Professional Documents
Culture Documents
Analysis of a
Rotating Disk
Eliannah Hunderfund
MANE 4240 Introduction to Finite Elements
Professor Ernesto Gutierrez-Miravete
Rensselaer Polytechnic Institute, Hartford, CT
May 17, 2015
Table of Contents
Abstract...........................................................................................................3
Introduction and Background...........................................................................3
System Overview.............................................................................................4
Figure 1......................................................................................................4
Figure 2......................................................................................................4
Figure 3......................................................................................................4
Figure 4......................................................................................................5
Theory & Methodology.....................................................................................5
Analytical Solution...........................................................................................6
Figure 5.........................................................................................................7
Figure 6.........................................................................................................8
Figure 7.........................................................................................................8
Figure 8.........................................................................................................8
Comsol Multiphysics........................................................................................9
Figure 9.........................................................................................................9
Figure 10.......................................................................................................9
Figure 11.......................................................................................................9
Figure 12.....................................................................................................10
Figure 13.....................................................................................................10
Figure 13.....................................................................................................11
Figure 14.....................................................................................................11
Figure 15.....................................................................................................12
Figure 16.....................................................................................................12
Figure 17.....................................................................................................13
Figure 18.....................................................................................................13
Figure 19.....................................................................................................14
Results...........................................................................................................14
Figure 20.....................................................................................................15
Figure 21.....................................................................................................16
Figure 22.....................................................................................................16
1
Figure 23.....................................................................................................17
Conclusion.....................................................................................................17
Appendix A....................................................................................................19
Appendix B....................................................................................................21
References.....................................................................................................22
Abstract
This project evaluates the finite element analysis of stress and displacement
induced through centripetal force in a thin rotating disk of uniform thickness
and constant angular velocity with a 1m outer diameter and a .1m inner
diameter made of Ti 6-4 Titanium. The stress and displacement were
determined using Maple software to obtain an exact solution for the 1D case,
and then compared to the FEM solution obtained through Comsol
Multiphysics software for the 1D and 2D case.
System Overview
The first case this paper will evaluate is a 1D rotating disk made of Ti 6-4
Titanium, with inner radius (a) of 0.1m and outer radius (b) of 1m, illustrated in
r(a)=0.1m
r(b)=1.0m
Figure 1.
Figure 1
The
properties
Titanium
below in
0.33
1.14 GPa
880 MPa
material
of Ti 6-4
are listed
Figure 2.
Figure 2
a=)
r
below.
Figure 3
a = - r
Eq. (1)
- r 2
F=ma=m )
Eq. (2)
In this project the angular speed of the disk will be evaluated at a constant
angular velocity of
The second case evaluated in this project will be a thin 2D rotating disk with
the same material properties and angular speed described earlier with a
thickness of 0.1m applied, as illustrated below in Figure 4.
t=0.1m
Figure 4
+
=
E
r2 r r r2
Eq. (3)
u ( a )=0
and
du
( b )=0
dr
Eq. (4)
When this boundary value ODE is solved for the 1D case, the displacement
will be known and a number of other values can be determined. Radial and
tangential strain can be derived using Equations 5 and 6, below.
Radial Strain:
r =
Tangential Strain:
du
dr
t =
Eq. (5)
u
r
Eq. (6)
From radial and tangential strain, the radial and tangential (hoop)
components of stress can be determined using Equations 7 and 8, below.
r=
Radial Stress:
Tangential Stress:
t=
E
+ v t
( 1v 2) ( r
E
+ v r
( 1v 2) ( t
Eq. (7)
Eq. (8)
Finally, from radial and tangential (hoop) stress, the Von Mises stress can be
derived which is critical to understand because the value of Von Mises stress
is compared to the Tensile Yield Stress to predict if the material will fail. The
Von Mises stress is described in Equation 9 for the 1D case (no axial stress).
The relationship of Von Mises Stress and Ultimate Tensile Yield Strength is
described in Equation 10.
vm, x , y
2
2
2
( r y ) + ( y 0 ) + ( 0 y )
Eq.
(9)
vm tensile , yield
Eq. (10)
Analytical Solution
To solve the boundary value ODE analytically described in Equation 3, Maple
software was used. The code input for the Maple Software is included as
Appendix A. The exact solution was obtained by solving the 2nd Order ODE
using boundary conditions and appropriate material properties and angular
velocity, as described in the System overview. The radial displacement was
solved to be the following, exact solution, descried in Equation 11.
u(r)
Eq. (11)
The graph for u(r) is in Figure 5 with radial position from r(a) o r(b) in meters
on the x axis and displacement in meters on the y axis. The displacement
increases from r=a to b. The maximum displacement at r(b) is equal to
0.0008291136940 m or 8.29114e-4 m.
