E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.

1051/e3sconf/202453303002
FORM-2024

Accurate model for simply supported plates

Huy Hoang Vu 1,*
1
Hanoi architectural university, Hanoi, Vietnam

Abstract. The reliability of a model in ABAQUS depends mainly on the

experience of the modeler. For the problem of elastic buckling of a simply
supported thin plate in compression, except for the example in ABAQUS’s
user manual, which only applies to rectangular plates that are symmetrical
in both directions, the current proposed models are personalized and
untested. Therefore, the purpose of this paper is to verify the currently used
models and thereby propose a general model for the problem. For this
purpose, different models are created in ABAQUS, and their reliability will
be asserted by comparing the results with the theoretical solution of
Timoshenko. The critical stresses and buckling coefficients of models are
the values to be obtained and used for comparison. The results show that
the model outlined in ABAQUS’s user manual and the model proposed in
the paper match the theoretical solution, while other models have
significant differences. The proposed model is more general than that of
ABAQUS’s manual, and it is applied to an example of a trapezoidal plate.
The knowledge obtained from the proposed model can be applied to the
modeling of other forms of structure.

Keywords. ABAQUS, modeling, buckling, simply supported, rectangular

plate, compression

1 Introduction
With the great development of structural analysis software, many complicated structural
problems can be simulated and solved with high accuracy. Therefore, the software can be
used to validate the analytical solution. However, to be verification data, the results
obtained from the software must be reliable, in which the user's experience in modeling is
one of the decisive factors.

*
Corresponding author: [email protected]

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons
Attribution License 4.0 (https://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

a) 3D pinned support c) “simply supported” on an edge of a plate

b) “pinned edge” of a plate d) “simply supported” on the circumference of a plate

Fig. 1. Types of support

The calculation model needs to be accurate and consistent with the actual structure.
Models for construction structures mainly use elements connected only through nodes, so
they are comparatively easy to manage. For the plate and solid element problem using
ABAQUS, the boundary condition is more complicated because it involves edges and
faces. For example, the 3D translational constraint of a node in Fig. 1a, when applied to an
edge in Fig. 1b, will produce a "pinned edge", similar to a fixed spindle with infinity
flexural stiffness, which allows the plate to rotate around this edge, but the edge cannot
display any deformation, even in the plane of the plate. This is inconsistent with the
"simply supported edge" in flexural plate theories, with the exact model shown in Fig. 1c,
where the "simply supported edge" would be prevented from out-of-plane displacement,
while in-plane displacements are free. However, if this "simply supported edge" is applied
to Fig. 1d, boundary conditions are insufficient and the plate is unstable in its plane.
Therefore, how to properly model restraints as well as loads in complex ABAQUS models
is a frequently asked question by structural engineers. For the problem of elastic stability of
thin plates, ABAQUS, in its manual [1], conducted a verification and validation problem
with a square plate under compression simply supported around the circumference.
However, the applicability of this example is very limited due to the requirement that the
plate be doubly axially symmetrical in terms of geometry, boundary conditions, and applied
loads. Except for this guide, models established by engineers and researchers are
personalized and unverified.
In order to provide more reliable experiences for engineers and researchers while
modeling in ABAQUS, this paper will model the buckling problem of a thin plate simply
supported around the circumference under compressive loads applied on two opposite
edges, then compare the outcome with the theoretical result conducted by Timoshenko [2],
thereby making comments and proposing an accurate model.

