december 19 2008 - pbworkssumit-tripathi-projects.pbworks.com/f/fem_plate.pdfmae 529: modal analysis...
TRANSCRIPT
MAE 529:
Modal analysis of
Rectangular plate
with all four edges
simply supported
December 19
2008 Project Report
By Sumit Tripathi
Contents Objective: ................................................................................................................................................ 3
Model particulars: ................................................................................................................................... 3
(a) Plate specifications................................................................................................................... 3
(b) Elements used in analysis ......................................................................................................... 3
(c) Assumptions/Methods ............................................................................................................. 3
Shell 93 Element (K I R C H O F F – L O V E)............................................................................................... 4
(a) Thickness ratio = ¼, thickness=16m .......................................................................................... 4
Mode shapes with Shell93 .............................................................................................................. 8
Mode frequency calculations ......................................................................................................... 12
(b) Shell 93 Element (Thickness ratio = 1/16) ............................................................................... 13
Mesh geometry ............................................................................................................................. 13
All 6 mode shapes with finest mesh ............................................................................................... 15
(c) Shell 93 Element (Thickness ratio = 1/64) ............................................................................... 17
All 6 mode shapes with finest mesh for shell93, thickness ratio 1/64 ............................................. 18
Mode frequency calculations ......................................................................................................... 18
Comparison of FEM and analytical solution for the three thickness ratios. ......................................... 19
Shell63 (K I R C H O F F – L O V E shell theory) ........................................................................................ 20
(a) Analysis with Shell63( elements (Thickness ratio=1/4) ............................................................ 20
Mesh geometry ............................................................................................................................. 20
Shell281 (M I N D L I N- R E I S S N E R ) .................................................................................................. 22
(a) Analysis with shell281 (thickness ratio ¼) ............................................................................... 22
Comparision Between Kirchoff- Love and Mindlin-Reissner shell theory ................................................ 23
Solid 186 ............................................................................................................................................... 24
(a) Analysis with Solid 20 nodes brick element (thickness ratio=1/4) ........................................... 24
Mesh geometry ............................................................................................................................. 24
Mode shapes with Solid186 ........................................................................................................... 26
Mode Frequency calculations ........................................................................................................ 28
(b) Solid186, 20 nodes brick element (thickness ratio=1/16) ........................................................ 30
Mesh Geometry ............................................................................................................................. 30
Frequency Calculations .................................................................................................................. 32
(c) Solid186, 20 nodes brick element (thickness ratio=1/64) ........................................................ 33
Mesh Geometry ............................................................................................................................. 33
All 6 mode shapes with coarsest mesh for solid186, 64/1 ...................................................... 34
Frequency Calculations .................................................................................................................. 36
Comparison of Analytical and FEM solutions for Shell93 and Solid186 for three thicknesses .................. 37
Analysis with Solid185 (thickness ratio 1/64) ......................................................................................... 39
Conclusions: .......................................................................................................................................... 40
Objective:
To perform modal analysis of a rectangular steel plate with all four edges simply supported. We
will perform the analysis using different types of elements and evaluate performance of each
element as compared to other in the context of mode shapes.
Model particulars:
(a) Plate specifications
(As per Person number 35904604)
E= 11 GPa
Density=220 Kg/m^3
Poisson’s Ratio= 0.15
(Close to Oak Wood)
Length / breadth = 1
2a= 128m
2b=128m
Thickness=16m ,4m & 1m
(b) Elements used in analysis shell93, shell63, shell281 & solid186, solid185
(c) Assumptions/Methods
All the graphs plotted in this analysis are log-log values.
Mode extraction method used: Block Lanczos
Note: (I have also used Subspace method , however its performance slows down as compared to Block
Lanczos with high mesh density)
Exact frequency calculation from FEM solutions
21
2
2
1
)(2
)(
nnn
nnnex
Where 21 ,, nnn are three frequencies associated with three finest meshes.
