Learning Forum

[AR]

Hello,

I am checking the structural integrity of a tool against a PSD. To do this I have performed a structural static analysis, a modal analysis and a random vibration.

Now I want to combine the results of the structural static with the random vibration using the solution combination. I have observed that it only allows me to extract the normal stresses and the shear stresses.

Is there any way to obtain the von mises equivalent stress?
Would it be correct if I export the three normal stresses and the three shear stresses and combine them node by node with an excel to obtain the von mises equivalent using the formula in the image?

If so, would there be any way to re-enter those results into ansys from excel to be able to represent them on my figure? If so, how could I do it?

[dlooman]

That wouldn't be correct because the PSD component stresses are unsigned.  That's another issue.  You are adding signed (static) results to unsigned (PSD) results.  The negative static stresses are cancelling out the unsigned psd stresses.  A third issue is that PSD stresses are probabilistic.  The 1 sigma psd stress or less occurs 67% of the time.  The 2 sigma stress 95% of the time, etc...  The static stress is probably determinate and expected 100% of the time.  If the static load is of the same type as the PSD (acceleration), you could add it to the PSD input curve at a low frequency.  The area you add to the curve should be the square of the static acceleration.

[AR]

Thank you for your response!

I’ve reviewed the information from Ansys and noticed that equivalent stress in random vibration analyses is calculated using the Segalman-Fulcher method not Von Mises method, meaning that combining results with Von Mises wouldn’t be correct either. So why does Ansys allow results to be combined between Random Vibration and Structural Static analysis if it doesn’t make sense to combine unsigned stresses (random vibration) with signed stresses (static)?

Therefore... How could I add the effects of the structural static with the random vibration? Would it be an acceptable approach to add the equivalent stress from the static structural with the equivalent stress from the random vibration (3-sigma)? I mean, in that case, I would add the Von Mises stress with stress values that cover 99.73% of the likelihood of the real stress falling within those limits. This seems like a conservative solution to me. Don’t you think?

Is there another way you could recommend to analyze the combined effect of both tests?

Best regards,

[dlooman]

If you are not using the PSD for a fatigue analysis (the more normal case) and simply want to do an allowable stress evaluation then I think it is reasonable to add the static stresses to the 3 sigma PSD stresses.  There is still the issue that negative static stresses will cancel out the unsigned 3 sigma stresses.  If there was a way to combine the absolute value of the static stresses with the 3 sigma stresses that would be more correct.  That's a feature of APDL POST1 load case combinations.  Some industries do treat the 3 sigma stress as an upper bound (which technically it isn't.)

[AR]

I will also have to calculate fatigue. For this, I have seen the Fatigue Combination module. From what I understand, this module combines the Fatigue Tools from both the static and random vibration analyses (in my case), combining only the damage. Is this correct?

Regarding the stress, I was referring to linearly summing or using SRSS in Excel to combine the Von Mises equivalent stress from the static and vibratory analyses. Since Von Mises stress is always positive, I would always be adding positive values. I had thought of doing this in Excel and then trying to input these values into ANSYS.

Can POST1 be used to combine different analyses results?

[dlooman]

Fatigue is probably based solely on the PSD results (S-N curve usually assumes some mean stress) so there's no combination needed for that.  If you have an S-N curve Mechanical will compute Fatigue damage or life for you.

Adding SEQV in EXCEL gets around the sign issue.  You could do something similar with SUMTYP,PRIN in the APDL POST1 module.  

[AR]

Thank you for the recommendation of the SUMTYPE command. I will look into it to see if it works for my case.

Regarding fatigue, I mentioned it because I want to conduct a fatigue analysis on both the structure and the random vibration analysis. To give you some context, I have a static analysis for a support structure of an airplane engine. In it, I apply the thrust force and the resulting torque. Afterward, I perform a modal and random vibration analysis to simulate the vibration spectrum generated by the engine when it is running.

On one hand, I obtain the stress from the static and random vibration analyses and combine them to evaluate the stresses the structure will endure. Additionally, I want to conduct a fatigue study both in the structural analysis to evaluate how the engine thrust force might affect after several tests, and in the random vibration analysis to assess fatigue due to the engine's inherent vibrations. Therefore, I would need to sum the fatigue effects of both analyses. This is why I want to use the Fatigue Combination module. Would this approach to combining both fatigue results be correct with the fatigue combination?

[dlooman]

Usually it's just the vibratory stress that is considered in a Fatigue analsyis.  So no need to combine the static stress with the PSD stress.  If you have an S-N curve defined in Engineering Data, the gui will support computing a damage factor based on the duration of the event or a life time.

[AR]

Ok, thank you Dlooman!