2 3
10
2
12 3
10000
0.3
Figure‐1 shows the resultant model in Ansys with loads and boundary conditions. The
deflection of the beam is shown in Figure‐2. Resultant vertical deflection (uy) and Stress (σx)
values calculated at each node are tabulated in Table‐1.
1.00
0.1 10
0.1 10
(1)
Figure 1
2‐Node Beam Element with Loads and Boundary Conditions
Figure 2
Deflection in 2‐Node Beam Element after Application of Load
Table 1
(2)
4‐NODE QUADRILATERAL ELEMENT
The same problem was solved in Ansys using 4‐Node Quadrilateral Element. For
meshing purpose, the whole beam was divided into 5 elements of similar size and shape. In
Ansys, this element is named as Plane42 Element. Unit thickness was assumed. The boundary
conditions were applied. The geometry, meshing and boundary conditions are shown in
Figure‐3. After running the solution, the resultant deflection profile is shown in Figure‐4.
Resultant vertical deflection (uy) and Stress (σx) values calculated at each node are tabulated in
Table‐2. (For Node Numbers Refer Figure 5).
In the given theoretical calculations, Vc is the vertical displacement at center of the right
end of the beam. In case of 4‐Node Beam, we cannot have a node at this point so the results
exactly at this location are not calculated. However, we can find the vertical displacement at
this point by averaging out the results at Node No 2 and 7. So
1.016
2
270
10 10
10
2 2
Figure 3
Beam Analysis with 5 4‐Node Quad Elements
(3)
Figure 4
Deflection Profile after Application of Load
Figure 5
Node Numbering for 4‐Node Quad Elements
(4)
Table 2
(5)
4‐NODE QUADRILATERAL ELEMENT (POOR ASPECT RATIO)
The above problem was repeated with same element type but with badly shaped
elements. For this, during meshing, the elements were designed to have poor aspect ratio. The
resultant meshed beam with deflection profile is shown in Figure‐6. Table‐3 shows the resultant
vertical deflection (uy) and Stress (σx) values calculated at each node.
In this case also, we cannot have a node at this point so the results exactly at this
location are not calculated. So, we find the vertical displacement at this point by averaging out
the results at Node No 11 and 12. So
0.99823
2
278.09
19.047 13.333
16.19
2 2
Figure 6
Meshing with Poor Aspect Ratio
(6)
Table 3
(7)
8‐NODE QUADRILATERAL ELEMENT
In Ansys, this element is named as Plane82 Element. The same problem was solved again using
8‐Node Quadrilateral Element. For meshing purpose, the whole beam was divided into 5
elements of similar size and shape. The geometry, meshing and boundary conditions are shown
in Figure‐7. After running the solution, the resultant deflection profile is shown in Figure‐8.
Resultant vertical deflection (uy) and Stress (σx) values calculated at each node are tabulated in
Table‐4. (For Node Numbers Refer Figure‐9). Now finding out the resultant parameters from
the analysis:
1.0105
300
10
Figure 7
8‐Node Quadrilateral Element
(8)
Figure 8
Deflection Profile of 8‐Node Quadrilateral Element
Figure 9
Node Numbering for 8‐Node Quad Element
(9)
Table 4
10
(10)
COMPARISON OF RESULTS
Discussion on Results:
A comparison of results obtained by using different types of elements and mesh shapes,
reveals following general trends: ‐
(a) Accuracy of the results increases with increasing the number of nodes in each
element.
(b) A poorly shaped mesh may give unpredictably wrong results.
With 2‐Node Beam element, we got extremely erroneous results for the stresses. The
main reason for that was the use of only one element for the meshing. As FEM approach
uses piecewise approximation to find the overall effect, using only one element and that
too with very simple assumptions enhances the probability of getting the results far
removed from the actual ones. In this case, although we got acceptable deflection
results; these results are less accurate than other two types (Q‐4 and Q‐8 elements).
The error in results shows a decreasing trend once we shift from Quad‐4 to Quad‐8
element. This is due to the increase in the number of nodes per element. More number
of nodes provides more number of calculation points. This makes the distance between
two calculation points smaller and thus chances of error reduce. Therefore, the overall
best results were obtained from 8‐Node Quadrilateral Element Mesh.
Another point to note here is that once the mesh shape was distorted while keeping all
other parameters same, the meshing resulted in poorly shaped elements with bad
aspect ratios. Using such type of mesh gives highly inaccurate results. The reason for
this is that once we use poor aspect ratios for element, we violate the basic assumptions
of that element which are based on the peculiar shape of that very element. So the
error is incorporated due to violation of shapes of element.
(11)