AQWA additional damping matrix problems

Discussion in 'Software' started by aparker, Jan 17, 2013.

  1. aparker
    Joined: Jan 2013
    Posts: 7
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Perth, WA

    aparker Naval Architect

    Good morning, I'm new to these forums but I was wondering if anyone can help me with an issue I'm having.

    I am modelling a system with three large vessels connected to each other with lines, and with the centre vessel restrained by fenders and fixed points. I am using a relatively short period wave spectra, however the accelerations I am getting in the time domain are quite high, which makes me think that I should add an additional damping matrix for each of the ships.

    Does anyone here have experience with populating the damping matrix with appropriate values? I am struggling to find a suitable method.

    Many thanks in advance.

    Alex
     
  2. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,148, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    What is your non-exceedence criteria?
     
  3. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    Is this academic work? e.g, can you get away with simplified methods?
     
  4. aparker
    Joined: Jan 2013
    Posts: 7
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Perth, WA

    aparker Naval Architect

    Jehardiman, I'm sorry but I do not have detailed wave information at the moment.

    CWTeebs, this isn't academic, however I'm open to simplified methods if it means that I can simulate some sort of additional damping.

    Thanks for the replies.
     
  5. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    Use FIDD if the quick and dirty approach will suffice. Proper time domain convolution requires specifying both damping and added mass matrices. They're related via a mathematical transform. Hats off to you if that's the approach you want to take.
     
  6. aparker
    Joined: Jan 2013
    Posts: 7
    Likes: 0, Points: 0, Legacy Rep: 10
    Location: Perth, WA

    aparker Naval Architect

    Thanks for the help CWTeebs. I think the FIDD should suffice; is there a relatively quick way to calculate the values for it?
     
  7. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    Compare most likely expected values from the spectra using standard values, compare to resonant frequencies of the structures for the degrees of freedom, use a percentage of the critical damping. I can't really say beyond that without seeing your AQWA setup, however. Are you following other 'best practices?' Did you run LIBRIUM before the time domain case?
     
  8. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,148, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    One must be careful when using that approach because you are not really interested in the significant response (i.e. 2 sigma) but rather some extreme response with a very low probability of non-exceedence. For multiple structures it is possible to have a fairly high RMS acceleration difference. This will drive your response spectra to large values and depending on how you transform from the frequency to the time domains may or may not give you a true maxima.

    aparker;
    If you are having trouble with the wave defraction damping matrix have you tried WAMIT to get that before using AQWA to force link the bodies.
     
  9. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    Well, that's why I said most likely expected, not significant, but I understand your point. The extreme event stuff always makes me cross eyed..."if the wave heights follow a gaussian distribution then the peaks follow a rayleigh distribution but if you are only in the sea for x hours and the zero crossing period is t and there's a 36% chance of hitting a mermaid except on tuesdays what's the most likely expected maximum acceleration in a ten year storm"... The real problem is that FIDD damps *all* frequency components and isn't really physically accurate.

    I'm guessing you are not an AQWA user. AQWA-LINE (linear) is the equivalent of low-order WAMIT and is required for time domain simulation (AQWA-NAUT, AQWA-DRIFT) when diffracting bodies are present.
     
  10. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,148, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    Yeah, we played with it 5 or 6 years ago and then rejected it because of our specific needs to get real "maximum load in life of project" answers and some issues we had with non-linear boundary interactions (though now I understand that you can write your own code for that) but I understand that schools are turning out student users now and a lot of companies are using it <shrug>. I have found that it is too easy to get pulled into the compelling output displays (cfd = colorful fanciful drawings...;) ) and think that a time domain solution is reality when it is only one posibility of an infinite number of soultions. I have also seen and critiqued time domain solutions "cherry picked" to show capability when returned response data did not support that conclusion. FWIW, we generally use WAMIT or some specific panel models we have written to get high fedelity RAO's and then keep it all in the frequency domain with a high non-exceedence. The structural and cost guys grumble, but I've got plenty of returned data to show that response to a significant event wave is driven by the wave energy and occurs much more frequently (i.e. an effective "100 yr event" occurs every 10 years).
     
    1 person likes this.
  11. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    Not sure what specifically you're referring to, possibly the user defined force DLL in the time domain.
     
