We put some animations on http://www.saic.com/nfa. There is an interactive animation of a foil on the site. If you do choose to view it, please note that it is big, over 500Mb compressed You'll have to download and install a viewer. The other animations show some breaking waves, stratified flows, and turbulence.

The analytical solutions of most of those problem are well known since WWII and available in the books from F.Ursell.

Unfortunately from those images we still do not know if the wave length and frequency obtained from the simulation matches the geometrical dimension and conditions of the simulations

http://www.worldscibooks.com/mathematics/2078.html

http://books.google.com/books?id=WtN...oC&dq=f+ursell

Ship Hydrodynamics, Water Waves and Asymptotics: Collected Papers of F. Ursell, 1946â€"1992. World Scientific, 1994. Vol. I, 568 pp.; Vol. II, 408 pp.

Hi Raul,

With all due respect, these numerical simulations are little bit more advanced than the analytic solutions that are proposed by Fritz Ursell. A lot of our papers can be googled on the web. We use a mixture of spectral methods, FDM, BIEM, VOF, and cartesian-grid methods. The results span the gamut from analytic, to semi-analytic, to brute-force cfd. The simulations of breaking waves, especially as it pertains to loading, does not lend itself well to analytical methods. The field has advanced a lot over the years. On the other hand, analytic solutions, such those developed by Professor Ursell, guide current research in free-surface hydrodynamics. Also see Michael Longuet-Higgins, C.C. Mei, Ken Melville, Dick Yue, etc.

Best regards,

Doug.

They are certainly very beautiful pictures. Unfortunately that doesn't mean much to many naval architects who often want good estimates of simple quantities such as total resistance. In many cases, CFD is not much better than the classical methods of Havelock and others when combined with a simple skin-friction line.

Roughly how long does it take for your code to produce a curve of total drag versus Froude number for a DTMB 5415 model hull, say for Fr=0.1, 0.11, ..., 0.99, 1.0?

Is SAIC prepared to offer a prediction of the total drag and dynamic trim and sinkage of model DTMB 5415 hulls *before* the conclusion of the series of tests currently being conducted around the world (and involving about 30 towing tanks)? Or will we have to wait until after the towing tank tests to see how well your (and many other) CFD codes perform when compared to experiments? ;-)

BGwMG

Dear Bikini Girl,

So now the 5415 is the Navy's latest planing attack boat?

For most resistance calculations, Das Boot is the the tool of choice, especially for optimization studies. However, NFA is carving out its own niche.

Regarding NFA's performance in blind studies, Tom Fu, et al., recently discussed some results. Google them and other results that are published in the ONR symposiums. Also, go to Korea this year to attend the next ONR symposium. There you'll find that NFA works on a supercomputer, and it ain't quick.

We're keen to perform additional blind studies. We're also keen to get paid for research. At present, we perform resistance calculations, impact loading, wave loading, wave predictions, forced-motion studies, and breaking-wave studies. For more information, please feel free to contact us at SAIC.

Thank you,

Doug.

Thanks for the reference to the ONR symps.

I agree that a real DTMB 5415 would not get up to Froude numbers quite that high, but does NFA fail at higher speeds? If so, why? (I'm not singling your code out for possible criticism: some codes have problems with "excessive" dynamic lifting effects).

You said that NFA is carving out its own niche. Did you mean in drag prediction as well as what it does now? I'm only asking because I'd love to see more CFD predictions of planing hull drag. I don't know why, but I find it discomfiting that Savitsky is still considered as the preferred method by many naval architects.

And thanks heaps for saving us all the fare to Korea to find out that NFA works on a supercomputer and that it ain't quick!

BGwMG

Dear Bikini Girl,

Some of my buddies have a very twisted sense of humor. Your name is consistent with that suspicion... Also, this forum is not the best platform for responding to serious and not-so-serious questions. For one thing, people are anonymous. Also, there is no stopping people from responding in my name or your name. I guess that makes things interesting... Also, it is pouring buckets here in San Diego, so I have time to respond to this forum.

