radiating panel: FDS vs. theoretical

[AACosta]

Hi all,

I'm trying to simulate a simple configuration of a radiating panel and 3 radiative sensors, and I want to compare FDS results against theoretical values of incident radiation calculated using:

View factor · 1· sigma · Tpanel^4 

The view factor computed is around 0.023 for the configuration presented in the file (sensor orientation (1,0,0), so the theoretical value is about 0.08 W/m2.

With FDS I get the following values depending on the scenario tested:

Differences are quite large and I don't understand why. 

I did another test positioning the sensor at a height of 1 m and I got a value of 0.26 W/m2 with FDS. The theoretical view factor is about 0.047. So if I do a simple calculation:

0.26 · 0.023/0.047 to estimate the value I should expect from FDS for the sensor at a height of 0.45 m I get : 0.128 W/m2. This value is similar to the one computed with FDS, so it makes me think that the "geometry" is well resolved but there might me something related to the radiation model that I am missing. 

Any help?

The code for scenario #11 is here:

&HEAD CHID='bb_4_m_ronan', TITLE='Radiating horizontal BB panel' /

------- 1MxxxCS -----------------------------------

&MESH ID='Mesh01', IJK=158,120,40, XB=-4.9,3,-3,3,0,2 /

MESH ID='Mesh01', IJK=79,60,20, XB=-4.9,3,-3,3,0,2 /

&TIME T_END=10, WALL_INCREMENT=1 /

---Strahlung-----

&RADI NUMBER_RADIATION_ANGLES = 400,TIME_STEP_INCREMENT=1,ANGLE_INCREMENT=1/

&DUMP DT_SLCF=5., DT_BNDF=5., DT_DEVC=.5, DT_HRR=5., WRITE_XYZ=.TRUE., SIG_FIGS=4, SIG_FIGS_EXP=2 /

&MISC SIMULATION_MODE='LES', TMPA=-273, HUMIDITY=0, Y_CO2_INFTY = 0 /

&SURF ID        = 'BBsurface'

      COLOR     = 'GREEN'

      TMP_FRONT = 227.15

  TAU_T=0

      EMISSIVITY=1

      HEAT_TRANSFER_COEFFICIENT=0/

&VENT MB='XMIN', SURF_ID = 'OPEN' /

&VENT MB='XMAX', SURF_ID = 'OPEN' /

&VENT MB='YMIN', SURF_ID = 'OPEN' /

&VENT MB='YMAX', SURF_ID = 'OPEN' /

&VENT MB='ZMIN', SURF_ID = 'OPEN' /

&VENT MB='ZMAX', SURF_ID = 'OPEN' /

&OBST XB=-2,2,-2,2,0.00,0.00, SURF_ID='BBsurface' /

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='GAUGE HEAT FLUX', ID='qdotgauge'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='RADIOMETER', ID='qdotradiometer'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='NET HEAT FLUX', ID='qdotn'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='RADIATIVE HEAT FLUX', ID='qdotr'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.0, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas1'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.3, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas2'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.6, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas3'/

&TAIL /

Thanks in advance,

[dr_jfloyd]

I looked at your case. You have a small view factor with a fairly oblique angle to the radiating surface; you need more angles and/or grid resolution.  The plot below shows the IOR=+1 flux at various elevations above the floor at the XMIN boundary as a function of the number of angles.

[AACosta]

Wow, thanks a lot. This is a very nice figure!

Why the radiative heat flux from FDS has this behaviour? 

Section 16.2 in the UG, should I have tried it (i.e. add slice (SLCF) files of the quantity ’INTEGRATED INTENSITY’)?

Thanks again!

El dia dijous, 1 d’abril de 2021 a les 14:22:13 UTC+2, dr_jfloyd va escriure:

[o...@aquacoustics.biz]

Dr. Floyd has provided an illustrative graph, but be aware that increasing the solid angle resolution has a significant computational cost.

[dr_jfloyd]

AAcosta - If you haven't already done so look at the discussion of the radiation solver in the Technical Reference Guide. I think in your case it is due to having the gauge far off to the side of the radiating plane with a very small view factor. Resolving that small view factor means you need more resolution in the solver, i.e. more angles.

Re Solver Time - If nothing else is changed the fraction of the computational cost consumed by the radiation solver will increase proportionately to the number of angles. The default mode of operation is that radiation solver is called every three timesteps (TIME_STEP_INCREMENT=3) and does 1/5 of the angles each time it is called (ANGLE_INCREMENT=5). In many cases the timestep being used by FDS is small enough that over 15 timesteps the average radiation source term does not change very much. You can compensate for the cost of more angles by increasing one or both of TIME_STEP_INCREMENT and ANGLE_INCREMENT so that the radiation solution needs more timesteps to update. It is often possible to do this without having a substantial impact on the results.  

