Beruflich Dokumente
Kultur Dokumente
by
Parth B Patel
Northeastern University
Department of Mechanical and Industrial Engineering
Boston, MA 02115
1|P age
1. Executive Summary
Here, in this project, the Mechanical APDL tool by ANSYS is explored and used to
evaluate stress- strain properties for different element/ mesh types in a cantilever beam and
in a thin regantangular plate with rectangular notches in it.
For the beam, different solid element type such as Q4 (Plane 182) and LST (Plane 183)
with Beam188 were used. Moreover, different mesh sizes were applied related to the edge
length as proposed by the professor. The detailed description of the problem is included into
the next section of this report. Both the elements are symmetric about their Global X axes. So,
in the rectangular plate only, the analysis was done by cutting the plate from the X axes and
apply applicable boundry conditions (BCs) at that end. The guidelines for applying the BCs are
outlined in the guideline section of the problem PDF. The units are kept in SI. In both the
cases, the material is linear elastic and cross sections stays plane were assumed to compare
the resultes with basic Mechanics problem (Euler Bernoulli / Timoshenko Beam theory). The
result table shows the results of this experiment. For the thin plate, Q4 and LST element types
are tested. In this problem, the deflection and different stress were determined for the given
problem. Again, the goal of this plate problem is to seek the effect of FEM solution parameters
to the predefined resultes at P1 and P2.
The results and numbers of this project are the same to Shaival Shah and Nikhil Patel as
we did modeling together, but explanations, wordings, and this final report is unique as it’s
made by myself.
2|P age
2. Brief Introduction of the Theory
The problem statement can be summerized showing the figure below. The cantilever beam is a
beam supported fixed at its one end at x = 0 which produces support reaction force and
moment. The most used theory in the beam mechanics is Euler Bernoulli (EB) theory. The
equations can be used for short and long beams and approximated for curved beams.
Initial Problem 1
As discussed before I used beam element to Plane 188 and solid elements to Q4 (plane 182)
and LST (Plane 183) elements for different mesh sizes. When we compare the results I get from
analysis with the EB equation (shown below), we can see some difference in them.
Bilinear quadrilateral and linear triangular elements are a 2D FE elements which are used to
solve mechanics of elastic materials’ for plane stress - plane strain sums. The bilinear
quadrilateral element is 4 nodes rectangle element. Linear trianglular element has 3 nodes. The
node number is given counterclockwise for any element type from any point. These elements
keeps the same E, v, and thickness as of whole.
3|P age
The Bilinear Quadrilateral element has its own limitations. For example, due to the bilinear
displacement, strains will vary in two directions. Also, the edges remain straight and no curving
occures (see figure below).
However, the good part of it, particularly for this problem is, the Q4 elements can be very stiff
for bending on developing normal and shear stresses. So, mesh convergence study proves
rectangular better on triangular mesh for this type of problems.
Image Reference:
https://www.researchgate.net/publication/320124305_A_smoothed_finite_element_method_
for_exterior_Helmholtz_equation_in_two_dimensions/figures
The thin plates probles are solved using plane stress theory.They are simple to solve because
acting forces are parallel to plane. The plane strain theory is used for thick members unlike thin
plates.
4|P age
The stress is 1000 Pa in positive X axes. The Degree of Freedom is set for the problem where we
cut it as explanied before. (see figure below where the plate is symmetric about the X axes.) It is
just to ease finding stresses at given points.
The force applied on the right face is as a stress. In order to observe the mesh convergance
study and to obtain monotonic convergence in the deformation results, the element edge
length was changed from 0.5a to 0.1a.
Initial Problem 2
Plates with rectangular openings develop stress concentrations under bending. While these
stresses can be determined using finite elements, in many problems this would be difficult
because a high density mesh would be needed in the neighborhood of every opening corner as
discussed before in this section.
