7th International Workshop on Premixed Turbulent Flames
Selwyn College, University
of Cambridge
Cambridge, England
July 28, 2000
Questionnaire Summary
This document summerizes
the inputs from a questionnaire sent to the participants of the 7th Workshop
prior to the meeting.
Each respondent's answers
are listed as separate paragraphs under the questions.
Number of respondents declairing their research emphasis as
|
|
|
|
|
|
|
|
|
|
|
|
|||
|
|
|
|
|
|
|
|
|
|
|
|||
|
|
|
I workd with co-flowing bunsen-type premixed turbulent burner. I
also worked with bunsen-type tunrbulent burner with perfolated plate installed
in a high-pressure bomb. Recently main emphasis is on stretch flames in
microgravity.
developed turbulence in turbulent V-flame, turbulent bluff-body flame turbulent Bunsen flame and turbulent swirl-stabilized flame
planar turbulent bunsen flame (fully developed turbulence) and lifted turbulent jet flames (fully developed turbulence)
Turbulent Bunsen flames and Stagnation-plate flames. They are fully-developed turbulent flows. More research regarding turbulent shear flows are in plan.
1) DNS of statistically planar premixed turbulent flame involving fully developed isotropic turbulence 2) Vortex/premixed flame interaction: A vortex pair colliding with a premixed flame 3) Vortex/premixed flame interaction: A vortex tube perpendicularlly inteacting with a premixed flame (flame propagation along a vortex tube)
DNS: very simple configurations, spherical kernel and planar flame in turbulence
Modeling: Complex geometry eg real industrial gas combustors
low-swirl burner, v-flames and large Bunsen flames with fully developed turbulence of 0.5 < u’ < 3 m/s.
Laboratory V-flames and SI, research engine.
Bunsen flame with fully developed turbulent flame
Calrification of hydrodynamic instability effects on turbulent flame
front configurations at high-pressure environment.
Of course our own experiments on detailed turbulent flame structure in high-intensive flames (Dinkelacker et al. 27. Symposium). Additionally the corresponding experiments of Tobias Plessing at the nice weak swirl flame of Robert Cheng (28. Symposium).
identify the physical mechanisms and suitable parameters that affect flame propagation in turbulent flows of different configurations. We believe there are other factors than turbulence intensity which are at least as important at least in some particular boundary and initial conditions.
LIF and DNS studies on flame structure of high intensity turbulent premixed flame.
Progress has been disappointing
The estabilishment of a thin-reaction zone regime to replace the "distributed reaction zone" regime. The important implcation is that thin flame is a useful model for combsution processes in most practical systems.
Recognition that reaction is confined to thin sheets even at rather
high turbulence intensities and therefore that thin reaction zones – flamelets
– are important in practical systems. Second most important is characterization
of flamelet instabilities and responses to flow induced perturbations.
This latter work is not complete.
Preheating zone broadening due to flow turbulence
What kind of structure and characteistics have the high intensity
turbulent premixed flames in both regions of the thickened flame (distributed
reaction zone) and the thickened wrinkled flame?
So many … Does the turbulent burning velocity exist? How does the flame "process" turbulence? What is the role of laminar instability?
There are many oustanding questions. We know a lot about stretch laminar flames but are they really the right model for turbulent premixed flames? Do the flame fronts have enough time to respond to local stretch and curvature effects cause by high frequency fluctuations? When does turbulence overwhelm flamr front instabiltiy?
Development of understanding of flamelet dynamics. This is needed to improve models for turbulent fluxes, ? and I0 and for length scales (such as the reciprocal of the flamelet crossing density).
What determines the wrinkling scale, total flame area, and flame surface
density for various conditions such as wide pressure range? It is an old
question but essential to predict the global burning rate.
No. But it was. I am working on a local reaction zone structure
of weakly strained flame in microgravityunder project of so-called High
Temperature Air combustion Technology. When highly preheated air is used
for, local reaction zone structure and flammability limit of flames are
totally different from conventional premixed nor non-premixed combustion.
Even at low combustion rate, flame could survive in that combustion mode.
This may be related to turbulent flame combustion of, say, SI engines.
Local heat release rate and mean heat release rate, How to model and calculate this effectively.
The work mainly consists of simulating turbulent premixed (or partially premixed) combustion using the G-mean, G-variance and sigma (flame surface area ratio), where the turbulent burning velocity is determined from $(\bar \rho s_T^0 ) |\nabla \tilde G| = (\rho s_L^0) \bar \sigma.$ Both the corrugated flamelet regime and the thin reaction zones regime are covered.
Not the major target, but is relavent. We have just measured quantites such as local flame front propagation velocity and flame surface density in some turbulent Bunsen flames.
Yes it is. I integrate the local reaction rates along the normal line perpendicular to the flame to obtain local burning rate. In addition, I integrate the local burning rate along the flame surface and to obtain global burning rate and flame length (or area) which result in mean local burning rate.
