3D Modeling of burning of porous forest fuels under cone calorimeter

[shanmugasundaram2706]

Dear FDS developers,
I am modelling the burning of live and fresh spruce needles in a cone calorimeter, and I am trying to validate the pyrolysis and char oxidation models using the HRRPUA and MLRPUA experimental data. The fuel particles are modelled as cylindrical Lagrangian particles. The porous sample holder (10 cm x 10 cm) with a fuel depth of 3 cm is modelled as a screen with a free area fraction of 0.5. A mesh size of x=5 mm, y=10 mm and z=5 mm is used in the fuel bed region. The kinetic parameters for pyrolysis and char oxidation are estimated using the simplified Arrhenius model available in FDS by fitting the TGA data. The model can therefore predict ignition time and peak HRRPUA and MLRPUA reasonably well. However, a strange, spurious numerical spike was observed at the end of char oxidation (when approximately 10% of the total mass remained), as shown in the attached figure.

During this time, no visible flame is observed. A local high-temperature zone (with T ~ 1450 °C) suddenly forms at the bottom of the sample holder and eventually spreads throughout the fuel bed. The char oxidation model was fitted using a general rate equation with N_s = 1 and n_(O2) = 0.78. This spurious spike was not observed in simulations run without the char oxidation reaction. Attempts were also made to replace the char oxidation model with the surface oxidation model available in FDS, and also running without SCREEN or porous walls. In both cases, the sudden spike at the end of char oxidation was still observed. What could cause this sudden spike? How can it be fixed? Could it be due to a collapse in heat capacity as the remaining particle mass becomes extremely small? The material lines for all the components are given below.

&DUMP DT_RESTART=5.0 /
&SPEC ID = 'WATER VAPOR'/
&SPEC ID = 'CARBON DIOXIDE'/
&SPEC ID='CARBON MONOXIDE'/
&SPEC ID='OXYGEN'/
&SPEC ID = 'Fuel gas', FORMULA = 'C2.1H6.2O2.2', RADCAL_ID = 'C2H6OH',
SIGMALJ = 4.53, EPSILONKLJ = 362.6, PR_GAS = 0.84/

&REAC FUEL = 'Fuel gas', HEAT_OF_COMBUSTION = 19190., SOOT_YIELD=0.02/
REAC FUEL = 'CARBON MONOXIDE', HEAT_OF_COMBUSTION = 10100./

&SURF ID = 'wet vegetation'
MATL_ID = 'dry pine'
COLOR = 'RED'
MOISTURE_FRACTION = 1.128
SURFACE_VOLUME_RATIO = 7272.
LENGTH = 0.09
GEOMETRY ='CYLINDRICAL' /

&PART ID='pine needles', SAMPLING_FACTOR=1, SURF_ID='wet vegetation', PROP_ID='wood image', SHAPE_FACTOR=0.32, QUANTITIES='PARTICLE TEMPERATURE','PARTICLE MASS','PARTICLE DIAMETER', STATIC=.TRUE. /

&INIT ID = 'pine needles init', PART_ID='pine needles', XB=-0.05,0.05,-0.05,0.05,0.1,0.13, N_PARTICLES_PER_CELL=1., MASS_PER_VOLUME=66.66 /

&PROP ID='wood image', SMOKEVIEW_ID='TUBE', SMOKEVIEW_PARAMETERS='L=0.09','D=0.00055' /

&MATL ID = 'MOISTURE'
EMISSIVITY = 0.98
DENSITY = 1000.
CONDUCTIVITY = 0.6
SPECIFIC_HEAT= 4.184
A = 600000.
E = 48200.
N_T=-0.5
NU_SPEC = 1.
SPEC_ID = 'WATER VAPOR'
HEAT_OF_REACTION= 2259. /

&MATL ID = 'dry pine'
EMISSIVITY = 0.933
DENSITY = 510.
CONDUCTIVITY = 0.135
SPECIFIC_HEAT_RAMP = 'd_ramp'
A = 89100.
E = 81460.
NU_MATL(1,1) = 0.4325
MATL_ID(1,1) = 'CHAR'
NU_SPEC(1,1) = 0.5675
SPEC_ID(1,1) = 'Fuel gas'
HEAT_OF_REACTION = 418./

&RAMP ID='d_ramp', T=20., F=1.3 /
&RAMP ID='d_ramp', T=70., F=1.8 /
&RAMP ID='d_ramp', T=120., F=2.3 /
&RAMP ID='d_ramp', T=170., F=2.8 /
&RAMP ID='d_ramp', T=200., F=2.0 /
&RAMP ID='d_ramp', T=500., F=2.0 /
&RAMP ID='d_ramp', T=1800., F=2.0 /