5|P age
3. Results and Discussion
Q1 Answer to Q1
1 2 4 8
VFEA / VExact
σFEA / σexact
6|P age
BEAM ELEMENT TYPE (BEAM 188) 1 ELEMENT
END DEFLECTION
7|P age
Q4 ELEMENT 1 ELEMENT
END DEFLECTION
8|P age
9|P age
Q4 ELEMENT 8 ELEMENT
DEFLECTION
10 | P a g e
11 | P a g e
Q8 ELEMENT TYPE 1 ELEMENT
END DEFLECTION
12 | P a g e
13 | P a g e
Q8 ELEMENT 8 ELEMENT
END DEFLECTION
14 | P a g e
15 | P a g e
VON MISES STRESS FOR THE DENSEST MESH FOR Q4 ELEMENT
16 | P a g e
CONCLUSION
We can conclude that the Q8 element is more suitable to perform the analysis for these types
of problems because values of displacement we obtained from Ansys analysis is closer to the
exact solution given by the EU theory. But when we refine the mesh for Q4 element type, the
deflection was more accurate. This is because the displacement function keeps the edges
straight during the deformation as discussed in the theory, this makes it very stiff for bending.
Concludes the most problems works decent than the Q8 element.
Also, this beam is short. And for the short beams, there are some limitations using the EB
theory as this theory is not using the shear stress. We have to use Timosenko Beam theory here
again.
17 | P a g e
Q2 Effect of Rectangular Holes on Stress Distribution in Plane Stress Problem
Q4 Element
Mesh size Node no of P1 uk1 Node no uk2
of P2
0.01 16 -1.5865E-09 43 3.4374E-07
-1.5900E-09
-1.6013E-09
-1.6000E-09
-1.6100E-09
-1.6200E-09
-1.6600E-09
Element Edge length
18 | P a g e
Plot uk2 for Q4 ELEMENT
3.5000E-07
Displacement of P1 in X-direction
3.4500E-07
3.4000E-07
3.3500E-07
3.3000E-07
3.2500E-07
3.2000E-07
(uk1)
3.1500E-07
3.1000E-07
3.0500E-07
3.0000E-07
2.9500E-07
0 0.01 0.02 0.03 0.04 0.05 0.06
Element Edge length
Plot sigma x, sigma y and tau xy at P1 and P2 as a function of mesh size for each element type
Q4 Element
Mesh Node no sigmaxP1 sigmayP1 tauxyP1 Node no sigmaxP2 sigmayP2 tauxyP2
size of P1 (Pa) (Pa) (Pa) of P2 (Pa) (Pa) (Pa)
0.05 16 115.07 407.58 132.77 43 3498.4 20512 -3793.8
0.04 20 89.62 412.55 99.542 55 2837.3 21607 -3044.7
0.03 26 63.658 419.16 79.889 72 2491.9 22236 -2496.3
0.02 37 43.932 412.46 50.496 106 2226.7 22685 -1678.4
0.01 72 21.592 415.24 27.138 207 1655.2 23054 -877.58
Sigma x at P1
90
75 63.658
60 43.932
45
21.592
30
15
0
0.05 0.04 0.03 0.02 0.01
Element edge length
Sigma y at P1
19 | P a g e
sigmayP1 for Q4 Element
425
419.16
420
415.24
sigmay at P1
410 407.58
405
400
0.05 0.04 0.03 0.02 0.01
Element edge length
Tau xy at P1
80
50.496
60
40 27.138
20
0
0.05 0.04 0.03 0.02 0.01
Element edge length
Sigma x at P2
2226.7
2500
2000 1655.2
1500
1000
500
0
0.05 0.04 0.03 0.02 0.01
Element edge length
20 | P a g e
Sigma y at P2
22000 21607
21500
21000 20512
20500
20000
19500
19000
0.05 0.04 0.03 0.02 0.01
Element edge length
Tau xy at P2
-2000
-2496.3
-2500
-3044.7
-3000
-3500 -3793.8
-4000
Element edge length
21 | P a g e
For Q8 Element type
Q8 Element
Mesh size Node no of P1 uk1 Node no of uk2
P2
0.01 16 -1.5463E-09 43 3.4797E-07
0.02 20 -1.5744E-09 55 3.4661E-07
0.03 26 -1.6023E-09 72 3.4446E-07
0.04 37 -1.6311E-09 106 3.4299E-07
0.05 72 -1.6738E-09 207 3.4033E-07