Shell 93 Element (K I R C H O F F – L O V E)
The element has six degrees of freedom at each node: translations in the nodal x, y, and z
directions and rotations about the nodal x, y, and z-axes. The deformation shapes are quadratic
in both in-plane directions. The element has plasticity, stress stiffening, large deflection, and
large strain capabilities.
Because of mid nodes, this element gives quadratic convergence.
I have modeled my plate so that all edges of middle plane have UX, UY and UZ equal to zero.
(a) Thickness ratio = ¼, thickness=16m
Shell93, Frequency Convergence thickness ratio=1/4
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 5.7621 5.7613 5.7612 5.7612 5.7612
Mode 2 14.1000 14.0830 14.0820 14.0811 14.0730
Mode 3 14.1000 14.0830 14.0820 14.0811 14.0730
Mode 4 21.2730 21.2420 21.2400 21.2390 21.2380
Mode 5 26.8290 26.6660 26.6550 26.6530 26.6526
Mode 6 26.9070 26.7410 26.7300 26.7280 26.7276
Convergence ex
feex
Mode 1 0.0001562174523 0.0000173574925 0.0000000000024 0.0000000000024
Mode 2 0.0019185674619 0.0007105805366 0.0006395224822 0.0005755702332
Mode 3 0.0019185674619 0.0007105805366 0.0006395224822 0.0005755702332
Mode 4 0.0016479894504 0.0001883416493 0.0000941708234 0.0000470854105
Mode 5 0.0066201698400 0.0005044335963 0.0000917151995 0.0000166754909
Mode 6 0.0067138367395 0.0005030181087 0.0000914578380 0.0000166286979
X Y Z Dispacement vector sum Convergence
h 1 0.5 0.25 0.125 Exact
Maximum displacements at nodes (m)
Mode 1 0.000254 0.000260 0.000260 0.000260 0.000260
Mode 2 0.000260 0.000265 0.000284 0.000271 0.000277
Mode 3 0.000260 0.000265 0.000284 0.000271 0.000277
Mode 4 0.000250 0.000255 0.000261 0.000261 0.000261
Mode 5 0.000266 0.000284 0.000287 0.000287 0.000287
Mode 6 0.000322 0.000367 0.000367 0.000367 0.000367
Convergence
Uex
UfeUex
Mode 1 0.0259777662565 0.0000192001217 0.0000192001241 0.0000192001217
Mode 2 0.0584063587257 0.0434340216736 0.0281366426405 0.0182269711294
Mode 3 0.0584530555793 0.0433723920178 0.0281251424212 0.0182379527486
Mode 4 0.0429234488318 0.0210027398037 0.0001933953941 0.0000017808047
Mode 5 0.0736331919200 0.0083287797669 0.0016831680365 0.0003401524255
Mode 6 0.1230199435372 0.0002904839281 0.0001452419640 0.0000726209820
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-12
-11
-10
-9
-8
-7
-6
-5
-4
-3
-2Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
However, if we try to make mesh dense in the areas of maximum displacements, we will get better
convergence.
New Mesh
h1
h2
h3
h4
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-6
-5
-4
-3
-2
-1
0Displacement vector sum Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(U
ex-U
fe||/
||Uex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
The above mesh have stiff elements. Since, shell93 have quadratic convergence it will show proper
results even with stiff elements.
Comparison between two meshes:
Clearly second mesh gives better convergence.
