Frequency Response values FREQi
Dear Altair community,
I just want to be sure about something in FRF FREQi specified values. I know that when you are doing a Frequency Response Analysis, then you are computing the structure's response at the frequency points specified in FREQi cards. But if I use the computed FRF in a Random Response Analysis which basically multiply the square of this FRF by a (PSD) input, then I know that if a computed value of the FRF (let's say for f=10) which is not a specified value in the PSD, then it will be computed either with linear or logarithmic interpolation. But on the contrary, if a PSD point is specified at f=10 whereas f=10 is not a specified value of the FRF specified in the FREQi card, then will it be just ignored? Only specified frequency values in FREQi are used, so if a point on the Input PSD does not line up with a point on the Transfer Function, that Input PSD point is ignored in the Response PSD calculation? Or is it interpolated too? I have found nothing about it in OptiStruct help.
Thank you /emoticons/default_smile.png' srcset='/emoticons/smile@2x.png 2x' title=':)' width='20' />
Answers

Hi,
Ideally, in simulation the forcing function or PSD should be spread across FRF response frequency.
For a broader understanding there can be four cases as follows:
1. The FRF response frequency and PSD frequency matches, then solver will use the same value of PSD.
2. The PSD frequency lies between two FRF response frequency, then solver will use interpolated PSD value.
3. The PSD start frequency is greater than FRF response start frequency, then solver will extrapolate the PSD values which might lead to negative PSD value and hence solver will terminate the run.
4. The PSD end frequency is lesser than FRF response end frequency, then solver will extrapolate the PSD values which might lead to an exponential increase in the PSD value will in turn increases the response exponentially.
Hence try to avoid (3) and (4).
Thank you
0 
Altair Forum User said:
Hi,
Ideally, in simulation the forcing function or PSD should be spread across FRF response frequency.
For a broader understanding there can be four cases as follows:
1. The FRF response frequency and PSD frequency matches, then solver will use the same value of PSD.
2. The PSD frequency lies between two FRF response frequency, then solver will use interpolated PSD value.
3. The PSD start frequency is greater than FRF response start frequency, then solver will extrapolate the PSD values which might lead to negative PSD value and hence solver will terminate the run.
4. The PSD end frequency is lesser than FRF response end frequency, then solver will extrapolate the PSD values which might lead to an exponential increase in the PSD value will in turn increases the response exponentially.
Hence try to avoid (3) and (4).
Thank you
Hi Prasanna,
Thank you for your answer, I understand. However, I am not sure to get the point (2).
PSD in only used at computed FRF values, isn't it?
For instance, in my example of PSD input (please see file attached) if I compute FRF at 2, 4 and 6Hz, then PSD response will be computed at 2, 4 and 6Hz, but the peak at 5Hz will be missed because there is no value at 5Hz for the FRF. Am I right?
0 
Hi Nathan,
Yes, the 5 Hz peak will be neglected here. That's why, it is advisable to use finer step size while applying loads and calculating responses for FRF simulation and Random Response analysis.
Thank you
0 
Altair Forum User said:
Hi Nathan,
Yes, the 5 Hz peak will be neglected here. That's why, it is advisable to use finer step size while applying loads and calculating responses for FRF simulation and Random Response analysis.
Thank you
Ok, thank you for your help! /emoticons/default_smile.png' srcset='/emoticons/smile@2x.png 2x' title=':)' width='20' />
0 
Altair Forum User said:
Hi Nathan,
Yes, the 5 Hz peak will be neglected here. That's why, it is advisable to use finer step size while applying loads and calculating responses for FRF simulation and Random Response analysis.
Thank you
Hello Prasannak,
Are you sure about this? why the software doesn't use the closed two points of the FRF result to interpolate, then use the interpolate FRF result to do the random analysis?
0