Title: Quantitative methods in fire safety engineering 5 U01620 9. Introduction to combustion modelling
1Quantitative methods in fire safety engineering 5
(U01620) 9. Introduction to combustion
modelling
- Stephen Welch
- S.Welch_at_ed.ac.uk
- School of Engineering and Electronics
- University of Edinburgh
2Introduction to combustion Scope
- Combustion processes
- Premixed
- Diffusion flames
- Combustion chemistry
- Equilibrium
- Finite rate
- Combustion models
- Mixing controlled
- pdf methods
3Introduction to combustion Scope
- Toxic minor species
- Carbon monoxide
- Smoke
- Flame spread
- Solid-phase pyrolysis
4Example application
5Combustion processes
- Premixed flames
- Deflagration
- Detonation
- Diffusion flames
- Laminar
- Turbulent
- Typical of natural fire
- Fairly long time scales
- Buoyancy dominated
6Turbulent diffusion flames
- Combustion chemistry/heat release
- Essentially mixing controlled
- Chemical kinetics can be assumed fast
- Useful simplifying assumption
- Species yields (e.g. toxic emissions)
- Products of incomplete combustion
- Non-equilibrium/slow chemistry important
- High heat loss
- Thermal radiation from carbon particles
- May influence flame spread
7Combustion chemistry
- Equilibrium
- Thermodynamic equilibrium
- Unphysically high yields of intermediates
- Finite-rate chemistry
- Typically described by Arrhenius expressions
- EA is the activation energy (typically very
high!) - Even for simple fuels can be hundreds of
reactions and dozens of intermediate species
8Methane mechanism
- Detailed scheme
- 149 elementary steps
- 144 reverse reactions
- 33 species
9Combustion modelling
- Want to describe
- Energy release
- Species yields, especially toxic products
- Distinguish controlling mechanisms
- Mixing control
- Chemical reaction rate
- Damkohler number
- Basis for simplified treatments
- Dagtgt1 (fast chemistry)
10Combustion modelling
- Transport equation for species
- Reynolds averaged form
- Chemical source term
- Highly non-linear in instantaneous gas
temperature - Problem of closure
11Chemical source term closure
- Consider simplified chemical reaction
- Fuel Oxidant Products
- k is rate constant for reaction
- Source term for fuel consumption
- If expanded get unknown cross-products
12Favre averaging
- Density-weighted time averaging
- simplifies mathematical description
- Reynolds average
- Favre average
- where
13Reynolds averaging
- Time averaging
- Reynolds (1895)
14Favre averaging
- Continuity equation
- Reynolds average
- Favre average
15Chemical source term closure
- Expand mean chemical source term
-
16Chemical source term closure
- Rate term can also be expanded
- cross correlations are unknown and highly
non-linear! -
17Does it matter?
- Consider
- temperature fluctuating between 500K and 2000K
(e.g. reactants and products) - Mean temperature 1250K
- Typical activation energy/R20,000K A1
- Rate from mean T
- Rate from exact T
- Ratio 0.5!
18Does it matter?
- Using mean temperatures
- Significant underestimate of reaction rates
- If we have to model finite-rate chemistry
- Need a method of accommodating influence of
temperature fluctuations - Solve additional transport equations for each
species of interest - Soon becomes intractable!
19Mixing controlled combustion
- If we cant model reaction rate, just assume its
infinitely fast - Constrained by microscopic mixing rate
- Need a model to describe mixing
- Eddy breakup (EBU version 1)
- Spalding (1971)
-
- k, ? turbulent kinetic energy and dissipation
- Yf fluctuation of fuel mass fraction
- CEBU empirical constant
20Mixing controlled combustion
- Eddy breakup (EBU version 2)
- Magnussen Hjertager (1976)
- For turbulent diffusion flame
- Assuming isotropic turbulence
- Usually taken to equal 4
21Eddy breakup models
- Reasonably successful across a range of
combustion systems - Relatively simple formulation
- Provides for distributed heat release
- e.g. flame lengthening under ventilation control
arises automatically - Tracks composition evolution in terms of
- Oxidant, reactant and products
- BUT, detailed chemistry neglected
- No good if require other minor species
- Toxic products of combustion, CO and smoke
22PDF methods
- Chemical source term closure requires
- Knowledge of all cross-correlations
- Can be solved using Monte Carlo method
- Introduce large number of fluid particles
- Track progress of each in time
- Adjust particle composition as it interacts
- Hence
- Obtain the joint-pdf numerically
- Hugely demanding computationally!
