Mechanisms of ConcurrentFlow Flame Spread Over Solid Fuels Under NASA Grants NAG31611 and NCC3671 Th

1
Mechanisms of Concurrent-Flow Flame Spread Over
Solid FuelsUnder NASA Grants NAG3-1611 and
NCC3-671(Thanks to CI and Boeing for travel
support)

• Linton K. Honda Paul D. Ronney
• Department of Aerospace
• Mechanical Engineering
• University of Southern California
• Los Angeles, CA 90089-1453 USA
• http//cpl.usc.edu
• Presented at the Twenty Eighth International
  Symposium on Combustion
• Edinburgh, Scotland, July 31 August 4, 2000

2
Motivation

• Flame spread over thin solid fuels - simple
  model for spreading flames (e.g. building fires)
• Well defined steady properties Spread rate (Sf)
  - Indicative of heat release and CO production
• Sf depends on properties of fuel, atmosphere and
  flow
• Opposed flow flame spread - reasonably well
  understood - generally steady
• Concurrent flow flame spread less understood but
  important for modeling fire spread in buildings,
  on walls, ...

3
Motivation (Continued)

• Models (Fernandez-Pello et al.) predict
  inherently unsteady spread due to continually
  growing flame length
• Unlikely that the flame length (L) can grow
  indefinitely due to heat and momentum losses
• Some studies suggest steady concurrent flame
  spread
• Tien (1983) - experiment, thin fuel
• Fernandez-Pello (1984) - experiment, thick fuel
• Ferkul Tien (1994) - computation, thin fuel

4
Hypotheses

• I. For narrow beds, flame length grows until
  boundary-layer thickness (?) width (W) of
  sample, where transverse heat and momentum losses
  will limit flame length (L) and spread rate (Sf)
  (convectively stabilized flames)
• II. For wide fuel beds, radiative losses from
  the fuel bed will limit L and Sf when radiative
  loss heat generation rate because loss L1
  whereas heat transfer to fuel bed Ln (n (radiatively stabilized flames)

5
Hypothesis I - cartoon

• Boundary layer thickness (d)
• sample width
• NO.

Boundary layer thickness sample width YES !!!

6
Model Flame Length (buoyant convection)

• Boundary-layer thickness (?) Nusselt (NuL)
• d LCGrL-c and NuL DGrLd, where GrLgL3/?g2
• (C, c, D, d from established literature values)
• For narrow sample widths, Hypothesis I states ?
  W
• For large W, Hypothesis II states heat generation
  radiative loss from the bed.
• where PlWºlg(Tf-Tv)/Wes(Tv4-T84) (Planck number)

7
Model Flame Spread Rate (thin fuels)
Sf,con is estimated by equating total heat
transfer to fuel bed (Q) NuL?g(Tf-Tv)W to heat
transfer needed to raise the fuel bed to the
vaporization temp. ?s Cp,s ?s (Tv-T8) W
Sf For narrow widths (convectively
stabilized) For wide widths (radiatively
stabilized)
8
Model - Transition Between Regimes

• Two types of transitions
• Convectively ? radiatively stabilized when L
  predicted by 2 stabilization mechanisms are equal
• Laminar ? turbulent at GrL 4 x 108
• (convective)
• 
  
• (radiative)

9
Model - Regimes
10
INTERNAL APPARATUS INTERFACES
Lid
Rack/Vessel Umbilical Cable
O-Ring Seal
Mixing fan
Aluminum Clamps
Window
Thermocouple Rack
Fuel sample
Pressure Vessel
Mixed Gas / Vacuum
Test Rack
Kanthal Wire Ignition system
Vacuum
Oxidant
Partial Pressure Gas
Diluent
Mixing System
Computer
Fuel
11
Test conditions

• Fuel sample Kimwipes tissue paper
• Thickness ?s?s 0.0018 g/cm2, single or double
  thickness
• Atmosphere 21 to 50 O2 with Ar, He, N2, CO2,
  SF6 diluent
• Pressure 0.25 to 3 atm
• Width 0.1 to 8.0 cm
• Length up to 120 cm
• Narrow widths, low pressures lead to extinction -
  use enriched O2 levels to avoid near-extinction
  conditions
• Enables observation of steady spread at GrW from
  300 (0.25 atm O2-He) to 3,000,000,000 (3 atm
  O2-SF6) within available flame spread distance (
  1.2 m)
• Determine Sf,opp from downward spread over very
  wide samples