-1.6000E-09
-1.6200E-09 -1.6311E-09
-1.6400E-09
-1.6600E-09 -1.6738E-09
-1.6800E-09
-1.7000E-09
Element Edge length
3.4797E-07
3.4800E-07 3.4661E-07
3.4600E-07
3.4446E-07
3.4400E-07 3.4299E-07
3.4200E-07
3.4033E-07
3.4000E-07
3.3800E-07
3.3600E-07
0.01 0.02 0.03 0.04 0.05
Element edge length
22 | P a g e
Plot sigma x, sigma y and tau xy at P1 and P2 as a function of mesh size for each element type
Q8 Element
Mesh Node sigmaxP1(Pa) sigmayP1 tauxyP1 Node sigmaxP2 sigmayP2 tauxyP2
size no of (Pa) (Pa) no of (Pa) (Pa) (Pa)
P1 P2
0.05 30 -0.50535 401.85 0.42905 84 1062.3 22741 63.154
0.04 38 -0.46486 406.17 0.52976 108 1032.9 22877 40.859
0.03 50 -0.34897 408.03 0.4228 142 1028.1 22908 28.859
0.02 72 0.24154 410.39 0.2586 210 1012.6 23052 14.454
0.01 142 -0.074202 412.22 8.52E-02 412 1003.7 23120 4.2651
Sigma x at P1
0.2
0.1
-7.42E-02
sigma x at P1
0
-0.1 0.05 0.04 0.03 0.02 0.01
-0.2
-0.34897
-0.3
-0.4 -0.46486
-0.50535
-0.5
-0.6
Element edge length
Sigma y at P1
408 406.17
406
404 401.85
402
400
398
396
0.05 0.04 0.03 0.02 0.01
Element edge length
23 | P a g e
Tau xy at P1
0.3 0.2586
0.2
8.52E-02
0.1
0
0.05 0.04 0.03 0.02 0.01
Element edge length
Sigma x at P2
1030
1020 1012.6
1003.7
1010
1000
990
980
970
0.05 0.04 0.03 0.02 0.01
Element edge length
24 | P a g e
Sigma y at P2
23000 22908
22877
22900
22800 22741
22700
22600
22500
0.05 0.04 0.03 0.02 0.01
Element edge length
Tau xy at P2
40 28.859
30
14.454
20
10 4.2651
0
0.05 0.04 0.03 0.02 0.01
Element edge length
25 | P a g e
2. Q4 Contour plot of sigma x obtained with mesh size of 0.1a for each element type
Contour plot of sigmay obtained with mesh size of 0.1a for each element type
26 | P a g e
Contour plot of tauxy obtained with mesh size of 0.1a for each element type
27 | P a g e
3. Q8 Contour plot of sigma x obtained with mesh size of 0.1a for each element type
Contour plot of sigma y obtained with mesh size of 0.1a for each element type
28 | P a g e
Contour plot of tau xy obtained with mesh size of 0.1a for each element type
In Q8, there is very less displacement on point 2 than to point1. Again, on refining the mesh
size, Q8 element with the triangular mesh is better over Q4 element. I would say, one had
better not to use Q4 for dense mesh size because it will give large stiffness. On the other hand,
Q8 elements allows curve between nodes and thus we can minimize the error in the shear
strain calculation.
29 | P a g e
On further Refining the Mesh?
When we are refining the mesh more and more, the results obtained is almost the same. With
such varying mesh sizes, we can see the trend of the finite element analysis. We can have more
accurate result, but it will take more computing power will increase the cost.
30 | P a g e
4. Summary and Conclusions
It can be concluded that the Q4 is accuarte for analysis to coarser mesh size due to its property
of stiffness but for smaller mesh sizes Q8 are used here to analysis, it is good to use Q8 element
but it will take more computational time for the processing. Hence, everything depends on the
model shape and mesh size, which changes problem by problem.
31 | P a g e
References
32 | P a g e