23Mixture fraction
- A variable can be defined as
- where, ?
- Assuming one-step chemistry
- the rate of change of each term is equal and
opposite - ? is not affected by chemical change
24Mixture fraction
- Call ? a conserved scalar
- Measure of the local mass which originated in
fuel stream - Value varies between 0 and 1
- Cannot be created or destroyed
- Affected by other mixing processes
- Convection and diffusion
- Can track in space by solving an additional
transport equation for ? - If we can relate other chemistry to this
parameter alone there is no need to close
chemical source term!
25Fast chemistry
26Laminar flamelet models
- Imagine the mixture consists of an ensemble of
laminar flamelets - Chemistry of each is defined for full range of
mixture fraction values found in a laminar
diffusion flame - Obtain instantaneous species concentrations
- Y(?) defined by the state relationship
- Need only model the evolution of P(?)
27Laminar flamelet models
- How can we determine P(?)!?
- Solve additional transport equations
- Mean
- Variance
- Ties in effect of turbulence
28Laminar flamelet models
- Prescribed pdfs, of fixed general shape
- Beta function
- Gaussian
- Shape evolves with local conditions
- Expressed in terms of computed mean and variance
of mixture fraction (beta function)
29Laminar flamelet models
- Flamelet chemistry
- Equilibrium (gross overprediction)
- Experimental measurements
- Opposed diffusion flame modelling
- Chemical kinetics limited only by knowledge of
mechanisms and computer power
30Laminar flamelet - example
- Compared mechanisms
- Simple
- 41 species
- 274 reactions
- Held et al.
- Complex
- 160 species
- 1540 reactions
- Seiser et al.
31Toxic minor species
- Represent detailed chemistry in flamelet
- Obtain by quadrature
- OK provided chemistry is sufficiently fast
- Probably not true for CO
- Definitely not true for smoke
32Carbon Monoxide prediction
- Yield is a balance between
- Rate of formation (relatively slow)
- Turbulent transport
- Oxidation
- Flamelet methods robust for many cases
- May be necessary to parameterise to include
- Heat loss
- Strain rate
- Degree of vitiation
- Requires huge libraries of pre-computed
flamelets! - Only works for pure fuels
33Smoke prediction
- Fast chemistry assumption invalid
- Construct flamelet descriptions of the rate of
formation - Solve additional balance equations for soot
particle number density and soot mass - Flamelet rates provide source terms
- Consider oxidation where necessary
- May be a mixing-controlled rate
34Smoke prediction
- Heat loss very important
- Multiple radiative loss libraries
- Looks up appropriate flamelet according to
overall energy balance - Function of local soot concentration
- Coupled problem
35Flame spread modelling
- Ignition criterion
- Surface temperature
- Requires very accurate modelling of solid phase
heat transfer - Critical accumulated flux
- Requires some approximations for unknown heat
losses from rear of specimen - Cone calorimeter data
- Take HRR direct from cone test
- Will only apply to that flux
- Further complicated by radiation from flames
36Critical accumulated flux
- Defined as
- qmin is a minimum heat flux below which no
ignition occurs
37Example application
- Flame spread over full-scale corner wall
- 10 different materials
- Cellulosics
- Plastics
- With/without fire retardant
38Heat release rate
39CO production rate
40Smoke production rate
41Sensitivity
42Sensitivity
43Sensitivity
44Sensitivity
45(No Transcript)
46Summary (1)
- Eddy breakup
- Mixing controlled
- Laminar flamelet
- Relaxes fast chemistry assumption
- Overcomes problem of turbulent closure of
chemical source terms - Slow chemistry processes, e.g. for toxic products
like CO and smoke, require solution of additional
transport equations
47Summary (2)
- Flame spread predictions require
- Accurate smoke predictions
- Radiative feedback to surface controls
volatilisation rate - Need a comprehensive model which links all
phenomena - This is the strength of CFD over simpler models
which are essentially more empirical
48References
- Chung, T.J. Computational Fluid Dynamics,
Cambridge University Press, 2002 - Cox, G Combustion Fundamentals of Fire,
Academic Press, 1995 - Moss, J.B. Turbulent diffusion flames, chapter
4 in above, 1995 - Wilcox, D.C. Turbulence modelling for CFD, DCW
Industries, 1998 - McGrattan, K. (ed.) Fire Dynamics Simulator
(Version 4) Technical Reference Manual, NIST
special publication 1018, 2004