2 Some solutions for modeling the buckling problem of a

thin plate simply supported around the circumference under
compressive loads applied on two opposite edges

1.1 ABAQUS
ABAQUS has tested the problem of a square thin plate simply supported around the
circumference under uniaxial compressive loads and presented it in Section 1.2.4. Buckling
of a simply supported square plate of the part “Analysis Test” of the “Abaqus Benchmarks
Manual” [1]. The length of the edge of the plate is taken as 2, the thickness of the plate is
0.01, and the slenderness of the plate is equal to 200. Thus, it can be considered a thin plate.
Since the plate is doubly symmetrical, only one-quarter of the plate is modeled (upper-right
quadrant of Fig. 2). The boundary conditions on the model are:

2
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

On edge x = 0 ux = ry = rz = 0
On edge y = 0 uy = rx = rz = 0
On edge x = b/2 uz = rx = 0
On edge y = b/2 uz = ry = 0
where the z axis is normal to the plane of the plate.
Loads are applied to the plate using two different methods: uniform edge loads and
thermal loads.
Calculation results using many element types have been considered, such as S8R5, S8R,
S9R5, STRI65, S4R5, S4R, S4, S3R, and STRI3. Results have been compared with those
calculated by the theory of thin plates presented by Timoshenko [2]. Comparisons show
that using different element types will not give completely consistent results. The results
when using elements S8R5 and S9R5 are the most accurate; using S3R is the most
misleading; with other element types, the error is acceptable.

Fig. 2. Calculation diagram in the example of ABAQUS [1]

This model is only used for plates with axial symmetry because at the axis of symmetry,
the in-plane displacement perpendicular to the axis is zero; thus, supports that prevent
in-plane displacement can be assigned to edges associated with the axes of symmetry.
Therefore, the model is not general and is not applicable to plates of any shape.

1.2 Study of Yanli Guo and Xingyou Yao on the buckling of thin
plates
In the study on buckling of thin plates [3], Yanli Guo and Xingyou Yao used ABAQUS
software to model those with round and rectangular holes. The outer edges are set up to
have no out-of-plane displacement, while in-plane displacement and rotation are free. The
two midlines passing through the hole are assigned supports that disallow in-plane
displacements perpendicular to the lines, as shown in Fig. 3.
Similar to the ABAQUS example, this model also takes advantage of the feature that at
symmetry axes, the in-plane displacements perpendicular to the axes are zero, so that
assigns supports that restrain these displacements. Obviously, this model is again only
applicable to doubly symmetrical plates.

3
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

Fig. 3. Calculation diagram in the example of Yanli Guo and Xingyou Yao [3]

1.3 Study of Kanta Prajapat, Samit Ray-Chaudhuri, and Ashwini

Kumar
In the study of the buckling of thin plates with edges partially restrained for in-plane
movement [4], Kanta Prajapat, Samit Ray-Chaudhuri, and Ashwini Kumar used ABAQUS
to model a rectangular plate with the following boundary conditions:
Loaded edge: u ≠ 0, v = 0
The edge opposite the loaded edge: u = 0, v = 0
Other edges: u ≠ 0, v = 0
where u and v are the in-plane displacements along and perpendicular to the direction of the
load.
The element used in the model is S8R5, with a size of 1/40 of the plate width.

1.4 Other models

In Ronald Wagner’s “ABAQUS tutorial” [5], the presenter shows how to use ABAQUS to
calculate the critical stress of a thin steel plate of size 100x100 and 0.1 thick with four
simply supported edges. The supports in the model are of two types: supports prohibiting
out-of-plane displacement and freeing in-plane displacement as well as rotation are
assigned to the four sides of the plate; supports that prevent in-plane displacement are
assigned to two adjacent nodes (nodes A and B in Fig. 4). “Edge Load” is applied to two
opposite sides of the plate. Ronald Wagner's model is shown in Fig. 4.

Fig. 4. ABAQUS model of Ronald Wagner

2 Methods

4
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

2.1 Buckling coefficient determination

The models in the above cases will be studied on a thin plate with dimensions axb =
1000x1000 mm, where b is the width of the plate (the length of the loaded edge) and a is
the length of the plate (the length of the non-loaded edge). The plate thickness t is 10 mm.
The material is steel, with the elastic modulus E = 210000 MPa and the Poisson ratio ν =
0.3.
According to Timoshenko [2], the critical stress of a flat plate is obtained from the
formula:

(1)

where, D is the stiffness of the plate

(2)

in which k is the buckling coefficient and is tabulated with respect to the ratio of the length
and the width of the plate a/b.
From Eq. 1

(3)

Eq. 3 is used to calculate the coefficient k for the studied cases, and the results will be
compared with the theoretical value of Timoshenko. For a square plate, the coefficient k,
according to the theoretical analysis, has a value of 4.0.

