Вы находитесь на странице: 1из 17

CYCLIC SYMMETRY MODELING

a report done by

Charisse Haines
Hannah Spiegel

In partial fulfillment of the requirements of the course

ME 422
Finite Element Analysis
Dr. Jones

DEPARTMENT OF MECHANICAL ENGINEERING


ROSE-HULMAN INSTITUTE OF TECHNOLOGY
Table of Contents
Table of Contents 2

Introduction 3

Validation Model 4
Verify Geometry 4
Verify Material Properties 4
Verify Element Type, Mesh, and Model Size 5
Verify Load Configuration 6
Verify Support Conditions 7
Verify Solution 7

Main Model 9
Verify Geometry 9
Verify Material Properties 10
Verify Element Type, Mesh, and Model Size 11
Verify Load Configuration 12
Verify Support Conditions 12
Verify Solution 12
Verify Results 13

Summary and Conclusions 16

Appendix A 17

Page 2
Introduction
The objective of this project is to study the cyclic symmetry feature in ANSYS. This study runs a
validation case, a simple circular disk, as a full 3D model and as a 30​o​ sector using the cyclic
symmetry feature. It compares the results of these two models to determine the accuracy of the
cyclic symmetry feature. This strategy decreases the computational time of the model by taking
advantage of the symmetry of the object being analyzed. The model is also compared to a
by-hand calculation of what the first natural frequency for a clamped disk should be (Appendix
A). Once the validation case has been verified, a more complicated case of a spoked wheel is
analyzed, in which a similar approach compares the accuracy of the 30​o​ cyclic symmetry model
to the full 3D model.

Page 3
Validation Model

Verify Geometry
The full model for the validation disk and its 30​o​ sector are created in SolidWorks and imported
to ANSYS.

Figure 1.​ Geometry views of the disk and its 30​o​ sector, created in SolidWorks.

The dimensions of the full disk are in Table 1, the 30​o​ sector is any 30​o​ of the disk.

Table 1. ​Dimensions for the full disk.

Outer Diameter 100 cm

Inner Diameter 30 cm

Thickness 0.5 cm

Verify Material Properties


The material for the disk and its 30​o ​section is aluminum with material properties of Young’s
Modulus E = 70x109 P a , poisson’s ratio ν = 0.3 , and density ρ = 2700 mkg3 .

Page 4
Figure 2. ​Material properties applied to the disk for both the full model and 30​o​ section.

Figure 3. ​The material applied to the full model (left) and the 30​o​ sector (right) in the ANSYS model.

Verify Element Type, Mesh, and Model Size


The full disk model uses quadrilateral elements with a max face size of 1 cm which when
changed by 20% changes the first natural frequency by less than 1% so it is said to converge. A
total of 14092 elements and 99564 nodes are in the model. The average element quality is
0.88355.

Page 5
Figure 4.​ Final mesh for the simulation for the full disk.

The partial disk model uses quadrilateral elements with a max face size of 0.5 cm which when
changed by 20% changes the first natural frequency by less than 1% so it is said to converge. A
total of 1820 elements and 13223 nodes are in the model. The average element quality is
0.89892.

Figure 5.​ Final mesh for the simulation of the 30​o​ section of the disk.

Verify Load Configuration


No load was applied to the full disk model or the 30​o​ sector in this simulation.

Page 6
Verify Support Conditions
A fixed support was added to the outer edge of the full disk and its 30​o​ sector. This represents the
clamping force that was used in computing the hand calculations (See Appendix A).

Figure 6. ​The fixed support constraint applied to the outer rim of the disk and its sector.

Verify Solution

Figure 7. ​Solution confirmation and simulation authorship confirmation for the disk (left and top) and its sector
(right and bottom).
Verify Results
The first mode of the solution is a breathing mode of the disk at 56.161 Hz for the full disk
model and at 56.088 Hz for the 30​o​ sector with cyclic symmetry applied.

Page 7
Figure 8.​ The first mode deformation for the full disk (left) and its sector (right).
The results for the by-hand analysis of the disk and its two ANSYS models are shown below.

Table 2.​ Comparison of frequency for the first mode from the full disk and cyclic symmetry model.

Model ANSYS Mode 1 Mathematical Result Difference (%)


Frequency (Hz) for Mode 1
Frequency (Hz)

Full Disk 56.161 55.91 ≤ 1%

30​o​ Sector 56.088 55.91 ≤ 1%

Because the differences between the analytical and simulated results are less than 1%, we
consider both models to be accurate and cyclic symmetry to be a valid tool for finite element
analysis.

Page 8
Main Model

Verify Geometry
The full model for the wheel and its 30​o​ sector are created in SolidWorks and imported to
ANSYS.

Figure 9.​ Geometry views of the wheel (right) and its 30​o​ sector (left), created in SolidWorks.

The wheel dimensions are in Table 3. Each sector used is a 30​o​ portion of the wheel with a spoke
through the direct center of the section.

Table 3. ​The dimensions for the full wheel model.

