[xerox]

Hi,

I am trying to simulate the single pipe flow of a heat exchanger with the LES tool provided by fluent.

I am especially interested in temperature and the pressure drop over a specific pipe length. In the past I have done successful analysis with a RANS model which could be validated but this is the first large eddy simulation.

I am facing the following problem:
With a RANS turbulence model I evaluated the static pressure at the In- and Outlet for the pressure drop.
If I do this with my LES data for a certain amount of time steps to get the mean value of the transient calculation I got a negative value for my Inlet (fluctuation from -100 Pa .. +30 Pa, mean value = -30 Pa) and 0 Pa for the Outlet. This would result in an non-physical increase of pressure. On top the magnitude (without the algebraic sign) would be way to less as expected.

If I do the same analysis for the dynamic pressure the values at the In- and Outlet are positive values without fluctuation, but the pressure drop is only about 10 Pa, which is also way to less as you would expect.

Here is my current initial situation:
- velocity Inlet with a given profile of a prior RSM simulation of the pipe using the spectral synthesizer for the fluctuating velocity algorithm with the single reynolds stresses
- calculation of steady RSM solution as initialization, then switch to LES
- dynamic smagorinsky Modell (dynamic stress and dynamic energy flux)
- SIMPLEC for pressure - velocity coupling
- very small timestep for CFL < 1
--> every timestep needs about 10..15 iterations to converge
- double precision, pressure based solver
- for the heat exchanger I used a wall with constant temperature

I am pretty sure the grid is fine enough as it suits the following conditions:
- y+ for the wall approx 0.7
- delta z + (for the streamwise direction) = 4, the fluent users guide recommends much coarser grid with delta z + approx 40 for LES
- aspect ratio for 75% of the cells is smaller than 3, the max aspect ratio is 6.7

Mean values of velocity or temperature at the Outlet match to prior RANS simulations but I do not see a way to tell the pressure drop with this data. Any help for my following questions would be kindly appreciated:

1) How do I analyse the pressure drop of a pipe flow with a Large Eddy Simulation? Would you recommend taking the mean static pressure or the dynamic one?
2) Would you recommend any changes to my initial setup? Any experience you want to share on how to make better progress?

If there are missing informations for my simulation please feel free to ask for it. As for now I tried to not forget important details.

Thank you very much in advance,
kind regards !

[SAIKRISHNA N]

I am also trying to simulate turbulent flow in straight circular pipe with LES. I am also facing same problem of very low pressure drop (just 10% of expected value). I have set up the case almost in the same way as Xerox, but with SST k-w model initially. I tried with Smogorinsky, dynamicKEqn and WALE models. I didn't get any improvement.
Anyone, please throw some light on what could be wrong here? why the LES result of pressure drop is so low while RANS give very good prediction?