Figure 5
The graph obtained for radial and tangential strain with radial position from
r=a to b on the x-axis and strain on the y-axis is in Figure 6.
Figure 6
The graph obtained for radial and tangential (hoop) stress with radial
position from r=a to b on the x-axis and stress in Pa is in Figure 7.
Figure 7
Finally, the graph for Von Mises stress with radial position from r=a to b on
the x-axis and Von Mises Stress in Pa is in Figure 8. The stress is greatest at
r=a and decreases as r approaches b. The maximum Von Mises Stress at r=a
is 2.846194981e8 Pa (284.6 MPa).
Figure 8
It is interesting to note that at
Mises stress at r(a) is equal to the Tensile Yield Strength of the material. So,
9
this reveals that the disk should be limited to less than 5575 rot/min with
some safety factor to prevent material failure. This relationship is described
in Equation 12.
At
= 5575 rot/min,
vm=880 MPa
tensile , yield
Eq. (12)
Comsol Multiphysics
The 1D case was evaluated using Comsol Multiphysics using the Coefficient
PDE, initial values equal to the boundary conditions described in Equation 4
and a Dirichlet Condition applied at r=a. The Comsol files are included in
Appendix B. The mesh sizes shown in Figures 9, 10 and 11 show what a
coarse, normal and extremely fine mesh look like, respectively, applied to
the 1D geometry. All cases would be considered sufficient for the 1D case,
with negligible difference in the results for displacement.
Figure 9
Figure 10
Figure 11
The displacement line graph for u(r) for a normal mesh size is in Figure 12. In
Figure 13 is the maximum displacement obtained at u(r=b) for coarse,
normal and extremely fine mesh sizes.
10
Figure 12
Mesh Size
Extremely
Coarse
Normal
Extremely Fine
Displacement at r=b
(meters)
0.00082931632651707
2
0.00082911399667980
8
0.00082911369400041
7
Figure 13
In Figure 14, the graph describes the displacement of the entire 1D disk in a
surface plot. Note, this is just an expansion of the axisymmetric u(r) function
shown in Figure 12.
11
Figure 13
The 2D case, described in Figure 4, was also evaluated using Comsol
Multiphysics using the Solid Mechanics input, material properties of Ti64,
initial values equal to the boundary conditions described in Equation 4, a
prescribed displacement of 0 at r(a) at one corner and a prescribed
acceleration described in Equation 1. The geometry of the disk is shown in
Figure 14
12
Figure 15
The 2D displacement was obtained using Comsol Multiphysics and shows
very different results than the 1D, because there is now a thickness which
must be considered. The line graph shows the displacement of the bottom
surface at y=0 from Figure 15. The maximum displacement now occurs at
r=0.92 and is 0.000633680287828534 m. The line graph for the u(r)
displacement is shown in Figure 16.
13
Figure 16
The Von Mises Stress was also obtained from Comsol Multiphysics and was
graphed in a line graph and surface graph in Figures 17 and 18, respectively.
In the 2D surface plot it is evident that the disk is stretching to become
thinner due to the centripetal force. The maximum Von Mises Stress is still at
r=a and is 237196407.729246 or 237.2 MPa.
14
Figure 17
Figure 18
Finally, the 3D surface plot of the Von Mises plot is shown in Figure 19 and
simply an expansion of the 2D Surface Plot, applied around 360 degrees.
15
Figure 19
Results
In the results section of this project the values obtained analytically, through
Maple, and using FEM, using Comsol Multiphysics, will be compared. The first
thing to be compared will be the displacement u(r) for the analytical 1D
solution, the extremely coarse, normal and extremely fine FEM 1D meshes
and the 2D extremely fine mesh. The graph for u(r) for each of these cases is
shown in Figure 20.
16
Displacement u(r)
0
0
0
0
0
u (meters)
0
0
0
0
0
0.2
0.4
0.6
0.8
1.2
r(meters)
u_exact_1D
2D_displacement_comsol
1D_displacement_comsol_normal
1D_displacement_comsol_extremely_coarse
1D_displacement_comsol_extremely_fine
Figure 20
Analyzing the graph in Figure 20, it is clear that there is a difference between
the 1D and 2D displacements. However, there is not a large difference
between the analytical and FEM 1D, nor between the mesh sizes. There is
one noticeable discrepancy between r=0.2m and r=0.4m for the 1D FEM
extremely coarse mesh, as would be expected for a large mesh size. In
addition, although the 1D and 2D results differ, the 1D results are
conservative showing greater displacement. This shows that FEM using
Comsol Multiphysics is a good way to obtain the approximate solution for the
case of a simple 1D rotating disk. The error between the exact 1D solution
and the 1D solution for FEM is shown in Figure 21.
