CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   EBU eddy break up unpremixed combustion for diesel (http://www.cfd-online.com/Forums/cd-adapco/64105-ebu-eddy-break-up-unpremixed-combustion-diesel.html)

hkoten April 29, 2009 07:29

EBU eddy break up unpremixed combustion for diesel
 
hi
how do you use ebu for unpremixed model,
did you close cfm or ecfm models in esice
hkoten@hotmail.com

hkoten April 29, 2009 07:42

who use EBU combustion models for unpremixed....

Pauli April 29, 2009 11:18

You can use EBU for un-premixed diesel combustion. You do not need es-ice.

Define the reaction, define the ignition, define the spray, initialize the scalars & you should be done.

If it doesn't work, read the methodology and user manuals. Then post specific questions & details regarding what went wrong.

hkoten April 29, 2009 11:26

thank you pauli
I am studying in-cylinder diesel combustion modeling
I meshed in esice,
in prostar start of inj. 700CA end of inj 735CA
and ignation in 705CA,

but temperatures arent increasing~1000K

sthamit September 1, 2010 15:47

Pro-Star: Shell Modeling
 
Quote:

Originally Posted by Pauli (Post 214652)
You can use EBU for un-premixed diesel combustion. You do not need es-ice.

Define the reaction, define the ignition, define the spray, initialize the scalars & you should be done.

If it doesn't work, read the methodology and user manuals. Then post specific questions & details regarding what went wrong.


Hi, I have been trying so many different reaction models but all of them have failed. How do you know that un-premixed combustion with EBU standard is the best model for diesel combustion? Could you tell me what's the best combination reaction model and type for Diesel combustion simulation in Pro-Star?

Thank you in advance.

Amit

Pauli September 1, 2010 17:55

I never claimed EBU was the best model. ECFM-3Z should be superior. Flamelet models should be superior.

What do you mean by failed? Did it not "burn"? Or do you not match a measured pressure trace?

sthamit September 1, 2010 22:06

Quote:

Originally Posted by Pauli (Post 273659)
I never claimed EBU was the best model. ECFM-3Z should be superior. Flamelet models should be superior.

What do you mean by failed? Did it not "burn"? Or do you not match a measured pressure trace?

Hi. I am not saying that you are claiming. I am asking basically. First of all EBU and ECFM-3Z are two different things. ECFM-3Z is the combustion model that best suits the diesel combustion whereas EBU is a reaction type. I have been trying different combinations under ECFM-3Z like partially premixed scheme or unpremixed scheme with EBU standard and other options but all of them ended up with almost zero Ignition delay. this is what failed mean...meaning....failing to match the results with empirical ones.

If you have more idea about it, could you please help.

Thanks

Pauli September 3, 2010 11:20

What ignition model are you using? Have you tried to vary any of the parameters?

I was using EBU-LATC with 4-step ignition. I was able to make ignition delay & pre-mix phase vary significantly by changing the various parameters.

sthamit September 3, 2010 12:40

Quote:

Originally Posted by Pauli (Post 273925)
What ignition model are you using? Have you tried to vary any of the parameters?

I was using EBU-LATC with 4-step ignition. I was able to make ignition delay & pre-mix phase vary significantly by changing the various parameters.


Hi,

Until now, I have used Premixed with CFM, Unpremixed and Partially Premixed schemes with EBU STANDARD, LATCT and Modwall. Unpremixed and Partially Premixed seem to have no Ignition Delay. Premixed with CFM is mainly meant for HCCI and SI combustion. I really have to try something new now. What reaction scheme did you use?

Thanks

Pauli September 3, 2010 13:48

I am using Turbulent Unpremixed/Diffusion, EBU-LATC & 4-Step Auto-Ignition.

Are you trying to model a diesel engine? If so, why are you using unpremixed or partially premixed? That makes no sense.

How are you defining ignition delay? I compare start of injection to initial rise in the heat release rate curve. With that definition I have 3-4 degrees delay in both my test data & simulation.

sthamit September 3, 2010 14:14

Quote:

Originally Posted by Pauli (Post 273942)
I am using Turbulent Unpremixed/Diffusion, EBU-LATC & 4-Step Auto-Ignition.

