Sie sind auf Seite 1von 24

FACULTY OF MECHANICAL ENGINEERING

UNIVERSITI TEKNOLOGI MARA (UiTM)

INDIVIDUAL ASSIGNMENT

MEC 628
MODAL TESTING AND ANALYSIS

NAME:
LUQMAAN HAKIEM BIN SULAIMAN (2015182365)

PROGRAM:
BACHELOR OF ENGINEERING (HONS) MECHANICAL (EM220)

GROUP:
EM2208E2

LECTURER:
DR. MUHAMAD NORHISHAM BIN ABDUL RANI
Table of Contents
1.0 Introduction ....................................................................................................................................................................... 2
2.0 MSC Patran ......................................................................................................................................................................... 3
3.0 MATLAB ............................................................................................................................................................................... 6
4.0 Validation ......................................................................................................................................................................... 16
5.0 Conclusion ....................................................................................................................................................................... 18
6.0 References ....................................................................................................................................................................... 19
7.0 Appendix A ...................................................................................................................................................................... 20

1
1.0 Introduction

The question is attempted by utilising two softwares:

1. MSC Patran 2017.0.2 Student Edition


2. MATLAB R2017a

The system is considered to be a 5 degree-of-freedom system with two free ends at masses m1 and
m5.

k1 (m1–m2) k2 (m2–m3) k3 (m3–m4) k4 (m4–m5)


m1 m2 m3 m4 m5

Figure 1.1: The free body diagram of the system

The mass and stiffness matrices are setup as follows;

𝑚1 0 0 0 0
0 𝑚2 0 0 0
𝑀= 0 0 𝑚3 0 0
0 0 0 𝑚4 0
[ 0 0 0 0 𝑚5]

𝑘1 −𝑘1 0 0 0
−𝑘1 𝑘1 + 𝑘2 −𝑘2 0 0
𝐾= 0 −𝑘2 𝑘2 + 𝑘3 −𝑘3 0
0 0 −𝑘3 𝑘3 + 𝑘4 −𝑘4
[ 0 0 0 −𝑘4 𝑘4 ]

2
2.0 MSC Patran

The main setup for the simulation was made using MSC Patran software according to the following
steps;

1. Create new file


2. Go to Home tab > Misc. section
3. Click Node size & Label Control buttons
4. Go to the additional bar > Select Node & Element
5. Go to Meshing tab > RHS Window
6. Action: Create, Object: Node
7. Go to Node Location List
8. Insert coordinates [0 0 0] [10 0 0] [20 0 0] [30 0 0] [40 0 0] & Apply
9. Action: Create, Object: Element, Shape: Bar
10. Let auto execute ON
11. Click initial node and then the next node. Repeat for all pair of nodes.
12. Action: Create, Object: Element, Shape: Point
13. Let auto execute ON
14. Click nodes consecutively, 1 until end.
15. Go to Properties tab
16. Action: Create, Object: 1D, Type: Spring
17. Property Set Name: spring_k1
18. Click Input Properties
19. Spring Constant: 50000, Dof at Node 1 = Dof at Node 2 = UX, click OK
20. Click Select Application Region
21. Select Beam Element on annexe side bar
22. Click on the bar which represents the spring with respective spring constant
23. Repeat steps 16-22 for all springs according to their spring constants
24. Action: Create, Object: 0D, Type: Mass
25. Property Set Name: mass_m1
26. Options: Lumped
27. Click Input Properties
28. Mass: 4, click OK
29. Click Select Application Region
30. Select Point Element on annexe side bar
31. Click on point elements which represents the mass with respective mass value
32. Repeat steps 24-31 for all masses according to their mass values
33. Go to Analysis tab > Analyze section
34. Click Entire Model
35. Go to RHS Window > Click Solution Type > Select Normal modes, Click OK
36. Click Subcases > Subcase Parameters
37. Number of Desired Roots: 5, click OK > Apply > Apply
38. Go to Access Results section > XDB
39. Go to RHS Window > Click Select Results File
40. Select the respective filename.xdb, click Apply
41. Go to Results tab
42. Identify the Freq. which is the natural frequency for each mode

3
Figure 2.1: The end graphical representation of the system to be analysed. The 5 masses are lumped
at the orange/red points, while the four springs are represented by the green lines in between.

