Discrepancy Between Natural Frequencies from Prestressed Modal Analysis in Ansys and Custom FEM Code in Octave

[omar ejjary]

Hi everyone,

I’m currently working on the modal analysis of two prestressed concrete beams. I'm using Ansys for the simulation, where I run a static structural analysis to apply prestress, followed by a prestressed modal analysis to extract the natural frequencies.

In parallel, I’ve developed a simplified finite element model in Octave, which doesn’t explicitly account for the prestress (just a classical undamped free vibration model using beam elements).

The problem is:
The natural frequencies I get from Ansys (with prestress included) differ significantly from those I compute in Octave. I'm trying to understand the root cause of this discrepancy.

One hypothesis I’m considering is that the issue might lie in the way I’ve defined the prestress in Ansys. I applied it as a compressive axial load on one side of the beam only, simulating the tendon effect — but I’m unsure whether this correctly represents a realistic prestress state across the entire beam. Could this asymmetric load application be distorting the stiffness in a non-physical way, thus affecting the modal results?

Here are some additional details:
Beam: rectangular concrete beam, ~4,8 m length
Prestress in Ansys: modeled via initial loading (tendon forces)
Octave model: standard Euler-Bernoulli beam elements
Boundary conditions: same in both models

Thanks in advance for your help!

 

[Stick.]

First, I marked this to be moved to the student forum as that is where all student posts should be made.

Second, run your Ansys model with no prestress (and any other necessary changes so it's closer to your Octave model) and then compare the results. It's hard to compare two models if they're assessing fundamentally different things, as least in initial stages while you're troubleshooting. Once you have two models that are in agreement, then start changing one and see if the changes are in a direction that you expect.

 

[GregLocock]

I would not be surprised if the prestress makes no odds, the bending stiffness of a beam with prestressed reo is no different to that of a beam with no prestress in the reo, if the beam is not cracked. The counter argument is an archery bow.

 

[IRstuff]

I would not be surprised if the prestress makes no odds, the bending stiffness of a beam with prestressed reo is no different to that of a beam with no prestress in the reo, if the beam is not cracked. The counter argument is an archery bow.

Obviously, not my wheelhouse, but that seems odd; the pretension ostensibly resists tension applied by a load, so it behaves as if it were "stiffer" so that would seem to me to have an effect on its natural frequency.

 

[GregLocock]

Hence my puzzlement. I don't see the preload popping up in the bending equation. Note I am talking about uncracked beams.

 

[omar ejjary]

Hello,
Without the prestressing force, when I apply external loads on the beam and perform a static structural analysis, the reaction forces at the supports are correct and consistent with manual calculations. This confirms that the definition of the beam and its parameters is correct.
However, in the modal analysis, the first mode is approximately 6 Hz higher than the value obtained from manual calculations and also from my FEM code in Octave.
Why ?