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-12
-11
-10
-9
-8
-7
-6
-5
-4
-3
-2Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/||w
ex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/||w
ex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Shell93, Frequency Convergence(new mesh) thickness ratio=1/4
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 5.7623 5.7611 5.7612 5.7612 5.7612
Mode 2 14.1610 14.0860 14.0830 14.0820 14.0815
Mode 3 14.1610 14.0860 14.0830 14.0820 14.0815
Mode 4 21.3860 21.2500 21.2420 21.2400 21.2393
Mode 5 27.1480 26.6880 26.6610 26.6550 26.6533
Mode 6 27.2370 26.7640 26.7360 26.7300 26.7284
Mode 1 0.0001909324456 0.0000173574939 0.0000000000010 0.0000000000010
Mode 2 0.0056457053582 0.0003195682279 0.0001065227427 0.0000355075810
Mode 3 0.0056457053582 0.0003195682279 0.0001065227427 0.0000355075810
Mode 4 0.0069054270379 0.0005022128755 0.0001255532189 0.0000313883047
Mode 5 0.0185610994087 0.0013024392596 0.0002894309465 0.0000643179881
Mode 6 0.0190298355170 0.0013332789139 0.0002857026245 0.0000612219910
Mode shapes with Shell93
Mode 1
h1
h2
h3
h4
ex
feex
Mode 2
h1
h2
h3
h4
Mode 3
h1
h2
h3
h4
Mode 4
h1
h2
h3
h4
Mode 5
h1
h2
h3
h4
Mode 6
h1
h2
h3
h4
All 6 mode shapes with finest mesh
1
2
3
4
5
6
Mode frequency calculations
Let us try to derive mode frequencies from fundamental frequency.
2^
2^
2^
2^0
b
n
a
mmn
Where,
m= number of crests /troughs in x direction (length of 1st axis),
n= number of crests /troughs in y direction(length of 2nd
axis)
a=half length, b=half breadth
for my particular case a=b=64; Hz7612.50
2nd
natural frequency
= 5.7612*(1^2 + 2^2)/(1^2 + 1^2)
=14.40 Hz, which is quite close to 14.0815Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
= 5.7612*(2^2 + 1^2)/(1^2 + 1^2)
=14.40 Hz, which is quite close to 14.0815Hz, the exact frequency from F.E. model for 3rd mode.
4th natural frequency
= 5.7612*(2^2 + 2^2)/(1^2 + 1^2)
=23.044 Hz, which is quite close to 21.2393Hz, the exact frequency from F.E. model for 4th mode.
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=( 5.7612*(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=26.479 Hz, which is quite close to 26.6533Hz, the exact frequency from F.E. model for 5th mode.
Analytical calculation for 6th mode remains same as 5
th mode.
(b) Shell 93 Element (Thickness ratio = 1/16)
Mesh geometry
Shell93,Frequency Convergence for thickness ratio=1/16
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 1.5551 1.5538 1.5537 1.5538 1.5538
Mode 2 3.9759 3.8972 3.8949 3.8949 3.8949
Mode 3 3.9759 3.8972 3.8949 3.8949 3.8949
Mode 4 6.2807 6.1816 6.1736 6.1735 6.1735
Mode 5 8.0391 7.7973 7.7827 7.7821 7.7821
Mode 6 8.0492 7.7994 7.7845 7.7838 7.7838
Mode 1 0.0008688656471 0.0000321802086 0.0000321802097 0.0000321802086
Mode 2 0.0207964260956 0.0005905158026 0.0000000000001 0.0000000000001
Mode 3 0.0207964260956 0.0005905158026 0.0000000000001 0.0000000000001
Mode 4 0.0173647505959 0.0013122649201 0.0000164033115 0.0000002050414
Mode 5 0.0330279183736 0.0019565110441 0.0000804045635 0.0000033042972
Mode 6 0.0341010411069 0.0020086045830 0.0000943639737 0.0000044332068
ex
feex
In above graph we can see that mode 2(green) and mode 3(red) totally coincided. Because mode 2 has
m=1 , n=2 whereas, mode 3 has m=2, n=1. Therefore, numerical values of frequencies for these two
modes are same.
All 6 mode shapes with finest mesh
1
2
3
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/16
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
4
5
6
Fundamental frequency = 1.5538 Hz
2nd
natural frequency
= 1.5538 *(1^2 + 2^2)/(1^2 + 1^2)
=3.88 Hz, which is quite close to 3.8949 Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
= 1.5538 *(2^2 + 1^2)/(1^2 + 1^2)
=3.88 Hz, which is quite close to 3.8949 Hz, the exact frequency from F.E. model for 3rd mode.
