Obtain ABSOLUTE acceleration value in a Response Spectrum Analysis

[pilafa]

Hi,

I am running a Response Spectrum Analysis (it us actually a DDAM analysis, a "special type" of Spectrum Analysis).

I understand that the acceleration values that I obtain as result in my structure are RELATIVE acceleration values (I see in the postproc that the acceleration at the base of the structure is 0).

I want to have an ABSOLUTE acceleration value at a certain point of the structure to compare with (absolute) accelerometer data.

In a time transient simulation, I could simply sum up the base input acceleration and the relative acceleration at a certain node to obtain the Absolute value.

But in a Response Spectrum Analysis, I don't see resulting acceleration values of the structure "in time", so how could I correct this RELATIVE acceleration to take into account the base acceleration?

Thanks

 

[FEA way]

Which software are you using ?

 

[driftLimiter]

Response Spectrum Analysis is enveloping the response quantities. The peak acceleration can be determined from the response spectrums, but you have zero information about what excitation is causing that, what the PGA is at the instant that the peak acceleration is unknown. Perhaps it cannot ever get larger than PGA + Max Spectral Acceleration, not sure how you aim to use this info. But if the excitation can be well defined, then a time history solution could provide a good result.

 

[pilafa]

@FEA way:
I am using Inventor-NASTRAN

@driftLimiter:
Imagine a simple frame fixed on the ground with a MOTOR on it.
The motor has a maximum allowable acceleration in its specs. I want to check that the acceleration "seen" by the motor will be lower than that. I then need the ABSOLUTE acceleration

When I postproc the results of DDAM, I see that the acceleration at the base of the frame is 0. I then deduce that we are looking at Relative acceleration.

To obtain the max "possible" absolute acceleration, I could do what you suggested: Sum up Max spectral acceleration and the Relative Acc obtain from the DDAM.
But, as you said, we have no information on the time, and those two accelerations will certainly not be in phase (not occurring at the same time...).
I except a very conservative value.

My question is: Is it possible to obtain this value with DDAM, or do I have to switch to full transient time history simulation??

Thanks for the help
Pierre

 

[driftLimiter]

I can't answer your question directly as I don't use that software. I'm pretty sure time history solution that includes a base input function would be the way to achieve this reault.

For us in the world of buildings we would select several real or synthetic excitation time histories and scale them so the peak acceleration is equal to or greater the peak of the spectral curve we are trying to design to. This usually involves widening and softening of peak acceleratios to achieve the correct scale factors. Run them all on your model and take the worst (or average) absolute acceleration.

Even when all that is done you still have a probabilistic solution that doesn't necessarily represent reality. Perhaps it is safest(and simplest) to add the PGA with the peak spectral acceleration after all they could be in phase :shrug:

 

[GregLocock]

Generating meaningful instantaneous acceleration peaks from psds and the like is not a trivial task. Given that you can run the model, do a time history.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[dmapguru]

Classical response spectrum analysis is based on the idea that you have a primary structure to which some transient excitation is applied. The primary structure forms a filter between the applied transient excitation and some response point on that primary structure, from which spectra (frequency domain) curves may be extracted. There is an interface in Nastran to do this where a virtual vibration table is created (by the software) at the one or more response points that you choose for a set of discrete frequencies that you also choose. When you generate the spectra, you may choose whether they are generated in an inertial frame of reference (absolute) or one relative to the response point. In other words, the spectra you generate, which may be displacement, velocity or acceleration spectra, are output including the movement of the response point or relative to it. These are called absolute and relative spectra.

You may not be the person who generated these spectra, but you should know in which frame of reference they were created. This distinction is necessary because the classical response spectrum method uses the Duhamel integral method to establish the (approximate) relationship between displacement, velocity and acceleration with the corollary that displacement and velocity are relative, but acceleration is absolute; the approximation is too wide of the mark at very low and very high frequencies.

When you move to use these spectra as inputs to a response spectrum analysis, you do not model the primary structure, it is a boundary condition to the response spectrum analysis; what you model are “appendages” to the primary structure where it is assumed that the presence or absence of these “appendages” has little effect on the behaviour of the primary structure. As you now know whether the input spectra are absolute or relative, this will govern your choice of whether to include the rigid body modes in the response spectrum calculation or not. For absolute applied spectra, the rigid body modes must be included (not just computed) in the analysis. For relative applied spectra, the rigid body modes are still computed, but they are filtered out (with a device like PARAM,LFREQ).

If you are using the large mass method for the response spectrum calculation, the point selected for base motion is free to move, but it will only move if you included the rigid body modes and you should be applying absolute spectra in this case.
Now in DDAM analysis, the assumption that “appendages” do not affect the behaviour of the primary structure is rejected. This follows the studies made by the US Navy during and after world war II where certain materiel, which should have survived under water explosion events, did not. It was determined that the degree to which the “appendages” interacted with the primary structure (and other appendages for that matter) was dependent on their location in the sea vessel (bulk head, deck, hull,…) as well as the type of sea vessel (ship, submarine,…) in which the materiel was installed. Consequently, DDAM analysis uses a set of input coefficients which depend on the configuration of the setup being analyzed. The latest set of these coefficients is confidential, and if you want to use them you have to license them from the US Navy and make sure they are installed on a secure server. Coefficients previously considered confidential have been declassified and released in a publication, but they apply to obsolete hardware.

The DDAM coefficients are applied to the input spectra you provide, which are relative velocity and absolute acceleration spectra if I remember correctly, and I think the spectrum it outputs, and that you eventually apply to your “appendage”, is always absolute acceleration; the coefficients will determine from which of your input spectra (velocity or acceleration) the output spectrum to apply was computed and how. Consequently, when using DDAM, you should always retain the rigid body modes in your analysis; if you have a large mass at the base motion point, it should then move with the acceleration of the applied spectrum. The response points in the appendage, when the rigid body modes are retained, will exhibit absolute acceleration.
DG