Best way to simulate swell: Model or CFD

Hi,
I would like to build a small cruising catamaran (8m). As I plan to design the hulls under water narrow and will get wider above water line there will be problems with swell, wave pounding onto the bridge and so on.

I already built a 1:10 model of it. What I am not sure about is which factor between model speed and real speed will show me the best fit to reality.

To improve the design and do not have to build for every designchange a new model I am looking for a cfd-software to try to simulate this. I had a look in the tools within CAE Linux and Acusolve (available at work) but could not decide which tool would fit best (and probably would not be too complicated).

Would be great to get some suggestions

 

Do you have any experience with CFD? CFD analysis of floating objects interacting with waves is a specialist area.

 

Slamming cannot be evaluated in the time domain CFD, it must be moved to the frequency domain using expected spectra and RAO's (WAMIT is a good program to get the free-free RAO's). Get and read a copy of Dynamics of Marine Vehicles by Bhattacharyya.

 

Slamming must be handled in the time-domain, and it can be done with CFD; It's just it's a massively difficult problem. There are a few methods to look at slamming:

A strip-theory method for motions, which is a linear, frequency domain system (not suitable but included for completeness).

A 6 DOF panel method, usually time-domain, but with very limited non-linear capability.

A volumetric 6DOF time-domain simulation, with some allowance for damping. These come with lots of levels of sophistication.

A full 6DOF CFD (eg. RANS) solution. I know only a few people with enough resource and capability to do this properly.

Your problem (of bridge-deck slamming) will be mainly driven by volumetric forces and wave impact forces. Given the sort of simulation time you're looking for (could be hours of simulated time), you're probably best to look for a primarily volumetric time-domain simulator. However, except for openDynamics (which is at a somewhat embryonic stage) I am not aware of any other open-source packages. Also since these packages require a considerable amount of experience to use properly, I would suggest that you seek help from professionals in that area.

The question of test-speed is more easily answered, as you need to test the model at the same Froude Number as the real vessel. Google and wikipedia are useful here. However, it is worth remembering that to perform useful model tests in waves, you will have to get the correct moments of inertia, which requires some careful calculation, and scaling via gyradii about each axis of rotation.

Feel free to contact me if you need any help.

Tim B.

 

Are you interested in the slamming loads, or whether or not there is any slamming at all (for a given condition)? It seems like the second question could be answered much easier than the first.

...sez me who has never solved either problem.

 

I totally disagree with this. I have spent the better part of my career answering this problem...the maximum motion-maximum load analysis. I will state catagorically that CFD (other than panel models to give RAO's for frequency domain) will not solve this problem. While it is true that you can run a time domain CFD analysis to get a single point solution; that answer is worse than irrelevent as, and I stated this in the thread about computers in NA, if gives a false sense of truth. A single CFD run and the resulting maximum motion or load is like having a single pixel of a picture and then saying what the picture looks like. And even if you run thousands of CFD runs, likewise you can never be sure of acheiving the "maximum" load because you cannot cover the infinite continious spectra. Even worse, you don't even have a clue about where you are relative to the maximum load.

This is why you need to throw it over into the frequency domain. There you can say that the motion and load given have a 99.995% probability of non-exceedence. Really, the physical value of the load is of much less importance than the probability of occurance. And a time domain analysis cannot give you that probability. CFD is a tool, but it is not the answer to all problems.

 

Hi,

Thanks a lot for all the answers till now.

I have no experience with CFD but lots of experience with strutural analysis based on FEM (mostly linear stuff, but also non-linear) as I work as engineer for this topic.

To refocus the question: I know that it is pretty complicated to analyse something like slamming on a bridge. So my real focus is to get an idea how to build the hulls that not too much swell is build up in the middle of the boat.

The only result should be how the bow wave builds up and how the hull shape influences this. The only case I will consider is the boat fixed and a steady flow of shallow water with cruising speed.

I know that the simulation is never the same as the reality, but it can give me an idea before I build a boat of the totaly wrong shape.

Additionaly it can be fun to do the simulation

 

Please read the advice I gave, as well as my comment on possibility.

Returning to probability... Using frequency domain data to get a probabilistic result is a valid method, but it will not give you any idea of the loads, because you simply don't get a sufficiently good idea of the flow conditions in the slamming region. Secondly, most results presented for frequency domain analysis have thier origins in linear theory. That's just going to give complete rubbish for this problem, as it is seriously non-linear. Consequently, you need to look at non-linear methods, and that means at least using a volumetric time-domain method.

In order to make the time-domain method work, you can do one of two things. For "small motions" analysis (ie. RAO generation), you feed a wave-spectrum consisting of a number of wave frequencies at unit height, then post-process the results with an FFT to produce the RAO. For large motions analysis, you use an estimated sea-state (or series thereof), and then record motions, slamming events, pressures and any other data you need. This may run into tens of thousands of simulations (with load conditions, speeds etc.), but you can then do probabilistic analysis on that data which will show you the loading and the frequency of occurrance. However, this is not the sort of task that a small NA firm can readily handle, purely because of the amount of computing resource required. Therefore, you need to look at the work that big maritime companies are doing (which is exactly that, though usually for capsize risk).

You need to consider the range of incident wave frequencies you are interested in, and apply suitable engineering judgement to limit the number of frequencies considered. For instance, on a 10m boat there is no point in working out the effect of a 0.1m (length) wave, we know that the effect will be as near to nothing as makes no odds, so why model it? Similarly, a very long wave (say 100m) is going to cause a response which tracks the wave, so there is no advantage there either. What we are usually interested in are the frequencies around the natural frequency of the boat in roll, pitch or heave; and that can be worked out using pencil and paper (it is well defined by the 2nd order differential equation for oscillatory motion). That is where the frequencies are clustered sufficiently to give a reasonable response curve.

Tim B.

 

Calvin,

The wave-pattern in calm water is another question, you'll probably want to look at the work done by Leo Lazauskas on this subject, and you may wish to look at the work by AF Molland in relation to catamaran resistance prediction.

Tim B.

 

Calvin, you may like to use the search button and find practical emperical formules on your questions on this forum discussed in other threads.

i'm more than fascinated with cfd and try'd a few programs but to do a bit more complex simulations aint easy for a beginner like me. astounding programs nevertheles, even simply watching flowlines in various regimes, better than ink in water or other messy expiriments. I'm no expert -like as said you have to be working with cfd- I still belive common sense like realising water only stands to a 120 degree wave -hitting most cats beams in bad conditions- before breaking up. incats giant nacelle is an solution igues but even with a big keel like stabiliser I hear people can get seasick. Boats are historical and for leisure , modern men fly