# periodic variation of velocity normal to flow direction using periodic BC

 Register Blogs Members List Search Today's Posts Mark Forums Read

August 29, 2012, 05:01
periodic variation of velocity normal to flow direction using periodic BC
#1
New Member

Nils Schueler
Join Date: Jul 2012
Location: Munich
Posts: 11
Rep Power: 14
Hello again,

I'm still trying to solve a turbulent cross flow over a tube bank with heat transfer. So far I have assembled the single domains exhaust (flowing around the tubes in x-direction), tubes and thermal oil (flowing through tubes in y-direction). In order to reduce computational costs I use periodic BC for the inlet and outlet of exhaust. Being aware that the periodic BC of CFX only applies to the momentum equations, I added an source term for the energy equations at the outlet of the exhaust, which adds the amount of heat being exchanged over the tube surfaces as proposed by stumpy:

Quote:
 Originally Posted by stumpy If your duct walls use a specified wall heat flux boundary, then you can use CEL to take the areaInt of Wall Heat Flux over the duct walls. Next apply a boundary source at the outlet to the energy equation that removes areaInt(Wall Heat Flux)@DuctWalls. This adds/removes the correct amount of energy before the profile gets wrapped around back to the inlet. This does not work with a temperature specified wall.
http://www.cfd-online.com/Forums/cfx...tml#post283314

This generally worked quite well: The solution converged. The pressure loss, the simulated heat transfer coefficients (which both are the results of most interest to me) and the absolute temperature changes of both thermal oil and exhaust are approximately as calculated by empirical correlations. BUT: Both temperatures are rising, where the exhaust should be cooled down by the calculated value...

Searching for the reason of this I found a periodically changing temperature (figure 1) and furthermore y+ (figure 2) at the outer surface of the tubes. Since the layers are of constant thickness along the tubes (fig. 3) I had a closer look at the velocity (fig. 4). The graph shows an periodically changing velocity distribution for which I can see no reasons.

Has anyone encountered a problem similar to this or any idea what can be wrong? Is this possibly related to stumpy's hint that his duct walls use a specified wall heat flux?

further BC:
exhaust - upper and lower wall (normal:y): symmetry
exhaust - side walls (normal:z): adiabatic free-slip wall
any other informations required?
Attached Images
 problem_yplus_temperature.jpg (34.4 KB, 17 views) problem_yplus.jpg (31.8 KB, 15 views) problem_yplus_ConstInflLayer.jpg (40.9 KB, 18 views) problem_periodicBC_velocity.jpg (32.0 KB, 17 views)

 August 29, 2012, 07:11 #2 New Member   Join Date: Jul 2012 Location: Germany Posts: 23 Rep Power: 14 Hi http://www.cfd-online.com/Forums/cfx...onditions.html maybe these velocity changes come from Dean-/Görtler/Floryan-eddy?(flow over/trough curved plates/channels ...) Last edited by bratzinger; August 29, 2012 at 07:37.

August 29, 2012, 08:31
#3
New Member

Nils Schueler
Join Date: Jul 2012
Location: Munich
Posts: 11
Rep Power: 14
Hello bratzinger,
thanks for the hint. I've never heard of these eddys. But why shall they occur in this direction. I mean if I understand it right the reason for theier occurance is curvature but the curvature is only in the x-y-plane (around the tubes). But if the reason for this phenomenon are eddys, their axes should be parallel to the x-y-plane where there is no curvature, or am I wrong?

Quote:
 Originally Posted by bratzinger solid domain boundary conditions: got problems with modelling these in a very short periodic model with my geometrie(tried serveral things). so i made them adiabatic, "said"/guessed that heat convection is dominant to heat conduction, created 2 or more periods(e.g. mirror the mesh in pre ) and then evaluated the middle one(its still way faster than simulating the hole geometrie). after a first run i checked if i was right and checked the gradients(->variable) and temperature profiles by hand, i mean eye ^^, through looking at there contour lines who should also be periodic(if there is a DT u can create a relative temperature via expressions that is displayed by the contour)
do you mean that you encountered the same phenomenon maybe because of a very short model in flow direction?

 August 30, 2012, 05:25 #4 New Member   Join Date: Jul 2012 Location: Germany Posts: 23 Rep Power: 14 Hi if i understand u correctly u got an air flow e.g. in picture 1 from the left side to the right around your pipes (-> if the fliow got enough space there could be a karman eddy street in xy-plane and maybe eddies in other planes caused be the curvature and turbulent instabilities?) and u got the oil flow from upfront to the back through your pipe with i guess transient/turbulent Re-numbers?(->periodicly occuring velocity changes near wall and therefore differences in heatflux/temperature/yplus?).

August 30, 2012, 07:21
#5
New Member

Nils Schueler
Join Date: Jul 2012
Location: Munich
Posts: 11
Rep Power: 14
That's correct besides that the oil is flowing through the tubes from backside to the front but that should not change anything.
Yes, the oil flow should be turbulent but actually the periodic changes do not occur IN the tube (see for example y+ plotted in figure 2 in the left tube) but outside the tube (right tube in fig. 2). So as I see it, the turbulence of the oil flow should not be the reason for the periodic changes, should it?

To be honest I don't have so much experience with turbulent flows so it did not came to my mind that the eddies in x-y-plane could cause eddies in other planes.

So maybe the periodicity isn't really the problem which causes the temperature to rise. Looking at the temperature distribution (see attachment) there is a hot spot in the wake of the second tube which seems unphysical. Maybe I should extend the domain to more than two tubes... What do you think?
Attached Images
 problem_periodicBC_temperature.jpg (29.9 KB, 3 views)