Hello friends,

I an adams bashforth time marching scheme. The solution of a flow variable is calculated from current solution and the derivatives from the previous three(time steps) and current times.Is it possible to derive it in terms of flow solution of previous time steps instead of the derivatives.

Thank you friends.

[hitzhwan]

Originally Posted by vasanth
;39133

Hello friends,

I an adams bashforth time marching scheme. The solution of a flow variable is calculated from current solution and the derivatives from the previous three(time steps) and current times.Is it possible to derive it in terms of flow solution of previous time steps instead of the derivatives.

Thank you friends.

Excuse me， how can set Adams-Bashforth scheme in the fluent？

[sbaffini]

There is no such thing in Fluent. There might be some chance by properly working with UDF, but the chances are actually very low and requires full proficiency with Fluent and all the details of its schemes, UDF and CFD as a whole. Thus, there is no real point in doing it.

[hitzhwan]

Originally Posted by sbaffini

There is no such thing in Fluent. There might be some chance by properly working with UDF, but the chances are actually very low and requires full proficiency with Fluent and all the details of its schemes, UDF and CFD as a whole. Thus, there is no real point in doing it.

But I see it in some articles.

[arjun]

Originally Posted by hitzhwan

But I see it in some articles.

Humm.



There are three ways to know for sure.


1. Save a case file in text format and open it in any text editor and search for the option.

2. Explore the Fluent's command prompt where the adavance settings are burried.


3rd and last leastest known method
Fluent also comes with a src folder (it came till i last checked). In this folder you can check header files and see what options really exist for the solver. A great amount of things one can learn from this source folder about Fluent.

[sbaffini]

Originally Posted by hitzhwan

But I see it in some articles.

Can you post at least one example?

[sbaffini]

Originally Posted by arjun

Humm.



There are three ways to know for sure.


1. Save a case file in text format and open it in any text editor and search for the option.

2. Explore the Fluent's command prompt where the adavance settings are burried.


3rd and last leastest known method
Fluent also comes with a src folder (it came till i last checked). In this folder you can check header files and see what options really exist for the solver. A great amount of things one can learn from this source folder about Fluent.

Now you're making me doubtful...

[arjun]

Originally Posted by sbaffini

Now you're making me doubtful...

I have never seen too but with Fluent we never know. They have so many options hidden.



Another interesting thing about Fluent is that they drop details from their documentation. So there are things in documents of version 6.3 that you won't find in current docs for example.

[sbaffini]

Originally Posted by arjun

I have never seen too but with Fluent we never know. They have so many options hidden.



Another interesting thing about Fluent is that they drop details from their documentation. So there are things in documents of version 6.3 that you won't find in current docs for example.

I used the trick to discover v2f and algebraic stress models back in the days when they were not available, but I also failed to activate the infamous RNG LES model, that is listed there as well.

But, while I haven't looked for it specifically, I think it would have hardly made any sense in the overall Fluent world, as the single place where it could have lived is in the NITA solvers (probably only the FSM and not PISO). And still, we are talking of an explicit method...

[arjun]

Originally Posted by sbaffini

I used the trick to discover v2f and algebraic stress models back in the days when they were not available, but I also failed to activate the infamous RNG LES model, that is listed there as well.

But, while I haven't looked for it specifically, I think it would have hardly made any sense in the overall Fluent world, as the single place where it could have lived is in the NITA solvers (probably only the FSM and not PISO). And still, we are talking of an explicit method...

RNG LES was very nice model. It gave me good results. Personally i really doubt that that option exist. It is not easy to have it in Fluent/Starccm types of solvers.

[FMDenaro]

Difficult to say what is exactly the time integration in the NITA formulation, just a look at the cited reference can suggest that a two-level method could be used

https://www.afs.enea.it/project/nept...th/node375.htm

[sbaffini]

Yes, indeed the original reference by Sung-Eun Kim (An Implicit Fractional-Step Method for Efficient Transient Simulation of Incompressible Flows) uses a 2nd order backward difference scheme, and that's the only 2nd order time scheme directly available in the Fluent GUI for the segregated pressure based solvers (not sure for the coupled pressure ones).

But we are talking of the implicit one, as the title suggests. And as this is pre-Ansys era, I expect the reference to be reliable info.

[LuckyTran]

Gents, I looked all up and down the .cas file and couldn't find Adams-Bashforth. I also checked the src dirs (yes it still ships with Fluent, great suggestion btw!) for both pbns and dbns and couldn't find code that retains multiple solution levels forward. We're already aware of course that the previous 2 timesteps can be stored for pbns and this is the 2nd order backward Euler.

One could change the Courant numbers in the multi-stage scheme, but this doesn't make it a multi-level solver nor does it change the solution weights to make it a new scheme as it should be for Adams-Bashforth

So... implementing Adams-Bashforth requires some major under the hood, or yoinking someone's udf. The best place to find these udf's is to ask the people that have used AB2 in Fluent in the past.