For an initial attempt, the following set of values was used;

Parameter Value
m1 4 kg (constant)
m2 3.2 kg (constant)
m3 3.2 kg (constant)
m4 4 kg
m5 4 kg
k1 50,000 N/m (constant)
k2 50,000 N/m (constant)
k3 50,000 N/m
k4 50,000 N/m
Table 2.1: Parameters used in MSC Patran for the first attempt

4
All natural frequencies are compared with the given values in the question. The error of each
frequency are calculated and summed for the total error. The set of values above yields a total error
of 84.33%.

Natural frequencies (Hz) Reference (Hz) Error (%)


RBM* RBM -
11.156 14.1 20.8794
21.933 25.9 15.3166
29.887 39.2 23.7577
36.755 48.6 24.3724
Total 84.3261
Table 2.2: The results from initial parameters. *RBM = Rigid body mode

Other values for the parameter are included in Appendix A which shows a method of only
increasing both stiffness value of k3 and k4 while maintaining both masses m4 and m5 as 4 kg. It
was discovered that the lowest total error found was 28.2023% for stiffness values of 120,000 N/m.
At this stage, it was concluded that repeating the steps manually in MSC Patran was too time
consuming for a try-and-error approach. In order to expedite the experimental process to obtain
the minimum error, MATLAB software was utilised as explained in the next section.

5
3.0 MATLAB

Since the final objective of the question is to reduce the error to as low as possible, a MATLAB
program was coded to directly relate the parameters involved with the total error. A portion of the
coding for finding the natural frequency involving matrix operations, Section B of the coding below,
was adapted from Dr. L. Chang [1].

% ***Section A % Sort eigen-values and eigen-vectors


m1=4; % constant [D_sorted, ind] =
m2=3.2; % constant sort(diag(D),'ascend');
m3=3.2; % constant V_sorted = V(:,ind);
m4=1;
m5=1; %Obtain natural frequencies and mode
shapes
k1=50000; % constant w1 = sqrt(D_sorted(1))/(2*pi)
k2=50000; % constant w2 = sqrt(D_sorted(2))/(2*pi)
k3=50000; % constant w3 = sqrt(D_sorted(3))/(2*pi)
k4=50000; w4 = sqrt(D_sorted(4))/(2*pi)
w5 = sqrt(D_sorted(5))/(2*pi)
% reference frequencies for each mode
w2_ref=14.1; % ***Section C
w3_ref=25.9; e2=abs(w2-w2_ref)*100/w2_ref
w4_ref=39.2; e3=abs(w3-w3_ref)*100/w3_ref
w5_ref=48.6; e4=abs(w4-w4_ref)*100/w4_ref
e5=abs(w5-w5_ref)*100/w5_ref
figure (1)
while m4<5 eT=e2+e3+e4+e5

% ***Section B plot3 (m4,m5,eT,'^b','MarkerSize',3)


A=[k1/m1 -k1/m1 0 0 0; xlabel('mass 4')
-k1/m2 (k1+k2)/m2 -k2/m2 0 0; ylabel('mass 5')
0 -k2/m3 (k2+k3)/m3 -k3/m3 0; zlabel('Total error')
0 0 -k3/m4 (k3+k4)/m4 -k4/m4;
0 0 0 -k4/m5 k4/m5]; grid on
[V,D]=eig(A); hold on

V; m4=m4+0.05
D; end

Section A contains the basic parameters involved including both the constant and variable masses
and spring constants, and the reference frequencies for each mode except the Rigid Body Mode. A
while loop for m4 was utilized to continuously plot points in the same graph in order to analyze the
pattern of the results and easily identify the areas with the lowest error.

6
Section B contains the coding for finding the natural frequencies by setting up the mass and
stiffness matrices. It was modified for the specific case of this question, a 5 degree-of-freedom
system in the linear translational direction.

