LGCIV2041: Numerical Analysis of Civil Engineering Structures – Assignment

Handout: 19.02.2024 Due: Sunday 31.03.2024

Professors: João Pacheco de Almeida, Hadrien Rattez
Teaching Assistant: Alexandre Sac-Morane

Instructions
• Assignment is graded to 100 points.
• The assignment is composed of 3 problems.
• The assignment should be solved in groups of 2 students.
• Please respect the due date indicated above, beyond which assignments will be accepted with penalty.
• Do not forget to download the supplementary files available in Moodle, “Python script to study frames.py” and
“Solution_Methods_and_Nonlinearity_TO_COMPLETE.py”.
• Hand in your solution together with the scripts and your input files.
Problem 1: Shear Locking, Reduced Integration, and Implementation of a Finite Element (30 points)
In the chapter of the lecture slides “A 2D Timoshenko Beam FE: Development, Characteristics, and Mesh Refinement”, a
Timoshenko beam theory was developed assuming a linear approximation both for the rotations 𝜃(𝑥) and the transverse
displacements 𝑢𝑦0 (𝑥). Similarly to the lecture, a cantilever is also herein investigated: 5 m long, rectangular cross section
of width b = 0.25 m and height h = 1.0 m, shear coefficient k = 5/6, subjected to a tip downward force of 20 kN and a tip
clockwise bending moment of 100 kN.m. Consider a Young’s modulus E = 30 GPa and a Poisson’s ratio  = 0.2, which
could represent concrete.

(a) Derive the exact solution for the transverse displacement field 𝑢𝑦0 (𝑥) and for the rotational field 𝜃(𝑥). Compare them
with the results obtained from the Python script distributed for the first exercise session (“Python script to study frames.py”),
i.e. using Euler-Bernoulli beam theory, using a mesh of 1 and 4 elements. Comment.
(b) Implement the stiffness matrix corresponding to the Timoshenko beam in the Python script distributed for the first
exercise session (“Python script to study frames.py”). Plot the transverse displacements and rotations for a mesh of 1, 4, 10,
and 100 elements, together with the exact solution. Comment.
(10-point bonus: Show and comment on the evolution of the main static quantities, namely the bending moments and shear
forces. Do they satisfy locally the equilibrium conditions? In particular: in the domain, between elements, and the natural
boundary conditions.)
(c) Consider the following alternative cantilever lengths: L=1 m, 10 m, and 100 m. Plot the transverse displacement along
the cantilever for a mesh of 100 elements, both for the Timoshenko and Euler-Bernoulli beam. Comment on the results,
indicating which ones you would “trust” as an engineer without access to the exact solution.

(d) Redo the analyses of question (c) with Abaqus and compare the results.

(e) Compute and show the stiffness matrix considering selective reduced integration, as discussed in the lecture. Implement
it in the Python script and plot again the transverse displacement for the same cases and mesh of question (c). Comment on
the results obtained.
Problem 2: Large Displacements and Solution Methods for Nonlinear Problems (30 points)
Consider the column represented in Figure 1, which was modelled and analysed as a rigid bar during the exercise session.

Figure 1. Bridge pier and structural model with global degrees of freedom, loading and properties.
This problem is a continuation of exercises session and studies the same structure. However, instead of a rigid bar, an elastic
Euler-Bernoulli beam is now used to model the behaviour of the column. Further, it is assumed that a nonlinear plastic hinge
develops at its base, which follows the moment-rotation response given by the Ramberg-Osgood law (similar to the exercise
session). The curve for a specific set of parameters My, 𝜃𝑦 , and 𝛾 is illustrated on the right-hand side of Figure 2.