4th natural frequency
= 1.5538 *(2^2 + 2^2)/(1^2 + 1^2)
=6.2152 Hz, which is quite close to 6.1735 Hz, the exact frequency from F.E. model for 4th mode.
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=( 1.5538 *(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=7.1412 Hz, which is quite close to 7.7821 Hz, the exact frequency from F.E. model for 5th mode.
(c) Shell 93 Element (Thickness ratio = 1/64)
Mesh geometry remains same.
Shell93,Frequency Convergence for thickness ratio=1/64
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 0.3964 0.3944 0.3942 0.3942 0.3942
Mode 2 1.0338 0.9893 0.9868 0.9867 0.9867
Mode 3 1.0338 0.9893 0.9868 0.9867 0.9867
Mode 4 1.6233 1.5825 1.5763 1.5760 1.5760
Mode 5 2.0809 1.9845 1.9752 1.9750 1.9750
Mode 6 2.0908 1.9857 1.9755 1.9750 1.9750
Mode 1 0.0057083417901 0.0005581489750 0.0000000000001 0.0000000000001
Mode 2 0.0477508239586 0.0026705795692 0.0001165729177 0.0000050885004
Mode 3 0.0477508239586 0.0026705795692 0.0001165729177 0.0000050885004
Mode 4 0.0300226600318 0.0041340845809 0.0002000363507 0.0000096791782
Mode 5 0.0536225981307 0.0048123629153 0.0001034916756 0.0000022256274
Mode 6 0.0586467264352 0.0054308421094 0.0002662177505 0.0000130498897
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/64
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
ex
feex
All 6 mode shapes with finest mesh for shell93, thickness ratio 1/64
1
2
3
4
5
6
Mode frequency calculations
Fundamental frequency = 0.3942 Hz
2nd
natural frequency
= 0.3942 *(1^2 + 2^2)/(1^2 + 1^2)
=0.9855 Hz, which is quite close to 0.9867 Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
= 0.3942 *(2^2 + 1^2)/(1^2 + 1^2)
=0.9855 Hz, which is quite close to 0.9867 Hz, the exact frequency from F.E. model for 3rd
mode.
4th natural frequency
= 0.3942 *(2^2 + 2^2)/(1^2 + 1^2)
=1.5768 Hz, which is quite close to 1.5760 Hz, the exact frequency from F.E. model for 4th mode.
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=( 0.3942 *(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=1.8118 Hz, which is quite close to 1.9750 Hz, the exact frequency from F.E. model for 5th mode.
Comparison of FEM and analytical solution for the three thickness ratios.
Thickness ratio = 1/4 Thickness ratio = 1/16
Thickness ratio = 1/64
It can be concluded from these
three graphs that as thickness
reduces and comes to moderate
level, shell93 element performs
better, since FEM solution
follows analytical solution more
closely.
2 2.5 3 3.5 4 4.5 5 5.5 612
14
16
18
20
22
24
26
28Comparision of analytical and FEM solution for Thickness ratio 1/4
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 62
4
6
8
10
12
14
16
18Comparision of analytical and FEM solution for Thickness ratio 1/16
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 60
2
4
6
8
10
12
14
16Comparision of analytical and FEM solution for Thickness ratio 1/64
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
Shell63 (K I R C H O F F – L O V E shell theory)
(a) Analysis with Shell63( elements (Thickness ratio=1/4) This element does not have mid nodes.
The element has six degrees of freedom at each node: translations in the nodal x, y, and z
directions and rotations about the nodal x, y, and z-axes. Stress stiffening and large deflection capabilities
are included.
