[kingjewel1]

Hi there,

I'm modelling a production of CO2 by the use of an additional variable and would like to use a steady state solution, as the system should be in equilibrium.The only reasonable results i've managed to produce so far are in transient form. Any ideas how I could do it in steady state?

It's CO2 source within an enclosure with velocity inlet and pressure outlet. How best would you model this given free range?
If not can you point me to a good source to read up on this as resources seem to be very scarce.

Best regards,
MFK

[ghorrocks]

Hi,

Obtaining a converged steady state solution is tricky if the time scales of the device have a wide range. For instance does the fluid turn over in a second or two, but the CO2 concentration take hours? In this case a steady state simulation will need to take careful consideration of the wide range of time scales.

Really it boils down to a convergence issue.

Glenn Horrocks

[kingjewel1]

Originally Posted by ghorrocks

Hi,

Obtaining a converged steady state solution is tricky if the time scales of the device have a wide range. For instance does the fluid turn over in a second or two, but the CO2 concentration take hours? In this case a steady state simulation will need to take careful consideration of the wide range of time scales.

Really it boils down to a convergence issue.

Glenn Horrocks

Thanks Glenn.

Complete fluid exchange is once every ten minutes. CO2 is released at a constant rate of 1.7E-6[m^-3 s^-1]. I'm not sure how to convert this to a flux. Could you point me in the right direction? In terms of mass flow rate this is 3.3E-6 [kgm^-3s^-1].

Best wishes,
MFK

[ghorrocks]

Hi,

It is not the flow rates which are important but the separation of time scales. You have said the fluid turns over in 10 minutes, but how long does the CO2 take to reach equilibrium from the starting condition? Seconds? Minutes? Days?

Assuming this is an issue then you will probably need to also include imbalances in your convergence critereon. Residuals alone may not indicate convergence.

Glenn Horrocks

[kingjewel1]

Originally Posted by ghorrocks

Hi,

It is not the flow rates which are important but the separation of time scales. You have said the fluid turns over in 10 minutes, but how long does the CO2 take to reach equilibrium from the starting condition? Seconds? Minutes? Days?

Assuming this is an issue then you will probably need to also include imbalances in your convergence critereon. Residuals alone may not indicate convergence.

Glenn Horrocks

Hi there,

I understand your point. The equilibrium is something that i'm trying to find out by doing the simulation itself. That way I can set the time set accordingly. Is there a better way to establish equilibrium?

[ghorrocks]

Hi,

But surely have you some feel for how long it will take? All we need is a guess, just to give us a guide as to whether there is a big range in time scales.

If you have no idea about the CO2 timescale then try to estimate it using a simplified analytical approach. Otherwise do a transient simulation and get it to work out the timescales but you may have to wait for weeks for it to finish.

Glenn Horrocks

[kingjewel1]

Originally Posted by ghorrocks

Hi,

But surely have you some feel for how long it will take? All we need is a guess, just to give us a guide as to whether there is a big range in time scales.

If you have no idea about the CO2 timescale then try to estimate it using a simplified analytical approach. Otherwise do a transient simulation and get it to work out the timescales but you may have to wait for weeks for it to finish.

Glenn Horrocks

Hi Glenn
The simulation is of a person breathing out a tracer (0.01kgs^-1) within a room (46m^3). Running transient isn't really an option because I don't have much time. Could you point me in the direction of calculating how long it might take for equilibrium via analytical methods or otherwise please?

[ghorrocks]

Now you have explained it better I have a better idea of what you are trying to do.

At steady state the inputs equals the outputs. So if 0.01kg/s is coming in from the person it must be going out somewhere. Work out where it goes out and what conditions are required to match the flow rates, and how long these conditions will take to be established.

An alternate approach: If the CO2 does not affect the flow field (ie the flow would be the same whether CO2 was there or not), then do a steady state model without the CO2, just the flow. This should be a straightforward steady state simulation. Then restart the simulation with the flow field as an initial guess but with the flow and turbulence equations turned off using expert parameters. As the only equation you are solving then is the tracer advection this should go really fast and you should converge it quickly.

Alternate approach 2: Even if the CO2 does affect the bulk flow field then still start with a fluid only steady state simulation. Use this simulation as the initial condition for the full simulation with the tracer activated. As an approximation of the steady state flow is being used as the initial guess it will accelerate convergence.

Alternate approach 3: In steady state simulations you don't need to use the same time steps for all equations. Put a much higher time step on the tracer equation and it will advance quicker and converge more rapidly - but make it too big and the simulation will go unstable so be careful.

Glenn Horrocks