17
Figure 21
Method
Exact_Maple
Comsol_Extremely_C
oarse
Comsol_Normal
Comsol_Extremely_Fi
ne
1D u(r) at
r=b
% Error
0.000829114
0.000829316
0.000829114
0.024363357
4.00451E-07
0.000829114
3.69068E-05
This is a primary example of when a smaller mesh size does not provide
more accurate results. Although it is miniscule, the percent error for the
extremely fine mesh size is actually greater than the normal mesh size. The
extremely fine mesh was not necessary to obtain a more accurate solution
for this simple case, and sometimes increasing mesh size can cause a less
favorable divide of the area of interest.
In Figure 22 is the comparison between the exact 1D solution and the 2D
FEM solution for displacement. This shows a 30% difference between the 1D
max displacement and the FEM 2D max displacement, but again this is
conservative error.
Method
1D_Maple
2D_Comsol_Extremely_Fine
% error of 1D_Maple compared to
2D_Comsol
Max Displacement
u(r)
0.000829114
0.00063368
30.84105912
Figure 22
Finally, the results for exact 1D Von Mises is compared to the FEM 2D Von
Mises using an extremely fine mesh in Comsol Multiphysics. The graph for
the two Von Mises values over r=a to r=b is graphed in Figure 23. The
difference here can be attributed the thickness which was not included in the
1D evaluation, however the error in this case is again conservative. So, the
1D case is appropriate to find a conservative solution with higher Von Mises
18
Stresses than expected. The shape of the graph and the point of max stress,
at r=a, is the same between the 1D and 2D cases.
5.00E+07
0.00E+00
r (meters)
2D_Von_Mises_Comsol
1D_Von_Mises_Maple
Figure 23
In Figure 24 is the comparison between the exact 1D solution and the 2D
FEM solution for Von Mises Stress. This shows a 20% difference between the
1D max stress and the FEM 2D max stress, but this error is conservative and
would not put the calculation at risk for providing a false positive margin to
yield strength of safety factor.
Method
2D_Comsol_Extremely_Fine
1D_Maple_Exact
% error of 1D_Maple compared to
2D_Comsol
19
Conclusion
For the problem of evaluating a simplified rotating disk to find displacement and
stress, the results obtained through this project show that FEM using Comsol
Multiphysics can obtain an extremely accurate solution in comparison. In addition,
for simple geometry, an extremely fine mesh size is not necessary. It is possible to
model the 1D rotating disk in less time than it takes to obtain the exact solution
analytically, and when this is expanded to a problem with complex geometry and
transient conditions the required work input to obtain an accurate solution will be
much less in FEM software than using analytical methods. In the case of a rotating
disk, it was also shown that the 1D case, which will generally be quicker and simpler
to evaluate, can show conservative and somewhat accurate results. For the
connection to an aerospace rotor, the simplified case would be useful as a first pass
to understand where stresses and displacements are in a preliminary design before
investing more time into a 2D or 3D analysis. In the case of complex 3D geometry in
rotors, the only method for predicting displacement and stress without actually
testing is FEM, so its accuracy is imperative.
20
Appendix A
Maple Code Input
>
>
>
>
>
>
>
>
>
>
21
>
>
>
>
>
22
Appendix B
1D_DISK_COMSOL.mph
2D_DISK_COMSOL.mph
23
References
http://solidmechanics.org/text/Chapter4_1/Chapter4_1.htm
http://www.ewp.rpi.edu/hartford/~sarric/SMS/Readings/32669_04.pdf
http://asm.matweb.com/search/SpecificMaterial.asp?bassnum=MTP641
http://homepages.engineering.auckland.ac.nz/~pkel015/SolidMechanicsBooks/Part_I
I/04_ElasticityPolar/ElasticityPolars_04_BodyForcesRotatingDiscs.pdf
http://www.doiserbia.nb.rs/img/doi/1450-5584/2006/1450-55840601065A.pdf
http://courses.washington.edu/me354a/Thick%20Walled%20Cylinders.pdf
http://www.hpmsl.neu.edu/content/2012Papers/Haghpanah-ASME-2012.pdf
http://www.solid.iei.liu.se/Education/TMHL61/formula_tables/formulas_table_solid_m
echanics.pdf
https://indigo.uic.edu/bitstream/handle/10027/10185/Chianese_Stefano.pdf?
sequence=1
24