[Dyls]

When modeling fully developed flow in a pipe, let's say, how should the pressure jump be calculated as? Ideally one would need to find the pressure drop that the flow feels through that particular section of pipe and then apply it to the interfaces. This critical pressure value should mean that the flow velocity is neither increasing to some value nor is it decreasing to 0.

Is this value typically found through trial and error? Also, if the pressure jump is increased above this value, the flow velocity would increase until there is a balance between the friction drag and the pressure (but not to infinity), correct?

Thank you for your time!

Dylan

[ping]

not sure what you are getting at here - you can use a pressure jump or a mass flow in a case like this so you need to specify either of these values from measurement or by doing a simple pipe flow calculation for a unit length of pipe

[Dyls]

Let me be more clear:

I am trying to model a basic pipe flow problem. I am keeping the flow laminar (so Reynolds number below 2300). I start with a inlet velocity just to initialize, then I attach the inlet and outlet with a fully developed, periodic interface that is mass-flow driven. I am unable to get this basic problem to become stable.

Does anyone have any suggestions? It seems that I am using the model incorrectly if I cannot get this basic water in a pipe problem. I would like to get this to work before moving onto a more difficult problem.

Thanks!

Dylan

[ping]

you don't need and should not use an inlet boundary type in this case - if you want to initialize the flow do it the correct was using initial conditions in the physics but it should not be needed

you should have three boundaries in your case being an inlet and and outlet and a wall but they should all be see to wall type boundaries since this is prevent problems with poor interfacing on the outer edges of the interface

then create the interface and set to fully developed and period then set either mass flow or pressure jump and input an appropriate value for the flow or pressure change

it is very simple if you follow these steps and it will converge very well

[Dyls]

Thanks for all the help with some of these more basic questions.

I am now finding that periodic flow works well for cylinders that are around 1 m diameter or smaller but when I experiment with much larger cylinders (diameters of 10m) I find that my viscosity is being limited (which I think means that my k^2/epsilon in the turbulent viscosity term is increasing to infinity).

The model I am working with is: steady, viscous, constant density gas, k-epsilon RANS turbulence model.

So can anyone tell me why turbulent viscosity is limited at this scale when using a fully developed periodic interface at inlet/outlet? Does it have to do with the coarseness of my mesh (perhaps my prism layers)? How can I reach this size of pipe without this problem?

Thanks again.

Dylan