Section C is the part for directly calculating the error of each natural frequency and its
corresponding total error. A 3-dimensional plot was used to more easily understand the
relationship between multiple parameters at the same time in one plot, which are m4, m5 and the
total error as three different axes, k3 as different marker color and types, and k4 in a separate
figure. The basic increment and range boundaries of values for each parameter are set as follows;

Parameter Increment Range boundaries


k3 (N/m) 10,000 50,000 to 100,000
k4 (N/m) 10,000 50,000 to 100,000
m4 (kg) 0.1 1 to 3
m5 (kg) 0.1 1 to 3
Table 3.1: The basic increment and range boundaries of values for each parameter

The increment stated above is used according to the judgment to find the lowest total error. As a
low error is discovered, the increment is made smaller by half to precisely identify the most
accurate parameter for the lowest total error.

As for the range boundaries, each increment is executed in accordance to the trend of finding the
lowest total error in the plot until a satisfactory guess that the next trend should continuously
increase is achieved.

The process of plotting the graph based on the coding above is as follows;
1. Run the current coding using given values
2. Right-click on m5’s value to use Increment Value and Run Section with 0.1 increments.
3. Click for a range around from 1 to 3
4. Increase k3 by 10,000 N/m
5. Change the color and/or marker type in Section C according to Table 3.2.
6. Repeat steps 1 to 5 for a k3 of up to 90,000 N/m

7
Marker color and type k3 parameter used

Magenta dot 10,000 N/m


Cyan dot 20,000 N/m
Red dot 30,000 N/m
Green dot 40,000 N/m
Blue dot 50,000 N/m
Black dot 60,000 N/m
Magenta ‘x’ 70,000 N/m
Table 3.2: The different marker colors and types used for a corresponding value of k3

Figure 3.1 shows the overall relationship between all 4 parameters for a constant value of k4. It
provides a basis for estimating the trend of the data in order to set the limits and to be able to
conjecture the next possible best parameter to modify, and whether to increase or decrease it.
Figure 3.2 shows that the total error reduces quite sharply as mass 4 increases from 1 kg until it
approaches around 1.5 kg where the error starts to rise in a logarithmic-like trend. Figure 3.3
shows that the total error reduces very slightly as mass 5 increases from 1 kg until around 1.5 kg to
1.7 kg.

8
Figure 3.1: The default view of the 3-D plot for k4 = 50,000 N/m showing all axes involved

9
Figure 3.2: The view of the 3-D plot for k4 = 50,000 N/m under the command of view([0 0]),
showing clearly the mass 4 and total error relationship

10
Figure 3.3: The view of the 3-D plot for k4 = 50,000 N/m under the command of view([-90 0]),
showing clearly the mass 5 and total error relationship

11
Various values were tested and the lowest total error was identified according to the view angle as
that of Figures 3.2 and 3.3. The values are sorted in Table 3.3 in an ascending order of k3 value, not
according to the sequence of attempt. As it was eventually found that a value of 47,000 N/m for k3
seemed to produce the lowest total error, the increments of k3 and k4 were reduced smaller to
increase the precision.

No k3 k4 Minimum total No k3 k4 Minimum total


(N/m) (N/m) error (%) (N/m) (N/m) error (%)

1 46,000 58,000 10.35 20 48,000 64,000 10.29


2 46,000 59,000 10.08 21 49,000 62,000 10.36
3 46,000 60,000 10.11 22 50,000 60,000 10.97
4 46,000 62,000 10.21 23 50,000 62,000 10.04
5 47,000 60,000 10.40 24 50,000 64,000 10.62
6 47,000 61,000 10.00 25 66,000 52,500 11.99
7 47,000 61,250 9.94 26 66,000 55,000 11.95
8 47,000 61,375 9.86 27 66,000 57,500 12.34
9 47,000 61,500 9.78 28 68,000 50,000 12.78
10 47,000 61,750 9.82 29 68,000 52,500 11.84
11 47,000 62,000 9.99 30 68,000 55,000 12.19
12 47,000 63,000 10.22 31 70,000 50,000 12.51
13 47,000 64,000 10.49 32 72,500 50,000 12.52
14 47,500 60,000 10.54 33 72,500 52,500 12.82
15 47,500 62,000 10.21 34 75,000 50,000 12.97
16 47,500 64,000 10.47 35 90,000 60,000 16.89
17 48,000 60,000 10.54 36 90,000 70,000 15.41
18 48,000 61,000 10.48 37 90,000 80,000 16.52
19 48,000 62,000 10.04
Table 3.3: The obtained minimum total error for various pairs of k3 and k4 values

