Department of Mechanical Engineering
MEC 313: Finite Element Techniques
A Computational Study of Clevis
Performance under Tension using
Finite Element Analysis
Safwan ZAYED
17/05/2023
2195
Executive Summary
This report presents the findings of a Finite Element Analysis (FEA) on Clevis under
tension, focusing on mesh selection, analysis techniques, and their comparison with
experimental measurements obtained through a Digital Image Correlation (DIC). The study
detailed the modelling strategy adopted, which included the steps taken to simplify the geometry
and the applied boundary conditions, the choice of using both a simplex tetrahedral mesh and a
complex hexahedral mesh, as well as the performance of both nodal and elemental analyses to
find the strain values along the area of interest on the Clevis mechanism
A mesh-independence study was performed on both meshes to verify their results. The
convergence graphs demonstrated that the complex hexahedral mesh outperformed the
simplex tetrahedral mesh, achieving convergence at a significantly smaller number of elements.
This advantage was attributed to the complex mesh's ability to capture geometric details and
conform to irregular shapes more efficiently, as compared to the simplex mesh.
The FEA results were also validated by comparing them to the DIC’s results. The
Comparisons between FEA results and DIC measurements revealed some discrepancies in
maximum strain values. These differences were attributed to factors such as assumptions and
simplifications in FEA, uncertainties in material properties, and inherent limitations of the DIC
technique. The overall contours of the strain variables, however, were similar for both the FEA
and DIC results.
The comparison between the nodal and elemental analyses highlighted variations in the
smoothness of the resulting strain contours. The nodal analysis produced smoother contours for
both mesh types. However, it was particularly noticeable in the simplex tetrahedral mesh, where
the results obtained through the nodal analyses produced significantly smoother contours when
compared to the elemental analysis.
Overall, this study highlighted the importance of mesh selection, and choosing suitable
analysis techniques on the produced results’ similarity to experimentally obtained data, and
resulting strain contours. It also highlighted the importance of considering the assumptions
made during each step in the modelling stage, and understanding their effects on the final FEA
results.
Contents
1 Introduction............................................................................................4
2 Modelling strategy and model construction...........................................5
2.1 Geometry....................................................................................................................................5
2.2 Mesh.............................................................................................................................................5
2.2.1 Simplex Mesh..................................................................................................................6
2.2.2 Complex Mesh................................................................................................................7
2.3 Boundary Conditions...............................................................................................................7
2.4 Material Properties..................................................................................................................8
2.5 Analysis Type.............................................................................................................................9
2.5.1 Nodal Analysis................................................................................................................9
2.5.2 Elemental Analysis........................................................................................................9
3 Results, Verification and Validation......................................................10
3.1 Mesh-Independence..............................................................................................................10
3.2 Validation..................................................................................................................................11
3.3 Nodal vs Elemental................................................................................................................12
4 Analysis and results discussion............................................................14
4.1 Mesh-Independence Discussion........................................................................................14
4.2 Validation Discussion............................................................................................................14
4.3 Nodal and Elemental Discussion......................................................................................15
5 Conclusions..........................................................................................16
References.....................................................................................................17
1 Introduction
FEA is a computational technique, which involves dividing a complex structure into
smaller elements and solving for the behavior of each element under a given set of
conditions. The results are then combined to give an approximation of the behavior of the
entire structure. A key challenges in FEA is the large dependence on computational
requirements. Therefore, it is essential to find ways to optimize the model to reduce
computational resources while still maintaining accurate results.
This study investigated the deformation of a Clevis mechanism, shown in Figure 1,
under a tensile test with a 50 kN tension force, using Finite Element Analysis (FEA). The
goal of the project was to create an FEA model of the Clevis mechanism and investigate its
deformation under tension, where the region of interest was around the Clevis’s hole.
Additionally, different measures which can reduce computational resuources were
investigated. The results of the FEA were then validated by comparing it to that of a Digital
Image Correlation (DIC), and verified through a mesh-independence study.
