|
[Sponsors] |
May 31, 2018, 04:14 |
Transient simulation
|
#1 |
Member
Mattia
Join Date: Mar 2018
Posts: 45
Rep Power: 8 |
Good morning, is anybody able to describe to me how a transient simulation works? I understand that it takes into consideration time, meaning that it should simulate how the system evolve in a certain time interval. That said, Let us suppose that I have two fluxes at different temperatures, that mix inside a chamber. In order to have information about the pressure drops and the heat exchanged by the two I should run the simulation for as long as it is needed for the two fluxes to reach the chamber and exit from the outflows?
I'm only used to steady-state problems, and using the SS I can't get stable solutions, so I'm turning to transient. Thanks and best regards |
|
May 31, 2018, 17:07 |
|
#2 |
New Member
Julia Hartig
Join Date: Jan 2018
Posts: 14
Rep Power: 8 |
Have you looked at the Fluent documentation? It's hard (at least for me) to answer such a general question in a way that would be very meaningful to you. You might find it easiest to start from here, where the equations for basic fluid flow are introduced.
Essentially the main difference (from a governing equations standpoint) between transient and steady-state is that all of the d/dt terms from the fluid equations drop out in a steady-state solution. The way steady-state solvers iteratively solve problems is different from the way transient solvers step through time, and I would again recommend just reading the Fluent documentation on that. Can you give some more information - are the two inlet fluxes of the same material and constant with time? I think for this problem you would get the same solution if you ran the steady-state solver to convergence as if you tried a transient simulation and calculated enough time steps to the point that the solution is no longer changing with additional time steps. What residual was preventing you from getting a stable solution, continuity? It's possible that something in your setup (mesh size, under-relaxation factor, etc.) is preventing you from achieving convergence. You should be able to solve your problem using the steady-state solver. |
|
June 1, 2018, 11:00 |
|
#3 |
Member
Mattia
Join Date: Mar 2018
Posts: 45
Rep Power: 8 |
Thamks I will get a look at the documentation. Since I'm using outflows as BC, I thought that the instability was due to the fact that the outflows were too close to the turbulent region. So I extended the pipes. The two fluxes mix inside the chamber, they have different temperatures but same mass-flow
|
|
June 1, 2018, 11:22 |
|
#4 |
New Member
Julia Hartig
Join Date: Jan 2018
Posts: 14
Rep Power: 8 |
That could be an issue, you always want to make sure that the outlet zones are sufficiently down stream so the flow is fully developed.
This setup seems kind of odd to me, are you modeling a real system? I assume there are some valves inside your mixing zone, otherwise I don't know what you're expecting to happen. I've attached a picture of what it seems like you are imagining would occur in the event that there are no valves, but I don't think such a solution is possible (it would most likely violate continuity, as it involves flow diverging from the center of the box). If this is a design you've made, I would highly recommend you include both of the mass inlets at the TOP and the two outflows at the BOTTOM. Is there any reason you're using outflow conditions? These are very unstable, it's often more realistic (and easier to converge) to use a pressure outlet condition. See attached Fluent slide for more information. |
|
June 1, 2018, 11:29 |
|
#5 |
Member
Mattia
Join Date: Mar 2018
Posts: 45
Rep Power: 8 |
Actually, that is the set up. It is a real device and there are no valves. Your idea of how the fluxes should behave is correct. I'm trying to create a model using as guidelines some experimental results. I need outflows because one of the conditions was that the mass flow must be the same on each inlet and outlet.
This is my very first attempt, maybe there is a better way to model the system. |
|
June 1, 2018, 12:17 |
|
#6 |
New Member
Julia Hartig
Join Date: Jan 2018
Posts: 14
Rep Power: 8 |
Alright, well in that case I'm not sure that I'll be able to help you very much. To me this seems like a pretty weird and not very effective design for a mixing valve, but maybe there's some reason why it needs to be set up this way. I think given this design you're going to have some trouble getting your simulation to converge regardless of how you set it up, but hopefully not.
Mass should be conserved via the continuity equation, so defining mass flow inlets AND mass flow outlets everywhere is kind of over-constraining the problem (basically, it adds an extra set of restrictions that effectively stiffen your solution and make it hard for the solver to relax to equilbrium). You may want to try experimenting with the boundary conditions by, say, calculating what velocity at the inlet/outlet corresponds to the mass flow inlet/outlet you are expecting (assuming you are working with incompressible flows which it sounds like you are) and trying different combinations of these boundary conditions to see if it improves the convergence behavior. Other than that, about all you can do is play with the solver settings. |
|
June 1, 2018, 15:17 |
|
#7 |
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,674
Rep Power: 65 |
This is going down a rabbit hole. You could use a flow-split outlet to make sure the mass flow out each outlet is the same. Regardless, switching to the transient solver in hopes that it stabilizes the solution doesn't make sense. If your solution should be steady (and I think it might be) then you should be stick with the steady solver. Even if the solution is unsteady, you would see snapshots of it even using the steady solver. If you have numerical stability problems, going to a transient solver only masks the issue.
|
|
June 4, 2018, 03:51 |
|
#8 |
Member
Mattia
Join Date: Mar 2018
Posts: 45
Rep Power: 8 |
I wasn't sure of the transient method, since I 've also thought that this is a steady state problem. I guess I'll try play with the BCs and see what happens.
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Continuing transient simulation after steady simulation | billy7590 | FLUENT | 0 | September 19, 2017 08:56 |
Transient RANS Simulation | Bazinga | Main CFD Forum | 5 | June 21, 2017 11:21 |
Transient simulation diverges | Graduation | ANSYS | 0 | February 28, 2012 07:13 |
How to integrate variable in a transient simulation? | nprace | CFX | 4 | January 9, 2012 08:59 |
Beginning a Transient Simulation | Norflow | CFX | 2 | October 7, 2011 00:38 |