Are you trying to model a diesel engine? If so, why are you using unpremixed or partially premixed? That makes no sense.

How are you defining ignition delay? I compare start of injection to initial rise in the heat release rate curve. With that definition I have 3-4 degrees delay in both my test data & simulation.


Hi Pauli,

Partially Premixed doesn't make sense to me either. I was just trying to see what happens when I use that option. Other than that, Unpremixed is the one that I think I should be using either with Modwall or LATCT reaction type because of the one chemical reaction requirement of the 4-step Auto Ignition Model.

Now coming to defining Ignition Delay, the way you are doing is what even I or anybody would do. i.e. by counting the no. of CAD for ID for any given fuel in the RHR graph of the real test and putting that value on 4-step autoignition model to imitate the nature of the graph or ignition. But the problem arises when you would like to simulate a new fuel with new CN that has not gone through experimental test.

What I meant to say here is, I want to simulate a diesel combustion with fuel of known CN in Pro-star and come up with results that match the actual experimental test data. Is it something possible to do with Pro-star you think?

Thanks

sthamit September 3, 2010 16:02

Quote:

Originally Posted by sthamit (Post 273944)
Hi Pauli,

Partially Premixed doesn't make sense to me either. I was just trying to see what happens when I use that option. Other than that, Unpremixed is the one that I think I should be using either with Modwall or LATCT reaction type because of the one chemical reaction requirement of the 4-step Auto Ignition Model.

Now coming to defining Ignition Delay, the way you are doing is what even I or anybody would do. i.e. by counting the no. of CAD for ID for any given fuel in the RHR graph of the real test and putting that value on 4-step autoignition model to imitate the nature of the graph or ignition. But the problem arises when you would like to simulate a new fuel with new CN that has not gone through experimental test.

What I meant to say here is, I want to simulate a diesel combustion with fuel of known CN in Pro-star and come up with results that match the actual experimental test data. Is it something possible to do with Pro-star you think?

Thanks


Also, what do you do with Tgradient and Start and End time? How do you setup Ignition Delay on 4-step Auto Ignition?

Pauli September 3, 2010 19:23

Calibrating a combustion simulation involves tuning the spray, combustion and ignition sub-models. It's all coupled and very non-linear. So until you've seen your model predict well over a range of conditions, you can not be sure about just how well its calibrated. (You could have gotten lucky!)

The end results is that if you take test data with two different CN fuels, you might not see similar changes in your CFD results. My feeling is that It depends on how well your original calibration was. In other words, your spray model may have been off & you inadvertantly compensated with some combustion or ignition parameter. What changed in the fuel to give it a different CN? Did that also change something which would impact the spray characteristics? If so, then you'll need to update your spray model tuning.

sthamit September 3, 2010 21:30

Quote:

Originally Posted by Pauli (Post 273967)
Calibrating a combustion simulation involves tuning the spray, combustion and ignition sub-models. It's all coupled and very non-linear. So until you've seen your model predict well over a range of conditions, you can not be sure about just how well its calibrated. (You could have gotten lucky!)

The end results is that if you take test data with two different CN fuels, you might not see similar changes in your CFD results. My feeling is that It depends on how well your original calibration was. In other words, your spray model may have been off & you inadvertantly compensated with some combustion or ignition parameter. What changed in the fuel to give it a different CN? Did that also change something which would impact the spray characteristics? If so, then you'll need to update your spray model tuning.


Thanks Pauli,

It feels really great to share info when I come across somebody who is equally interested in the subject matter as I do.

What you are saying is right as spray modeling does play a very important role in combustion modeling. This afternoon I was trying to use 4-step auto ignition but I got confused with how to adjust the start and end time of ignition in terms of crank angle degrees as the option I see is in terms of time only. Could you guide me a bit please? Also, what ignition temperature/sec should I use? About spray modeling I will let you know what I have so that you will have an idea as well.

Thanks

sthamit September 7, 2010 12:46

Quote:

Originally Posted by Pauli (Post 273967)
Calibrating a combustion simulation involves tuning the spray, combustion and ignition sub-models. It's all coupled and very non-linear. So until you've seen your model predict well over a range of conditions, you can not be sure about just how well its calibrated. (You could have gotten lucky!)

