CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Re: Detonation in hydrogen-air mixtures. (http://www.cfd-online.com/Forums/fluent/76833-re-detonation-hydrogen-air-mixtures.html)

vikram June 6, 2010 15:28

Re: Detonation in hydrogen-air mixtures.
 
Hello all,

I was wondering if anybodycould help me with the simulation of detonation in a hydrogen-fueled detonation (pulse) engine using fluent. I am not able to get the required pressure spike for my model. The following are its technical specifications:

The ambient conditions outside the tube are assumed to be 1 bar and 298 K. A high-pressure, high-temperature driver gas (driver pressure, pdriv = 30 p0, and driver temperature, Tdriv = 2500 K) consisting of water vapour is used in a small region next to the closed end of the tube (length of driver gas, Ldriv = 0.005 Lt).

I used Fluent's patch option to patch the driver (H2O) and driven gas (fuel mixture) zones in the cylinder. I am not getting the required pressure spikes for a detonation, though the mixture seems to undergo combustion. I am using a stoichiometric hydrogen-oxygen fuel mixture for simplicity along with inviscid flow conditions. I don't seem to be getting a DDT, though the paper I read guarantees a detonation for the above conditions. Is there something wrong with my species definition? Should I rather use a pressure based solver with a spark ignition system? What reaction mechanism does Fluent use for hydrogen-oxygen combustion? The paper had used a Jachimowski reaction mechanism (23 reactions in all). Do I have to set ignition conditions for the mixture based on what mechanism Fluent uses?

I would really appreciate it, if anybody could kindly help me. Thank you for your time.

Sincerely,

Vikram.

ssixr October 14, 2010 02:41

hello vikram,
i am also trying to simulate same model from the same paper. as far as i know your approach except for flow model...flow in detonation is turbulent.so you should choose turbulence model like k-epsilon for ddt..hope it helps you...are u working in pulse detonation engine area?

truly
sudhir

sreerao October 14, 2010 05:57

hi Vikram

Would you be able to give me the details of the paper, which you are following.

I agree with Sudheer. However, better check chemistry. More over, check the values in the materials panel as mostly we take default values as granted, which may not be true in reality.

Sree

vikram January 15, 2011 07:01

@ sudhir: Thank you for your reply and advice. Yes, I think the turbulent model will better fit the simulation. Yes, I was working on PDEs. What are you doing? Are u working on the same?

@ Sree: Thank you for your suggestion. The title of the paper is "Computational study of NOx formation in hydrogen-fuelled' pulse detonation engines" by S. YUNGSTER, K. RADHAKRISHNAN and K. BREISACHER. I will send you the paper if you give me your e-mail ID. Are u working on PDEs?

ssixr January 17, 2011 11:42

@vikram,
you can also try inviscid flow for ideal case without a need of high performance computers..yes, im also working on a 2D model of PDE with simple geometry and boundary conditions. can u send me tat paper which you mentioned? my mail id:sudhir86@gmail.com.
thank you

gsatr1989 November 28, 2011 22:22

Modelling in GAMBIT
 
Hi,

I am starting with a project on PDE to analyse DDT with varying cross sections and optimum CJ velocity lengths. I decided to start with simple direct detonation modeling based on the paper stated above.

I guess I'm having a hard time understanding about the mesh itself. I'm meshing in gambit. How do I differentiate the two fluid zones. I know how to assign fluid zones to a particular face or volume, but here the two fluid zones are right next to each other. So do I draw an edge between the two faces and set the boundary to interface? I think the patch option in FLUENT initializing will help you set the respective properties but I'm not getting how to define the two different zones separately from the meshing point of view in GAMBIT.

Any help would be greatly appreciated

ssixr November 30, 2011 12:34

hello gsatr1989,
for learning about patching, you can go through VOF tutorial of fluent.. about your two fluid zone i am not understanding... please tell me where you are applying this two fluid zones and what is the boundary conditions applied ?

gsatr1989 December 2, 2011 05:19

Hi,

I am looking to study initiation in PDEs. So as first step, I am considering an already existing high pressure high temperature (close to Tadiabatic) region composed of H2O vapor to act as the initiator. So a small region at the beginning of the PDE tube requires to be patched with the the conditions stated above.

After that, the rest of the tube length needs to be initialized with H2 and O2 mixture at a given equivalence ratio and at ambient temperature and pressure.. So both zones are gaseous, so I cant use VOF.

So I expect the high energy 'driver' zone at higher temperature and pressure shall propagate through the length of the tube due to pressure difference. In doing so, the energy with the wave can behave like an ignition and cause detonation and the detonation wave will be carried out due to heat produced behind the wave.

The only BC here is the pressure outlet at the right end of the tube where the combusted wave needs to exit. The left end of the tube contains the driver gas and hence this end is closed.

I am using species transport for combustion with 1 step reaction. I was able to patch the values, but I am not able to see the combustion happening.. I hope I have expressed my case clearly. Any help will be appreciated.

