
[Sponsors] 
February 17, 2016, 01:16 
Rayleigh Benard 3D coding with finite differences

#1 
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 
I am developing a simple fortran solver in finite difference. My domain is 2pi*2pi*1 as taken by S. Chandrasekhar. My equations are having non dimensional parameters as Reynolds Number, Rayleigh Number and Prandtl No.
The code is running well but w velocity contours(heating at bottom and cold at top) are coming only positive and negative alternatively and i am not getting rolls as expected. Can someone explain it.? Thanks in advance 

February 17, 2016, 04:18 

#2  
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 3,081
Rep Power: 35 
Quote:
please, provide much more details about your code such as formulation, type of discretization, grid arrangement, values of the nondimensional parameters, etc 

February 17, 2016, 06:07 

#3 
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 
I am implementing Pressure velocity method, and using pressure poisson equation for divergence free flow. Ra=1800, Re=1.0, Pr=1. Grid is uniform in x and y while clustered with cos function in z direction(gravity). discreatisation is done utilizing fourth order Central differencing for diffusion terms, while third order Kawamura and K. Kuwahara upwinding for convective terms. thanks for reply.


February 17, 2016, 06:11 

#4  
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 3,081
Rep Power: 35 
Quote:
if you use a second order discretization and a colocated arrangement, you could have experienced a decouplingmode solution in the pressure equation. Could you post some figures of velocity and pressure fields? 

February 17, 2016, 06:21 

#5  
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 
Quote:
Here are the figures encl : 

February 17, 2016, 06:42 

#6 
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 3,081
Rep Power: 35 
it is a decoupling along the gravity axis, you can see the pressure field oscillating ... why don't you try using a staggered grid?


February 17, 2016, 06:52 

#7 
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 
Yes i am using collocated grid till now. Would you please suggest me some writeup on staggered grid implementation in view of this problem.


February 17, 2016, 07:01 

#8  
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 3,081
Rep Power: 35 
Quote:
a second order discretization on staggered grid is a very old and well suited formulation, you can find that in any CFD textbook. For example, see the book of Peric & Ferziger 

February 17, 2016, 07:15 

#9 
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 

February 22, 2016, 01:17 

#10 
New Member
Madhup
Join Date: Feb 2016
Posts: 6
Rep Power: 3 
Now i applied staggered grid, please see wvelocity profile , i can see rolls still. Will they apper after long time ?


February 22, 2016, 05:32 

#11 
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 3,081
Rep Power: 35 

Tags 
finite difference method, fortran code, prandtl number, rayleighbenard 
Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Rayleigh Benard convection in a bottom heated square cavity  arackal.rahul  Main CFD Forum  2  June 30, 2015 08:14 
Rayleigh benard  afnene  FLUENT  0  June 12, 2015 11:29 
conservative finite differences and finite volumes  Joachim  Main CFD Forum  7  January 23, 2014 17:32 
Is CFD coding in finite element method a formidable job?  bearcat  Main CFD Forum  0  June 18, 2010 20:53 
Finite differences  Zdravko Stojanovic  Main CFD Forum  8  March 24, 2001 23:31 