At any rate, NFA is a cartesian-grid based method. A panelized geometry is used as input to NFA. This minimizes the time that is required to grid. The interface tracking that is used in NFA is volume of fluid (VOF). The cartesian-grid based formulation in combination with VOF provides a stable and turn-key platform for simulating ship waves over a wide range of Froude numbers. Planing hulls do not necessarily cause more difficulty. We have, for example, simulated Boston Whalers going at warp speed.

Getting back to your comment about NFA's cpu requirements, NFA is implemented on a supercomputer using SHMEM. It also works on a cluster using MPI. Given that background, let's discuss how NFA is used in practice. We grid up the hull using panels. This takes a couple hours if we have the nurbed surface. The panelized geometry is used as input to NFA. NFA calculates internally all the metrics associated with the geometry. We then submit multiple cases to the supercomputer using anywhere from 128 to 256 processors. We have concurrently run 8-12 jobs on a Cray XT3. We often get results by the next morning, depending on how busy things are on the Cray. How many codes can claim that kind of turn around with so few man hours without crashing? Also, supercomputers are getting faster and faster. We'll be on a Cray XT4 in few months that is 5 times faster than the Cray XT3!

NFA is under continuous development. It can't do everything. Currently we are improving the efficiency of the Poisson solver. We also need to develop techniques to handle free motions and various types of wave radiation issues. There is a long list of things to be done.

Also, NFA uses some approximations. We don't, for example, resolve the hull boundary layer at present. NFA does have a Smagorinsky turbulence model, but it is primarily for the wavy portion of the flow. NFA calculates wavemaking drag and form drag directly. Skin friction is estimated using a variation of the ITTC formulation. The ITTC trick is used in Das Boot. NFA does not use a correlation allowance.

Aside from ship waves, we also simulate turbulent wakes with and without stratification and with and without a free surface. If possible, we would also like to simulate ship waves for the movie industry. That is part of reason why we posted NFA's web site to this forum.

Best regards,

Doug.

This is a good point. It is not really worth doing those heavy calculation if not sure that amplitude and frequency of the waves (and forces) are within the expected range.

The 5415 data are actually well known. To cite a few. I think for marine application the software vendor should guaranty at least those results. And the customer is entitled to ask for the results and computing time.

Few links:

http://www.nmri.go.jp/cfd/cfdws05/go..._variables.htm

http://books.nap.edu/openbook.php?re...10189&page=474

https://pronet.wsatkins.co.uk/marnet/

Hi Ron,

You're right with respect to documentation. Along with our coauthors, we will be presenting several papers at the ONR symposium in Korea. These papers will provide the latest and greatest validation studies. Other references are provided below.

Best regards,

Doug.

Dommermuth, D.G., O'Shea, T.T., Wyatt, D.C., Ratcliffe, T., Weymouth, G.D., Hendrikson, K.L., Yue, D.K.-P., Sussman, M., Adams, P., and Valenciano, M. (2007) An application of cartesian-grid and volume-of-fluid methods to numerical ship hydrodynamics. In Proc. 9th Inter. Conf. on Num. Ship Hydro., Ann Arbor, MI, To appear.

Dommermuth, D.G., O'Shea, T.T., Wyatt, D.C., Sussman, M., Weymouth, G.D., and Yue, D.K.P. (2006) Numerical simulation of ship waves using cartesian-grid and volume-of-fluid methods. In Proc. 26th Symp. On Naval Hydro., Rome, Italy, To appear.

Dommermuth, D.G., Sussman, M., Beck, R.F., O'Shea, T.T., & Wyatt, D.C. (2004) The numerical simulation of ship waves using cartesian grid methods with adaptive mesh refinement. In Proc. 25ndt Symp. on Naval Hydro., St. John's, NewFoundland, Canada, To appear.

Rottman, J. W., Dommermuth, D.G., Innis, G.E., O'Shea, T.T., & Novikov, E. (2002) Numerical simulations of wakes in weakly stratified fluids. In Proc. of 24th Symp. on Naval Ship Hydro., Fukuoka, Japan, 517-533.

