periodic, unsteady, non-constant mass flow rate
 

Hello.

In a periodic, unsteay, 3-D flow simulation, I'd like to set a non-constant mass flow rate condition varying with time, for example;

Mass flow rate at periodic boundaries = function of time = m * Sin(a*t)

Would you please send me the method?

Thank you.

Lee

Hi,

I'm interested to know whether you found a way to do this. I want to do something very similar.

Hello,

I'm also very interested if anyone could help us with this question.

Thanks!

Dear friends,

If you cannot hook any UDF for this purpose, you can write a journal file for that.

Bests,

Thanks for the response. But then how exactly do you read in the journal file and where?

First you have to prepare your desired journal file, then you can easily read it in FLUENT (file->read->journal)

Bests,

Hi Amir,

I will write down exactly what I'm doing:

I have a 2D channel. I want to vary the velocity with time in this channel. I can do this by having a very long channel and a profile file which I read in. The profile file contains this:

((rampup transient 3 0 0)
(time
0.000000e+000 5.000000e+000 1.000000e+001)
(u
3.524500e-002 2.014000e-001 2.014000e-001)
)

Once that is read in, I can go to my velocity inlet boundary and pick this profile instead of 'constant'. That works fine.

However, I don't want to use such a long channel. I can get fully developed flow by using a short section of channel and using periodic boundary at inlet (mesh/modify-zones/make-periodic). Once I do this my inlet velocty boundary only lets me input a constant value of mass flow - i.e. I can't read in my profile as I did before.

If I can do this with a journal file, then where do I specify the journal file to give me the velocity variation? There seems to be no option in the velocity boundary once it's periodic. Also, if that is possible, then what is the format of the journal file?

I would appreciate more information.

Many thanks,
C

Hi,
The journal files help you to automate manual procedures. Consider you've set a value for mass flow rate and after one time step you've change the mass flow rate and run for another time step. you can automate this procedure with a journal file. You can generate a journal file in two way: 1) TUI commands 2) FLUENT journal writing
Finally, you'll need to extend journal file for your time zone with a simple fortran code ... (for more info refer to manual)

Bests,

hi,
not a good solution!
if you want to have a fully developed flow inside a duct or pipe and you want to model short length, forget about periodic BCs and simply set inlet and outlet as pressure BCs. then by a udf you can change pressure difference between inlet and outlet and then massflowrate would change. I have done it in another software(OpenFOAM) and the results were exactly the same as analytic solution.
be aware that pressure inlet sets total pressure and pressure outlet static pressure.
yours,
mohammad

Dear Mohammad,
In OpenFOAM, you have to set boundary conditions for both velocity and pressure, so you've set zeroGradient for velocity @ inlet and outlet and implicitly generate a periodic zone, am I right? But here, there is no guarantee for velocity gradients @ pressure boundary conditions and some deviations may exist even for short pipes.

Bests,

hi,
no. I think using pressure BC in fluent is similar to fixedValue of p and zeroGradient of U in OpenFOAM. if I remember correctly, one of my friend do exactly the same problem in fluent. and another one solved this in cfx. all used pressure-pressure BC and got the answer.
yours,
mohammad

So, why did you talk about short section?! Or, why ability of translational periodic boundary condition with pressure loss is added to FLUENT? (As what you said, it can be easily handled via pressure-pressure BC)

consider a duct and some small boxes inside it. boxes are at the centre and they are placed in the duct with some space between each other(spaces are even). then you can run the flow with a periodic BC in fluent with just solving for a piece of duct containing one box. in this case you have the wake of the box in velocity at the inlet as well as outlet(as you set peridic BC). but if you forget periodic and simply set pressure BC, you have a fully developed flow at the inlet section with no wake effect in velocity. but you have a wake in outlet and as inlet and outlet were not set periodic, this change in velocity profile doesn't change the velocity profile at the inlet.
this is an example of why fluent added periodic BC!!

