CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   are there any advantages of writing codes than simulation in fluent? (http://www.cfd-online.com/Forums/main/110646-there-any-advantages-writing-codes-than-simulation-fluent.html)

dssrkanth December 17, 2012 04:53

are there any advantages of writing codes than simulation in fluent?
 
hello everyone

i have been developing a code for a pulsatile flow in a furrowed channel. can be this done better on fluent?? or developing a code gives us more accurate and flexible options to solve the problem??

Hope to get a reply for this

cdegroot December 17, 2012 21:45

Depends what you want to do. If your code would just end up doing the same thing that Fluent would do, why spend the time? But if you want to do something Fluent can't do or implement some models Fluent doesn't have then writing your own code makes sense. Also, maybe you just want to write a code for fun; that would be okay too.

Sören Sander December 18, 2012 05:55

Imho you should always start by using a implemented code (get an overview, learn about the solver). If your results do not fit the experimental data, you can define more precise functions afterwards. fluent offers possibilities to adopt e.g. your boundary conditions by writing so called UDF's. These will allow you to adopt your case to your needs in at least 90 % of the cases.

cdegroot December 18, 2012 08:23

Yes, a general purpose CFD code like Fluent or CFX will be sufficient for almost all problems you will encounter, unless you have very specialized requirements.

FMDenaro December 18, 2012 12:52

and don't forget, commercial codes must be paid ...

Engr.RZA December 18, 2012 13:14

Developing a CFD code is not an easy task..mostly done in post graduate and Phd studies..I would recommend you to use Fluent / CFD ACE+ with user defined functions for boundary conditions if needed..

SergeAS December 18, 2012 14:42

Quote:

Originally Posted by FMDenaro (Post 398114)
and don't forget, commercial codes must be paid ...

This is especially sensitive when you need to use in the calculation thousands cores

mturcios777 December 18, 2012 14:49

Though there is a bit of a learning curve, OpenFOAM is a nice bridge between an integrated code and building your own. www.openfoam.com

dssrkanth December 19, 2012 09:44

thanks for the reply all
 
i am a post graduate student. i wanted to learn how to implement codes. so i have done my project on simulation of pulsatile flows. but when i am attending interviews people are asking me why have i done my project on coding. when i said its a part of learning curve, they didnt accept. is the accuracy of fluent good in the case of pulsatile unsteady laminar flows??
and by the way can we use K- epsilon and k omega model for laminar flows??

Sören Sander December 19, 2012 10:39

If your flow is laminar you should neigther use k-e nor k-w (you should not use a turbulence model at all). Especially k-e model will give you wrong results. k-w might be a possible solution if your flow is laminar in one region and turbulent in another.

dssrkanth December 19, 2012 10:42

got the reply i needed
 
thank you for ur reply

cdegroot December 19, 2012 11:26

Quote:

Originally Posted by dssrkanth (Post 398252)
i am a post graduate student. i wanted to learn how to implement codes. so i have done my project on simulation of pulsatile flows. but when i am attending interviews people are asking me why have i done my project on coding. when i said its a part of learning curve, they didnt accept. is the accuracy of fluent good in the case of pulsatile unsteady laminar flows??
and by the way can we use K- epsilon and k omega model for laminar flows??

If you ever interview for a position as a CFD code developer they will appreciate the fact that you have written your own code. Maybe you should look into how Fluent discretizes the governing equations. Maybe you are doing something better and you don't know it. For example, you are looking at a transient flow; do you use something higher than second order in time?

dssrkanth December 19, 2012 12:46

no,its first order accurate in time but i am using a third order accurate solution in space using deferred correction procedure.

cdegroot December 19, 2012 12:51

Fluent is second order in space and time so it sounds like you can pitch that as the advantage of your code.

dssrkanth December 19, 2012 12:54

thanks a lot for the advise, but is fluent second order in time?

FMDenaro December 19, 2012 13:11

Quote:

Originally Posted by dssrkanth (Post 398293)
thanks a lot for the advise, but is fluent second order in time?

yes, you can use second order accuracy both in time and in space, for example this is the standardo for LES.
However, I think that also a simple Fulent user would be greatly appreciated if he has had experience in programmig CFD codes .... I see too many often people that use Fluent without any knoledge of the basis of CFD, with dramatic results.....

cdegroot December 19, 2012 13:21

Absolutely. A lot of people don't see how experience with CFD coding helps one as a CFD user, but it absolutely does. You have much more knowledge and appreciation of the discretization issues of stability, accuracy, etc. Next time someone asks why you wrote your own code you should mention this as well.

dssrkanth December 19, 2012 13:24

actually i mentioned it in the interview that i had, but he didn't seem to like it.... but i personally feel that writing a code can make me a adapt to any software quite comfortably

cdegroot December 19, 2012 13:24

You are right about that.

Martin Hegedus December 20, 2012 14:10

My post does not address pulsating flow, but is more a general comment, and my opinion, on the interview process.

It depends on who you were interviewing with.

Just because one can code does not mean one knows where to apply CFD and how to set it up for a specific case. There are a lot of choices to make, unstructured, structured, number of grid points, targeting flow features, turbulence models (laminar, SA, SST, transition models, curvature corrections, LES, etc.), incompressible , compressible (for example, a slatted and flapped airfoil at Mach 0.2 does has supersonic flow at leading edge for high alpha), explicit, implicit, central-difference, flux-vector splitting, flux-vector splitting, flux-differencing. Choices, choices, choices. And, sometimes, CFD just doesn't take one where they want to go. Thus one needs to know when they have maxed out on the capabilities of CFD and move on.

If one has coded something up to help them understand these choices, that's great! However, if ones coding/research has narrowed their focus such that they need to be educated about those choice on the job, then there is an employer cost associated with it. And, it's not just time spent, but also very much about errors and wrong assumptions made. And, sorry to say, egos.

Then there is all of the marketing hype behind CFD. One needs to be able to efficiently figure out what's true and what's hand waving. Thus both narrow and broad knowledge is required.

Just because one understands the foundation for the errors and uncertainties in CFD does not mean they are able to use that knowledge to understand the uncertainties and trade offs in the big picture.

All of this is very important not only in getting the best possible answer out of CFD but also in being aligned with other team players and determining cost quotes.


All times are GMT -4. The time now is 05:26.