Mesh geometry
Shell63,Frequency Convergence for thickness ratio=1/4
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 6.3196 6.3296 6.3329 6.3338 6.3341
Mode 2 15.7370 15.7990 15.8260 15.8330 15.8354
Mode 3 15.7530 15.8020 15.8270 15.8330 15.8349
Mode 4 25.2860 25.3060 25.3290 25.3350 25.3371
Mode 5 31.2480 31.5200 31.6320 31.6610 31.6711
Mode 6 31.3370 31.5630 31.6420 31.6630 31.6706
Mode 1 0.0022951033191 0.0007163564100 0.0001953699300 0.0000532827082
Mode 2 0.0062170636135 0.0023017975491 0.0005967623275 0.0001547161589
Mode 3 0.0051717891533 0.0020773574684 0.0004985657925 0.0001196557902
Mode 4 0.0020175004819 0.0012281447122 0.0003203855772 0.0000835788463
Mode 5 0.0133601957466 0.0047719332416 0.0012355898572 0.0003199295166
Mode 6 0.0105335362119 0.0033975812444 0.0009031545080 0.0002400790464
ex
feex
Clearly order of convergence for shell93 is better than shell63.
Mode 5- Convergence Shell63/shell93 = 0.0003199295166/0.0000643179881 = 4.97
This implies that shell93 converges faster than shell63, which was expected as shell93 have mid nodes.
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-5
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1Frequency Convergence comparison b/w 6 modes with shell63--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Shell281 (M I N D L I N- R E I S S N E R )
(a) Analysis with shell281 (thickness ratio ¼) This elements also have mid nodes hence exhibits quadratic convergence. It is an 8-node element with
six degrees of freedom at each node: translations in the x, y, and z axes, and rotations about the x, y, and
z-axes.
The modeling of shell281 is governed by the first order shear deformation theory (
Mindlin-Reissner shell theory).
Mesh geometry remains same as in case of Shell93.
Shell281,Frequency Convergence for thickness ratio=1/4
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 5.6975 5.6966 5.6965 5.6965 5.6965
Mode 2 13.8330 13.7600 13.7570 13.7570 13.7570
Mode 3 13.8330 13.7600 13.7570 13.7570 13.7570
Mode 4 20.7010 20.5690 20.5610 20.5610 20.5610
Mode 5 26.1890 25.7610 25.7310 25.7290 25.7289
Mode 6 26.2800 25.8380 25.8080 25.8060 25.8059
Mode 1 0.0001755463934 0.0000175546441 0.0000000000053 0.0000000000053
Mode 2 0.0055244602749 0.0002180708005 0.0000000000002 0.0000000000002
Mode 3 0.0055244602749 0.0002180708005 0.0000000000002 0.0000000000002
Mode 4 0.0068090073442 0.0003890861341 0.0000000000002 0.0000000000002
Mode 5 0.0178843100021 0.0012492920678 0.0000832861378 0.0000055524091
Mode 6 0.0183734589601 0.0012455644068 0.0000830376271 0.0000055358418
ex
feex
Comparision Between Kirchoff- Love and Mindlin-Reissner shell theory
Kirchoff- Love shell theory –shell93 Mindlin-Reissner shell theory –shell281
Mode 6- Convergence Shell93/shell281 = 0.0000130498897/ 0.0000055358418= 2.357
It is evident from above graph that shell281 have faster convergence than shell93.
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell93--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/||w
ex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2
0Frequency Convergence comparison b/w 6 modes with shell281--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/||w
ex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Solid 186 SOLID186 is a higher order 3-D 20-node solid element that exhibits quadratic displacement behavior.
The element is defined by 20 nodes having three degrees of freedom per node: translations in the nodal x,
y, and z directions. It also has mixed formulation capability.
