CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Les and diesel (http://www.cfd-online.com/Forums/openfoam-solving/57962-les-diesel.html)

wizard1301 February 12, 2009 09:55

I would like to use a LES appr
 
I would like to use a LES approach instead of the RANS one to simulate the diesel spray


is there someone who knows where I could start?

ville February 12, 2009 11:03

You could start from the Xoodl
 
You could start from the Xoodles solver and add the dieselSpray stuff from that RANS solver to Xoodles.
All you need to do in principle is:

1) add relevant include files to Xoodles (that's basically the spray class)
2) add spray momentum source to UEqn.H
3) add the lines where you create the dieselSpray
4) take the input files from the dieselSpray (is that the name of the RANS solver!!? Don't remember,
something like that) tutorial where the inputfiles injectorProperties and sprayProperties are
5) I think you need to change the thermodynamic approach to hcombustionThermo (instead of hhu) for some reason.. BTW, COULD SOMEONE TELL WHY THIS NEEDED to be done?? Perhaps it was only an instruction I got some years ago since I didn't want to study combustion..
6) I think if you can figure out those things you
are nearly done - it shouldn't take long since it is mostly copy-pasting to Xoodles from an existing solver ..

7) And remember to modify the files in Make : take example from the dieselSpray tutorials.

8) the aachenbomb case gives you the input files.

Good luck!

-Ville

ville February 12, 2009 11:08

Hi, By the way, I have a ques
 
Hi,
By the way, I have a question about the thermodynamic approach in the main top level solver and then e.g. spray class: could someone
please explain about the thermodynamic approach the spray class expects? So does the spray class expect the actual solver to use the hcombustionthermo approach or is the class 'blind' to the thermodynamic approach otherwise than the information that goes to the constructor?

Best regards,
Ville

niklas February 13, 2009 04:14

I guess Im the one to answer t
 
I guess Im the one to answer this...

but your question is abit tricky and I think what you really is after is something else.

the spray class depends on that the gaseous properties are the same as those defined
in sprayThermoTypes.H.

The liquids have real multicomponent treatment, this also have some implications on the gaseous treatment. It needs to be a multicomponent mixture.
It should be pretty easy to use these to sum it up and collect them in a fuel source term though.
but you will still need the individual component concentrations to evaluate the evaporation rate.

There are a few places where it also assumes concentrations to be the same in the entire cell, so if you want to split the cells into burnt/unburned fractions, you need to rewrite some of the code. Mainly how it estimates evaporation times and saturation, so its nothing serious.

so I would say there is a slight thermo-dependency.

ville February 13, 2009 04:50

Thank you very much, Would
 
Thank you very much,
Would it please be possible to still clarify with a few words a little bit more on my confusion - especially question 2) ..

1) a thermodynamic approach is chosen in the top-level code to define the gaseous phase thermodynamics -> I would now understand that depending on what kind of a spray I'm simulating, the thermo-dependency comes to more/less into play. Thus, if I just study a non-evaporating cloud, 'solid-particle' cold-injection type situation with no reactions etc, then the spray class will only need the knowledge of gas density and velocity which go to the constructor (I guess that's all the class needs to know about the flow field to update parcel positions).

2) I still don't quite get why the original hhuCombustionThermo from Xoodles needs to be changed to hCombustionThermo if Xoodles is modified to be a spray solver? Is there some restriction on using
e.g. the basicThermo approach (as in coodles) to
create a cold spray injection LES solver starting
from the coodles?

Best regards,
Ville

niklas February 13, 2009 06:05

1. yup. 2. hhu means that t
 
1. yup.

2. hhu means that the thermodynamics is split into 2 fractions of burned and unburned gas, you also have 2 temperatures. There is no implementation of multicomponent mixtures for hhu.

I dont know what you mean with restriction, but if you look in the
src/thermophysicalModels/combustion/hCombustionThermo/hCombustionThermos.C
you will see all the available combinations and it should be possible, in theory, to modify the speciesProperties typedef in sprayThermoTypes.H to the type of thermo you would like to use.

ville February 13, 2009 06:59

Right, think I got it, thanks
 
Right, think I got it, thanks once again!
-V

flying June 23, 2009 10:01

I am also interested in the using LES approach instead of the RANS one to simulate the diesel spray.

But it is very sad that I am not sure about how to solve "Y" because I don't know the equation when I choose LES. I used the method as below to solve the Y, but I am not sure about it. could anyone make sure my method?

solve
(
fvm::ddt(rho, Yi)
+ mvConvection->fvmDiv(phi, Yi)
- fvm::laplacian(turbulence->muEff(), Yi)
==
dieselSpray.evaporationSource(i),
// + kappa*chemistry.RR(i),
mesh.solver("Yi")
);

In fact, There is no combustion in my case, so I didn't care about the chemistry and I delete "kappa*chemistry.RR(i)".If I care about the Chemistry, how do treat the term of "kappa*chemistry.RR(i)", which is generated in the RAS?

flying June 23, 2009 10:01

Thanks and best wishes!

az1362f July 9, 2009 10:41

Hello Dear Ville;

I have some question about LES Spray simulation,please describe these points by more details:

first you said :"2) add spray momentum source to UEqn.H"

1-where is UEqn.h file? because there is no UEqn.h file in Xoodles directory. is it in Xifoam solver folder?

second you said:"4) take the input files from the dieselSpray (is that the name of the RANS solver!!? Don't remember,
something like that) tutorial where the inputfiles injectorProperties and sprayProperties are"

2-what do you mean? do you mean just copy and paste aachenbomb case in Xoodles tutorial folder?

third you said:"5) I think you need to change the thermodynamic approach to hcombustionThermo (instead of hhu) for some reason.. BTW, COULD SOMEONE TELL WHY THIS NEEDED to be done?? Perhaps it was only an instruction I got some years ago since I didn't want to study combustion.."

how can I change hhucobustionthermo to hcombustionthermo?
you mean simply change it in thermophysicalproperties file of acchenbomb case?
does it need to use chemkin folder in aachenbomb case for Xoodles solver?

fourth you said:"3) add the lines where you create the dieselSpray"


you mean just add these lines to Xoodles solver:

Info << "Evolving Spray" << endl;

dieselSpray.evolve();

or something else.
I try to make this solver but i face more problem.
Please help me.

Thanks
regard

sandy July 23, 2009 16:46

How to set BC about LES??

flying July 25, 2009 06:06

There is example in tutorial!

flying July 25, 2009 06:07

Hello Dear Ville and az1362f:

Do you care about the question below?


I am also interested in the using LES approach instead of the RANS one to simulate the diesel spray.

But it is very sad that I am not sure about how to solve "Y" because I don't know the equation when I choose LES. I used the method as below to solve the Y, but I am not sure about it. could anyone make sure my method?

solve
(
fvm::ddt(rho, Yi)
+ mvConvection->fvmDiv(phi, Yi)
- fvm::laplacian(turbulence->muEff(), Yi)
==
dieselSpray.evaporationSource(i),
// + kappa*chemistry.RR(i),
mesh.solver("Yi")
);

In fact, There is no combustion in my case, so I didn't care about the chemistry and I delete "kappa*chemistry.RR(i)".If I care about the Chemistry, how do treat the term of "kappa*chemistry.RR(i)", which is generated in the RAS?

Thanks and best wishes!

Xinguang Cui

sandy July 31, 2009 09:13

Hi Cui, do you know how to specify the inlet value of k to oneEqEddy model?

flying July 31, 2009 11:43

I am really sorry. I don't know how to give it.


All times are GMT -4. The time now is 19:59.