Dear Mohammad,

That's exactly what I was trying to say! In your example, consider a special case; if we consider a control volume in which the block is located in the middle and consequently, inlet and outlet are in the middle of two blocks, it makes sense that as a result of periodic velocity profile, normal velocity gradients are zero on both inlet and outlet, we also know the pressure difference; here we've set zeroGradient for velocity and fixedValue for pressure @ inlet and outlet which seams proper to use pressure-pressure BC upon your definitions, how do you justify this? :confused:

Bests,

hi,
I didnt understand your answer very well, but I think for any case setting pressure-pressure BC (which means zero Gradient of velocities) results a fully developed velocity profile at inlet. but setting periodic BC with pressure gradient results exactly the same inlet velocity profile as the outlet profile. this is the differernce of these two options. simply you can create a simple model and run it in two ways and see the differences in results. even if you want to solve this example in OpenFOAM setting pressure fixedValue and velocity zeroGradient results false results. you must use directMapVelocity or cyclic+fan BC to achieve correct results(periodic results)
yours,

I elaborate it now!
In a control volume in which block is located in a middle, velocity gradients are zero @ both inlet and outlet and we reach closure for defining boundary condition (both velocity and pressure) and we don't need extra BCs, so why pressure-pressure BC cannot lead to physical results; that's exactly the same case you defined for pressure boundary condition!

hi,
no. for the control volume you specified, if the space between box and inlet,outlet boundaries is large enough, both pressure-pressure and periodic results the same. but if the space is not so large, which causes a wake effect on outlet, defining pressure boundary condition is false because the pressure has some gradients at the outlet because of box wake. if you ignore these pressure gradient, you've accepted a simple fully developed flow and ignored the box wake effect on boundaries. the same is right about the inlet and pressure gradient on it because of box stagnation point. although wake effect is more effective relative to stagnation effect.
yours,,

I guess you mean pressure variation @ inlet and outlet plane not normal to them because obviously we have normal pressure gradient in fully developed flow. Upon your definition of pressure BC, it should be lead to physical result if we have zero normal gradient of velocity and if we can set a pressure value over that (negligible in-plane pressure variation) which both can be seen even in large wake.(with approximation)

Bests,

hi,
if you neglect pressure variation in plane of boundary, writing bernouli+headloss equation you see that for these two streamlines all variables are the same:
1-streamline starting from inlet and ending at outlet. start point and end point are both far from box location and both are far from the wake.
2-streamline starting from inlet and ending at outlet. start point and end point are both near the centre of duct. end point is in the wake of box.
for both, pressure diff. is the same. headloss is the same. so velocities should be the the same. as equations dont have anything to set at inlet for velocity but a simple fully developed flow profile, the this type of boundary forces outlet velocity to be the same as inlet, afully developed flow with no effect of wake. so by setting pressure-pressure bc you force the wake to end before the outlet. and this is wrong!
but if you specify periodic, these two streamlines have all variables the same and similarly the velocities should be the same. but now the software places velocity profile of outlet at the inlet and continue the solution. so instead of changing outlet velocity according to inlet velocity(in pressure-pressure case), the periodic case would change the inlet velocity according to outlet velocity. which is the thing we want and THIS is physical.
yours,

Dear Mohammad,
I'll be glad if you can elaborate these points which may lead to clear this discussion:
As you know, each zero velocity gradient is not necessarily fully developed condition which doesn't have wake effect!!! If you think that pressure BCs necessarily lead to fully developed flow, so they have restrictions more than zero normal gradient, is it the case?
I'm not sure about the procedure which FLUENT uses in periodic zone calculation but I think it's better to focus on this question:
As you said, @ pressure BCs, normal velocity gradient is zero; fully developed flow is just an example but not all the cases! Let's move backward; we know the solution; we have a periodic flow which velocity gradient are zero @ inlet and outlet so it's compatible with pressure BC; otherwise we've reached two different results for a same case and BCs. (This doubt can be removed if we accept that pressure BC is not just a fixedValue @ pressure and zeroGradient @ velocity)

