How to set up a 1-D simulation

February 1, 2002, 11:11

Hi, everybody!

I want to simulate a column of fluid which is forced by a momentum influx at the top (high z) parallell to the wall. The actual case is a column of water in the sea forced by the wind at the surface. In my case the surface stress is 0.1 Pa and set with COefficient FIXFLU. The fluid is linearly stratified to begin with, and I want to see how the stratification is altered due to the turbulent mixing that results from the near-surface current profile. This is a simple standard case to compare different turbulence models, using the data from an experiment by Kato & Phillips.

I have written my own code to do this as well as used a 1-D code for boundary layers. However, I have failed miserably in PHOENICS. The set-up is as follows:

1. CARTES=T, NX=NY=1, NZ=200, total extent in the z-direction 50 m, time step 30 s, LSWEEP=10.

2. Standard k-e model with built-in sources and sinks (including buoyant destruction) and logarithmic boundary conditions.

3. An initial linear density in C1 (representing salinity) and a simple linear equation of state (coded in GROUND) yielding a constant initial buoyancy frequency of 1E-2 s^-1. All other variables are 0 to begin with. The initial conditions have been checked and are correct.

So, how do I set this up? Since I do not want any lateral boundaries I have used XCYCLE=T and set the momentum influx only in the x-direction. However, I have had all kinds of problems, which I have tried to solve in lots of different ways. In short, if I manage to produce any results at all, these are clearly erroneous. Does anyone have a recipe or perhaps some helpful hints? I use PHOENICS 3.2 under WinNT.

Thanks,

Olof

February 1, 2002, 13:06

The single-slab solver is used in PHOENICS to solve fully-developed couette and non-couette flows, as documented on line at:

http://www.cham.co.uk/phoenics/d_polis/d_enc/fdf.htm)

Library examples for both Couette flows and flows with a pressure gradient in the main flow direction can be found in the advanced turbulence library (turbulent examples) and in the core library (laminar examples).

From the above you should be able to determine the required PIL commands for your 1d solution with the single slab solver, but remember you must solve on a single iz slab, so choose ix as the 1d direction. The Couette example (library case 505) is a good starting point for your case because that is simply a moving wall driving the flow.

February 1, 2002, 11:11	How to set up a 1-D simulation	#1
Olof Liungman Guest Posts: n/a	Hi, everybody! I want to simulate a column of fluid which is forced by a momentum influx at the top (high z) parallell to the wall. The actual case is a column of water in the sea forced by the wind at the surface. In my case the surface stress is 0.1 Pa and set with COefficient FIXFLU. The fluid is linearly stratified to begin with, and I want to see how the stratification is altered due to the turbulent mixing that results from the near-surface current profile. This is a simple standard case to compare different turbulence models, using the data from an experiment by Kato & Phillips. I have written my own code to do this as well as used a 1-D code for boundary layers. However, I have failed miserably in PHOENICS. The set-up is as follows: 1. CARTES=T, NX=NY=1, NZ=200, total extent in the z-direction 50 m, time step 30 s, LSWEEP=10. 2. Standard k-e model with built-in sources and sinks (including buoyant destruction) and logarithmic boundary conditions. 3. An initial linear density in C1 (representing salinity) and a simple linear equation of state (coded in GROUND) yielding a constant initial buoyancy frequency of 1E-2 s^-1. All other variables are 0 to begin with. The initial conditions have been checked and are correct. So, how do I set this up? Since I do not want any lateral boundaries I have used XCYCLE=T and set the momentum influx only in the x-direction. However, I have had all kinds of problems, which I have tried to solve in lots of different ways. In short, if I manage to produce any results at all, these are clearly erroneous. Does anyone have a recipe or perhaps some helpful hints? I use PHOENICS 3.2 under WinNT. Thanks, Olof

February 1, 2002, 13:06	Re: How to set up a 1-D simulation	#2
Mike Malin Guest Posts: n/a	The single-slab solver is used in PHOENICS to solve fully-developed couette and non-couette flows, as documented on line at: http://www.cham.co.uk/phoenics/d_polis/d_enc/fdf.htm) Library examples for both Couette flows and flows with a pressure gradient in the main flow direction can be found in the advanced turbulence library (turbulent examples) and in the core library (laminar examples). From the above you should be able to determine the required PIL commands for your 1d solution with the single slab solver, but remember you must solve on a single iz slab, so choose ix as the 1d direction. The Couette example (library case 505) is a good starting point for your case because that is simply a moving wall driving the flow.