Hi,

Finally, I made up my mind to give up BVM. I won't be able to finish BVM model in the current time frame and the facilities and support available for me at present from Uni. So, I'm going to concentrate on EDM as of now.

EDM

I tried using Methane Air WD1 mixture as combustion material for the model, used a value of 25k W as ignition energy (added as a function so as to supply this energy to domain at 10 degree BTDC, just for one degree as I thought that is only needed for a combustion initialization) However, in my case combustion is not happening. I do not know why. I defined the default domain boundary as the source of this energy.

Does the values/settings I used look sensible? Do I need to use a different mixture? Can I do it as a single step reaction? (I don't have to worry about combustion products at this stage of project, what I need is a combustion simulation only) I couldn't figure out how to deal with smeared volume concept, though I will try it again tomorrow.

I know its a lot of questions. (probably some of them are very simple) Its all because I'm so confused now. Please help me with these queries.

post your ccl

Hi George,

Attached is the CCL file generated for the model, saved as TXT file.

Please have a look and give your suggestions.

EXPERT PARAMETERS:
coupled scalars = f why?
ggi permit no intersection = t
include pref in forces = t
min timesteps = 1
model coefficient relaxation = 1.0 why?
pressure diffusion scheme = 2
solve energy = t
solve masfrc = t
solve reaction progress = t
solve turbulence = t
solve volfrc = t
stress diffusion scheme = 2
tbulk for htc = 300.0
tef numerics option = 1
transient initialisation override = t
END

you dont need to redefine options are true by default :confused:

coupled scalars = f why?....because, in literature it says that the setting if made to 'f' will help the system from failure of coupling procedure due to oscillatory convergence or divergence. I'm not sure if this is applicable in my case.

model coefficient relaxation = 1.0 why?....the literature suggests to choose some value above 0.1 for better convergence of k-e model. so, i chose 1.

these setting might be wrong. what will be more sensible values for these settings?

for expert parameters dont change things unless you know what they are doing.

you want coupling of scalars get rid of that option

for transient initialisation override if you use a global initialisation this is not needed

use a coefficient of 0.75 if you have difficulties. think of relaxation as a numerical delay. if for exmaple for a particular iteration the computed value for temperature is 1 [K] then the number used for the next iteration is 1[K] * 0.75

Thank you George. I'm going to try the changes you suggested and try a new simulation tomorrow morning.

To be frank, I'm not sure about all the settings I made in expert parameters. May be I get rid of all the settings I made without knowing what exactly it is.

Hi George,

Please have a look on the attached ccl file.

Hi George,

Its running now.


got this error...probably with the mesh..

ERROR #002100012 has occurred in subroutine cVolSec. |
| Message: |
| A negative ELEMENT volume has been detected. This is a fatal |
| error and execution will be terminated. The location of the first |
| negative volume is reported below. |
| Volume : -0.1465E-08 |
| Location : ( -0.11466E-01, -0.48278E-01, 0.31179E-02)

slow the mesh movement compared to the timestep used and check the mesh quality during squashing. as i told you stop simulation, re-mesh and restart.

Are you doing moving mesh? This is a common problem for moving mesh simulations. If not then you have a serious problem with your mesh.

Hi Glenn,

Yes, this is moving mesh....infact a piston motion inside cylinder.

As George suggested, I will try reducing the time steps further.

At the moment I'm using cfxmesh in CFX. I tried using sweep and hexa, but didn't work well for me.

cfxmesh model was running fine. I made some slight changes in the model last day (in fact I decreased time step, so what I expected is a more stable simulation) and now it started misbehaving. Since I have interfaces on cylinder, the mesh concentration is very high near interfaces, and in some parts it seems distorted even if I re-meshed it quite a few times.

Please provide me tips to improve my simulation stability.

in v11 cfxmesh can do only tetra witout periodicity; how did you manage to create a sweep or hexa mesh :confused:

since the files you send me the other day did you create a new model with symetry?

Because, in Meshing mode in CFX, under the definition part, we can choose 'method', which gives u options to make hex/sweep/tetra/cfxmesh/automatic. I do not know if except tetra we can use the other meshes without periodicity. May be that was the reason my simulation failed when I tried to use the other types of meshes. The model I sent you, I think I made it using cfxmesh as that only worked fine in my simulation.

I will try reducing time step when I reach Uni today. Happy if that solves my problem with simulation.

No, I didn't try symmetry. To be frank I do not know how to create rotational symmetry. I never tried it before. Could you brief me how it is.

Hi George,

You had mentioned about no oxygen in the model, just the fuel. Could you tell me how to specify air/oxygen in my case where it is a premixed combustion model.

Also, why the EDM/FRC model is preferred over the EDM model?

well at the inlet you spesified the mixture to be
ch4 = 1
co2 = 0
h20 = 0
o2 = 0.232
n2 = constrant = (1 - all the above)

you cant have a mixture that adds up to more than 1; you need to know the mass of fuel that is injected and with that work out the mass composition of each component in the mixture

as for the mesh problem try a coarser mesh on the cylinder

I think this is better to be answered here so others will read and benefit. if you will use one inlet boundary to specify the air/fuel ratio you will need to link the inlet mass fractions with the timestep of your cycle. from what you said for your setup one would presume that you have indirect injection therefore if we assume the fuel is homogeneously mixed with the air you can write mathematical relations/functions that will fully describe the fuel (in your case ch4) mass fraction with time in your mixture.

In my case, as I used the methane-air mixture from the library, do I need to write functions again? Or, are they already defined by CFX?

I wonder if writing the relations for chemistry is needed only when we develop a new mixture using CFX-RIF

george, to be honest i think you are trying to do more than what you can achieve with your timeframe. your questions lack the fundamental understanding of what you do and I (or anyone else in this community ) cant explain you everything what there is to learn in order to perform your simulation. you are paying someone money to study your masters degree they should have provided technical support and your supervisor should have had the capability to help you with the simulation. I'm afraid my suggestion is to request for an extension on your project and chose something that your university can help you with. Furthermore I suggest to redo all tutorials including the CFD meshing tutorials. Personally I'm happy to help you with pointers however I'm not going explain how to setup your model from scratch.

best of luck with this... :(

Hi George,

Thank you for your tips and suggestions.

It is a sad truth that I never got enough support from my Uni.

I hope those who read this thread in future regarding EDM can find it useful to answer their queries.

An extension for the project or course is impossible for me at this moment. The only thing I can do is to submit with whatever I have achieved yet. (pass or fail, I leave it to God and destiny!!)

Thanks for your time; sorry if I asked quite lot questions. As I said in the beginning I'm just a beginner in the area of combustion modeling, and to master the whole process in 1 year time without enough support to answer my queries is a herculean task I believe.