.43
-25878.44
6934.11
25878.
44
6934.1
1
2819.8
7
K21c
14370505.
92
52364839.
91
52364839.
91
19576763
2.43
25878.44
-6934.11
25878.
44
6934.1
1
1409.9
4
K22c
14370505.
92
52364839.
91
52364839.
91
19576763
2.43
-25878.44
6934.11
25878.
44
6934.1
1
1409.9
4
FOR THE
TABLE 15 - K
MATRIX BEFORE
TABLE 16 - K
MATRIX AFTER
14370505.
92
52364839.
91
52364839.
91
19576763
2.43
25878.44
-6934.11
MATRIX
3 RD ELEMENT
G AUSSIAN ELIMINATION 3 RD
ELEMENT
25878.
44
6934.1
1
2819.8
7