Figure 2. Ramberg-Osgood law used for the moment-rotation relation of the plastic hinge.
The Python script provided for the exercise sessions (‘Solution_Methods_and_Nonlinearity_TO_COMPLETE.py’) allows
solving the structural problem when linear geometry is considered. Both the classical Newton-Raphson and a displacement-
control method were implemented, which can be selected in the part of the script after the comment ‘INPUT
PARAMETERS: START’.
(a) Compute the force-displacement response using the Python script and compare it, in the same graph, with the response
obtained: (i) during the exercise session, with the rigid bar assuming geometric linearity; (ii) with the distributed Abaqus
input file. Comment.
(b) You are now asked to extend the provided script to account for large displacements according to the corotational
formulation, as addressed in the lectures. In particular: complete the two parts of the code corresponding to the computation
of the element resisting forces in the global reference system p and the element stiffness matrix in the global reference
system k, for large displacements. Using displacement-control, push until a lateral displacement lat = (-)1.2 m. Provide a
physical explanation for the difference in the results with respect to linear geometry. Hand in the completed Python script.
(c) Compare the force-displacement curve with large displacements obtained in (b) with the response obtained: (i) during
the exercise session, with the rigid bar considering nonlinear geometric effects; (ii) with the distributed Abaqus input file,
but now also considering nonlinear geometric effects. Comment.
(d) Identify and comment on the limitations of using the classical Newton-Raphson method in capturing the response, both
with and without geometric nonlinear effects, with respect to that of the displacement-control method. Illustrate your
response with the required plots.

Problem 3: Material Nonlinearity (40 points)

The Single-edge notch-bending test (also called three-point bending test) for fracture toughness testing is a common test to
assess the ability of fractures to propagate in a material. A simplified geometry of the test is shown in Figure 3. The test is
performed by applying a punctual force P at the top of the specimen, while the bottom part is maintained by two supports,
to propagate the initial sharp crack.
In this problem, we will investigate different ways to model the mechanical behavior of a specimen made of steel (E=210
GPa and =0,3) subjected to this test using an elasto-plastic constitutive law in Abaqus.

Figure 3. Single-edge notch bending test seen from below (Source: Wikipedia).

In the following, we consider W=25 mm, B=10 mm, a=W/2 and the crack is modelled as approximated as a triangular
crack with a maximum width of a/10.
(a) As a first step to model the test, simulate the structure depicted in Figure 3 with Abaqus using beam elements. The load
is modelled as a point load with a maximum of 10 kN. The notch is ignored in this question.
(a.1) Which type of elements and beam theory would you choose for this analysis? Justify.
(a.2) What is the displacement obtained for the node in the middle and maximum von Mises stress for different
meshes? For which mesh size we can consider that we have obtained an accurate value.
(b) Simulate the structure in 3D (with the notch) and perform in this case as well a mesh convergence analysis. How does
the displacement compare to (a)? Why?
(c) To decrease the computational time for further calculations, you consider using reduced integration. Simulate the
structure with quadrilateral elements in 3D with two elements in the height of the beam using reduced integration. Plot the
deformed shape of the elements near the boundary conditions using a large deformation scale factor. What do you observe?
How can you prevent this instability?
(d) In a second step, you use an elasto-plastic constitutive law (with a von Mises yield function) for the 3D geometry
considering a yield strength of 200MPa (no hardening or softening are considered here). What is the displacement
(maximum one in the middle of the beam) required to reach plasticity? At the point where plasticity is attained what is the
stress tensor and verify that the yield function F=0, in Abaqus the variable AC YIELD = 1? Plot the force-displacement
(maximum one) curve of the specimen.
(e) You now consider a hardening for the constitutive law with the values define in Table 1. Plot the force-displacement
curve of the specimen in 3D. How is this force-displacement diagram changing while refining the mesh? What is the value
of displacement at which stress the first point reaches plasticity for the different meshes? Why?
Yield stress (MPa) Plastic strain (-)
200 0
240 0.0235
280 0.0474
340 0.0935
380 0.1377
400 0.18
Table 1. Hardening rule (linear approximation between points)

(f) To obtain a more realistic response compared to the experimental results, you consider softening for the material with
the values of Table 2. Abaqus simulation should have convergence issues, what problem are you facing and why? To
overcome this problem, you consider a ramp of displacement applied at the bottom (the previous force application zone)
with a final value of 20% of the height. How is this force-displacement diagram changing while refining the mesh in this
case? Same question with the plasticity field? How can you explain this difference by observing the final deformation of
the whole specimen and how does it differ from the deformation of the specimen with hardening.

Yield stress (MPa) Plastic strain (-)

200 0
160 0.0235
120 0.0474
80 0.0935
40 0.1377
Table 2. Softening rule (linear approximation between points)