2.2 Model selection

In addition to the three models of ABAQUS, Yan Li Guo, and Ronald Wagner, five
more models were used for comparison.

a) Proposed model b) Remove in-plane translational restraints at B

c) Remove in-plane translational restraints at A and B
d) Using “Rigid Body” e) Add longitudinal restraints to edges

Fig. 5. Investigated models

5
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

The first model proposed in this paper is based on the model of Ronald Wagner, with
some modifications. This model keeps the edge supports and the bidirectional in-plane
restraint at corner A while adjusting the support at corner B from preventing displacement
in two directions to preventing displacement in just one direction (see Fig. 5a).
The second and third models are shown in Figs. 5b and 5c. These two models are
derived from the “Proposed” model, but some supports at corners are removed. These two
forms of support do exist in practice, e.g., a plate lays freely on a square rim, but their
computational models lack the in-plane boundary conditions required for model stability.
The fourth model uses the “Contraint-Rigid Body” function of ABAQUS. This feature
is popularly used in ABAQUS to substitute a distributed load on a cross-section for a
concentrated force or moment, or to assign restraint to the whole cross-section. As can be
seen in Fig. 5d, the load uniformly distributed over the edge of the plate is replaced by a
concentrated force. Through the “Rigid body” definition for the edge of the plate, this
concentrated force will be evenly distributed over the edge of the plate, similar to the edge
load.
The final model adds longitudinal restraints to the edges of the slab (see Fig. 5e). These
restraints do not in principle affect the rotation of the edges.

3 Results
Models are established based on the ABAQUS User's Manual [6]. The calculation results of
the critical stress σcr and buckling coefficient k from the studied models are listed in Table
1.
Because the models shown in Figs. 5b and 5c are unstable, their results are not included
in Table 1.
Table 1. The critical stress σcr and buckling coefficient k

Yan Li Ronald Rigid Longit

Vars Units ABAQUS Proposed
Guo Wagner Body Rst
σcr MPa 75.96 75.93 75.39 75.93 161.25 103.38
k 4.00 4.00 3.97 4.00 8.50 5.45
Note:
Columns “ABAQUS”, “Yan Li Guo”, and “Ronald Wagner” are the results of the
models of ABAQUS and the corresponding authors.
The “Proposed” column is the result of the proposed model, as shown in Fig. 5a.
The column “Rigid Body” is the result of the model using ABAQUS's “Rigid Body”
feature (Fig. 5d)
The column “Longit Rst” is the result of the model in Fig. 5e.

4 Discussions
The buckling coefficient k obtained from the model of ABAQUS, YanLi Guo, and the
proposed model has a value of 4.0 and is consistent with the theoretical value. In fact, the
value of k is not exactly equal to 4 (k of ABAQUS is 4.00195 and of the other two models
is 4,00068), due to the fineness of the mesh. That explains why the critical buckling stress
of the ABAQUS model is slightly higher than that of the other two models. The larger the
number of elements, the smaller the size of the elements, and the closer the value of k is to
the theoretical solution. It can be concluded that these three models match the theoretical

6
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

model of the problem of the buckling of a rectangular thin plate with simple supports
around the circumference subjected to a uniform unidirectional compressive load.
The buckling mode shapes corresponding to these three models are shown in Figs. 6a to
6c. The buckling forms of YanLi Guo and the proposed model in Figs. 6b and 6c are in
half-wave form, which is consistent with the theoretical model. The buckling form of
ABAQUS in Fig. 6a is a quarter-wave form because the ABAQUS model is built only in
one quadrant.

