One step towards perpetual motion

[Claus Andersen]

I'd like some insight in a small study I've run with SYRTHES/Saturne the past few days.

The base of a pin has a heat flux imposed via SYRTHE:. 2W/((0.002m)² • Pi) ~ 160kW/m² 

The entire pin sits in a square duct of 0.01m • 0.01m • 0.04m with water flowing past it at 1m/s - so that's 0.098kg/s (to which Saturne agrees in the listing file)

Now, this is the part where I might be using the wrong method: To get the mean temperature of the outlet I use Paraview to 'slice' the very end of the flow domain - next I apply the 'Integrate variables'. The obtained temperature value is then simply divided by the area.

The problem is, according to this, I get 3W of heat into the water.
0.00200074/0.0001 = 20.0074C at the outlet - 20C at inlet.
Cp • dt • q m = should be 2W
4182J/(kg•K) • 0.0074K • 0.0998kg/s = 3.1W
 
This is a specific Code_Saturne questions because none of my normal alarm bells go off with the calculation.

The calc. runs for 500 iterations and outputs 25 (real) seconds of simulation. Residuals remain stable, and relatively(?) low, here at final step:

c  Pressure     0.13167E-02     290   0.98388E-06   0.11337E-01 c  VelocityX    0.44936E-02     254   0.98831E-06   0.69313E-01 c  VelocityY    0.17894E-02     257   0.96946E-06   0.57092E-02 c  VelocityZ    0.19605E-03     355   0.97597E-06   0.33200E-02 c  TempC        0.69555E-01     152   0.99791E-06   0.97833E-03
 
Are the values used for this test simply too low and Im nitpicking, or am I making some mistake somewhere that Im unaware of?

I've attached an image of the slice+integrated variables and a cut where you can see the pin and the flow domain - listing file is 2.6Mb, let me know if excerpts are needed. I've even gone as far as to upload a vid of the simulation to youtube: http://www.youtube.com/watch?v=agTq7YuhTmc
 
If you can't find nothing wrong, Im patenting this energy generating flow tube :) Screw prior art.
 
Regards,

Claus

[Yvan Fournier]

Hello,

It would be interesting to use the postprocessing of the coupled surface (extract SYRTHES_faces_1 block in ParaView, and the Wall temperature and heat flux should be available on it).

As the temperature gradient along the pin's axis is pretty high near the base, I would suspect some spatial interpolation error issues between the fluid and solid mesh (especially as Code_Saturne uses face-centered values, and Syrthes uses vertex values). Another source of error could be related to turbulence wall laws. Running your calculation with a different turbulence model to see if the difference still occurs would be an interesting test.

Best regards,

  Yvan

[Claus Andersen]

I think part of the problem was, that I underestimated the flow velocity, i.e. running a laminar simulation when in fact it would turn turbulent at the constriction around the pin. Had a little time to launch a new simulation where I halved the mass flow and the result is closer to reality now:
4182J/(kg•K) • 0.0109K • 0.0499kg/s = 2.275W

I'll see if I can increase the mesh density around the base of the pin, in case the interpolation is a problem that needs looking into.
 
Regards,

Claus