12
Video
Large GrW
Small GrW
13
Results temperature measurements

• Distance between leading and trailing edge ?
  length
• Correlation with video when 900oC used for flame
  edge

14
Results Flame Spread Rate (air, 1 atm)

• Small width Sf W2.8 - similar to CL scaling
  (Sf W3, Sfcon/Sfopp GrW1)
• Large width Sf W.5 - similar to RT scaling
  (SfW0, Sfcon/SfoppGrW-1/3PlW1)
• Transition at GrW 30,000 - close to prediction

15
Results Flame Spread Rate (dimensionless)
16
Results flame spread rate (dimensionless)

• Low Grw - CL (Grw1) relation fits well for all
  atmospheres

17
Results Flame Spread Rate (dimensionless)
18
Results flame spread rate (dimensionless)

• Low Grw - CL (Grw1) relation fits well for all
  atmospheres
• High Grw
• Bending towards horizontal - transition to
  radiatively-stabilized flames
• Compares well with radiative stabilization
  prediction (W0), asymptotic limit depends on
  atmosphere

19
Results Flame Spread Rate (dimensionless)
20
Results flame spread rate (dimensionless)

• Low Grw - CL (Grw1) relation fits well for all
  atmospheres
• High Grw
• Bending towards horizontal - transition to
  radiatively-stabilized flames
• Compares well with radiative stabilization
  prediction (W0), asymptotic limit depends on
  atmosphere
• Transition Grw between 5,000 and 200,000
  depending on PlW and thus lg - fits prediction
  well

21
Results flame length
22
Results flame length

• Low Grw follows convective stabilzation /
  laminar flow prediction
• High Grw follows radiative stabilization /
  turbulent flow prediction for most atmospheres

23
Relation between L/W Sf,con/Sf,opp

• L ? NuL ? Sf
• Combining predictions for L/W Sf,con/Sf,opp
  yields

• (convective stabilization)
• (radiative stabilization)

 
24
Results relation between L/W Sf,con/Sf,opp
25
Results relation between Sf,con/Sf,opp L/W

• CL Small Grw consistent, though somewhat high

26
Results relation between L/W Sf,con/Sf,opp
27
Results relation between Sf,con/Sf,opp L/W

• CL Small Grw consistent, though somewhat high
• CT never close

28
Results relation between L/W Sf,con/Sf,opp
29
Results relation between Sf,con/Sf,opp L/W

• CL Small Grw consistent, though somewhat high
• CT never close
• RT
• Small Grw - similar to CL but Sfcon/Sfopp
  predictions dont match
• Large Grw good agreement

30
Results relation between Sf,con/Sf,opp L/W

• CL Small Grw consistent, though somewhat high
• CT never close
• RT
• Small Grw - similar to CL but Sfcon/Sfopp
  predictions dont match
• Large Grw good agreement
• Moral only CL at low GrW, RT at high GrW are
  consistent with BOTH measured Sf and L data

31
Regimes of upward flame spread
32
Conclusions

• Concurrent-flow flame spread is not inherently
  unsteady accelerating - transverse
  conductive/convective losses and surface
  radiative losses limit spread rate over
  sufficiently long samples
• Predictions of simple model are generally
  consistent with experimental results for upward
  flame spread over thermally-thin fuels over
  107-fold range of GrW.
• Future work
• Forced flow
• Thermally-thick fuels

33
Backup 1 Flame Length
forced-convection flame spread ?LAReL-a and
NuLBReLb buoyant-convection dominated spread
?LCGrL-c and NuLDGrLd, where GrL?gL3/?g2
Hypothesis I states ?W (forced
convection) (buoyant convection)
Hypothesis II states that Q(NuL?g/L)(Tf-Tv)
equals the radiative loss from the bed
(H)??(Tv4_T84) (forced convection)
(buoyant convection)
34
Backup 2 Flame Spread Rate (thin fuels)
Sf,con is estimated by equating Q to the rate of
fuel bed enthalpy increase Substitution for
Each Type Yields - Sf/Sf,opp
35
Backup 3 Transition Between Regimes
Convectively-Stabilized Flames (buoyant)
(forced) Radiatively-Stabilized
Flames (buoyant) (forced) Transition
Occurs when Lengths are Equal (buoyant)
(forced)