a) ABAQUS b) YanLi Guo c) Proposed

d) Ronald Wagner e) Rigid Body f) Longitudinal restraint

Fig. 6. Buckling mode shapes of the investigated models

As pointed out above, the models of ABAQUS and YanLi Guo require the plate to be
symmetrical about two axes, so they cannot be applied to any shape or loading type. Thus,
the proposed model in Fig. 5a is the most accurate and general. This model commits the
boundary conditions in the normal direction of the plate, similar to those in Fig. 5c. In the
plane of the plate, the supports are installed as a simple beam, with one pinned and one
roller support, ensuring the minimum number of boundary conditions for the plate to be
in-plane stable, as shown in Fig. 7a. This modeling approach ensures that the edges do not
move in the out-of-plane direction while they can rotate freely around the edge; therefore, it
matches the requirement for a simple support on the circumference of the plate. Meanwhile,
in the plane of the plate, every point on the edge can freely move, not only in the direction
perpendicular to edges but also in the longitudinal direction of edges, as shown by the
dashed line in Fig. 7a.

a) Proposed model b) Support at B is removed c) Supports at B and A are both removed

Fig. 7. Models and corresponding in-plane deformations or movements

Compared to the proposed model, Ronald Wagner's model in Fig. 4 has an excessive
constraint at the corner B in the direction of edge AB, which prevents deformation of edge
AB in the longitudinal direction. As a result, the critical stress σcr and buckling coefficient k
of this model are smaller than those of the proposed and theoretical models. The difference

7
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

is insignificant because the additional boundary condition affects primarily the AB edge
rather than the whole plate.
As for the models lack of restraints as shown in Figs. 5b and 5c, ABAQUS still
proceeds, but the critical buckling stresses are not reliable as the differences from the
theoretical solution are magnificient, and the buckling form is not the out-of-plane
half-wave but is in-plane rigid body translation and/or rotation (the model in Fig. 5b rotates
around point A as presented by the dashed line in Fig. 7b, and the model in Fig. 5c rotates
around the center of the plate as presented by the dashed line in Fig. 7c). Therefore, these
models are dismissed.
Using a Rigid body similar to the model shown in Fig. 5d has the advantage of
assigning a distributed load to an edge of the plate. But at the same time, the approach
makes the edge a rigid, undeformable edge that is only able to translate or rotate. The
in-plane displacements of the points on the edge are constrained, analogous to adding
excessive constraints to the model in Fig. 5e. The results show that the critical stress σcr and
the buckling coefficient k are both much higher than the theoretical solution, meaning that
these two models are incorrect.
It should be noted that, although the models of Ronald Wagner, the Rigid Body, and
adding longitudinal restraints are incorrect, their buckling forms in Figs. 6d to 6f are still
half-wave shapes, similar to those of the proposed model (Fig. 6c). The difference in the
buckling form of these models is so small that it is difficult to distinguish. Therefore, it is
necessary to correctly assign the boundary conditions and loads from the beginning of
modeling because it cannot rely on deformation to find modeling errors, which is the
method often used when checking the accuracy of the model in practice.

5 Application of the proposed model

The proposed model is applied to solve the buckling problem of an isosceles trapezoidal
steel plate simply supported around the circumference and subjected to compression on two
opposite sides. The dimension of the plate is (900–1100) x 1000 mm (see Fig. 8a). The
plate thickness t is 10 mm. Boundary conditions and loads assigned to the plate are shown
in Fig. 8b. The four sides of the plate are assigned supports to prevent out-of-plane
displacement (along the z-axis). The front corner of the plate is assigned an additional
constraint that disallows displacement in the x and y axes (in the plane of the plate). The
left corner of the plate is also assigned an additional constraint, but it only prevents
displacement in the y-axis. The load on the right edge of the plate (side length 1100 mm)
has a value of 1 N/mm, while the load on the left edge (edge length 900 mm) has a value of
1 N/mm x 1100 mm/900 mm = 1.222 N/mm. Elastic modulus of steel E = 210000 MPa and
Poisson ratio ν = 0.3.
Calculation results by ABAQUS are shown in Fig. 8c.

