relative motion of two structures in Ansys AQWA

Hello!
My name is Jan and I'm a naval architect student from Germany.
At the moment I have a problem, and maybe anyone out there can help me.
In my AQWA-scene is one free-floating body, and one body which can only move in z-direction (I damped the others DOF's) next to each other.
Because I am searching for the relative motion, I want to evaluate stochastically the both RAO's for heave motion. So I take the RAO's from AQWA to Excel, multiply the square with my Jonswap, and I have the response spectra of the movement.
But the problem is, I have not adjusted the phase-shift which results from the distance of the two structures.
Does anybody have an idea how to do that in AQWA, or “by hand” in excel?
I would be very pleased if someone knows the answer.
Kind regards,
Jan

 

Welcome to the forum Jan.

What you are trying to do is rather complex. It is time domain situation that you have. Since whatever distance separates the two floating structures, each shall experience a given RAO. If the 2 structures were placed next to each other, the heave generated by the amplitude of the wave is instantaneous for both. If you now placed the 2 structures 1km apart, clearly the response relative to each other would be different. If the 2 structures experienced the same wave spectrum, but separated by a distance, it is purely the difference in time for each structure to react to the given sea spectrum.

Therefore you need to define your spectrum with respect to time, and use one as the origin of time to ascertain the "lag" in motion and hence the difference in response.

You should be able to extract the velocity function of the wave profile from the Jonswap. With the known distance part, the time 'lag' can be found. Since the wave profile, with a velocity in the direction (axis) x, and amplitude H, shall be of the general form:

y = H/2 . sin.(qx + wt)

where t = time.

 

Hello Ad Hoc, thanks for the introduction and the fast reply!

To solve this in time domain, I would have to transform my RAO’s by a inverse fourier transformation first, right? Then the comparison would be easier, but I see the possibility that this will be like “cherry picking” the right values in the two resulting graphs.
I initially searched for a way to do it al in frequency domain. Maybe the RAO of the second structure can be “corrected” by the phase shift (resulting by the distance) and then the RAO can be compared directly. Or a RAO of relative-motion can be received someway. Just an idea, but maybe someone can imagine something like that?

 

Perhaps you're over thinking this. If a wave with a defined profile, if that wave strikes both structure but at different intervals, then it should be straight forward to obtain the second derivative of the waves profile equation to obtain "t" with respect to your motions.

However, if the structures are further apart than a wave length, then it does indeed become extremely complex.

I've not done higher maths like FFT for years, but i suspect that if the 2 structures are further apart than the wavelengths your investigating then indeed a FFT would probably be the way to go to transform your algorithms into a decimation in time. That level of maths is beyond me now...even struggled with it 25 years ago when at Uni!

For that maths you will probably get a fuller response from JEH
http://www.boatdesign.net/forums/profile/jehardiman.html
or Leo
http://www.boatdesign.net/forums/profile/leo-lazauskas.html

 

Even today during the studies FFT forces struggling, I think that’s an all-time non-favourite .
But maybe I can avoid it. The distance in my case is about 0.5m, far away from the wave length, mostly.
So your idea could be the better one, but can you give me a hint how to extract the velocity function from my Joswap, and how to deal with this result to receive the movement of my two bodies?
In my opinion I need the time-difference to know, how much later my second body reacts, but how can I use this in combination with my RAO’s ?

 

From a standard JONSWAP I assume you will calculate the value of Mo, to obtain the significant wave height?

From this, you can also calculate the probably of waves of ‘X’ metre occuring and then ‘X + 1’m and so on also the period of each, via Mo and M1. Thus, you can work backwards to obtain wave profiles, with the given depth; crudely of course. Once you have that and the general form, of a wave, you should be able to define the amplitude, and hence heave, with regards to time “t”.

 

Look into using the module AQWA-FER (Frequency Evaluation of Response) to calculate response with input spectra. You'll need to resort to editing text files with a restart from the AQWA-LINE job.

EDIT:
What version of AQWA do you have? On my ANSYS v13 I found an example AQWA-FER file called 'AFTAKBUY.dat':

C:\Program Files\ANSYS Inc\v130\aqwa\training\aftakbuy.dat

It's a little cumbersome to use, because you have to step outside the Workbench environment and you also have to point the AQWA-FER job file to point to the directory containing the .RES and .HYD files from your previous AQWA-LINE analysis.

I've used FER to recalculate RAOs for wave energy converters in spectra with damped constraints, an oh-so-fun endeavor when done by hand (sarcasm).