Figure 1: Clevis mechanism
2 Modelling strategy and model construction
2.1 Geometry
The geometry of the Clevis mechanism was simplified by leveraging its inherent
symmetry. Specifically, given the symmetry of the mechanism about its y- and x-axis, only a
quarter of its geometry was used during the analysis. By modelling only fourth of the
geometry, the computational effort was reduced, without significantly reducing the physical
accuracy of the simulation. The simplified geometry is shown in Figure 2.
Figure 2: Simplified geometry used for FEA
2.2 Mesh
Meshing is a critical process in FEA, whereby the geometry of an object is divided into
finite elements that can be analysed numerically. The quality of the mesh is crucial, since an
appropriate mesh size and element type must be chosen to balance the competing demands
of accuracy and computational efficiency.
To optimize the computational efficiency of the FEA analysis, the geometry of the
clevis was divided into multiple sections around the hole, as shown in Figure 3. This allows
for the creation of mesh elements that conform better to the local curvature, resulting in
improved element quality, reduced distortion, and better approximation of the curved
surface. Additionally, sectioning enabled greater control over mesh density, allowing for
localized refinement near the area of interest to capture necessary details.
Figure 3: Sectioning of the geometry
Two types of meshes were compared: a simplex mesh with tetrahedral elements, and
a complex mesh with hexahedral elements. For both meshes, sections 1 to 5 had the same
element size, and the element size for section 6 was 1.3 times that of the other sections.
This was to done to refine the mesh near the area of interest, while minimizing
computational loss.
2.2.1 Simplex Mesh
Figure 4 shows the geometry after applying a simplex mesh. A 3D simplex mesh
consists of tetrahedral elements, where each element is a four-faced polyhedron composed
Figure 4: Tetrahedral simplex mesh
of straight edges and triangular faces. The patch conforming method was used to make the
simplex mesh. It consists of dividing the object into smaller patches and decomposing each
patch into tetrahedra that fit together smoothly. This technique is suitable for modelling
irregular surfaces without gaps/overlaps.
2.2.2 Complex Mesh
Figure 5 illustrates the complex mesh with hexahedral elements applied to the
geometry. Hexahedral complex meshes can often provide a more compact representation of
the domain due to their ability to capture the geometric details and conform to irregular
shapes more efficiently. The Sweep Method was employed to generate the complex mesh
with hexahedral elements. This method involves extruding a base shape along a defined
path to create structured mesh elements.
Figure 5: Hexahedral complex mesh
2.3 Boundary Conditions
The boundary conditions’ function is to represent the interactions between the different
components in the design. It was critical that any assumptions made during the definition of
the boundary constraints do not lead to substantial deviation from the real testing conditions.
Figure 5 highlights all the boundary conditions applied in this study.
In this project, the tensile forcing was assumed to be a static. A static-load assumption
can be made when the loading is applied slowly, so that no vibrations or dynamic effects
develop in the structure. Since the loading rate was slow for the tensile test, this assumption
was accurate to be used in the FEA. The value for the forcing was 12.5 kN, a quarter of the
original forcing, as only fourth of the geometry was simulated.
Two displacement boundary conditions were set to account for the geometry’s
simplification. When utilizing symmetry, the displacement boundary conditions are set to the
portion of the surface that is symmetrical to its mirror image, constraining displacement and
forces to zero. Each displacement boundary condition accounted for the symmetry across
different axes.
(a) (b)
Figure 6: Illustration of the boundary conditions ((a): Force and fixed support, (b):
Displacement constraints)
In this study, the base of the geometry was assumed to be fixed, which is reasonable
given the nature of the tensile test where the mechanism is secured from one end and
subjected to a force from the other.
2.4 Material Properties
The materials used in this project were Aluminium Alloy, and Stainless Steel. The
values for these elements were given, and are shown in Table 1. Certain assumptions about
the material properties had to be made. The manufacturing imperfections, for instance, were
assumed to be negligible. This may not always be accurate, as different manufacturing
conditions can significantly alter the material’s performance. However, since it’s not possible
to control the manufacturing conditions in the FEA, it was necessary to make this
assumption.