Bests,

hi,
consider the inlet thread of cells. we have zero gradient of velocity on it and constant pressure in its plane. so tell me what is the difference between this thread of cells and a duct of the same length having fully developed flow in both inside it? both have zero normal gradient velocity and constant pressure in plane normal to axis. so fixedvalue for pressure and zerogradient for velocity at inlets leads to fully developed flow for internal flows.
having zero normal gradient of velocity cannot lead to pressure bc, but the only thing that can lead to pressure bc is constant pressure without any variations in plane. so with considering these variations, we cant use pressure bc and so there would be no doubt about two results with one problem.
yours,

Ok, so we cannot use pressure BC to generate fully developed flow. (I checked that)

Thanks for changing your idea! :D:cool:

hi,
which idea do you mean that I changed?
I told that for a fully developed flow you can use pressure BC and I persist on it. as I said the relation between first thread of inlet and a duct in physics, setting pressure-pressure BC results fully developed flow at inlet. think about that again and you will accept my idea.
the main reason that I recommend pressure-pressure BC for simple fully developed flow, is you can use udf in it and the other reason is in this example, the periodic BC would result a false velocity and pressure profile after too many iterations(also it may result good profiles in some few iterations but after too many iterations it would result false profiles) because the errors that must exit the domain from outlet, would enter the domain from inlet again and it causes a main error.
I persist that pressure-pressure can be used in fully developed flow in a duct.
and I persist that presure-pressure cannot be used for flow in a duct containing boxes. but periodic could be use.
yours,

Dear Mohammad,
you can easily check it in FLUENT!!

Bests,

Dear Mohammad,
we can analytically show that pressure BC is not appropriate for fully developed flow because of pressure inlet BC issue. As you know, there shouldn't be any in-plane pressure variation @ inlet and outlet boundaries in fully developed flow. We don't have any problem with pressure outlet which needs static pressure but @ inlet if we set constant total pressure, static pressure varies in-plane which violate fully developed flow!
If you insist on using pressure BC, you'll need an extra UDF which can set a variable total pressure as a function of velocity in order to achieve constant static pressure. ;)

Bests,

لطفا با هم دوست باشید بچه ها!

hi,
to Amir:
I didnt say you can use "pressure inlet". I said you can use pressure boundary condition for inlet. so perhaps using pressure outlet for both inlet and outlet results good. but if pressure outlet for inlet boundary results false, you may use udf to get the velocity and by adding dynamic pressure to static pressure, using "pressure inlet" would be a good idea.
BTW I'm glad to see that you've changed your idea and accepted the correctness of pressure-pressure boundary for fully developed flows. ;)
yours,

and to momech:
بچه وقتی دو تا مهندس با هم بحث میکنن یک بچه نمیپره وسط!!

Dear Mohammad,

We have a general discussion on appropriateness of two kind of boundary conditions which is available in FLUENT; if we use OpenFOAM, I'll completely agree with you that pressure-pressure BC is proper for fully developed flow because we can set static pressure, but in FLUENT it's not; if you want to write a UDF to use your favorable pressure BC, I can also write another UDF which can generate fully developed flow without restrictions of pressure or periodic BCs! (piece of propaganda for myself :D)
In contrast of you, I don't used to insist on my wrong suggestion but here, your suggestion doesn't make sense. :D So in FLUENT the best choice would be PERIODIC BC and not PRESSURE one. :cool:
It's better to refer to manual(7.15):
"The second type of periodic BC allows a pressure drop to occur across translationally periodic boundaries, enabling you to model fully-developed periodic flow"
Now it's completely clear and other friends who visit this thread can simply realize who is wrong. :D

محمد نظرت چیه این حرکت محسن رو به عنوان یک پیام بازرگانی در نظر بگیریم؟ اینجوری خیلی هم شخصیتش خرد نمیشه