Yes! We have investigated all of these using DNS
Yes! We have investigated the relationship bewteen turbulent burning speed, flame surface density, burning rate integral, and consumption speed. By the use of OH-PLIF and LDV on a low-swirl burner, we are able to demonstrated that flowfield divergence is just as important as the turbulence intensity in determining the combustion intensity.
Yes. Burning rate integral, mass burning rate in engines, mass flux measurements (for burning rate), and flamelet surface normal and flamelet crossing density measurements via crossed-plane tomography (for ?). High-resolution instantaneous temperature gradient measurements via crossed-plane Rayleigh imaging are beginning.
Yes. Turbulent burning speed (velocity, the global rate) is a parameter
that I have been investigating lately. I am interested in the pressure
dependence of turbulent burning velocity. I use turbulent Bunsen flame
stabilized in a high-pressure chamber. Laser tomography and OH-PLIF and
their image processing are employed determine the turbulent burning velocity
based on the mean flame-cone angle method. I am also trying to clarify
the smallest scale of flame wrinkles and relations with pressure and the
turbulent burning velocity.
Although it has been an open issue for long time, I would like to
raise a reaction zone broadening due to turbulence. Flame is quenched or
broaden, Zeldovich number may have a hint on this.
All these factors are important. I don’t think there are answers for these questions in general consensus. And i don’t have one either. Certainly, upstream conditions are important, particularly the flow field.
When the turbulence is strong and the Kolmogorov scale is smaller than the flame thickness (thickend wrinkled flame), flame is highly wrincked and local flame interaction becomes important. In this situation, what kind of mechanism occupies heat and mass transport and reaction rates.
We need much more information on instability meahcnisms (thermo-diffusion, Derrius-Landau etc) and how they coupled to the incoming turbulence
Experimentally, we do not have an accurate method to measure the local reaction rates in fully developed turbulent flames. Measuring flame front broadening is even more problematic because of the ambiguity introduced by the "tilt" effects. Quantifying the effects due to flame front instability in a fully developed turbulent flame is another difficult problem. We are also lacking information on how a laminar flames response to fludcuations of different discrete frequencies.
Almost all issues are open. Available burning velocity correlations suggest that u’/SL is an important parameter and I would agree. However, the burning velocity is an ill-defined quantity, and our measurements of burning rate integral suggest that available correlations of burning velocity over predict combustion rates. Therefore these correlations must be viewed with skepticism. Our measurements in V-flames and engines show that the flamelet surface normal distribution and ? are not solely dependent on u’/SL. It is my opinion that they depend on the flame configuration and perhaps other flow parameters as well as u’/SL.
Recent studies of laminar flame response to perturbations such as strain and flame-vortex interactions and studies of stability reveal a variety of interesting flamelet phenomena. This information has not been integrated fully into theories and models of turbulent flames and the phenomena’s significance to turbulent flames is not fully known. For example, there is conflicting evidence regarding the importance of diffusive instabilities in turbulent flames. It seems likely to me that the importance of these phenomena will very from flame to flame depending on the values of important parameters. We don’t know currently what these parameters are.
Effects of transient flame stretch and curvature. Time variation of
local burning velocity of the flame under development of flame instability
especially of hydrodynamic is related to this matter. In this case, certain
time is needed to develop the flame instability fully, so that the stage
of the development must be a parameter.
All. Also flame instability and its scale may should be added.
The flame displacement speed is important. I don’t think there are answers for these questions in general consensus. And i don’t have one either.
For wrinkled flames, these parameters are identical or relatively easy to relate with each other. But for thickened wrinkled flames and thick flames, I do not know how I relate them. We need information of flame structute and velocity fluctuation in the flame zone as well as the local diffuion rates and reaction rates for obtaining new parameters.
We need high-quality visualization of the flame/tubulence interactions high-quality measurement (quantitative)
For wrinkle flames, we need to merge the point measurements (i.e. crossing frequencies, flame crossing angles, and orientations) with a global parameter such as the turbulent burning velocitye. These point mesurements
Of the parameters mentioned, we have measured crossing frequency, crossing angle, flamelet normal vector (in 3 dimensions) and burning rate integral. Furthermore we have used our data for flamelet normals and crossing density to find ?; these quantities are related. I believe that quantities such as burning velocity, consumption speed and displacement speed are ill-defined such that values for these quantities in one configuration cannot be meaningfully compared to values measured in another flame configuration. We, as a research community, need to discuss our definitions for these quantities and their physical significance in order to reach agreement on their meaning and uses.
It depends on the problem. When the flame stability on the burner is
discussed, the global flame velocity is should be known. When the total
heat-release performance in the chamber is discussed, the consumption speed
is parameter. When the local flame structure is modeled, the crossing frequency
and flame orientation etc. are important. The relation of these parameter
is, of course, just we should investigate. It seems to me that the configuration
dependence is mainly the stage of development of flame instability and
the turbulent flame front itself.