Material Young’s modulus Yield strength Ultimate tensile
strength
(GPa) (MPa) (MPa)
Aluminium alloy
68.9 260 310
(Clevis)
Stainless Steel
200 230 600
(Pin and attachments)
Table 1: Material properties used in FEA
2.5 Analysis Type
In FEA, the strain components along the x and y axes are determined by differentiating
the displacement field with respect to the respective coordinates, utilizing strain-
displacement relations. Similarly, the shear strain is calculated using a similar approach,
relating the displacement gradients to the shear strain components. The strain values
obtained from the FEA were used to evaluate the structural performance of the Clevis
mechanism. Two analysis types were performed and compared, including nodal and
elemental analysis.
2.5.1 Nodal Analysis
Nodal analysis is used to determine the average values of the different strain
components. In nodal analysis, the displacement values of the nodes surrounding each
integration point are used to calculate the strain values. These values are then averaged
over all the nodes within the integration point to obtain an averaged value for each
component of strain. By averaging the values, nodal analysis can provide a more accurate
and smoother estimate of the strain field [1].
2.5.2 Elemental Analysis
Elemental analysis is another approach used in FEA to determine the strain values in
an element. Unlike nodal analysis, elemental analysis provides unaveraged results for each
strain component at specific locations within the element. The resulting strain values are
reported at predetermined locations, such as the element's integration points. Elemental
analysis provides detailed information about the strain distribution within the element, which
can be useful in understanding how the structure behaves under different loading conditions,
but they may be more prone to noise/errors at individual integration points [1].
3 Results, Verification and Validation
3.1 Mesh-Independence
Figures 7 and 8 show the mesh-independence study performed on both the simplex,
and complex meshes, respectively.
Simplex Mesh
1.40E-03
1.20E-03
1.00E-03
Nodal Results
8.00E-04 eyy
exx
6.00E-04 exy
4.00E-04
2.00E-04
0.00E+00
0 100000 200000 300000 400000 500000 600000 700000
Element Size
Figure 7: Mesh-independence graph for simplex mesh
Complex Mesh
1.40E-03
1.20E-03
1.00E-03
Nodal Results
8.00E-04 eyy
exx
6.00E-04 exy
4.00E-04
2.00E-04
0.00E+00
0 20000 40000 60000 80000 100000 120000
Element Size
Figure 8: Mesh-independence graph for complex mesh
3.2 Validation
Figure 9 compares the results obtained from the simplex and complex meshes with the
results obtained using the DIC.
Simplex (Nodal) Complex (Nodal) DIC
Eyy
Maximum = 1.2538E-3 Maximum = 1.2833E-3 Maximum = 0.79E-3
Exx
Maximum = 1.1014E-4 Maximum = 1.9606E-4 Maximum = 1.65E-4
Exy
Maximum = 4.7951E-4 Maximum = 5.2121E-4 Maximum = 10.4E-4
Figure 9: Comparison of simplex, complex, and DIC results.
3.3 Nodal vs Elemental
In Figure 10, the nodal and elemental results are shown for the simplex mesh, as to
compare them.
Nodal (Averaged) Elemental (Unaveraged)
Eyy
Exx
Exy
Figure 10: Comparison of nodal and elemental results for simplex mesh
Figure 11 highlights the differences of results obtained through the nodal and
elemental analysis, using the complex mesh.
Nodal (Averaged) Elemental (Unaveraged)
Eyy
Exx
Exy
Figure 11: Comparison of nodal and elemental results for complex mesh
4 Analysis and results discussion
4.1 Mesh-Independence Discussion
The convergence graphs for the simplex and complex meshes, shown in Figures 7 and
8, respectively, provide valuable insights into the convergence behaviours of the mesh
types. The results reveal a significant difference in convergence characteristics between the
two mesh types. The complex hexahedral mesh achieved convergence much faster, with a
smaller number of elements of around 60,000, while the simplex tetrahedral mesh required a
finer mesh with a total number of elements of approximately 400,000 for convergence.