  12. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,148, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    Perhaps, I really haven't dug into the DLL capability, and something like that may work for aparker also. The issue we were having was that as the bodies moved relative to each other, the response RAOs, damping and force geomentry changed significantly (orders of magnitude). So we needed a real-time recursive force, response, and damping matrix determination.

    This was not new or unexpected as we have had a history trying to solve the multi-body interaction problem since the mid-1960's. We made a really concentrated multi-year effort in the late 1980's and developed a significant amout of CFD code with some pretty unique constraints and ran large scale testing in the old CBI wave tank. What we found was that there multiple bifurcation points (saddles and cusps) in the solution space often dependent on the total position deritives (i.e. x, x-dot, x-double dot, etc) as well as wave length and phasing relative to the bodies. Our problem is a real mess. Every now and then somebody says "AQWA can solve that", but for multiple free-free bodies non-rigidly connected in less than wave length proximity I have yet to see a solution match up with the returned data.
     
  13. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    I'm a bit confused. The conversation has seemingly jumped from AQWA time domain to frequency domain statistical techniques to viscous CFD.

    What program are you referring to? Was this when you were experimenting with AQWA 5 or 6 years ago? Or is this what you're currently doing with WAMIT and your custom codes?

    The hydrodynamic interaction of multiple moving diffracting bodies is fairly sophisticated in AQWA. The hardest case is the drift analysis (low frequency plus wave frequency) of moored hydrodynamically structures (interacting QTFs of up to 3 bodies). You're right, movement from the design position invalidates the solution, and AQWA summarily flags warnings upon movement, and finally terminates the solution if movement exceeds a certain threshold.

    From what I've seen, the time domain stuff is absolutely necessary in only a few scenarios and adds complexity that can be hard to justify. We stick with frequency domain statistical stuff as much as possible. An example is slack mooring system design.

    Hmm. AQWA's time domain solvers are solid, but, AFAIK, no inviscid irrotational potential flow code can truly handle the buildup of the radiated waves between structures, and as you said, the hydrodynamics are ultimately linearized and calculated about the mean wetted surface (the AQWA theory is the same as WAMIT with few exceptions).
     
  14. jehardiman
    Joined: Aug 2004
    Posts: 3,762
    Likes: 1,148, Points: 113, Legacy Rep: 2040
    Location: Port Orchard, Washington, USA

    jehardiman Senior Member

    Both AQWA 5 or 6 years ago and our custom codes. Basicly, a close moor/station keeping, pier moor/maneuvering, or underway replenishment all have a similiar problem. As the bodies approach or seperate, the interaction dramaticly changes, but not consistantly. The managitude and type of change will vary with seaway and other factors. Over years of experience we have determined that the significant design criteria is the available system energy (in the body and the environment). This can, as aparker said in his original post and you allude to, lead large accelerations and forces when snubbing up or fendering when trying to transfer from energy/frequency to force/time.

    From what I have seen of it, AQWA is a fairly capiable program in the limited CFD world (inviscid irrotational potential flow) and I know a lot of people use it just for the floating body-mooring catenary tools. But from my unique perspective, a lot of what it does can be accomplished by simple experienced inspection. Like any diagnostic tool, it is up to the user to determine if its accuracy warrents belief in its precision. My big problem is when someone who hasn't wrung out half a dozen PhD's on the topic comes to my sponser and says "Look at my shiney new tool (i.e. we just hired a grad that used AQWA in school) and pretty pictures" and I have to wade in and sort it all out. One of the things we (US Navy RDT&E) do is every few years we revisit current technology and see if it applies to old problems. So far this problem is one of those that hasn't yeilded to new approaches.
     

  15. CWTeebs
    Joined: Apr 2011
    Posts: 232
    Likes: 15, Points: 0, Legacy Rep: 171
    Location: Maine

    CWTeebs AnomalyGenerator

    jehardiman: I had no idea AQWA produces pretty pictures, and I'm an expert in the tool! You have a broad depth of experience and I'd like to talk more other tools you've explored, but I'm afraid we've hijacked Mr. Parker's thread and we should therefore take this sub-conversation in private.

    aparker: any progress?
     
Loading...
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.