Heat exchanger storage tank problem

[Poofij]

Hello,

I would like to ask for help with set up simulation of heat exchanger. Its storage tank with stainless steel coil inside.  My main problem is validation of that project becouse i constantly get too high temperature of warter inside tank in compare to laboratory results. I dont know where i made a mistake, is it in boundry condition or in other settings?. In my simulation i run transient solution mode with time step 0.5 and inner iterations 1 (maybe its too inaccurate but i supose if i increase time step i will get even higher temperature), turbulence k-epsilon and advection ADV2 (i have tried both ADV2 and ADV5 but there was no big difference). I set water and ethylene glycol as variable materials and i create air surrounding tank. In boundry condition i set:

-temperature at air surface;

-"volume flow rate" and transient temperature at inlet of coil;

-"unknown" at outlet of coil;

-pressure and transient temperature at inlet of water tank (at bottom);

-transient "volume flow rate" at outlet of tank (i set negative values to imitate water consumption at particular time).

I sent cfz file and picture where i signed conditions.

Pease, if anyone could help me i will be very grateful.

Thank you.

[Jon.Wilde]

Hi,

I would avoid using an 'unknown' condition. Just use a P=0.

It is better to use P=0 at all outlets and flow rates and temps at the inlets if you can.

Try a smaller timestep (maybe 0.25 or 0.1) and try 3 inner iterations.

Ensure you have sufficient mesh - it looks like it might be too coarse. Ideally we would want 2 elements through the thickness of the solid coil.

ADV5 will likely be better here.

I hope that helps,

Jon

[Poofij]

Thank you very much for answer,

In my case i have to simulate about 12 hours realtime Tank work and unfortunately at 0.5 time step and 1 inner iteration it took about 6 days, is it normal that it is so long or its my computer/net fault? Maybe i should try to run only half of my model and use "slip/symmetry" boudry condition, do you think it may work?

And i have last question, wouldn't it be better to use SST k-omega turbulence model in this case or it doesnt really matter?

Thanks for help.

[Jon.Wilde]

Hi,

The only way to know is to try it. Run for a short time and see if there is a difference in results. Long runtimes when we are trying to capture so much real-time information are common, the key is to use this to find the optimum choices so that youcan run faster as you move forwards. (A better PC always helps of course!).

Use 3 inner iterations either way - this lets the solver better converge on each timestep before moving on.

Yes, SST might predict the heat transfer better.

If you can run a 1/2 model then definitely do - I was not sure as it looked like a coil.

Kind regards,

Jon

[Poofij]

Thank You for help, i'm appreciate.