Sussman, M. & Dommermuth, D.G. (2001) The numerical simulation of ship waves using cartesian-grid methods. In Proc. of 23rd Symp. on Naval Ship Hydro., Nantes, France, 762-779.

Dommermuth, D.G., Rottman, J.W., Innis, G.E., & Novikov, E.A. (2000) Numerical simulation of the wake of a towed sphere in a weakly stratified fluid. IAHR 5th Inter. Symp. on Stratified Flows, Vanouver, Canada.

Dommermuth, D.G., Innis, G., Luth, T., Novikov, E., Schlageter, E., Talcott, J. (1998) Numerical simulation of bow waves. . In Proc. of 22nd Symp. on Naval Ship Hydro. Washington, D.C., 508-521.

Dommermuth, D.G., Gharib, M., Huang, H., Maheo, P. & Novikov, E.A. (1996) Turbulent free-surface flows: A comparison between numerical simulations and experimental measurements. In Proc. of 21st Symp. on Naval Ship Hydro. Trondheim, 249-265.

Dommermuth, D.G. & Mui, R.C.Y. (1995) The vortical structure of parasitic capillary waves. In Proc. of 1995 Inter. Mech. Eng. Congress and Expo., San Francisco.

Dommermuth, D.G. & Mui, R.C.Y. (1994) The vortical structure of a near-breaking gravity-capillary wave. In Proc. of 20th Symp. on Naval Ship Hydro., Santa Barbara, 530-550.

Dommermuth, D.G., Novikov, E.A., & Mui, R.C.Y. (1994) The interaction of surface waves with turbulence. In Proc. of the ASME Symp. on Free-Surface Turbulence, Lake Tahoe.

Dommermuth, D.G. & Mui, R.C.Y. (1994) Numerical simulation of free-surface turbulence. In Proc. of the 12th US National Congress of Applied Mech., Seattle, Appl. Mech. Rev., 47 6(2), 163-165.

Dommermuth, D.G., Mui, R.C.Y., & Loeser, D.J. (1994) The numerical simulation of fully-nonlinear three-dimensional ocean waves including the effects of shear currents. In 1994 Offshore Mech. and Arctic Eng. Conf., Houston, 37-44.

Dommermuth, D.G. & Novikov, E.A. (1993) Direct-numerical and large-eddy simulations of turbulent free-surface flows. In Sixth-International Conference on Numerical ship Hydrodynamics, Iowa City, 239-270.

Dommermuth, D.G. & Yue, D.K.P. (1992) The formation of U-shaped vortices on vortex tubes impinging a free surface. In Proc. of 18th Symp. on Naval Ship Hydro., Ann Arbor, 727-788.

Dommermuth, D.G. & Yue, D.K.P. (1988) The nonlinear three-dimensional waves generated by a moving surface disturbance. In Proc. of 17th Symp. on Naval Ship Hydro., The Hague.

Dommermuth, D.G. & Yue, D.K.P. (1987) Nonlinear three-dimensional wave-wave interactions using a high-order spectral method. In Symp. on Nonlinear Wave Interactions in Fluids, Boston.

Dommermuth, D.G. & Yue, D.K.P. (1986) Study of nonlinear axisymmetric body-wave interactions. In Proc. of 16th Symp. on Naval Ship Hydro., Berkeley.

Dommermuth, D.G., Rottman, J.W., Innis, G.E., & Novikov, E.A. (2002) The numerical simulation of the wake of a towed sphere in a weakly stratified fluid. J. Fluid Mech., 473, 83-101.

Longuet-Higgins, M.S. & Dommermuth, D.G. (2001) Vertical jets from standing waves. II. In Proc. R. Soc. Lond. A 457, 2137-2149.

Longuet-Higgins, M.S. & Dommermuth, D.G. (2000) On the breaking of standing waves by falling jets. Phys. of Fluids, 13(6), 1652-1659.