(a) Analysis with Solid 20 nodes brick element (thickness ratio=1/4)
Mesh geometry
While creating volumes for solid brick elements area is extruded in two equal halves so that one middle
plane is created. All the edges of this middle plane have UX, UY, UZ=0
Solid186,Frequency Convergence thickness ratio=1/4
h 1 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 5.6668 5.6339 5.6302 5.6297
Mode 2 13.6600 13.5380 13.5290 13.5283
Mode 3 13.6600 13.5380 13.5290 13.5283
Mode 4 20.3220 20.0560 20.0370 20.0355
Mode 5 25.5400 25.1590 25.1370 25.1357
Mode 6 25.6200 25.2540 25.2300 25.2283
Mode 1 0.0065844770441 0.0007405034974 0.0000832785088
Mode 2 0.0097364027904 0.0007182592222 0.0000529863361
Mode 3 0.0097364027904 0.0007182592222 0.0000529863361
Mode 4 0.0142976710614 0.0010212622187 0.0000729473013
Mode 5 0.0160866403013 0.0009288873665 0.0000536365409
Mode 6 0.0155255790277 0.0010180707559 0.0000667587381
Here, I used only 3 mesh refinements, because the forth mesh refinement was exceeding the ansys limit
of number of elements for ‘ANSYS’. However, it is not required to refine mesh any further since
frequencies and mode shapes are already converged.
ex
feex
For solid elements above mesh gives much better performance than the mesh having more number of
volumes in initial geometry.
I tried many combinations, some of them is as follows
Both of these geometries gave erroneous results. The best one is the previous mesh with no partitions on
edges.
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-5
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1Frequency Convergence comparison b/w 6 modes with solid186--ratio xi=1/4
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Mode shapes with Solid186
Mode 1
h=1
h=0.5
h=0.25
Mode 2
h=1
h=0.5
h=0.25
Mode 3
h=1
h=0.5
h=0.25
Mode 4
h=1
h=0.5
h=0.25
Mode 5
h=1
h=0.5
h=0.25
Mode 6
h=1
h=0.5
h=0.25
All 6 mode shapes with finest mesh for solid186, 4/1
1
2
3
4
5
6
Mode Frequency calculations
Using,
2^
2^
2^
2^0
b
n
a
mmn
and Hz6297.50
2nd
natural frequency
= 5.6297*(1^2 + 2^2)/(1^2 + 1^2)
=14.074 Hz, which is quite close to 13.5283 Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
= 5.6297*(2^2 + 1^2)/(1^2 + 1^2)
=14.074 Hz, which is quite close to 13.5283 Hz, the exact frequency from F.E. model for 3rd mode.
4th natural frequency
= 5.6297*(2^2 + 2^2)/(1^2 + 1^2)
=22.5188 Hz, which is close to 20.0355 Hz, the exact frequency from F.E. model for 4th mode.
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=( 5.6297*(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=25.8751 Hz, which is quite close to 25.1357 Hz, the exact frequency from F.E. model for 5th mode.
Analytical calculation for 6th mode remains same as 5
th mode.
(b) Solid186, 20 nodes brick element (thickness ratio=1/16)
Mesh Geometry
h1
h2
h3
h4
Note: Thickness is refined only once
Solid186,Frequency Convergence for thickness ratio=1/16
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 1.5538 1.5531 1.5523 1.5523 1.5523
Mode 2 3.8916 3.8883 3.8864 3.8862 3.8862
Mode 3 3.8916 3.8883 3.8864 3.8862 3.8862
Mode 4 6.1641 6.1565 6.1515 6.1511 6.1511
Mode 5 7.7800 7.7557 7.7505 7.7501 7.7501
Mode 6 7.7815 7.7569 7.7516 7.7510 7.7509
Mode 1 0.0009663080592 0.0005153642983 0.0000000000002 0.0000000000002
Mode 2 0.0013955952471 0.0005464315447 0.0000575191100 0.0000060546431
Mode 3 0.0013955952471 0.0005464315447 0.0000575191100 0.0000060546431
Mode 4 0.0021191098043 0.0008835514528 0.0000706841163 0.0000056547294
Mode 5 0.0038623323670 0.0007268754677 0.0000559134975 0.0000043010382
Mode 6 0.0039448971625 0.0007710817709 0.0000872922759 0.0000098821444
ex
feex
All 6 mode shapes with finest mesh for solid186, 16/1
1
2
3
4
5
6
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2Frequency Convergence comparison b/w 6 modes with solid186--ratio xi=1/16
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
Frequency Calculations
Hz5523.10
2nd
natural frequency
=1.5523*(1^2 + 2^2)/(1^2 + 1^2)
=3.8807 Hz, which is quite close to 3.8862 Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
=1.5523*(2^2 + 1^2)/(1^2 + 1^2)
=3.8807 Hz, which is quite close to 3.8862 Hz, the exact frequency from F.E. model for 3rd mode.