12
After sweeping the ranges of values for both m4 and m5 for each pair of k3 and k4 in Table 3.3, the
lowest error discovered is 9.78% for k3 = 47,000 N/m, k4 = 61,500 N/m, m4 = 2.1 kg and m5 = 1.2
kg. The pattern of the trend for this particular set of stiffness values exhibit a similarity to that of
the preliminary results. The lowest total error value is noted for all Figures 3.4, 3.5 and 3.6.

Figure 3.4: The default view of the 3-D plot for only k3 = 47,000 N/m and k4 = 61,500 N/m showing
all axes involved

13
Figure 3.5: The view of the 3-D plot for only k3 = 47,000 N/m and k4 = 61,500 N/m under the
command of view([0 0]), showing clearly the mass 4 and total error relationship

14
Figure 3.6: The view of the 3-D plot for only k3 = 47,000 N/m and k4 = 61,500 N/m under the
command of view([-90 0]), showing clearly the mass 5 and total error relationship

15
4.0 Validation
A validation process of the results obtained from MATLAB is done by using the selected parameters
in MSC Patran again. The following results are obtained;

Natural frequencies (Hz) Reference (Hz) Error (%)


RBM RBM -
14.1 14.1 0
25.912 25.9 0.046332
35.412 39.2 9.66327
48.566 48.6 0.06996
Total 9.779556
Table 4.1: The results obtained using MSC Patran for k3 = 47,000 N/m, k4 = 61,500 N/m, m4 = 2.1
kg and m5 = 1.2 kg

The total error calculated using the values displayed in MSC Patran differ very slightly from that of
MATLAB. A rounding-off to the nearest 1/100th of the error produces a same value of 9.78%. The
five mode shapes are identified from the .fo6 results file and tabulated in Table 4.2. Point ID
represents the each mass with its respective numbering. The points for mode shape 1 all have the
same value of 2.701716E-01 since it is of the rigid body mode, which means that all points move as
a rigid body i.e at the same pace and direction.

Point Mode shape 1 Mode shape 2 Mode shape 3 Mode shape 4 Mode shape 5
1 2.701716E-01 3.259106E-01 -2.297822E-01 1.340313E-01 5.028788E-0
2 2.701716E-01 1.212612E-01 2.574878E-01 -3.968012E-01 -3.243225E-02
3 2.701716E-01 -1.443032E-01 3.079405E-01 3.295946E-01 1.233853E-01
4 2.701716E-01 3.497015E-01 -1.941398E-01 -8.591143E-03 -4.930949E-01
5 2.701716E-01 4.129458E-01 -4.021231E-01 -2.525191E-01 6.036121E-01
Table 4.2: The five mode shapes of the system obtained from the .fo6 file

16
Sketches of the five mode shapes are as follows;

Figure 4.1: Mode shape 1

Figure 4.1: Mode shape 2

Figure 4.1: Mode shape 3

Figure 4.1: Mode shape 4

Figure 4.1: Mode shape 5

17
5.0 Conclusion

In conclusion, it was found that the best set of parameters within the tested range is k3 = 47,000
N/m, k4 = 61,500 N/m, m4 = 2.1 kg and m5 = 1.2 kg producing a total error of only 9.78%. The
value was validated using both MSC Patran 2017.0.2 Student Edition and MATLAB R2017a.