Local flamelet structure.
Laser imaging diagnostics is the method i am using.
For wrinkled flames, I integrate the local reaction rates along the normal line perpendicular to the flame to obtain local burning rate. In addition, I integrate the local burning rate along the flame and to obtain global burning rate and flame length (or area) which result in mean local burning rate. In some cases, I simply integrate the local reaction rates in volume to obtan the turbulent burning rate. For thickend wrinkled flames and thick flames, it is very difficult to evaluate local burning rate and flame length (or area).
Flamelets! BML – S equation (in LES)
Years ago we have measured local reaction rates (crossing frequencies) through the flame brush and integrated the results to infer a turbulent rate burning. We have progressed to integrating flame surface density at different C values and obtained consistent results when comparing the scalar results with turbulent burning velocity. We found that the knowledge on the flamr flowfield is very imporant to give real physical meaning to the locat reaction rate.
We want a measure of the mean combustion rate (product formation rate) "per unit area of the turbulent flame brush" and have argued for several years that the quantity to measure is the burning rate integral, BT. We have used mass flux data to determine BT. We can now measure ? as a function of position across the turbulent flame brush with cross-plane tomography. We are now working on estimating I0 from instantaneous temperature gradient data to be obtained by crossed-plane Rayleigh imaging. With I0 and ? measured we have the mean local product formation rates and by line integration we find BT.
Precise observations of the development of flame
instability related to the local reaction rate and the integration to the
global burning velocity are planed.
Yes
They are linked as far as the transport equation tells. How close is the critical question.
Yes. Analyzing the production rate of turbulent scalar flux from our DNS database, the velocity-reaction rate correlation term takes a part in production of the scalar flux.
Yes! Yes! Yes!
They are linked for certain class of problems. I belive that under certain conditions (e.g. very thin flamelets), they may be considered separately.
Yes, they are linked by common physical processes. Flamelet wrinkling (flamelet dynamics) increases ? and introduces intermittency that is the major contributor to turbulent fluctuations and fluxes.
I have no idea.
Not much. just trying to gather as much information as possible.
The change of local reaction rate due to stretch is more important than the increase of flame area in the case of vortex/flame interaction.
Definition of the pdfs for DNS, elucidation of tansport mechanism
We are quite convinced that the flamelet model extends beyond the Klimov-Williams criterion. Another observation is that flame configuration (i.e. transient or stationary, normal or incline) contribute to the scatter in the turbulent burning velocity data.
Burning velocity data in the literature over estimate mean rates of product formation. This is clear when such data are compared to data for BT.
Flamelet surface normal data obtained in engines and V-flames from crossed plane imaging measurements can be fit to the same probability density (PDF) form; this form has one fitting parameter. A very significant consequence is that ? = Anc. nc is the flamelet crossing density along a line normal to <c> constant surfaces. A is a function of the fit parameter and varies nearly monotonically from 1 (the laminar flame limiting value) to 2. (Note: Corrsin showed in the fifties that ?=2n for an isotropic distribution of surface normals.) nc is easily measured!
To reasonable approximation nc = Nc<c>(1-<c>)/l. l is the flame brush thickness, and Nc is the line integral of nc across the flame brush. With this approximation, ? is a known function of <c>, the fit parameter, and the length scale l/Nc.
I have not yet investigated the local reaction
rate in my research precisely. Probably, it will be essential in order
to validate our global burning-velocity correlation in terms of local reaction
rate variations under the flame instability at high-pressure.
Yes, we will. To model combustion of SI engine,
initial condition of mixture such as temperature, composition, non-homogenity
of mixture employed by fundamental researches are limited.
Yes, although the so far studied Bunsenflame might not be a perfect example for a practical problem, the used model definitely allows the application to practical problems.
Surely, this is our goal and i believe we are approaching, yet not making it happen now...
No. The model does not contain the effects of local flame interaction as well as the corrrelation with the reaction rate. In addition, effects of high pressure, intrinsic instabilities and inhomogeneous mixture.
Yes SI Engines, GT engines, off-shore explosions.
May be. It depends on the type of problems.
Yes. With our crossed-plane imaging data, we have estimated mass burning rates in a SI engine and compared the results with burning rates estimated from pressure trace data. We assumed I0 was one and neglected the effects of any cycle-to-cycle variations. Our estimates are high by a factor of approximately three. This difference we attribute to the assumption that I0 = 1 and to the neglect of cycle-to-cycle variations. We are investigating ways to account for the latter effect.
It seems 'No' because the precise mechanism to determine the local reaction rate and global burning rate is still unknown. I think that the lack of the knowledge about pressure effects in terms of chemical reaction, flame-vortex interaction, flame instability etc. is the problem especially for SI engine and gas turbine combustor models.