Outer Diameter 150 cm

Thickness of Outer Rim (out of plane) 1 cm

Diameter of Inside of Outer Rim 140 cm

Diameter of Outside of Inner Rim 40 cm

Inner Diameter 30 cm

Thickness of Inner Rim (out of plane) 1 cm

Thickness of Spoke (out of plane) 0.5 cm

Spoke Dimension 20​o

Page 9
Verify Material Properties
The material for the full disk, as well as the 30​o ​sectors, is aluminum with material properties of
Young’s Modulus E = 70x109 Pa, Poisson’s ratio ν = 0.3 , and density ρ = 2700 mkg3 as seen in
Figures 10 and 11.

Figure 10. ​Material properties applied to the wheel for both the full model and 30​o​ section.

Figure 11. ​The material applied to the full model (left) and the 30​o​ sector (right) in the ANSYS model.

Page 10
Verify Element Type, Mesh, and Model Size
The full wheel model uses tetrahedral elements with a max face size of 3 cm which, when
changed by 20%, changes the first natural frequency by less than 1%, and so is said to converge.
A total of 12453 elements and 26330 nodes are in the model. The average element quality is
0.33925. The final mesh is in Figure 12.

Figure 12.​ Final mesh for the simulation for the full wheel..

The partial wheel model uses quadrilateral elements with a max face size of 1 cm which when
changed by 20% changes the first natural frequency by less than 1% so it is said to converge. A
total of 2504 elements and 15611 nodes are in the model. The average element quality is
0.52585. The converged mesh can be seen in Figure 13.

Figure 13.​ Final mesh for the simulation of the 30​o​ section of the wheel.

Page 11
Verify Load Configuration
No load was applied to the full wheel model or the 30​o​ sector in this simulation.

Verify Support Conditions


Figure 14 shows a fixed support was added to the outer edge of the full wheel and its 30​o​ sector.

Figure 14. ​The fixed support constraint applied to the outer rim of the wheel and its sector.

Verify Solution

Figure 15. ​Solution confirmation and simulation authorship confirmation for the wheel (left and top) and its sector
(right and bottom).

Page 12
Verify Results
The first mode of the solution is a breathing mode of the wheel at 23.945 Hz for the full disk
model and at 23.601 Hz for the 30​o​ sector with cyclic symmetry applied as seen in Figure 16.

Figure 16.​ The first mode deformation for the full disk (left) and its sector (right).

The results for the ANSYS modal simulation for the wheel and its cyclically symmetric sector
are below in Table 4.

Table 4.​ Comparison of frequency for the first mode from the full wheel and cyclic symmetry model.

Model ANSYS Mode 1 Frequency Difference (%)


(Hz)

Full Wheel 23.945


≤ 1%
30​o​ Sector 23.601

Because the differences between the two results are less than 1%, we consider both models to be
accurate and cyclic symmetry to be a valid tool for finite element analysis.

With both models being valid, further analysis is done on the harmonics of the 0th mode in the
cyclic model by finding their corresponding modes in the full wheel model. These results can be
seen visually in Figure 17 and numerically in Table 5.

Page 13
Figure 17.​ The first three harmonics of the 0th mode for the 30​o​ Sector of the wheel (Right) and their corresponding
modes for the full wheel model (left).

Page 14
Table 5. ​Comparison of full wheel modes and harmonics of the first mode from the cyclic symmetry model.

Full Wheel Full Wheel 30​o​ Sector 30​o​ Sector Difference (%)
Mode Frequency (Hz) Harmonic Frequency (Hz)

1st 23.945 1st 23.601 ≤ 1%

13th 110.02 2nd 107.63 1.10%

20th 232.97 3rd 229.93 ≤ 1%

The mode shapes in the full model and the harmonics of the first mode in the cyclic model match
well. The frequencies are within reasonable error and the visualization of the mode shapes shows
that the full disk and its sector produce very similar results from modal analysis in ANSYS using
both methods.

Page 15
Summary and Conclusions
The displays for the mode shapes using the two methods, full model and cyclic symmetry, are
slightly different. The cyclic symmetry feature defaults the modes to be displayed primarily by
their mode number and harmonics, while the regular modal analysis displays the results by
increasing frequency. The display of modes by their harmonic can be useful for analysis, but no
matter which method is used, the results produced are the same. The full model has a slightly
different mode shape than the cyclic for the highest frequency shown due to mode localization.
The full model’s imperfect mesh around the wheel makes this mode appear to only have some of
the spokes moving while the perfectly symmetric cyclic model shows all of the spokes moving.
In reality, the mode shape will be more similar to the full model in that it will not have a mass
perfectly distributed between each of its spokes; the mode will show the spokes moving different
amounts. However, the imperfections in the full wheel model do not match the imperfections that
the real wheel will experience. This is why the industry standard is to use the perfect cyclic
model to predict the mode shape, under the assumption that the wheel was perfectly
manufactured. The cyclic symmetry function allows for the model to be run faster, which can
allow for a better refined mesh to be run in the same amount of time as the full model would run
with a less accurate mesh. This can be useful for the analysis of many axisymmetric parts such as
turbines and complicated wheel shapes.

Page 16
Appendix A
Hand Calculations for Validation Model

Page 17