18
6.0 References

Dr. L. Chang (2018). An example of programming in MATLAB to obtain natural frequencies and
mode shapes of MDOF systems. Department of Mechanical and Nuclear Engineering, The
Pennsylvania State University. Retrieved from
http://www.mne.psu.edu/chang/me370/eigen_matlab.pdf

19
7.0 Appendix A

Table A1: Results of the initial attempts using MSC Patran by increasing both k3 and k4
simultaneously
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 11.156 14.1 20.8794
3 3.2 50,000 21.933 25.9 15.3166
4 4 50,000 29.887 39.2 23.7577
5 4 36.755 48.6 24.3724
84.32612
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 11.632 14.1 17.5035
3 3.2 60,000 23.06 25.9 10.9653
4 4 60,000 31.421 39.2 19.8444
5 4 38.268 48.6 21.2593
69.57244
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 11.991 14.1 14.9574
3 3.2 70,000 24.091 25.9 6.98456
4 4 70,000 32.465 39.2 17.1811
5 4 40.122 48.6 17.4444
56.56757
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.269 14.1 12.9858
3 3.2 80,000 25.034 25.9 3.34363
4 4 80,000 33.227 39.2 15.2372
5 4 42.141 48.6 13.2901
44.85681

20
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.488 14.1 11.4326
3 3.2 90,000 25.884 25.9 0.06178
4 4 90,000 33.861 39.2 13.6199
5 4 44.2 48.6 9.0535
34.1678
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.581 14.1 10.773
3 3.2 95,000 26.275 25.9 1.447876
4 4 95,000 34.159 39.2 12.8597
5 4 45.225 48.6 6.94444
32.02506
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.665 14.1 10.1773
3 3.2 100,000 26.642 25.9 2.864865
4 4 100,000 34.452 39.2 12.1122
5 4 46.241 48.6 4.85391
30.00832
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.932 14.1 8.28369
3 3.2 120,000 27.88 25.9 7.644788
4 4 120,000 35.655 39.2 9.04337
5 4 50.17 48.6 3.230453
28.2023
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 13.035 14.1 7.55319
3 3.2 130,000 28.37 25.9 9.53668

21
4 4 130,000 36.297 39.2 7.40561
5 4 52.062 48.6 7.123457
31.61894

Table A2: Results of the initial attempts using MSC Patran by varying m5 for selected spring
constants
Natural
Mass Spring Reference
No. frequencies Error (%)
(kg) (N/m) (Hz)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.593 14.1 10.6879
3 3.2 120,000 27.144 25.9 4.803089
4 4 120,000 34.873 39.2 11.0383
5 5 49.701 48.6 2.265432
28.79473
Natural
Mass Spring Reference
No. frequencies Error (%)
(kg) (N/m) (Hz)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.705 14.1 9.89362
3 3.2 130,000 27.682 25.9 6.880309
4 4 130,000 35.402 39.2 9.68878
5 5 51.547 48.6 6.063786
32.52649

Table A3: Results of the initial attempts using MSC Patran by varying m4 and m5 for selected spring
constants
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.725 14.1 9.75177
3 3.2 120,000 27.88 25.9 7.644788
4 5 120,000 35.187 39.2 10.2372
5 4 47.458 48.6 2.34979
29.9836
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.826 14.1 9.03546

22
3 3.2 130,000 28.369 25.9 9.532819
4 5 130,000 35.876 39.2 8.47959
5 4 49.163 48.6 1.158436
28.20631
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 12.923 14.1 8.34752
3 3.2 130,000 29.054 25.9 12.17761
4 5 130,000 37.452 39.2 4.45918
5 3 48.109 48.6 1.01029
25.9946
Natural
Mass Spring
No. frequencies Reference (Hz) Error (%)
(kg) (N/m)
(Hz)
1 4 50,000 RBM RBM
2 3.2 50,000 13.234 14.1 6.14184
3 3.2 130,000 29.622 25.9 14.37066
4 5 130,000 40.476 39.2 3.255102
5 2 50.855 48.6 4.639918
28.40752

23

Das könnte Ihnen auch gefallen