hi,
OK, lets have a conclusion. first of all a problem was defined as a pipe and a periodic flow in it. and you recommend writing a journal and solve the problem for many many flowrates using periodic BC and write each case and data. so having huge space filled with cases and datas just for a very simple problem. then I recommend to change periodic boundary to pressure-pressure boundary(I mentioned "pressure inlet" at nowhere, I just said pressure. even I warn about total pressure in pressure inlet!). then you disagree my recommend because you thought pressure BC doesnt mean something like zeroGradient velocity in OpenFOAM. now you tell that the only problem is that the "pressure inlet" gets total pressure not static pressure(what I have told in first post of my own. it seems you reach this very late!!! :) :D ). Ok it seems a victory for me and it is a good victory for start!!
then you asked about why fluent gives an option about periodic BC and you thought that you can ignore pressure variation of wake in plane of outlet! now it seems that you agree with me about the sentence I told:"if you ignore these pressure gradient, you've accepted a simple fully developed flow and ignored the box wake effect on boundaries." this can be the second victory!! :) :D
then you told about total pressure and static pressure and when I told that you can use a simple udf, strangely you said "boundary conditions which is available in FLUENT"!!! I must remind you that udf is one of fluent tools, not one of openfoam tools, not one of cfx tools, and even not one of jet-audio tools!! udf is available in fluent and anyone can use it easily! the first problem was a periodic flow in a pipe. if you dont want to use any udf, how can you solve the problem? with journal?!! then I would say that journal is not a thing that is available in fluent!!! (your third defeat?? :D )
I think this is enough for me and ac2011 can choose between my way with a simple udf and and your way with lots of gigabytes of hard disk and a very very very hard post processing.

and about momech:
عامو محسن کیف کردی؟ مدل احمدی نژادی کوبوندمشا! همون که خیلی دوستش داری!!

محمد خون خودت کثیف نکن! ایی فک کرده اینجا مثل فروم اپن فم هست که هرچی غلط غلوط هم بگه هیچکی نیس
جوابش بده، امیر تا حالا شرایط مرزی فلوینت رو دیدی یا فقط شنیدی!؟ ا
:d

Dear Mohammad,
I'm very sorry because it seems that you got angry of this discussion and you don't want to apologize; just try, it's not very difficult. :cool:
And about this discussion, because I know you, that it's very difficult for you to change your idea, I tried to help you find your mistakes yourself; it's completely evident in one of your posts that you don't even know about normal pressure gradient in fully developed flow :eek: and after I lead you to the main idea; in-plane pressure variation; you tried to justify your wrong suggestion but you didn't have any success! :D
And in first post, you just wanted to warn about difference between total and static pressure value and you didn't aware of in-plane pressure variation!!! :D (If you were aware of this, you had mentioned about a UDF)
Finally,I tried to have a general comparison between these 2 BCs in steady state flow which your suggestion couldn't handle such simple case. But I should say that in unsteady flow, if we use journal file, it wouldn't need any huge space, we can save data in any time step we want in journal file!! it seems that you don't know journal file well either. :)
UDF is an auxiliary program but journal is not a program it's used just for automation! So they are very different for amateurs.
Final suggestion: I can continue this discussion if you think you need it and you can learn more, but if you struggle more, apologizing seems more difficult, I though that this discussion is scientific so I didn't count your defeats! because it would need more than one post! :D
And about post processing which you said it's very hard!!!! It's very simple; if you have a question about that, you can ask.

محسن! من فقط داشتم سعی می کدم که تو رو از یک کارگر افغانی که دسته بیلش شکسته به یک پیام بازرگانی ارتقاء بدم ولی می بینم که لیاقتشو نداشتی. به هر حال اینجا ما کاری واست نمی تونیم انجام بدیم باید بری ابزار فروشی! :D

سی ای بچو که از صب تا حالا داشته زور میزده که جواب انگلیسی بده؛ بترس از خشم محمد، (به اندازه موهای ریشش که ازموهای کله ی منو تو بیشتره ران گرفته!) میاد یه لقمت میکنه ها از ما گفتن بود بعد گریه نکنی.

