FEA3D : Aluminium BCFWTT RHS beam around test weld

FEA = Finite Element Analysis modelling

This is the previous steel BCFWTT with the material changed to Aluminium (Aluminum) properties

• Elastic "Young's" modulus = 70MPa
• Poisson's Ratio = 0.33

This is a linear-elastic FEA model.

Sunday 04 September 2022 - repeating previous caveat - there has never been an independent review of this FEA model.

BCFWTT = Beam-Configuration Fillet-Weld Tensile Test
As seen physically in the test rig .

Comment in retrospect...

In most ways this is an exercise in pointlessness.

The Finite Element Analysis solution applied is linear-elastic.

Getting away from specialist terminology - that means things (stresses and strains in this case) vary in a simple directly proportional way to inputs applied.

Thus - double the load(s) applied to a structure - you will double the stresses and strains.

Thus; whatever the complexity of the linear-elastic FEA output - and a general method to predict the spectrum of stresses flowing across a structure of any shape was a "holy grail" until a few decades ago - if you increase the forces applied by 1.23X, at any given location in the structure the local stress is 1.23X and the local strain is 1.23X.

So in a linear-elastic FEA solution, if you make the Elastic modulus ("Young's modulus") 1/3rd of the previous value (E(steel)=210GPA; E(aluminium)=70GPa), all the deflections increase by 3X.

Which is exactly what is seen.
That can be looked for and seen comparing the BCFWTT solutions for steel and aluminium (this solution; this page).

• The exaggerated deflection view of the loaded structure - the FEA output offers 13X for aluminium and 40X for steel (40/13~=3)
• The rotation of the Rectangular Hollow Section (RHS) beam wall is 1-in-168 for steel and 1-in-54 for aluminium (168/54~=3)
• the top surface of the RHS deflects upwards by the rotation about the weld (the "pull" on the beam and the "anchoring" of the weld are not in-line, producing a Moment ("twist") causing this), which is 0.0001749m (17.49e-5) for aluminium and 5.85E-05m (5.85e-5) for steel (17.49/5.85~=3)
• etc

The effect of changing the Poisson's Ratio [Wikipedia] (Poisson's Ratio (steel) is taken as 0.3; Poisson's Ratio (aluminium) is taken as 0.33) is too small to see in any obvious way in the results, comparing the simulations for steel and aluminium versions of the Beam-Configuration Fillet-Weld Tensile Test.
Being a linear-elastic solution - if the Poisson's Ratio were unchanged, the stresses at locations for objects of the same geometry and loading would be identically the same.

The way I visualise "linear-elastic" for a linear-elastic finite element solution is - the modelled structure (the structure represented in a parallel world of numbers and equations) is completely unchanged (shape, size) by the applied stress, and the answer is how much the structure would have developed stresses and strains. That hypothetical structure stays completely unchanged (shape, size) by any applied condition. Which is why the predicted answer is always simply proportional to the size of the inputs (load in this case).

Pertinent comment would be that there is no point looking for non-linear effects with a linear model.

The model - how it is constructed

All explanation of the model - see the previous steel BCFWTT model , which is otherwise identical apart from material properties.
ie geometry, constraints, loadings, etc are identical.

Without explanation : redisplayed here:

What has also not been changed is the load applied.

This would give a shear stress of 560MPa on the longitudinal leg of the fillet weld, given

• 63105N longitudinal load (there is/are no other loading(s))
• 1.126869e-04 m^2 area of weld leg - given 6mm fillet leg and 18.8mm length (the plane-of-symmetry imposed makes that a half-length)

The loadings in the steel BCFWTT FEA model are what is known to cause the physical samples to break.
The breaking load and therefore stresses for the steel samples are not what theory predicts.

This makes it difficult to know what load should be correctly applied to the Aluminium BCFWTT model, so that the stresses at the instant of breakage can be examined.
Given no physical test has yet been done.

The simplest choice is the one applied : leave the loading the same as for the steel BCFWTT model.
Being a linear-elastic model, strength is not a property represented (the model material implicitly has infinite strength).
Given the major imponderables, the advantage gained is that the most comparable steel and aluminium BCFWTT test simulations are obtained.

Model's output

All outputs have displacement magnified 40X.
This is the same as for the steel BCFWTT model , producing a more exaggerated portrayal of deformations than usual
(the FEA software offers a magnification of about 13X, given the larger deformations it finds)
but instructing 40X deformation magnification gives easy comparability to the aluminium and steel BCFWTT models.
Being mindful that the very exaggerated deformations possible amount to distortion of the view.

The model is built in SI units, so the output values are SI units.

Displacement in y

von Mises stress

The black regions around plane where weld metal joins the top surface of the RHS is where the model stresses exceed 1GPa.

Rotation of the top surface of the RHS beam

The "Angular displacement" ("rotation") predicted is calculated as seen:

Between Node13 and Node1727, that is a rotation of 1.05degrees, or "1-in-54".

Between Node13 and Node1728, that is a rotation of 1.00degrees, or "1-in-57".

Summary comment

The deformations of the Aluminium FEA model are significantly greater than for the steel FEA model.
Which is expected given the aluminium's Elastic (Young's) modulus is 1/3rd that of steel.

Given the FEA simulation is linear-elastic, any cascading effect ("non-linear effects") of deformations on stresses produced is not taken into account, so no comment is better.

(R. Smith, 04Sep2022, 05Sep2022 (l-e realism, edit final comment))