&MATL ID = 'CHAR'
EMISSIVITY = 0.75
DENSITY = 220.575
CONDUCTIVITY = 0.05
SPECIFIC_HEAT_RAMP = 'c_ramp'
A = 361500.
E = 112600.
N_S = 1.2
N_O2 = 0.78
SPEC_ID ='CARBON MONOXIDE','CARBON DIOXIDE','OXYGEN'
NU_SPEC = 0.78,1.82,-1.65
MATL_ID = 'ASH'
NU_MATL = 0.05
HEAT_OF_REACTION= -12000./
&RAMP ID='c_ramp', T=200., F=0.85 /
&RAMP ID='c_ramp', T=300., F=1.08 /
&RAMP ID='c_ramp', T=400., F=1.33 /
&RAMP ID='c_ramp', T=500., F=1.59 /
&RAMP ID='c_ramp', T=600., F=1.87 /
&RAMP ID='c_ramp', T=700., F=2.16 /
&RAMP ID='c_ramp', T=800., F=2.46 /
&RAMP ID='c_ramp', T=900., F=2.77 /
&RAMP ID='c_ramp', T=1000., F=3.10 /
&RAMP ID='c_ramp', T=1100., F=3.45 /
&RAMP ID='c_ramp', T=1200., F=3.81 /
&RAMP ID='c_ramp', T=1300., F=4.18 /
&RAMP ID='c_ramp', T=1400., F=4.56 /
&RAMP ID='c_ramp', T=1500., F=4.96 /
&RAMP ID='c_ramp', T=1600., F=5.37 /
&RAMP ID='c_ramp', T=1700., F=5.79 /
&RAMP ID='c_ramp', T=1800., F=6.23 /

&MATL ID = 'ASH'
EMISSIVITY = 0.75
DENSITY = 11.03
CONDUCTIVITY = 0.1
SPECIFIC_HEAT_RAMP = 'a_ramp' /

&RAMP ID='a_ramp', T=200., F=1.09 /
&RAMP ID='a_ramp', T=300., F=1.24 /
&RAMP ID='a_ramp', T=400., F=1.36 /
&RAMP ID='a_ramp', T=500., F=1.45 /
&RAMP ID='a_ramp', T=600., F=1.55 /
&RAMP ID='a_ramp', T=700., F=1.62 /

[drjfloyd]

This isn't a complete input file so no one can run this and see what is happening. You also have not provided the specific outputs that you are plotting. For us to look at this we would need a complete input file. We would then have to add outputs to let us understand what is happening in the particle bed. This is something you should be making the first attempt at. I suggest you first run a virtual TGA and make sure the results look OK. Then suggest you just make one particle and device outputs so you can track what is happening in depth in the particle.

[shanmugasundaram2706]

Dear Jason Floyd,
Thanks for your quick response. Please find the attached FDS input file. HRRPUA (kW/m2) plotted above is the sum of HRR (kW) and HRR_oxidative (kW) (obtained from HRR file) divided by cross sectional area of fuel bed.

spruce_needle.txt
The TGA simulation is run for heating rate of 10 K//min and the result is also shown below.

Please let me know if anyother information is needed.

Thanks,

Regards,
Shanmugasundaram D

[drjfloyd]

The case you attached has very few outputs that allow you to develop a deep understanding of what is happening in your simulation.

Run a case with a single particle with a known exposure and heavily instrument it so you understand the in-depth temperatures and material composition. Look to see if you notice anything that might explain what you see in the full case.

If you don't see anything there repeat this with your full case. Add more devices so you can monitor the heat flux to particles, the in-depth temperatures, etc. See if you can determine what is happening.

[shanmugasundaram2706]

Dear Jason Floyd,
Based on your suggestion, I have done a single particle simulation by imposing external heat flux of 50 kW/m2. I am observing a similar behaviour as that of full case at the end of char oxidation. The input file of this simulation and the results comparing both the cases are also attached below. Please suggest me how to fix this issue. Can we impose some limitation on particle temperature or do we have any other way to solve this issue? Thanks.
single particle.txt

[drjfloyd]

&SPEC ID = 'Fuel gas', FORMULA = 'C2.1H6.2O2.2', RADCAL_ID = 'C2H6OH',

Does RADCAL_ID='C2H6OH' exist in the list of available inputs in 14.3.1 of the User's Guide?

&REAC FUEL = 'Fuel gas', HEAT_OF_COMBUSTION = 19190., SOOT_YIELD=0.02/
This is a simply chemistry reaction. It does not use the primitive species of OXYGEN. I suggest you review the User's Guide chapters on species, gas phase reactions, solid phase reactions, and the examples on modeling vegation in the wildland chapter. If you want your char reaction to use the primitive species of OXYGEN you should make this REAC also use primitive species and approriately define all needed primitive species with their initial mass fractions and background species choice. Or you could define a second REAC for the species of C with yields that match your CHAR yields. This will create a second PRODUCTS species. Then for your CHAR material you would consume AIR and make PRODUCTS. The .out file would show you the mass based stoichiometry to help you with the inputs.

