I have written a Fortran program to simulate flow around airfoil:

My problem is: Determination of separation on naca airfoil at low Reynolds numbers (1e3 to 1e5)

Flow caracteristics: * Steady, 2D * Incompressible, viscous and Laminar

Mathematical model: * Continuity equation * 2 momentum equations

Discretization method: * Structured C-grid generated by "mesh pilot" software * the farfield boundary is 15xcord * transformation of model from physical (x, y) to computational domain (ksi, eta) * Finte volumes discretization * Staggered grid for velocity and pressure * Simpler Algorithm * Second order Upwind scheem for convective terms and central differences for diffusive terms

Solution of Algebraic equations: * Sweeping * Thomas Algrithm (TDMA)

My program diverges after about 5000 itterations althought i've used low relaxation parameters: relaxation for velocity 4e-1 and for pressure 1e-1.

Is that because i've not well imposed boundary conditions or because of other problems?

At those Reynolds numbers, the flow is going to separate very soon after the maximum thickness. The wake flow will be unsteady. I don't have a clue why your solver is diverging - it could be anything from a boundary condition to a coding error to the phase of the moon. But I would definitely consider using an unsteady solver for the problem you are seeking to simulate.

Discretization method: * Structured C-grid generated by "mesh pilot" software * the farfield boundary is 15xcord * transformation of model from physical (x, y) to computational domain (ksi, eta) * Finte volumes discretization *

The book by Blazek is an excellent reference about the Finite volume method, thereis no need to transform from physical to computational domains, the finite volume method works directly with the physical domain, I guess you have mixed up different concepts in your code, check the book, it has a CD with fortran code hope that will help

I have set the following boundary conditions: on airfoil (low side of comput domain): u=v=0 farfield (upper side of comput domain): u=1, v=0 inflow (left side):du/dksi=dv/dksi=0 outflow (right side): du/dksi=dv/dksi=0 u(ksi, eta) and v(ksi,eta) the compnents of velocity.

Is that the correct way to set boundary conditions?

What are you doing with pressure on the boundary?

I'm using SIMPLER Algorithm which allow to find pressure from velocity

[morteza08]

Originally Posted by Bounecer
;47729

I have written a Fortran program to simulate flow around airfoil:

My problem is: Determination of separation on naca airfoil at low Reynolds numbers (1e3 to 1e5)

Flow caracteristics: * Steady, 2D * Incompressible, viscous and Laminar

Mathematical model: * Continuity equation * 2 momentum equations

Discretization method: * Structured C-grid generated by "mesh pilot" software * the farfield boundary is 15xcord * transformation of model from physical (x, y) to computational domain (ksi, eta) * Finte volumes discretization * Staggered grid for velocity and pressure * Simpler Algorithm * Second order Upwind scheem for convective terms and central differences for diffusive terms

Solution of Algebraic equations: * Sweeping * Thomas Algrithm (TDMA)

My program diverges after about 5000 itterations althought i've used low relaxation parameters: relaxation for velocity 4e-1 and for pressure 1e-1.

Is that because i've not well imposed boundary conditions or because of other problems?

hi i am an aerospace engineer....i have to simulate the flow over an airfoil, i have to write a code and i have not enough time to do that...
can i borrow your code?
i prefer spending my time on a pre-written code and try to modify it for my purpose..

thanks