Why Material Test Data After Necking is Not Replicated in the Solver Material Data?
Altair EmployeeThis article is translated from the following one:
Assume we obtain some test data, as shown in the following figure. After some necking point in the curve, the stress drops significantly. Here is my question: Should we apply the stress drop to the material property? The answer is no. This is the geometry property but not material property. Through this article, we will address this issue in depth.
Stress in material property in reality does not drop, but stress can often be seen to drop after necking in a 'stress/strain' plot.
The header title is slightly misleading. Obviously, it is wrong. But, indeed, strangely, it is also true.
To explain: we do the following material tensile test:
We use /MAT/LAW2,
Here is the yield stress formulation: σ: yield stress, εp: plastic strain, and we assume a=90MPa, b=223MPa and n=0.368 as aluminium. The most important point of this formulation is that the stress always increases when the plastic strain increases.
This figure shows the image of the above material formulation. The stress keeps increasing. If in doubt, you can calculate it in Excel or a similar tool.
And, this is the result of our tensile test simulation. This shows the force-stroke curve. We can see the force drop after stroke=30mm.
Why does it happen?
The answer is necking. When necking occurs, the section becomes narrower and then the object cannot react against an increasing force anymore.
However, of course, the stress in the necking area keeps increasing during the simulation.
We now understand the behavior from the necking causing the force to drop. But, why do we see stress also dropping when clearly it should still be increasing?
Indeed, it comes from a traditional methodology of testing. When we do a tensile test, we express the value that we divide the force by, as the coupon initial section area, Force/initial_Area as "Stress".
This is the cause of the plot leading us to wrongly think that "the stress drops after necking".
We may obtain some Stress-Strain curves from material test engineers. It is a good idea to check how they defined stress and strain before we apply the Stress Strain curve to our material model.
Is it unfair that we treat the test data as stress before necking?
It might feel unfair as the header title. However, I cannot say "absolutely", it is almost always OK.
To understand this, we check the stress triaxiality. When the stress field is one axial tensile field, the triaxiality should be 1/3. We can see the beautiful 1/3 field in a wide area before necking.
And the magnitude of the stress is also uniform.
The stress field has the same orientation and the same magnitude. Thus, all elements in the uniform area have the same stress value that we obtain by dividing force by the initial coupon section area, Force/initial_Area. We can say that the test data and the material property are close enough to identical.
However, after necking, the triaxiality and the stress field becomes varying. What we really should know is the stress and the strain relationship for every element. Even though we can calculate Force/Area_initial, to which element we should apply the value? It is impossible to figure out the material property from this stress field.
Download the exampleThe example Radioss data above: radioss_model.7z
Find more posts tagged with
Comments
Altair EmployeeHello, please remind that I am also the one of the Radioss expert.
Altair EmployeeTitle:
Material Test Data after necking is not replicated in the material data
Assume we obtain some test data, as shown in the following figure. After some necking point in the curve, the stress drops significantly. Here is my question. Should we apply the stress drop to the material property? The answer is no. This is the geometry property but not material property. Through this article, we will address this issue in depth.
Stress in material property in reality does not drop, but stress can often be seen to drop after necking in a 'stress/strain' plot.
The header title is slightly misleading. Obviously, it is wrong. But, indeed, strangely, it is also true.
To explain: we do the following material tensile test,
We use /MAT/LAW2,
Here is the yield stress formulation. σ: yield stress, εp: plastic strain, and we assume a=90MPa, b=223MPa and n=0.368 as aluminium. The most important point of this formulation is that the stress always increases when the plastic strain increases.
This figure shows the image of the above material formulation. The stress keeps increasing. If you still have doubt, you can calculate it in Excel or similar tools.
And, this is the result of our tensile test simulation. This shows the force-stroke curve. We can see the force drop after stroke=30mm.
Why does this happen!?
The answer is necking. When necking occurs the section becomes narrower. And then, the object cannot react against an increasing force anymore.
However of course, the stress in the necking area keeps increasing during the simulation.
Ok, we now totally understand the behavior from the necking causing the force to be dropping. But, why do we see stress also dropping when clearly it should still be increasing? Indeed, it comes from a traditional methodology of testing. When we do a tensile test, we express the value that we divide the force by, as the coupon initial section area, Force/Area_initial as "Stress".
This is the cause of the plot leading us to wrongly think that "the stress drops after necking".
We may obtain some Stress Strain curves from material test engineers. It is a good idea to check how they defined stress and strain before we apply the Stress Strain curve to our material model.
Is it unfair that we treat the test data as stress before necking?
You might feel unfair as the header title. However, I cannot say "absolutely", it is almost always OK.
To understand this, we check the stress triaxiality. When the stress field is one axial tensile field, the triaxiality should be 1/3. We can see the beautiful 1/3 field in a wide area before necking.
And the magnitude of the stress is also uniform.
The stress field has the same orientation and the same magnitude. Thus, all elements in the uniform area have the same stress value that we obtain by dividing force by the initial coupon section area, Force/Area_initial. We can say that the test data and the material property are close enough to identical.
However, after necking, the triaxiality and the stress field becomes varying. What we really should know is the stress and the strain relationship for every element. Even though we can calculate Force/Area_initial, to which element we should apply the value? It is impossible to figure out the material property from this stress field.
Download the example
The example Radioss data above: radioss_model.7z
@altair_fukuoka I have asked a Radioss expert to do an editorial review of this discussion before it is published. Apparently, there is a regular Tuesday meeting to discuss Community questions and Pierre Christophe Masson will bring the editorial review up at that time.