Variable Density - Free Surface with FIDAP

April 6, 2000, 08:34

<HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=windows-1252"> <META NAME="Generator" CONTENT="Microsoft Word 97"> <TITLE>I use FIDAP 8</TITLE> <META NAME="Template" CONTENT="\\Isf-n1\data\APPS\office97\Office\html.dot"> </HEAD> <BODY LINK="#0000ff" VLINK="#800080">



I use FIDAP 8.52 to study weld pools. A weld pool is essentially a FREE-Surface system. FIDAP works quite well in almost all aspects of the phenomena involved (Marangoni, buoyancy, electromagnetic body force, phase change etc.)

But there is a problem with the volume expansion caused by the phase change and heating.

In order to illustrate the question, consider a glass of water, which is totally full. When we heat the bottom of the glass, the water expands, the deformed free surface is the result (a sketch is given below). The above situation is considered as a TRANSIENT, COMPRESSIBLE(VARIABLE-DENSITY), FREE problem, the density being a function of temperature (expressed through a volume expansion coefficient).

It seems, that FIDAP fails to represent correctly the pressure gradient caused by density variations due to the heating. Instead, zero pressure results, if the problem is specified as COMPRESSIBLE(Variable-Density) and FREE.

However, if the outlet is fixed in space (like an outlet from a pipe) and the problem is NOT specified as FREE, but is still COMPRESSIBLE, the pressure is correct, and the amount of fluid passed across the outlet in course of reaching the steady-state satisfies the mass balance within a tolerance of 0.01 % !

The segregated solver with pressure-projection algorithm and kinematic update of the free surface is used.

I guess, there is a defect in FIDAP, when COMPRESSIBLE and FREE keywords are specified simultaneously.

Any other ideas how to model the "water expanding from a glass"?

Thank a lot,

Vitaliy.

__ Sketch of the problem (the gravity is zero for simplicity) :


************wall (no-slip)

***********_______________

Time=0*:**½****************

**********½**************** <- outlet

bottom -> ½**************** (free surface)

heat flux ½**************** T=const

(constant)½_______________*

*************symmetry

Time > 0 :*_______________

**********½***************.*

**********½***************.**

**********½***************.***

**********½***************.****

**********½___________________*

Time-> ¥

(steady-state):

***********_______________

**********½***************.**

**********½***************.****

**********½***************.******

**********½***************.*******

**********½______________________*
</BODY> </HTML>

April 8, 2000, 05:33

Are you using the mixed, discontinous in the PRESSURE command or anything else? you should use it instead of the PENALTY method.

April 8, 2000, 17:46

Hello:

Free surface problems are always tricky, especially when setting up the driving force of the surface deformation. We can't establish this a defect from the above information and we will take a look at this through our regular support channels and help the user accomplish his modeling objective. A brief resolution will be posted here once the support process is complete.

Let me stress that the best place to track supsected new defects that are obviously not yet confirmed or documented in the forum at Fluent's webiste is through your support engineer.

April 9, 2000, 10:43

Yes, the pressure is MIXED=0.1E-8 and DISCONTINUOUS.

A far as I know, the PENALTY can not be used because of : 1) the segregated solver invokes MIXED automatically; 2) a free surface analysis is not recommended with PENALTY; 3) compressible flow analysis (which can be solved by FIDAP only with the segregated solver ) is also not possible with PENALTY.

April 9, 2000, 11:24

Hi Costas,

thank you for attention and responce. I will take into account your comment. Indeed, my personal assumption about defect is not yet a subject for discussion here.

Best regards, =Vitaliy=

April 11, 2000, 06:32

Hi Vitaliy, Would you please let me know the sequence you run the problem i.e. is it one run or two?? if you can send a copy from your FIPREP file that will be better. As a matter of fact my work is in the free jet impingement including free surface using FIDAP. There are so many things to be done here: 1- The gravity, body forces and surface tention value

2- The time step

3- The relaxation factors and the numbers of kinematic updates

4- You meight need to use CONTINUOUS instead of DISC

5- You meight need to solve the problem as a FIXED problem first and then use the READ RESTART to solve the FREE surface problem reading the temperatures from the previous run. These are some thoughts I'm not sure if it meight help but I have gone through all of them to solve my problem. Finally make sure that you are getting a converged solution with the segregated solver which meight take several hundreds of iterations. =Ahmed=

April 11, 2000, 09:37

Hi Ahmed,

Since the problem is transient, it is solved in a one run. Well, I solved it as FIXED and found , that the mass balance (the amount of fluid passed across the top of the glass) is satisfied. After the flow has reached steady-state, the temperature does not change any more (a constant temperature gradient was established, that's all), and the velocity is zero everywhere: no more expansion -> no more driving force for the momentum. You see, it has no sense to use such solution for RESTART with FREE surface.

