Learning Forum

[Monta]

Hey everyone,

I am modeling a Heat Exchanger (see photo).

[Rob]

Plot contours of velocity on the mid plane (same view as the mesh) every 10-20 iterations and see what's happening starting from the last data set (eg 1010, 1020, 1030 iterations), you'll want 10-20 images. There's a bit of a jump in cell size into the shell side, but I don't think that's too much to blame.

[Monta]

I plotted th velocity contours for 1000 and 1020 iterations for Inlet & Outlet of the shell side. I see that some differences in the velocity field and a small difference between maximum velocities.


Furthermore I checked the mass imbalance for the hot flow in pipe and cold flow in shell side.
for hot water in pipe side is mass imbalance ~ 10^(-11) kg/s ( mass flow rate is 0.72 kg/s)
for cold water in shell side is mass imbalance about -0.01 kg/s (mass flow rate is 3.9 kg/s)
Now I probably know that the problem is in the shell side (cold water) but I am not quite sure what could the best step to do next in order to find out the issue! Any ideas?
Thanks for help.

[Rob]

And you have back flow on the shell side.
Going through the data:
Mass is conserved so that suggests the overall solution is more-or-less converged
Residuals have done something (I suspect they're bouncing around at 2e-3 ? )
Monitor is showing a small change
Flow shows some sign of change over a number of iterations.
What you probably have is a system with some shedding/flow separation, and it's probably in the outlet. Extend the outlet by 3-4 diameters and see how that changes the convergence.

[Monta]

I actually extended the inlet/ outlet by 10-diameters (not shown in the figures) & still get a reversed flow at the outlet of hot water. I decided now to study the shell side flow separetly & energy off. The problem converges but the residuals of continuity & epsilon don't even reach 10^(-2).

[Rob]

That's looking like it's transient. Look for flow separation in the domain (the outlet looked that way) and see if the separation & reattachment points are moving. The bulk flow is probably not changing (much) but some details are.