4th natural frequency
=1.5523*(2^2 + 2^2)/(1^2 + 1^2)
=6.2092 Hz, which is close to 6.1511 Hz, the exact frequency from F.E. model for 4th mode
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=(1.5523*(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=7.1346 Hz, which is quite close to 7.7501 Hz, the exact frequency from F.E. model for 5th mode.
Analytical calculation for 6th mode remains same as 5
th mode.
(c) Solid186, 20 nodes brick element (thickness ratio=1/64)
Mesh Geometry
h1
h2
h3
h4
Solid186,Frequency Convergence for thickness ratio=1/64
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 0.3937 0.3943 0.3942 0.3942 0.3942
Mode 2 0.9875 0.9868 0.9867 0.9867 0.9867
Mode 3 0.9875 0.9868 0.9867 0.9867 0.9867
Mode 4 1.5842 1.5762 1.5761 1.5761 1.5761
Mode 5 1.9816 1.9752 1.9748 1.9747 1.9747
Mode 6 1.9816 1.9752 1.9748 1.9747 1.9747
Mode 1 0.0014205266089 0.0001014661870 0.0000507330938 0.0000253665472
Mode 2 0.0008209348515 0.0001216199773 0.0000608099883 0.0000304049938
Mode 3 0.0008209348515 0.0001216199773 0.0000608099883 0.0000304049938
Mode 4 0.0051392678127 0.0000634477505 0.0000000000003 0.0000000000003
Mode 5 0.0035111411209 0.0002700877785 0.0000675219446 0.0000168804862
Mode 6 0.0035111411209 0.0002700877785 0.0000675219446 0.0000168804862
ex
feex
The FEM calculations had problem with 64/1 . The coarse mesh gave erroneous results because the
elements were cuboids with very high aspect ratios on faces.
All 6 mode shapes with coarsest mesh for solid186, 64/1
1
2
3
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-14
-12
-10
-8
-6
-4
-2Frequency Convergence comparison b/w 6 modes with solid186--ratio xi=1/64
log(h)
log(||(w
ex-w
fe||/
||wex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
4
5
6
We can see that mode shapes 5 and 6 are flipped as compared to previous solutions.
However, the problem was resolved with first mesh refinement on length and breadth section
leaving thickness unrefined throughout.
All 6 mode shapes with finest mesh for solid186, 64/1
1
2
3
4
5
6
Frequency Calculations
Hz3942.00
2nd
natural frequency
=0.3942*(1^2 + 2^2)/(1^2 + 1^2)
=0.9855 Hz, which is close to 0.9867 Hz, the exact frequency from F.E. model for 2nd
mode.
3rd natural frequency
=0.3942*(2^2 + 1^2)/(1^2 + 1^2)
=0.9855 Hz, which is close to 0.9867 Hz, the exact frequency from F.E. model for 3rd
mode.
4th natural frequency
=0.3942*(2^2 + 2^2)/(1^2 + 1^2)
=1.5768 Hz, which is close to 1.5761 Hz, the exact frequency from F.E. model for 4th mode
5th natural frequency
The plane of symmetry in this case switches to half diagonals of the square plate.
=(0.3942*(3^2 + 2^2)/(1^2 + 1^2))/sqrt(2)
=1.8118 Hz, which is close to 1.9747 Hz, the exact frequency from F.E. model for 5th mode.
Analytical calculation for 6th mode remains same as 5
th mode.
Comparison of Analytical and FEM solutions for Shell93 and Solid186 for
three thicknesses SOLID186
4/1 16/1 64/1
SHELL93
We can clearly see that trend for Solid186 and Shell93 are almost similar.