The end results is that if you take test data with two different CN fuels, you might not see similar changes in your CFD results. My feeling is that It depends on how well your original calibration was. In other words, your spray model may have been off & you inadvertantly compensated with some combustion or ignition parameter. What changed in the fuel to give it a different CN? Did that also change something which would impact the spray characteristics? If so, then you'll need to update your spray model tuning.

Hi,

Although I have been trying, I still don't have an idea on how to set Tgradient in 4-step Shell Autoignition. Also, how to set start and end time of the ignition in terms of crank angle degrees rather than in seconds.

I would appreciate your help.

Thanks in advance

sthamit September 7, 2010 14:08

Quote:

Originally Posted by Pauli (Post 273942)
I am using Turbulent Unpremixed/Diffusion, EBU-LATC & 4-Step Auto-Ignition.

Are you trying to model a diesel engine? If so, why are you using unpremixed or partially premixed? That makes no sense.

How are you defining ignition delay? I compare start of injection to initial rise in the heat release rate curve. With that definition I have 3-4 degrees delay in both my test data & simulation.


Hey Pauli,

If I am modeling a diesel engine, what should I use? Premixed or Unpremixed or what?

Please help. Thanks

Pauli September 8, 2010 11:12

To enter values in crank angle, you need to tell prostar to use crank angle. That is done with the trelation command. Or in the Nav Center, it is under the first entry: Select Analysis Features -> Time Relationships. Note: The time relationships sub-panel is not active until you switch to transient mode.

Diesel engine is unpremixed.

sthamit September 8, 2010 12:47

Quote:

Originally Posted by Pauli (Post 274469)
To enter values in crank angle, you need to tell prostar to use crank angle. That is done with the trelation command. Or in the Nav Center, it is under the first entry: Select Analysis Features -> Time Relationships. Note: The time relationships sub-panel is not active until you switch to transient mode.

Diesel engine is unpremixed.


Thank you very much man.
In my case, I have been using time relationship in angle with transient mode fortunately but the problem is it is not firing.
Okay, if I am running a case for which SOI is at 718 CAD (in prostar or 1 CAD before TDC in real engine), and the RHR trace obtained from the real test shows that the combustion starts 4 CAD after TDC. In such situation, what value should I enter in the start and the end boxes?
In addition to this, the other problem I am facing is the TGRADIENT (K/s). Currently I am using the default value of 1.0e+06. I wonder if this is the right number to be entered in the TGRADIENT box !!
Most importantly, if we are specifying the start of combustion using the start and the end boxes then it means that we are defining the Ignition delay. In such case how will the cetane number plays its role as this is what determines the delay.

I have been trying different values but it is misfiring every time. I would appreciate your guidelines deeply.

Thanks a lot.

Pauli September 8, 2010 13:51

The start & end of ignition boxes only tell the code when to activate the ignition sub-model. They have nothing to do with ignition delay. Set start of ignition crank angle to something just before start of injection. Set end of ignition to the end of your simulation or end of injection. Limiting the time the ignition sub-model is active is done to minimize computational overhead (no point checking for ignition when it can't/shouldn't occur).

The effect of cetane number is decribed in the methodology manual. See the section for the ignition model (equation 11-85 in the Star 3.26 manual).

The default Tgradient value should work fine for initial runs. You can tune it later.

sthamit September 8, 2010 13:56

Quote:

Originally Posted by Pauli (Post 274489)
The start & end of ignition boxes only tell the code when to activate the ignition sub-model. They have nothing to do with ignition delay. Set start of ignition crank angle to something just before start of injection. Set end of ignition to the end of your simulation or end of injection. Limiting the time the ignition sub-model is active is done to minimize computational overhead (no point checking for ignition when it can't/shouldn't occur).

The effect of cetane number is decribed in the methodology manual. See the section for the ignition model (equation 11-85 in the Star 3.26 manual).

The default Tgradient value should work fine for initial runs. You can tune it later.


Thank you man.

Let me try what you said and will get back to you very soon.

Thank you once again.


All times are GMT -4. The time now is 22:40.