Grid Sensitivity E_COEFFICIENT

[Marion]

Hey! I am modeling an experimental fire test and am having problems with the grid sensitivity. 

I have a setup with an upholstered chair with a HRRPUA = 7792 kW/m² and a sprinkler with the following parameters:

&SPEC ID='POLYURETHANE', FORMULA='C25H42O6N2'/
&REAC FUEL='POLYURETHANE', SOOT_YIELD=0.131, HEAT_OF_COMBUSTION=23500. /
&VENT XB=1.80,2.20, 1.60,2.00, 0.60,0.60, COLOR='RED', SURF_ID='BURNER' / 
&SURF ID='BURNER',HRRPUA=7791.625, RAMP_Q='HRR1', COLOR='RED', E_COEFFICIENT=0.03 /
&RAMP ID='HRR1', T=0.0, F=0.00 /
&RAMP ID='HRR1', T=25.0, F=0.00 /
&RAMP ID='HRR1', T=50.0, F=0.01 /
&RAMP ID='HRR1', T=75.0, F=0.01 /
&RAMP ID='HRR1', T=100.0, F=0.02 /
&RAMP ID='HRR1', T=125.0, F=0.02 /
&RAMP ID='HRR1', T=137.5, F=0.02 /
&RAMP ID='HRR1', T=150.0, F=0.03 /
&RAMP ID='HRR1', T=175.0, F=0.17 /
&RAMP ID='HRR1', T=200.0, F=0.51 /
&RAMP ID='HRR1', T=225.0, F=0.85 /
&RAMP ID='HRR1', T=250.0, F=1.00 /
&RAMP ID='HRR1', T=251.0, F=0.96 /
&RAMP ID='HRR1', T=275.0, F=0.64 /
&RAMP ID='HRR1', T=300.0, F=0.42 /
&RAMP ID='HRR1', T=325.0, F=0.28 /
&RAMP ID='HRR1', T=350.0, F=0.23 /
&RAMP ID='HRR1', T=375.0, F=0.19 /
&RAMP ID='HRR1', T=400.0, F=0.18 /

&SPEC ID='WATER VAPOR' /
&PART ID='WATER_DROPS'
SPEC_ID='WATER VAPOR'
DIAMETER=2000.
SAMPLING_FACTOR=1
QUANTITIES(1:2)='PARTICLE DIAMETER','PARTICLE VELOCITY' /
&PROP ID='ZSTP15'
SMOKEVIEW_ID='sprinkler_pendent'
PART_ID='WATER_DROPS'
PARTICLES_PER_SECOND=5000.
FLOW_RATE=50.6
ORIFICE_DIAMETER=0.0127
SPRAY_ANGLE=0.,105.
OFFSET=0.20 / 

Based on a grid width of 10 cm, I calibrated the E coefficient to 0.03 with results of experiments with this setup. I then conducted a grid sensitivity study and found that even with a 1.25 cm grid there were still significant differences in the simulation results compared to a 2.5 cm grid. I then repeated the grid sensitivity study with an E coefficient of 2.0, believing that 0.03 might be too low a value. However, the results still showed significant deviations depending on the grid size.

Is there a specific grid size at which the submodel is validated to reduce the heat release rate through sprinklers, or is it necessary to use an even finer grid than 1.25 cm? or maybe it's because my HRRPUA keeps rising after sprinkler activation?

I also performed a sensitivity analysis of the other physical or numerical parameters. 

[Kevin McGrattan]

There is only one set of validation experiments in the FDS Validation Guide that invoke E_COEFFICIENT, namely the UL/NFPRF experiments. There are 5 rack storage fire simulations P-1 through P-5, that use it. The validation is indirect, in that we only consider the sprinkler activation pattern and not the HRR, which was not measured. 

In your case, you also have to consider other parameters that might affect your outcome. For example, the rate of spread in the horizontal and vertical direction of the water droplets. The default values were estimated for relatively hard surfaces like boxed commodities, not upholstered furniture.