However, if we plot the difference of analytical and FEM solutions for Solid186 and
Shell93, we can compare them much better.
2 2.5 3 3.5 4 4.5 5 5.5 612
14
16
18
20
22
24
26
28
SOLID186 Thickness ratio 1/4
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 62
4
6
8
10
12
14
16
18
SOLID186, Thickness ratio 1/16
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 60
2
4
6
8
10
12
14
16
SOLID186 Thickness ratio 1/64
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 612
14
16
18
20
22
24
26
28
SHELL93, Thickness ratio 1/4
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 62
4
6
8
10
12
14
16
18
SHELL93, Thickness ratio 1/16
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
2 2.5 3 3.5 4 4.5 5 5.5 60
2
4
6
8
10
12
14
16
SHELL93, Thickness ratio 1/64
Modes
Natu
ral fr
equeccie
s (
Hz)
FEM
Analytical
Clearly Shell 93 difference always remains below Solid186, which concludes that performance of shell
element is better.
2 2.5 3 3.5 4 4.5 5 5.5 60
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Difference with analytical solution, Thickness ratio 1/4
Modes
ab
s((
Wa
na
lytic-W
fem
) / W
an
aly
tic)
SHELL93
SOLID186
Analysis with Solid185 (thickness ratio 1/64) SOLID185 is used for 3-D modeling of solid structures. It is defined by eight nodes having three degrees
of freedom at each node: translations in the nodal x, y, and z directions.
Mesh geometry remains same as in case of Solid186.
Solid185,Frequency Convergence for thickness ratio=1/64
h 1 0.5 0.25 0.125 Exact
Frequencies for all 6 modes with different mesh (Hz)
Mode 1 1.0438 0.7490 0.6134 0.5393 0.4498
Mode 2 3.0023 2.1138 1.7000 1.4702 1.1832
Mode 3 3.0023 2.1138 1.7000 1.4702 1.1832
Mode 4 4.1908 2.9999 2.4550 2.1575 1.7998
Mode 5 6.5835 4.5773 3.6415 3.1187 2.4569
Mode 6 6.5835 4.5773 3.6415 3.1187 2.4569
Mode 1 1.3204845469648 0.6650443678293 0.3637015048852 0.1989021951836
Mode 2 1.5374413046126 0.7865114844253 0.4367818731777 0.2425627705564
Mode 3 1.5374413046126 0.7865114844253 0.4367818731777 0.2425627705564
Mode 4 1.3285398787336 0.6668384991440 0.3640749742987 0.1987746464560
Mode 5 1.6795868380279 0.8630322524045 0.4821471057460 0.2693593790169
Mode 6 1.6795868380279 0.8630322524045 0.4821471057460 0.2693593790169
Clearly, solid185 has
very slow
convergence as
compared to solid186
for 64/1 (low
thickness ratio).
Although, it gave
reasonable
convergence result
with 16/1 &
4/1 ,
(not presented in this
analysis for solid185)
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4Frequency Convergence comparison b/w 6 modes with solid185--ratio xi=1/64
log(h)
log(|
|(w
ex-w
fe|| /
||w
ex||)
Mode 1
Mode 2
Mode 3
Mode 4
Mode 5
Mode 6
ex
feex
The poor convergence of SOLID185 in the case of 64/1 can be explained by presence of stiff
elements in modeling of mesh for 64/1 . Since, this element does not have any mid node the shape
function will be of first order only (it will not have quadratic elements), which will cause non-
smoothening of curves at the stiff corners. This will lead to numerical imprecision.
Conclusions:
With Block Lanczos method, we can achieve faster convergence to FEM problems as compared
to subspace.
Shell elements give better performance than solid elements.
Shell93 gives better performance with moderate thick plates.
Mindlin-Reissner shell theory exhibits better results as compared to Kirchoff – love for square
plate with all edges simply supported.
Elements, which do not have mid nodes give poor convergence with stiff mesh elements.
Elements, with mid nodes exhibit much better results with stiff mesh elements.