[AACosta]

Dear Dr. Floyd,

I tested the simulation just changing the number of radiation angles (N=1600) and I got a value slightly different from yours (I guess so, according to the figure you shared). 

Apparently, you get almost no difference at z = 0.45 m between FDS and "exact" if N= 1600. However, I got these results:

Sensor orientation (1,0,0,) : 0.09492 kW/m2

Sensor orientation (1,0,0.3,) : 0.08652 kW/m2

Sensor orientation (1,0,0.6,) : 0.07352 kW/m2

I got the same results (but faster) with N= 1600 and TIME_STEP_INCREMENT=1 instead of 15.

So, my doubt is: why did I get different results? Am I missing anyth.?

Moreover, why the solution obtained using N=1600 angles does not "converge" to the exact solution when we increase the z value if we are increasing the view factor? An oscillating behavior is still apparent, isn't it?

I used exactly the following lines of code:

&HEAD CHID='bb_4_m_ronan', TITLE='Radiating horizontal BB panel' /

------- 1MxxxCS -----------------------------------

&MESH ID='Mesh01', IJK=79,60,20, XB=-4.9,3,-3,3,0,2 /

&TIME T_END=10, WALL_INCREMENT=1 /

---Strahlung-----

&RADI NUMBER_RADIATION_ANGLES = 1600,TIME_STEP_INCREMENT=15,ANGLE_INCREMENT=1 /

&DUMP DT_SLCF=5., DT_BNDF=5., DT_DEVC=.5, DT_HRR=5., WRITE_XYZ=.TRUE., SIG_FIGS=4, SIG_FIGS_EXP=2 /

&MISC SIMULATION_MODE='LES', TMPA=-273, HUMIDITY=0, Y_CO2_INFTY = 0 /

&SURF ID        = 'BBsurface'

      COLOR     = 'GREEN'

      TMP_FRONT = 227

  TAU_T=0

      EMISSIVITY=1

      HEAT_TRANSFER_COEFFICIENT=0/

&VENT MB='XMIN', SURF_ID = 'OPEN' /

&VENT MB='XMAX', SURF_ID = 'OPEN' /

&VENT MB='YMIN', SURF_ID = 'OPEN' /

&VENT MB='YMAX', SURF_ID = 'OPEN' /

&VENT MB='ZMIN', SURF_ID = 'OPEN' /

&VENT MB='ZMAX', SURF_ID = 'OPEN' /

&OBST XB=-2,2,-2,2,0.00,0.00, SURF_ID='BBsurface' /

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='GAUGE HEAT FLUX', ID='qdotgauge'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='RADIOMETER', ID='qdotradiometer'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='NET HEAT FLUX', ID='qdotn'/

&DEVC XYZ=-4.9,0,0.45, IOR=+1, QUANTITY='RADIATIVE HEAT FLUX', ID='qdotr'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.0, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas1'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.3, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas2'/

&DEVC XYZ=-4.9,0,0.45, ORIENTATION=1,0,0.6, QUANTITY='RADIATIVE HEAT FLUX GAS', ID='qdotrgas3'/

&TAIL /

Thanks for yuor help,

El dia diumenge, 4 d’abril de 2021 a les 13:07:47 UTC+2, dr_jfloyd va escriure:

[Tim O'Brien]

Have you run a verification study on your computational platform using the NIST suit?  Different hardware, and software (particularly for compiled FDS) can produce surprisingly different results in extended iterative calculations.  While a particular platform should produce consistent results for a particular model you should reasonably expect variations in results between platforms.  In my experience radiation calculations are particularly sensitive.  Does your platform meet the NIST verification thresholds for radiation calculations?      

On Thursday, April 1, 2021 at 2:58:59 AM UTC-4 AACosta wrote:

Hi all,

I'm trying to simulate a simple configuration of a radiating panel and 3 radiative sensors, and I want to compare FDS results against theoretical values of incident radiation calculated using:

View factor · 1· sigma · Tpanel^4 

 

The view factor computed is around 0.023 for the configuration presented in the file (sensor orientation (1,0,0), so the theoretical value is about 0.08 W/m2.

 

With FDS I get the following values depending on the scenario tested:

 

--
You received this message because you are subscribed to a topic in the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/fds-smv/8ebGSip7W2s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to fds-smv+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/c5aaa4dd-3767-406c-bf4b-e77dd34f4223n%40googlegroups.com.

[dr_jfloyd]

AACosta,

Probably slightly different input files. I may have simplified your case some before running it. I did not save those files from two weeks ago.

o...@aquacoustics.biz,

I would not expect a case with just radiation to show much if any difference between platforms. With no participating media, each update to the radiation solution simply starts from the initial conditions. We routinely run on different platforms during development and don't typically see noticeable differences. At least not with the current generation of Intel compliers.