CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   3D Nonhydrostatic Model Needed (http://www.cfd-online.com/Forums/main/3696-3d-nonhydrostatic-model-needed.html)

Guo July 12, 2001 17:12

3D Nonhydrostatic Model Needed
 
Hi, there,

I am wondering if somebody can tell me where to buy a code for simulating a 3D geophysical flow(like estuaries or lakes) with free surface. The code should have a nonhydrostatic pressure resolution and be not too stiff at time step requirement. I know that some commercial software can do the job, but it turns out to be too CPU-consuming. The CPU required to simulate a physical process may need 20 times of the real physical time. That is really impractical.

Thank you for your attention!

John C. Chien July 12, 2001 18:06

Re: 3D Nonhydrostatic Model Needed
 
(1). Can you use faster computers to do simulation? (2). I know that writing your own code can also speed up the computations. (3). Commercial codes as some have mentioned are for companies with money to spend. (4). Why not contact the person (a student) who is currently writing his code and making it available at 10~15 dollars a piece, here. Check the posted message.

Steven Jachec July 13, 2001 08:58

Re: 3D Nonhydrostatic Model Needed
 
Dear Guo,

There are number 2D & 3D hydrostatic models available on the market to simulate tidal circulation, or flow within lakes. For example, 2D models include: RMA-2 (finite-element), MIKE21 (finite-difference) while 3D models include RMA-10, MIKE3, and DELFT3D (finite-difference). But, again these are hydrostatic. If you are modeling tides, this not a bad assumption. If you really need a nonhydrostatic model, I suggest a Boussinesq model, which is a hot-topic of research, especially when applied to short-crested waves that are considered shallow water waves (d/L < 1/20) like tides are. An example of this type of model is FUNWAVE (2D) created by Univ. of Delaware (James Kirby).

Hope this helps.

Guo July 13, 2001 11:14

Re: 3D Nonhydrostatic Model Needed
 
(1)Thank you! (2)I need a real 3D model that is not CPU intensive.(3)There are some commercial software that can do the job, but they are not very efficient.(4)Using faster computer or parallel computing may solve this problem, but I am more interested in a numerical technique, such as implicit method to relax the time-step requirement, to speed up the simulation. (5)I don't want to rebuild the wheel if there is one already. I want to be a slave if there is a good master. (6)Specifically, if someone has a code based on terrain-following coordinate system that solves geophycical flow problems using fully 3D equations without any simplification and wants to sell the software, let's have a talk. (7)Thank you for your attention!

Steven Jachec July 14, 2001 11:44

Re: 3D Nonhydrostatic Model Needed
 
Guo,

You could also try using the Priceton Ocean Model (POM). It is a 3D hydro model that includes (I believe) a sigma coordinate system (i.e., contours to the bottom). Best of all its shareware. Check out http://www.coastal-guide.com/software.htm for coastal engineering related software.

Hope this helps ...again

Guo July 14, 2001 13:44

Re: 3D Nonhydrostatic Model Needed
 
(1)Thank you for the information.(2)The Princeton Ocean Model is a hydrostatic one. What I want is a nonhydrostatic model.

Herve July 16, 2001 09:22

Re: 3D Nonhydrostatic Model Needed
 
Hi,

I did my PhD in the field of river and estuarine hydrodynamics, and did use CFX-4 (www.cfx.aeat.com) to do so. There's now a new free-surface feature in CFX 4-4 which should allow the deformation of the free surface (mesh-adaption). This didn't exist per se when I did my research. On the other hand I tested the code against a detailed experiemnt from the former Hydraulic Station (HR Wallingford) and obtained very good result.

Guo July 16, 2001 11:09

Re: 3D Nonhydrostatic Model Needed
 
(1)Thank you.(2)I checked the website and found one example on channel flow simulation. But the free surface simulation is said to be of one dimensional. I don't know if that is appropriate for my research. (3)I will do some further searching.

joe July 17, 2001 07:24

Re: 3D Nonhydrostatic Model Needed
 
It seems that you want fast non-hydrostatic code which simulate free surface flows in sea or estuaries. In my opinion, there is no one like it yet.

Do you know why hydraulic engineers do not use 3D CFD package such as Fluent, PHOENICS, etc, to model sea or estuaries ? It's because solving full Navier-Stokes equation is impossible and time-consuming. In addition, detailed physical information is not necessary at all in geophysical fluid flows. What is different between the velocity - 5.10m/s and 5.00m/s ? No difference. Both are wrong.

Without using good mainframe computer it is impossible to model 3D free surface flow. Do you think 3D hydrostatic sigma-coordinate model is not expensive ? Since the vertical scale is much less than the horizontal scale, it's very stiff at time.

Within 100 years, it may be possible to see 3D non-hydrostatic geophysical flow code.

Herve July 17, 2001 11:44

Re: 3D Nonhydrostatic Model Needed
 
I have used CFX to model two 1-km reach of rivers with reasonable sucess. The answer was fully-3D and certainly helped shed some light on areas of interest regarding erosion processes notably. CPU time on Sun Ultra station were of the order of 100 hours +. But it is possible... No later than today I have looked at a sort of bay which is 700 m x 700 m using the free surface deformation patch in CFX 4.4...

Herve

Guo July 17, 2001 14:17

Re: 3D Nonhydrostatic Model Needed
 
Hi,

Thank you for supplying me the performance information. I think what Joe said is partially correct. At least, a large scale(tens of KM) simulation for Geophysical Flow using 3D nonhydrostatic model is very CPU intensive and may need parallel computing. But a smaller scale(less than 1KM) can be simulated using a reasonable amount of CPU time. I think the performance can be improved by using some numerical skills, although I don't have a clear picture of that.

Herve July 18, 2001 03:34

Re: 3D Nonhydrostatic Model Needed
 
I think we all agree. You can only use CFD for localised detailed hydrodynamics, whicvh is what I did. You can say use a 2D or quasi-3D code at very large scale and use it to provide BC to a fully 3D code for a specific localised investigation, e.g. of a part of the coastline, a jettie, an oil rig etc. But it is possible to do so and this is very useful where detailed 3D flow structures need to be represented.

Herve

Guo July 18, 2001 17:25

Re: 3D Nonhydrostatic Model Needed
 
Hi, Herve,

Is that possible to know the following details of your simulation of 1KM river, say, the grid resolution, the real time simulated by 100 hours of CPU, and the maximum flow speed for your BCs?

Herve July 19, 2001 11:17

Re: 3D Nonhydrostatic Model Needed
 
Grid resolution was about 0.4 to 0.6 m if I remember correctly, larger in regions of low gradient of the variables; inlet velocity of the order of 0.65 m/s. I cannot remember the real time sorry!

I also modelled a scaled physical experiment (size 24x10x0.2) using 250,000 elements and matched the exp. results very well, inc. shear stress.


All times are GMT -4. The time now is 22:53.