CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Bound node loop. (http://www.cfd-online.com/Forums/fluent/27772-bound-node-loop.html)

Jack Keays July 5, 2000 13:25

Bound node loop.
 
Hi,

I am trying to run a 3d simulation of a turbomachine in which I have a sliding interface. However, when I try iterate I immediately get an error message saying

"Error: Failed creating bound node loop"

Does anyone know what this is?

Could anyone help me please.

Jack.

Jonas Larsson July 6, 2000 05:05

Re: Bound node loop.
 
I've also had numerous problems related to sliding interfaces in turbomachinery applications. You have to do everything in the right order. One of the things I found was that I had to do a "grid check" every time just before I defined the sliding interface. Another alternative is to first define them as translational, delete them, and then redefine them as rotational... wonderfully intuitive tricks :) This was in Fluent 5.2.3, it might have been fixed in later versions, I don't know. Another gotcha is that if you run in parallell you have to partition with "Don't partition across zone boundaries" activated.

Jack Keays July 6, 2000 09:39

Re: Bound node loop.
 
Hi,

I tried some of your tricks, and they didn't worry initially. I got the bound node error again. However, I checked the grid and then I got a meesage saying that it was "deleting wall zone-12 (not referenced by grid) and also "deleting wall zone-13 (not referenced by grid)". These are the rtwo wall zones created when I defined the sliding interface. Are they necessary for the program to run? I am not sure. I am currently running the program just to see if it is capable of running...so I will not know for a while if they deletion of these two wall zones causes problems. Also, when you say first define the sliding interface a translational, what do you mean? DO you mean to define the fluid zone containing the rotating parts as undergoing translational movement within the boundary conditon window? Then to create the interface. Then to redefine the movement as rotational in the B.C. menu, then delete the interface and create it again?? I hope I understand you. Its a crazy world when you gotta do such things to get a problem to work!! Thanks for your help Jonas. Also, thanks for the forum, its a life line to people like me at times. :o)

Jonas Larsson July 6, 2000 10:05

Re: Bound node loop.
 
Here are a few notes that I wrote to myself on how to get my problematic sliding-interface case to run in Fluent 5.2.3. Hope it helps, otherwise I'd really recommend you to contact fluent-support and try to talk to someone who has actually run a complicated sliding-mesh turbomachinery case:

===

Define all periodic boundary-zones as walls in Gambit. Defining them as periodic does not work

Setup

Merge all mesh parts using tfilter tmerge3d

Load the case in Fluent and do the following:

1. Scale your grid to SI units - this can not be done afterwards!

2. Define the rotational axis for the first fluid zone as the x-axis.

3. Define the periodic-boundaries around this fluid zone as periodic (previously defined as walls in Gambit).

4. Make these periodic boundaries periodic with the text-menu command /grid/modify- zones/make-periodic. You might need to adjust the tolerance with /grid/modify-zones/??? In order to get all nodes to match. If the grid comes from Gambit this is usually not necessary though.

5. Do points 1-4 for the next fluid zone.

6. Define the sliding interface between the two fluid zones with Grid/Define Interface. Before you define the interface you have to do a Grid/Check. This somehow corrects a problem which otherwise prevents you from defining the interface. Remember to mark the interface as rotational.

7. Repeat the above points for all fluid zones/interfaces, remembering to do a Grid/Check each time before you define a new interface. It is important that you do all this in the correct order, defining only one interface at a time. Do not try to cut corners by doing many things in each step.

Jack Keays July 7, 2000 09:25

Re: Bound node loop.
 
Thanks Jonas. Do you, or any of the other people know what was meant by "failed creating bound node loop". I just want to know what it means exactly. What is a bound node loop? Is it something to do with matching nodes across the sliding interface due to the non-conformality of the grid?? TH=hanks again.

John C. Chien July 9, 2000 14:57

Re: Bound node loop.,"User-friendly"?
 
(1). I have written many Navier-Stokes codes before. But even if I tried to read your message and notes a couple of times, I still could not understand why the users must be put through this kind of torture. (2). The other day, an engineer showed me a model which he was trying to repair. He was trying to show me that the code he used, can generate the mesh automatically and now he has 6 million cells in it.(not to mention the several licenses he needs) He is going through it patiently to look for bad cells. I think, he is a subcontractor, and the job is ideal for him. He kept saying that he was flexible. (about the assignment and the arrangement) (3). I have not talk to him for about a month since then. And I am sure that he was using the same code. (and they are facing a lot of other real technical problems) (4). Suggestions to the vendors: keep the code simple. write the code such that only well trained researchers can use the code. (this is because, in his PhD training, he is well trained and enjoy the trouble to handle those trivial things) give a warning when the below average engineers are trying to use it. I don't know whether the vendor of a code can be sued by the employer for long term phychological and emotional sickness through the use of the code. (torture) (5). Based on my experience, most engineers in a company love to use simple 1-D codes, performance codes, but they don't like to use these complicated codes. As a result, these jobs are being handled by subcontractors, temporary engineers, etc. (6). I think, if Jonas is a temporary engineer (trying to earn some money with not much choice), and has to run similar cases over and over again (different conditions, parameters), I think, he is going to go crazy at the end of the day, week or month. (7). Jonas, your experience and message can very well become another person's nightmare. (just a joke!)


All times are GMT -4. The time now is 17:33.