Dommermuth, D.G. (2000) The initialization of nonlinear waves using an adjustment scheme. Wave Motion. 32, 307-317.

Novikov, E.A. and Dommermuth, D.G. (1997) Distribution of droplets in turbulent spray. Phys. Rev. E., 56(5) 5479-5482.

Longuet-Higgins, M.S. & Dommermuth, D.G. (1997) Crest instabilities of gravity waves. Part 3. Nonlinear development and breaking. J. Fluid Mech., 336, 33-50.

Mui, R.C.Y., Dommermuth, D.G., & Novikov, E.A. (1995) The conditionally-averaged vorticity field. Phys. Rev. E. 53(3), 2355-2359.

Mui, R.C.Y. & Dommermuth, D.G. (1994) The vortical structure of a near-breaking gravity-capillary wave. J. Fluids Eng., 117, 355-361.

Novikov, E.A. & Dommermuth, D.G. (1994) Conditionally-averaged dynamics of turbulence. Modern Phys. Lett. B, 8(23), 1395-1401.

Dommermuth, D.G. (1994) Efficient simulation of short- and long-wave interactions with applications to capillary waves. J. Fluids Eng., 116, 77-82.

Dommermuth, D.G. (1994) The initialization of vortical free-surface flows. J. Fluids Eng., 116, 95-102.

Dommermuth, D.G. (1993) The laminar interactions of a pair of vortex tubes with a free surface. J. Fluid Mech., 246, 91-115.

Dommermuth, D.G. (1992) The formation of U-shaped vortices on vortex tubes impinging on a wall with applications to free surfaces. Phys. Fluids A 4(4), 757-769.

Liu, Y., Dommermuth, D.G., & Yue, D.K.P. (1992) High-order spectral method for nonlinear wave-body interactions. J. Fluid Mech., 245, 115-136.

Dommermuth,D.G. & Yue, D.K.P., Lin, W.M., Rapp, R.J., Chan, E.S., & Melville, W.K. (1988) Deep-water breaking waves: a comparison between potential theory and experiments. J. Fluid Mech., 189, 423-442.

Dommermuth, D.G. & Yue, D.K.P. (1987) Numerical simulations of nonlinear axisymmetric flows with a free surface. J. Fluid Mech., 178, 195-219.

Dommermuth, D.G., & Yue, D.K.P. (1987) A high-order spectral method for the study of nonlinear gravity waves. J. Fluid Mech., 184, 267-288.

"Some of my buddies have a very twisted sense of humor. Your name is consistent with that suspicion... " Your buddies like The Cramps? Tres Cool!

You'll have to excuse the tongue-in-cheek comments, but I find CFD funny.

For example, using the ITTC line is understandable because it is well-understood within the naval architectural community. But it is also very amusing. The line was proposed and agreed to as an interim solution in 1957 so that delegates could get off to lunch. It is not a physics-based friction line, although it does claim some descent from Schoenherr. I would expect something a little more "realistic" from CFD codes that cost thousands of dollars.

An endless source of amusement is that linear codes still do as well as CFD on far-field wave patterns, but in a timy fraction of the time. As far as I can judge, most modern ship hydrodynamics codes do no better than Noblesse's (linear, inviscid) code FARWAV, arguably the best performer in the 1991 Wake-Off.

Supercomputers are indeed getting faster, but that's not a huge comfort if the final answer is no better than other codes can do in less time on a $400 PC (which are also getting faster and cheaper).

Good luck in the movie industry! I look forward to seeing a Bollywood musical where NFA provides the beautiful scenery and remarkable colors, and advertising execs jump from behind bushes and sing and dance its praises.

Why not get out your brolly now and have a little practice splashing through the rain on the way back to your car in San Diego? ;-)

BGwMG.

