Learning Forum

[majam]

Hello,

I am simulating the mass transfer of O2 from air to water in an Eulerian multiphase bioreactor system. To calculate the mass transfer coefficient I need the concentration profile of o2 that is transferred from air to water over time and I would simulate that on a frozen air-water system. 
The problem I am dealing with is that the convergence of the system is dependent on lowering the under-relaxation factor (especially when the simulation is close to saturation concentration of oxygen ) and at the same time the calculated concentration at each time step heavily depends on the URF (and at the end the concentration profile affects the calculated kla value ).

Should I change the no of iterations so that it reduces the URF effect?

What is the right approach here, should I just choose the URF that fits the profile with experimental data? 

Thanks,

[Rob]

I'm not sure I understand the question. Mass transfer is determined by the model options and flow field. If the system becomes stiff near the saturation values why are you changing URF? Timestep is usually more reliable. If you do reduce URF you also need (many) more iterations to ensure the time step is converged. I assume you're also looking at monitors and not just residuals? 

[majam]

Close to the saturation if I do not decrease URF then the DO concentration will jump to values higher than the maximum DO concentration in water which is defined based on Henry's constant.

The slope of the concentration profile gives the mass transfer coefficient and as you can see it is different with 2 URFs. Fluen also gives mas transfer rate ( which varies over time due to the change in O2 concentration in liquid).

[Rob]

How good is your convergence? For transient systems we tend to alter the time step and leave UR factors alone. 

[majam]

I use a 0.5ms time step, and the convergence looks good, except in some cases close to saturation simulation diverges that's when I reduce URF bc in each time step it adds a huge value to DO and then it goes above the saturation.

[majam]

This is an example of concentration over shoot.

[Rob]

That looks weird, did the time step converge, and what are you monitoring? What value does it reach?

[majam]

I donot use too many iterations per timestep (10 to 20 iterations-simply bc it takes forever to solve) but the residuals are le-6,e-8 for both species eqs, so I assume it is converged.

[majam]

It is solving species eqs for O2 in air and in water. I am monitoring the concnetration of O2 in water. and it should at max reach to 0.008 g/l.

[Rob]

What about the other equations? 

What does the concentration reach? Does it then flatline or continue changing?

[majam]

flow continuity and turbulence is already converged and reached to Steady state then I freeze them and just solve the species eqs.

For the cases that overshoot it just keeps increasing.

[Rob]

What mass transfer settings did you use? 

[majam]

-I have O2-n2 in the air-mixture, O2(l)-water in liquid phase mixture. 

-species-mass-transfer model in the mass transfer tab,  I use Henry's law, define the Henry coefficient, use a UDF for the mass transfer model in the Liquid phase, zero resistance in the gas phase, and use a UDF for the interfacial area.

[Rob]

And what happens to the UDF maths at an O2 concentration of about 0.0025 or time of 20s? 

[majam]

the UDF values are steady and within the range. since in the UDF, I defined kl(mass transfer coefficient in liquid phase) based on energy dissipation rate and a(interfacial area) is a function of air volume fraction and constant bubble size. They look ok.  

 

Another problem is for the systems that dont overshoot and run til the steady state value, the max dissolved oxygen concentration that they reach is close to 0.0065. However, based on Henry's constant and absolute pressure it should go up to 0.008. I can not figure out why that happens

 

 

 

[Rob]

Back calculate the maths: what pressure/temperature would give the lower value? Ie is the maths correct and there's a error in the solver set up (boundary or operating condition would be my starting point) OR you've used the wrong macro for pressure.