The reason you see the temperature spike is the char reaction is exothermic. If the temperature is spiking too high, the solution is not to attempt some ad hoc restriction on temperature which would violate conservation of energy, but rather look to see if changes are needed to your char oxidation reaction.

[shanmugasundaram2706]

Dear Jason Floyd,
Thank you for your valuable input on how to correctly use primitive and lumped species. I will check the user guide and try rerunning the cases with the corrections.

Regards,
Shanmugasundaram D

[rmcdermo]

Maybe I missed it in this thread, but please make sure you are using the latest version of FDS. In v6.10.0 we introduced this

A new model for oxidative solid phase reactions has been implemented which enforces consistency between the surface mass flux of oxygen and the rate of oxygen consumption in the solid phase reactions. See Solid Phase >> Pyrolysis Models >> Solid Fuels in the FDS Tech Guide.

The point of this new approach was to deal with surface temperature spikes.

[shanmugasundaram2706]

Dear Randy Mcdermot,
I am using the latest version of FDS, 6.10.1. Thanks for pointing it out. I have also looked at this new oxidative solid phase reaction model in the work 'Numerical simulations of flame spread in pine needle beds using simple thermal decomposition models' by Mueller et al. I believe that correcting the usage of 'primitive' and 'lumped' species will solve this problem. I will get back to you with the updated results.

Thanks!

Regards,
Shanmugasundaram D

[shanmugasundaram2706]

Dear Jason Floyd and Randy Mcdermot,
I tried rerunning the single particle case by correcting the char oxidation reaction and including lumped species, writing it as Char + 7.2 AIR → 8.15 PRODUCTS + 0.05 ASH, rather than Char + 1.65 O₂ → 2.6 CO₂ + 0.05 ASH. While doing so, the magnitude of abrupt second spike was decreased. Before extending this lumped approach to the full case, I simulated a case by increasing N_PARTICLES from 1 to 100. In this case, the abrupt spike was not observed; however, char oxidation happens very slowly at the end, and the overall char burnout time increases. This is evident in the oxidative HRR plot, where HRR asymptotically approaches zero (See Case 3). This observation is also profoundly evident in the full case. Similarly, with N_PARTICLES set to 100 and SURFACE_OXIDATION_MODEL set to T, with the kinetic parameters adopted from the FDS example (A = 465 and E = 68,000), the simulation also shows slow char oxidation at the end.



single_particle.txt
100_particle.txt

Please help me answer the following questions:

1. Why could the overall char oxidation process not be modelled correctly with a greater number of particles? What possible changes do I need to make to the present model to avoid this very slow char oxidation behaviour at the end?
2. Kindly clarify why the second spike was not observed when O₂ was replaced with the lumped species 'AIR' in the char oxidation reaction for N_Particles= 100 and full cases.
   The plots and their associated input files are attached below for your reference.

Thanks,

Regards,
Shanmugasundaram D

[rmcdermo]

I think there are issues with radiation absorption that are affected with multiple particles. @ericvmueller can better answer this. But in the end, this is all research, and at least I cannot just give you an answer without doing the research myself, which I do not have time for. What I can say is that I do not trust the models until they have matched data on TGA, DSC, and MCC, and it does not look like you have compared against all of these.

[ericvmueller]

I can take a closer look at the case when I get a chance (when I return from traveling) and see if I can offer any advice.
However it looks like your 1 and 100 particle cases have different INIT volumes defined. Check you have both the same mass and same volume of cells occupied by particles (after the simulation has initialized, not just based on the input file). Otherwise differences may well occur.

[shanmugasundaram2706]

Dear Randy Mcdermot and Eric Mueller,
Thank you for your quick response. I will correct both cases to have the same mass and volume of cells, and I will let you know if there is any difference in the results. Your input on how to avoid slow char oxidation at the end would be really helpful.

Thanks,

Regards,
Shanmugasundaram D

[shanmugasundaram2706]

Hi Eric Mueller,
I have simulated a single particle case by keeping the individual particle mass same between both 1 particle and 100 particles cases. With the single particle case, the char oxidation HRR goes to zero whereas it is not the case with 100 particles case. The plots and the associated input files are attached below for your kind reference. Please let me know if you require any additional details. Please help to resolve this issue.

single_particle_same_mass.txt
100_particle.txt

Thanks,

Regards,
Shanmugasundaram D

[drjfloyd]

Your char oxidation rate is a function of the available oxygen. In one case you have 0.66 kg/m3 of wood and in the othe case you have 66 kg/m3 of wood. Do you expect that the 66 kg/m3 case is going to have as much oxygen available as the 0.66 kg/m3 case?