The time step is small-fixed at the beginning and variable (truncation error = 1.0E-4) later on , and is mainly related to the energy equation.

The relaxation factors seems to be OK, since convergence is rather fast.

The kinematic updates (ONLY kinematic are used in a transient analysis) are = 50. But the SURFACE converges already with a few (normally 1) updates. Really, nothing is moving (to remember: only if FREE is specified), so the surface tolerance is about 1.0E-8 !!! after the first update.

CONTINUOUS ?? My previous experience tell me : better DISC. When the CONT was set in the problem under consideration, no influence on the surface behaviour was found. Instead, spurious velocities and pressure near the wall to be heated were observed.

You will find more details in the input files, that I sent to you. The test case is based on the example ex26sg.FDREAD from the FIDAP Example Manual.

I can describe it here too, if some body else has an interest.

Regards, - Vitaliy-

May 2, 2000, 16:56

We have reviewed Vitaliy's problem. This is not really a bug. It is related to the fact that FIDAP can't support the mass constraint for confined compressible flows and is written for unconfined flows (such as flow in a pipe). This feauture is not difficult to implement and Vitaliy's question has reinforced the belief that we should provide this enhancement to FIDAP in the near future.

Sincerely,

COstas Dimitropoulos

April 6, 2000, 08:34	Variable Density - Free Surface with FIDAP	#1
Vitaliy Pavlyk Guest Posts: n/a	<HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=windows-1252"> <META NAME="Generator" CONTENT="Microsoft Word 97"> <TITLE>I use FIDAP 8</TITLE> <META NAME="Template" CONTENT="\\Isf-n1\data\APPS\office97\Office\html.dot"> </HEAD> <BODY LINK="#0000ff" VLINK="#800080"> <FONT FACE="Arial"> I use FIDAP 8.52 to study weld pools. A weld pool is essentially a FREE-Surface system. FIDAP works quite well in almost all aspects of the phenomena involved (Marangoni, buoyancy, electromagnetic body force, phase change etc.) But there is a problem with the volume expansion caused by the phase change and heating. In order to illustrate the question, consider a glass of water, which is totally full. When we heat the bottom of the glass, the water expands, the deformed free surface is the result (a sketch is given below). The above situation is considered as a TRANSIENT, COMPRESSIBLE(VARIABLE-DENSITY), FREE problem, the density being a function of temperature (expressed through a volume expansion coefficient). It seems, that FIDAP fails to represent correctly the pressure gradient caused by density variations due to the heating. Instead, zero pressure results, if the problem is specified as COMPRESSIBLE(Variable-Density) <U>and</U> FREE. However, if the outlet is fixed in space (like an outlet from a pipe) and the problem is NOT specified as FREE, but is still COMPRESSIBLE, the pressure is correct, and the amount of fluid passed across the outlet in course of reaching the steady-state satisfies the mass balance within a tolerance of 0.01 % ! The segregated solver with pressure-projection algorithm and kinematic update of the free surface is used. I guess, <U>there is a defect in FIDAP</U>, when COMPRESSIBLE and FREE keywords are specified simultaneously. Any other ideas how to model the "water expanding from a glass"? Thank a lot, Vitaliy. *__ Sketch of the problem (the gravity is zero for simplicity) :* </FONT><FONT FACE="Courier New" SIZE=2> ************wall (no-slip) **********_______________ Time=0:</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>************ ******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>************ <- outlet bottom -> </FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>************ (free surface) heat flux </FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>************** T=const (constant)</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>_______________* *************symmetry Time > 0 :_______________ *******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>************. ********</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********. ********</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********.* ********</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********. *******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>___________________ Time-> <FONT FACE="Symbol">¥</FONT> (steady-state): *********_______________ ******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********. ********</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********. ******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********.** ******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>***********.*** *******</FONT><FONT FACE="Symbol" SIZE=2>½</FONT><FONT FACE="Courier New" SIZE=2>______________________ </FONT></BODY> </HTML>

April 8, 2000, 05:33	Re: Variable Density - Free Surface with FIDAP	#2
Dr. Ahmed Hassaneen Guest Posts: n/a	Are you using the mixed, discontinous in the PRESSURE command or anything else? you should use it instead of the PENALTY method.