Thanks
KArthik

vikram December 2, 2011 09:34

hey karthik.. try running the simulation with a detailed reaction mechanism consisting of elementary reactions like the one in the article mentioned above.. highly simplified chemistry models are important for mathematical analysis and
algorithm development, but they are of limited applicability. For example, one-step chemistry models cannot model the initiation, branching and recombination steps characteristic of the combustion process at the detonation front and therefore cannot reproduce the interactions between the chemical kinetics and fluid dynamics. This coupling determines the dynamics of detonation wave propagation.. also are u using a laminar or turbulence model?

ssixr December 5, 2011 11:44

hey karthick,
if i am not wrong, you are following ebirami and merkle's paper:)
i also tried that once.... chk whether you gave adiabatic wall condition for all the three remaining surfaces or not....you can create that high pressure and high temperature zone with water vapor as species using patch option. you can also specify the H2 and O2 fraction using initialize option..... give small small time steps in order of micron (10^-6 or less).... it ll be better if you run this simulation in any hp/dell or some workstations...
hope this helps you...

truly
sudhir maheswaran

gsatr1989 December 8, 2011 08:36

Hi

Thanks for the reply sudhir. I actually took the idea from Yungster, Radhakrishnan in Combustion Theory and modelling

http://www.tandfonline.com/doi/abs/1...tctm20#preview

Basically I was able to patch the regions separately, to the temperature, pressure and species config that i wanted. The driver region is patched right next to the closed end of the tube, so there is essentially no inlet. The model banks on the pressure gradient between the driver and driven regions to propagate forward. I was able to run the simulation for non-reactive species transport and match it with theoretical calculations for shock wave flow through tube.

Here is the brief of my FLUENT setup apart from the driver and driven gas properties:
-->left end closed, no inlet, right end open and set to pressure outlet with atm pressure and backflow species conc in ratio of 1:2
-->Turbulence (k-e realizable) and species transport enabled with turbulence chemistry run according to eddy dissipation concept and I have included a chemkin file which contains the 23 rxn mechanism.
-->I am running it transient with 1E-6 step and 1000 time steps with 10 iterations each.

I have 3 clarifications: There is no gap between the driver region and left end wall, So will there be some flow recirculation ( the model is experiencing this). And predominantly I am getting some 20%of backflow at the pressure outlet.
Should I patch initial velocity to the driver gas. so as to initiate the rightward movement ( maybe the CJ velocity for that T and P) Further should I consider other model for combustion say Parially pre-mixed?

Is patching a driver gas region same as a Pressure inlet condition...but then what are the inlet species and filled species for the tube.

I read another post where sudhir talks abt using AUTODYN...can you elaborate or give me a source for this

Thank you again and again any help will be appreciated..
Karthik

tsram90 March 29, 2012 23:13

@gsatr1989, @ssixr

I am trying to study combustion of h2 in a SI engine. I am not able to get the combustion, Which I thinkis due to improper parameters can you send the case,mesh file you have used to my Id sankarram90[at]gmail.com

I just want to check all the parameters you have given and models selected.
Thanks a lot in advance

gsatr1989 April 1, 2012 10:12

Hi,

If you are using FLUENT, for SI, you can use the Premixed model since it uses a decent code for the flame characterization. Or you can go with the generic species transport where you can choose the volumetric mode and specify even a detailed chemistry with a chemkin .inp file. Any of the 3 turb-chem interactions should do. You need to setup ignition model and patch the volume with initial properties.

I simulated DDT with a driver flame which was already fully developed and my model dealt with detonation for the major part. And I simulated an open system. Give more details abt geometry, setup details

aero_peace October 13, 2014 04:11

Hello there,

Has anyone been successful at creating a CJ detonation for stoichiometric H2/O2? I am having a rather difficult time attempting this. I am using a 18 step mechanism developed by Petersen and Hanson for ram accelerator combustion. Likewise, I am patching a thin region of 30 atm and 3500 K at the thrust wall to initiate a shock induced detonation (note the region also consist of the mass fraction of the species in a post detonation equilibrium state as predicted by NASA CEA). It seems that no matter how fine my grid is (10 micro meters with level 3 adaptive mesh refinement), or how small my time step is (1.0e-8~1.0e-11) the solution will not converge. It appears that that the chemical kinetics are too stiff for fluent. Can any light be shed on this matter? It would be greatly appreciated.

Thanks,

sanjeetlimbu April 26, 2015 23:56

3 Attachment(s)
Hi
I am trying to do auto ignition for nheptane/Ar/O2/N2 mixture... with adiabatic wall
Can anyone suggest how to do ..ignition

1. I tried the laminer - species transport: It shows no rise in temp / pressure (I did the cold flow calculation + patch temp 1400K and species reactant), as there is no combustion even after patch temp... just increase in temperature as in attached snapshot

2. I tried the flame let model in Partially premixed but that show error in the unsteady initialisation that the mixture fraction limit is grater than the largest domain fraction even as I entered 0 in it

I tried using the cold flow+ patch at 1400K the body -part in mesh , but no combustion observed , only the temp raised....

I tried then 2900K/40atm patch but no raise observed in products _ CO2 or H20 in mole fraction...

In both cases the temperature just climbs to the patch value and there is no temp /pressure raise after than


All times are GMT -4. The time now is 02:41.