|
[Sponsors] |
Should I do multi cycles test when using RANS model ? |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
January 12, 2017, 04:31 |
Should I do multi cycles test when using RANS model ?
|
#1 |
New Member
Shen shiquan
Join Date: Jul 2016
Location: The State Key Laboratory of Engines (Tianjin University)
Posts: 12
Rep Power: 10 |
Hello everyone, As the title says I am doing a simulation of two-stroke marine diesel engine by using RANS k-ε model ,when I run my case in 14 cycles ,I found some output data such as swirl ratio ,TKE and EPS in the cylinder vary with the cycles. Especially the first cycle has great difference with the later cycles. The question is : if i want to analyze the velocity field in the cylinder ,should I do simulation of multi cycles and just pick one stable cycle in the last or do average of the last stable cycles? PS: I use RANS K-ε model in the Converge .
I am looking forward to your reply,thanks |
|
January 12, 2017, 12:10 |
|
#2 |
Member
Jerome Le Moine
Join Date: Jan 2016
Location: Convergent Science, Madison WI
Posts: 50
Rep Power: 10 |
Shen shiquan,
Thank you for your question. For gasoline cases, it is very common that the initial conditions play an important role in the results of the first cycles. So we recommend running at least 3 cycles in order to obtain solutions that are independent of the initial conditions. From the plot of the swirl ratio, it seems after the second cycle the solution remains unchanged from one cycle to the other. For your study, you could pick results from one cycle after the second cycle or you could average the results from cycle 3 to the last cycle. The most accurate way will be the last of the two options. |
|
January 12, 2017, 21:34 |
|
#3 | |
New Member
Shen shiquan
Join Date: Jul 2016
Location: The State Key Laboratory of Engines (Tianjin University)
Posts: 12
Rep Power: 10 |
Quote:
But another question is: if i want to analyze the velocity field in the cylinder ,should I do average of 3D data files from different and stable cycles ? or just pick one cycle's 3D output files for analysis ? PS: I use RANS K-ε model ,not LES model. |
||
January 13, 2017, 10:47 |
|
#4 |
Member
Jerome Le Moine
Join Date: Jan 2016
Location: Convergent Science, Madison WI
Posts: 50
Rep Power: 10 |
In your case, I would average the 3D data files from different and stable cycles because you have the data available to you. Even if you are running with RANS k-epsilon model, there may be some small variations in velocity field so the most accurate way is to average the 3D data over multiple engine cycles.
|
|
January 13, 2017, 20:58 |
|
#5 | |
New Member
Shen shiquan
Join Date: Jul 2016
Location: The State Key Laboratory of Engines (Tianjin University)
Posts: 12
Rep Power: 10 |
Quote:
|
||
January 21, 2017, 09:51 |
|
#6 | |
Senior Member
Tobias
Join Date: May 2016
Location: Germany
Posts: 284
Rep Power: 11 |
Quote:
Regarding the Main Question: So far the max. number of cycles of a singel simulation i have ran were 5. But, during the setup i spend a lot of time to check how much fuel i need to add premixed into the intake port, to match the target Lambda with the first cycle. I am working with DI and PFI gasoline models, using RNG k-e model. PFI is more difficult of course. There i run 1 cycle (-540 to +180), than map the spray, wallfilm and flow field of the intake port region only. Than basically rerun the case and thats my new 1st cycle. Then i continue with restart mapping from +180 BDC. About your plot, i would say you should check the initial conditions. The first cycle looks way too different. |
||
January 30, 2017, 21:49 |
|
#7 |
New Member
Michael Cao
Join Date: Jan 2017
Posts: 1
Rep Power: 0 |
If you want to average the 3D data,you need to write your own matlab program ,may be u can get these programs from your partners who is doing LES simulation, because they need to do average from different cycles. Good luck
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
2D RANS Model? | wanted2105 | Main CFD Forum | 0 | July 15, 2016 16:44 |
RANS model | Cris2015 | Main CFD Forum | 11 | September 11, 2015 06:58 |
RANS simulation of Caradonna blade model | deeps | SU2 | 4 | July 10, 2014 08:07 |
using interface with Euler multi phase model | karimi | Main CFD Forum | 0 | May 22, 2011 02:39 |
RANS model for near wake region | frederic felten | Main CFD Forum | 0 | April 10, 2003 17:24 |