Hello Amir Okhovat ,

your conversion from mole fractions to ppm seems to be right on a first glance. I am not an expert in combustion modelling, but I know that good prediction of NOx results, in example often strongly goes in hand with the used reactions mechanism and there are special mechanisms for better predictions of NOx values.

Maybe you should reformulate your question and give a bit more details about your simulation (i.e. Turbulence Modell, Burner Type, Reaction Mechanism etc.) and how much deviation there is from experiment to get more detailed answers from experts on this area.

Best regards

Christian

Hey Amir Okhovat,

Christian Schubert is totally right with the calculations.

NOx chemistry is really complicated and is strongly dependent of the actual process you are looking at. It is not easy to get quantitatively good results even if you know your process really well

Results are strongly dependent on the reaction mechanisms, the turbulence-chemistry interaction and the choice of your boundary conditions. More information will is needed if you want help.

For detailed NOx-mechanisms you could have a look at the contribution of Glarborg et al.: Article Modeling nitrogen chemistry in combustion

He also gives a brief overview about important process parameters affecting NOx emissions and reaction mechanisms for NOx.

Best regards

Nico

Hello Dear Christian Schubert , Nico Schmitz

Thank you to reply my question. I'm simulating a real geometry of a non premix burner to predict its nox emission. a figure attache to this comment show the geometry and its boundary conditions. i used non-premixed combustion model in Ansys Fluent. in the front of burner is placed a test flame tube. the capacity of burner is 11MW and diameter of test falme tube is 1.2 m and its length is about 7m. stack diameter is about 80 cm. I calculate area weighted average of NO mole fraction and the value is multiple 10e6 to convert to ppm.

I use realizable k-epsilon turbulent model, in wall y+ is less than 350, radiation is DO, to moding combustion in species module i use non-premix combustion model, turn on thermal nox, fuel is Natural Gas (0.884 CH4, 0.05 C2H6, 0.012 C3H8, 0.005 C4H10, 0.045 N2, 0.005 CO2) air mass flow rate is about 4 kg/s and gas mass flow rate is about 0.2 kg/s.

okay, so you are using non-premixed combustion for turbulence-chemistry interaction. By choosing this model, you also assume that the chemical reaction time scale is much faster than the mixing time scale (Damköhler-Number).

What reaction mechanism do you use? Does it imply NO-chemistry or are you using the internal methane-air-2step mechanism with thermal-NO in post-processing?

Nico Schmitz

I'm using Non premixed combustion model. methane-air 2 step used in species transport module. I use PDF tables to calculate fractions.

But you are free to choose a different thermodynamic database. This could change things.

In our experience we try to avoid using non-premixed combustion approaches for NO prediction and also post-processing for NO. We choose mechanisms with NO chemistry and use the species transport model in combination with eddy dissipation concept or other turbulence chemistry interaction schemes for staged or flameless combustion. But this is highly dependent on geometries, flame structures and combustion regimes.

Keep in mind that steady state RANS calculations are a big simplification of the real problem we are looking at (highly turbulent, unsteady reaction flow) and may never give a perfect prediction.

It is hard to estimate what is best for your burner, not knowing it in detail. I would recommend you to try different approaches and a brief comparison.

Wowww

So with non premix combustion model there is not correct prediction for NO and i must use eddy disipation concept model. My geometry grided with 15000000 tetrahedral mesh that convert to 3500000 polyhedral mesh.

Can i message you in whatsapp messenger?

Possibly you will get good predictions with other models, but keep in mind that a quantitative estimation of the NO emissions is not an easy task. You will have to try out different models. A brief comparison of your burner with those in literature could also help a lot.

Calculations with big reaction mechanisms will last at least weaks, but this is reasonable for such complex problems.

Nico Schmitz

thank you to answer my questions clearly

I have a question

In air methane 2 step can i add some component to my fuel. natural gas analysis show that NG consist of 0.883 CH4, 0.05 C2h4, 0.012 C3H8 and etc as i said in previous messages.

How can i add this component to fluent library??

Amir Okhovat,

you cannot add these components to the methane 2 step mechanism. It is only made for CH4 and has no reactions for other fuel components.

Try to use an available reaction mechanism like DRM-19 (http://combustion.berkeley.edu/drm/) which implies other fuel components. Simplification of your NG (to CH4, C2H6, N2) composition could also be a way to avoid big reaction mechanisms.

Hello Amir Okhovat ,

Something to have in mind for your conversion from mole fractions to ppm is that experimental NOx measurements are commonly reported on a dry basis.

If that is your case, you must use the following expression:

ppmNOx(dry) = n_NOx / (1 - n_H2O) *1e6 , where n is the mole fraction

Best regards,

Alex

Alex M. García

thank you. I'm trying to run with GRI-MECH30 reactions with eddy dissipation concept to calculate NO.

I calculated both dry and not dry formoula but i have error about 200%.....

If you're using non-premixed combustion models in FLUENT (I guess you chose chemical equilibrium), then it doesn't make sense to include NOx species in the tabulation procedure, as your calculated NO concentrations will be off the chart. If you want to determine NO emissions with this kind of model, you need to use the NO post-processing to obtain reasonable numbers.

You can also, of course, use a rather detailed mechanism like GRI 3 in combination with EDC, but this is going to be quite expensive computationally.

Jörg Leicher thank you to answer my question

I used Non premix combustion model to determine flow and temperature field and in post processing I turned on NOx in Ansys Fluent to calculate NOx emission in my system but in LPG I have an error about 300% (I calculated NOx in the stack outlet about 28ppm but in experiment it's about 100 ppm).

I used EDC in Ansys Fluent but reactions not occurred and temperature field didn't raise. I considered a wall with a temperature of 2000K as ignitor.