This discrepancy in convergence behaviour can be attributed to the inherent
dissimilarities between the mesh types. The complex hexahedral mesh, known for its
superior capability to capture geometric details and conform to irregular shapes, exhibited a
higher level of accuracy per element. Consequently, it achieved convergence at a coarser
resolution, necessitating fewer elements overall. This shows the advantage of using
hexahedral elements for problems involving complex geometry, as they offer more efficient
and accurate representation of the domain. Conversely, the simplex tetrahedral mesh,
although providing flexibility and ease of mesh generation, necessitated a finer element size
to achieve convergence, which suggests that it demands greater computational resources,
as compared to the hexahedral complex mesh.
4.2 Validation Discussion
The strain contours obtained from both the simplex tetrahedral mesh and complex
hexahedral mesh exhibited overall similar patterns to the Digital Image Correlation (DIC)
results, across all strain values, as shown in Figure 9. The percentage difference between
the maximum strain value along the y-direction obtained from the FEA and DIC
measurements was approximately 58% for the simplex mesh and 62% for the complex
mesh. While for the strain in the x-direction, the percentage difference was 33% and 19%,
for the simplex and complex meshes, respectively. As for the shear strain, the difference
was 54% and 50%, respectively.
The observed differences between the FEA and DIC results can be attributed to various
factors. Firstly, it is important to note the inherent limitations of the FEA method. Finite
element analysis relies on assumptions and simplifications in terms of material behaviour,
boundary conditions, and mesh discretization. These assumptions can introduce
discrepancies between the idealized FEA model and the actual behaviour of the tested
specimen. Furthermore, the DIC technique itself has certain limitations that can contribute to
the observed differences. DIC measurements rely on image analysis, which can be affected
by factors such as image resolution, image quality, and the presence of noise. These factors
can play a crucial role in obtaining accurate DIC strain results.
The similarity in strain contours between the FEA results from both simplex tetrahedral
and complex hexahedral meshes and the DIC measurements is encouraging. However, the
observed percentage differences indicate that further analysis and consideration of the
aforementioned factors are necessary for a better match between the two methods.
4.3 Nodal and Elemental Discussion
Figures 10 and 11 showcase the obtained results from the nodal and elemental
analyses, for both mesh types. The observed differences in the resulting contours between
elemental and nodal analyses was found to be more pronounced in the case of the simplex
tetrahedral mesh, whereas the complex hexahedral mesh exhibited relatively similar
smoothness for both analyses. Specifically, for the simplex mesh, the nodal analysis yielded
smoother contours compared to the elemental analysis, while the complex mesh
demonstrated consistent smoothness in both analyses.
These differences can be attributed to several factors. For instance, nodal analysis
calculates the strain values based on the displacement values of the nodes surrounding
each integration point. The averaging effect of nodal analysis can help smooth out local
variations in strain within the element, resulting in smoother contours. On the other hand,
elemental analysis provides unaveraged strain values at specific locations within the
element, potentially capturing more localized variations and resulting in less smooth
contours.
The variation in smoothness between the simplex and complex meshes can be
attributed to their inherent characteristics. The simplex tetrahedral mesh is composed of
irregular and varying tetrahedral elements, which can lead to larger variations in strain
values between elements. This variation can become more apparent in elemental analysis,
resulting in less smooth contours. In contrast, the complex hexahedral mesh with its regular
and structured elements provided a more consistent strain distribution, resulting in similar
smoothness for both elemental and nodal analyses.
5 Conclusions
In conclusion, this FEA project emphasized the importance of mesh selection, analysis
techniques, and their comparison with experimental measurements. The complex
hexahedral mesh outperformed the simplex tetrahedral mesh in terms of convergence rate
with fewer elements. Discrepancies were observed when comparing FEA results with Digital
Image Correlation (DIC) measurements, with differences in maximum strain values. This
discrepancy may stem from the methods’ inherent limitations. Additionally, contour analysis
showed variations in smoothness between elemental and nodal analyses, with nodal
analysis producing noticeably smoother contours for both mesh types.
References
[1] K.-J. Bathe, Finite element procedures, 2nd ed. Englewood Cliffs, N.J.; London:
Prentice-Hall, 1996.