April 8, 2000, 17:46	Re: Variable Density - Free Surface with FIDAP	#3
Costas Dimitropoulos Guest Posts: n/a	Hello: Free surface problems are always tricky, especially when setting up the driving force of the surface deformation. We can't establish this a defect from the above information and we will take a look at this through our regular support channels and help the user accomplish his modeling objective. A brief resolution will be posted here once the support process is complete. Let me stress that the best place to track supsected new defects that are obviously not yet confirmed or documented in the forum at Fluent's webiste is through your support engineer.

April 9, 2000, 10:43	Re: Variable Density - Free Surface with FIDAP	#4
Vitaliy Pavlyk Guest Posts: n/a	Yes, the pressure is MIXED=0.1E-8 and DISCONTINUOUS. A far as I know, the PENALTY can not be used because of : 1) the segregated solver invokes MIXED automatically; 2) a free surface analysis is not recommended with PENALTY; 3) compressible flow analysis (which can be solved by FIDAP only with the segregated solver ) is also not possible with PENALTY.

April 9, 2000, 11:24	Re: Variable Density - Free Surface with FIDAP	#5
Vitaliy Pavlyk Guest Posts: n/a	Hi Costas, thank you for attention and responce. I will take into account your comment. Indeed, my personal assumption about defect is not yet a subject for discussion here. Best regards, =Vitaliy=

April 11, 2000, 06:32	Re: Variable Density - Free Surface with FIDAP	#6
Ahmed Hassaneen Guest Posts: n/a	Hi Vitaliy, Would you please let me know the sequence you run the problem i.e. is it one run or two?? if you can send a copy from your FIPREP file that will be better. As a matter of fact my work is in the free jet impingement including free surface using FIDAP. There are so many things to be done here: 1- The gravity, body forces and surface tention value 2- The time step 3- The relaxation factors and the numbers of kinematic updates 4- You meight need to use CONTINUOUS instead of DISC 5- You meight need to solve the problem as a FIXED problem first and then use the READ RESTART to solve the FREE surface problem reading the temperatures from the previous run. These are some thoughts I'm not sure if it meight help but I have gone through all of them to solve my problem. Finally make sure that you are getting a converged solution with the segregated solver which meight take several hundreds of iterations. =Ahmed=

April 11, 2000, 09:37	Re: Variable Density - Free Surface with FIDAP	#7
Vitaliy Pavlyk Guest Posts: n/a	Hi Ahmed, Since the problem is transient, it is solved in a one run. Well, I solved it as FIXED and found , that the mass balance (the amount of fluid passed across the top of the glass) is satisfied. After the flow has reached steady-state, the temperature does not change any more (a constant temperature gradient was established, that's all), and the velocity is zero everywhere: no more expansion -> no more driving force for the momentum. You see, it has no sense to use such solution for RESTART with FREE surface. The time step is small-fixed at the beginning and variable (truncation error = 1.0E-4) later on , and is mainly related to the energy equation. The relaxation factors seems to be OK, since convergence is rather fast. The kinematic updates (ONLY kinematic are used in a transient analysis) are = 50. But the SURFACE converges already with a few (normally 1) updates. Really, nothing is moving (to remember: only if FREE is specified), so the surface tolerance is about 1.0E-8 !!! after the first update. CONTINUOUS ?? My previous experience tell me : better DISC. When the CONT was set in the problem under consideration, no influence on the surface behaviour was found. Instead, spurious velocities and pressure near the wall to be heated were observed. You will find more details in the input files, that I sent to you. The test case is based on the example ex26sg.FDREAD from the FIDAP Example Manual. I can describe it here too, if some body else has an interest. Regards, - Vitaliy-

May 2, 2000, 16:56	Re: Variable Density - Free Surface with FIDAP	#8
Costas Dimitropoulos Guest Posts: n/a	We have reviewed Vitaliy's problem. This is not really a bug. It is related to the fact that FIDAP can't support the mass constraint for confined compressible flows and is written for unconfined flows (such as flow in a pipe). This feauture is not difficult to implement and Vitaliy's question has reinforced the belief that we should provide this enhancement to FIDAP in the near future. Sincerely, COstas Dimitropoulos

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Bubblze induced turbulence at a free surface ????	thomas	Main CFD Forum	6	April 21, 2012 10:48
CFX convergence issues with free surface	adenlan	CFX	3	September 2, 2011 07:43
gravitational force for free surface flow	Jongtae Kim	Main CFD Forum	1	July 2, 2000 12:57
Any Body got a Free Surface Solution in FIDAP	Dr. Ahmed Hassaneen	Main CFD Forum	2	November 8, 1999 15:59
Radial jet with free surface using FIDAP	Dr. Ahmed Hassaneen	Main CFD Forum	0	October 28, 1999 14:20