Dear all,

I am simulating a 2D laminar flow passing a bluff body in a channel. I just wonder if I know the flux rate/ bulk velocity, how can I accelerate the flow to achieve the flux rate? Any literature about this acceleration method?

Many thanks in advance! Rui

continue with last post. Two periodic boundaries are set up at the two ends of the channel.

Use the g_x (body force) term in your x-momentum equation to impose a force driving the fluid through your mesh (this assumes that your primary flow direction is x). Run several different values for g_x and calculate the resulting u_x (average velocity in the x direction for each value of g_x. Plot u_x vs g_x, pick off the value of g_x that corresponds to the u_x you want. That solution is the one you are looking for.

But is there any automatic method to change the pressure gradient, g_x, corresponding to the bulk velocity we obtained after several time steps?

Plotting out the g_x VS U_x will be very painful.

Anyway, thank you very much, otd for the reply!

If you specify the body force (acceleration), the pressure gradient will emerge automatically as the solution to the pressure correction equation. You shouldn't need to specify any pressures (it's all relative in incompressible flow anyway - only the pressure gradients count).

Note that u_x is the component of velocity in the x (primary flow) direction. Specify g_x (this is one value over the entire mesh). Run your calculation to steady state, integrate the flow at any value of x to get the value of volume flux (V_f) = integral (from y_min to y_max)of u_xdy. Due to conservation of mass, this is the same for any value of x.

That gives you one point g_x(1), V_f(1).

Repeat for g_x(2) to get V_f(2), (3), .... Three points isn't in my opinion a very difficult graph. You may need 4 or 5 points to be sure you bracket your desired flow rate. Enter the curve at the flow rate of interest; that corresponds to the correct value of g_x*. Repeat your CFD calculation using g_x* to get the final pressure field.

Since you have a bluff body in the flow field, you'll need to examine the effect of the length (upstream and downstream) of your computation region on the answer. Experiment to be sure you have a long enough computational domain.

Hi otd,

Thank you very much for the help. It is very valuable!

What I am thinking is that is there any way that every time step, or every few time steps, we obtain a V_f(n), corresponding to the g_x(n), in next step, we could change g_x(n) to g_x(n+1), replying on V_f(n), g_x(n) and the ideal flux rate V_f*, according to g_x*? That is g_x(n+1)=function(g_x(n),V_f(n),V_f*) repeatedly, we can get a g_x(i) which is close to g_x*.

I think there is some way we could do that as in commercial code, STAR-CD you can specify a flux rate V_f* for a bluff body in the channel, then the code can automatically accelerate the flow without getting a graph of g_x(i) V.S. V_f(i).

Many thanks! Rui

Having your code iterate on g_x to seek the solution you want automatically is a whole different level. I have no experience with that.

If you're using a commercial code, check with your technical support folks.

Good luck!

Yes, I can check the technical support. Cheers, Rui