@ محمد
کیف کردی؟ تریپ مشایی ازت حمایت کردما برو حالش ببر!!:d

Dear friends
i do not know why you guys make it too complicated, this problem has a very simple solution , you just need to be familiar with AVL Fire v8.3

all the best,

Mohsen

hi,
I like this post!!! :) :)
and about Amir, it seems that he doesnt want to accept his mistakes. however I even used analytical equations to show him his mistake, but he is grouching about this. so I think its better to finish this discussion and let the thread starter to choose between udf and journal!!
and about Amir's last post, he said some jokes about journal and some other things. it seems he never learns correct solution methods!!
after all I should say that the first problem in this thread is one of "viscous flow theory" coarse project and I and my friends have done this. I did it by openfoam and other students with fluent.
and the last thing I'm so curious about the easiness of post processing of Amir method. can you tell me how you can plot a velocity vs. time in some points in pipe by your method? it is one of two main results of this problem in analytic solution. if you tell me more about that, then I WOULD LEARN IT AND I WOULD ALSO THANK YOU! I know you have learned journal writing and you like using it anywhere you can, even if it is not correct!

امیر برو عامو برای چی چی مردم رو میذاری سر کار؟ میخوای با جورنال بشینه ران بگیره؟
اگه به نصایح من گوش نمیدی خوب حداقل به نصایح محسن گوش بده. بره با فایر ران بگیره بهتره تا با جورنال که!!!

yours,

First of all, I'll try to answer to the post which may have some valuable information, so I discard momech posts. :rolleyes:
Good idea, it seems that you need more time to understand my suggestion because you don't have enough information about journal file and its capabilities.
It's better; just ask what ever you don't know before talking about! ;)
What you want doesn't need any special thing, you can simply use solve->monitor->surface and prepare your files, during run of journal file, these files are also generated!!! Just think that a guy run a case time step by time step! post processing and other thing is not different. (Journal just automate what we want to do)
But I also want to answer in another way, it's very easy and I've done several times even for moving points in dynamic mesh. First create desired points before using journal file. As you know, when we create virtual geometries (here point) you can find its ID in (surface->manage) you'll need that for post processing; then in report panel you can write any variable you want (for point you can use vertex minimum or maximum doesn't have any difference) you have to save results in a file and in next time steps you can simply add data to that file.
I prefer not to discuss about capabilities of a feature before I know about it. :cool: If you've used journal file for different cases; you would agree with me.
محمد جان این اصلا چیز سختی نیست. از تو تعجب می کنم که فکر می کنی استفاده ازش مشکله! در ضمن اینجا فروم فلونته و افرادی که میان اینجا می خواهند که مشکلشون رو با این نرم افزاری که احیانا کلی واسش هزینه کردن حل کنن!

hi,
you said sth about journal capabilities and then use a simple monitor! I think its a kind of cheat!
and its time to end.....

آخه امیر این چه لجیه؟ برای چی با جورنال اخه؟ وقتی با فشار میشه، وقتی حتی اگه نخوای از فشار استفاده کنی میشه از massflowinlet استفاده بشه، استفاده از جورنال چه توجیهی داره؟
تازه من اینو چک نکردم ببینم با جورنال میشه یا نه؟ اخه باید طرف تابع بنویسه. تو جورنال تابع مثلا سینوسی بنویسه؟
کلا راه منطقی ای نیست..

Dear mohammad,
What cheat, look, journal file is not a special thing!!
In another words, think you want to solve this case one time step (1TS) then you change mass flow rate and run another 1TS and so on. We just want to automate this procedure. During this solution, every post-processing file which defined are also generated (I checked that).
Be more unprejudiced; we both use UDF for years and also have experience of using journals, on my opinion, writing a journal is much easier even for me who is familiar with UDF!!!
Additionally, we don't need to write something like sin(t) in journal! You can simply write a simple code (at most 7-8 lines) and generate a journal file with any resolution. I can explain more if it's necessary.

Bests,

Dear mohammad,

Regardless of these discussions which I think that can help other friends, I prefer to recommend easiest procedures because not all of the users are familiar with C programming and all other UDF issues and I'm sure you conform this.
I didn't change my idea because it's completely true and it doesn't violate analytic results (periodic with journal), but if we accept that you were aware of in-plane pressure variation issue @ inlet and using UDF to handle that, you have to announce @ first because other users of this forum may be less experienced than you. So I want you to be more careful in replying posts.

Bests,