Dear Bikini Girl,

Linear potential-flow codes with linear free-surface boundary conditions miss a lot of physics. Wave phase is off, just to name some problems. Euler codes work because they capture the nonlinear effects such as flow separation off of transom sterns, spray separation off of planing hulls, and wave breaking. Froude's hypothesis has a sound basis, so whether you like it or not, things like the ITTC line also work, especially when the Reynolds number is order 10^9. Use a RANS code. Spend 2-3 months gridding. Then spend another month nursing the code so that it does not crash. Get an answer that has mucho numerical damping to stabilize the numerics. You're right that traditional CFD is fraught with problems... There are plenty of researchers working on alternatives. Come to Korea to see the recent advances in the field.

Best regards,

Doug.

Hi! I'm truly impressed by Your's CFD performance, I am a beginner of this discipline, so although I have same good references about numerics I was not able to find any kind of material about marine CFD best practice, in particulary how to set up the correct waves to perform breaking waves analisys. For instance I wanna ask you some question about the simulations available at your web page: a) Is the sphere simulation performerd for any kind of validation? b)In the moving hulls simulation are the hulls moving forward or is the water flowing against them? c) What are you trying to state in the forced motion simulation? I'm also interested in simulate solitons propagation in shallow water (e.g. harbour) but I was not able to state the right initial condition. I'm a chemical engineer and a free time sailor, so I beg Your pardon if my question are so naif.

Any help will be appreciated!

Hi Giovanni,

If you are Italian, you have some very good resources at INSEAN just outside of Rome. The research that is performed there is top notch.

The sphere simulation was set up for some experiments that were attempted at David Taylor Model Basin. It did not quite come together in the laboratory. There are some analytic solutions. Francis Olgilvie and Fritz Ursell published papers in this area. Their theories are linear with some nonlinear extensions. They were published in the Journal of Fluid Mechanics.

The ship is fixed and the water is moving in our simulations.

The forced-motion simulations were performed to calculate forces for input into a specialized type of seakeeping code.

With respect to solitons, try C.C. Mei's book on ocean surface waves. He has some good analysis. I recently saw a beautiful soliton generated in the wave tank at Oregon State (http://wave.oregonstate.edu/).

Best regards,

Doug.

Thank you very much for your kind suggestions. I have another question, which is the best turbolence model for this kind of application? I don't believe that k-epsilon would perform well in open flow scenario, so what do you suggest? In your simulations what turbolence model did you used? Thanks in advance Giovanni

Hi Doug,

Are surface tension effects included in the 5613/head wave simulations? I see some very high wavenumber disturbances in the head waves upstream of the hull and was wondering if the code was trying to produce capillary waves or if the disturbances are generated by the numerics/boundary conditions.

Tom

Hi Giovanni,

We alternate between using a limited Quick FDM scheme and a Smagorinsky turbulence model for the free-surface flows. Limited Quick is more stable with minimal numerical damping. Breaking waves are driven by inertial effects, so limited Quick works. For turbulent wakes and stratified flows, a mixed model (Bardina) is used to model the SGS stresses in the momentum equations. The SGS density term is modeled using a combination of a similarity model and an Eddy diffusivity model. This is all within the context of LES.

-Doug.

Hello Tom,

We use 2nd-order reconstruction and advection. It is less susceptible to generating flotsam than first-order schemes. Even so, the flow is turbulent and when the length scales get grid scale, the free surface fragments and breaks up.

-Doug.

Thank you for your kind answer.

How do you know it is flotsam and not jetsam?

So, there is resolved (simulated and not modeled)turbulence in the incoming wave?

Also, just out of curiosity, is surface tension included in your model? If not, is it practical to put it in there (from both physical modeling and grid resolution perspectives)?

Tom

Hi Tom,

We doing some studies of pure breaking waves to answer the flotsam versus jetsam question. We hope to compare to experiments that are being performed. We want to resolve the inertial range.

The incoming waves are virtually irrotational except near the interface. Turbulence occurs where the wave breaks and where the flow separates. This is also part of the pure breaking wave study.

There is no surface tension in the flow simulations of the ship because it is not practical. We have used surface tension in 2d simulations using level-set methods. Mark Sussman and others use height functions to calculate the curvature in VOF formulations. This appears promising for applications involving sheet breakup.

-Doug.