8
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

b) Boundary conditions and loads

a) Geometries

c) Calculation result

Fig. 8. Dimensions, calculation scheme, and calculation result of the trapezoidal plate

The eigenvalue determined by ABAQUS has a value of 693.07. The greatest stress is on
the left side of the plate (the bottom of the trapezoid, size 900 mm). The critical stress can
be determined by the stress on this edge:

(4)

The critical stress of a 1000-mm-square steel plate is 75.93 MPa (see Table 1). To
compare with the square plate, the average stress at the midpoint of the trapezoidal plate
will be determined. The length of the median of a trapezoid is bavg = (900 + 1100) / 2 =
1000 mm. The internal force on the median is equal to Ncr,avg = 1N/mm x 1100mm /
1000mm = 1.1N/mm. The average critical stress is:

(5)

As can be seen, the critical stress of the trapezoidal steel plate (76.24 MPa) is greater
than that of the corresponding square plate (75.93 MPa).
Therefore, the mean buckling coefficient of the trapezoidal plate must be greater than
4.0.

(6)

9
E3S Web of Conferences 533, 03002 (2024) https://doi.org/10.1051/e3sconf/202453303002
FORM-2024

6 Conclusions
Research shows that the modeling of a plate is much more complicated than that of a line
element. Even small changes in the assignments of restraints and loads can result in a large
error in the calculation results.
For the problem of buckling of a thin plate under uniform compression on two opposite
sides with simple supports on the circumference, the accurate model is shown in Fig. 5a,
conforming to the following principles:
● Constraints that resist out-of-plane translational displacement of the plate are
arranged over the entire circumference of the plate.
● In the plane of the plate, arrange a point where translation is prevented in both
directions and another point that is prohibited from displacement in one direction,
similar to the principle of modeling a simply supported beam.
● An external force is applied on the edge of the plate on the principle that it does
not interfere with the deformation of the plate or its edges; e.g., for a distributed
load, the ABAQUS's edge load can be used.
Because it is difficult to use the deformed shape to evaluate the accuracy of the model,
the modeling process needs to be elaborately conducted, and at the same time, the
constraints as well as the loads must be determined exactly while modeling.

References
1 http://dsk-016-1.fsid.cvut.cz:2080/v6.12/books/bmk/default.htm?startat=ch01s02ach16.html,
(07/05/2023)
2 S.P. Timoshenko, J.M. Gere, Theory of Elastic Stability, second ed., McGraw-Hill, New York,
(1961)
3 Yanli Guo, Xingyou Yao, Buckling Behavior and Effective Width Design Method for Thin
Plates with Holes under Stress Gradient, ResearchGate, (2021), DOI:10.1155/2021/5550749
[https://www.researchgate.net/figure/ABAQUS-boundary-and-mesh-for-the-simply-supported-plate-with-hole-a-Thin-pl
ate-with_fig4_355379008]
4 Kanta Prajapat, Samit Ray-Chaudhuri & Ashwini Kumar, A study on buckling of simply
supported plates with edges partially restrained for in-plane movement, Springer Link,
(2020), DOI:10.1007/s42452-020-03756-1
[https://link.springer.com/article/10.1007/s42452-020-03756-1]
5 https://www.youtube.com/watch?v=1UWD_8GSHsY, Linear buckling analysis of an
isotropic plate with simply supported boundary edges, (07/05/2023)
6 Dassault Systèmes Simulia Corp, Abaqus/CAE User’s Manual, Dassault Systèmes Simulia
Corp., Providence, RI, USA, (2011)

10