alright - so I just had a quick look, but what I can definitely say is that the operations log message is miss-leading (already known - an improved error reporting for this is underway). The problem is not that the machine is running out of memory - in fact, your core count is not efficient. With 400k cells, an 8 core machine is more than enough.
The problem seems to be numerical instability. I didn’t find anything obvious in the solver log message. So the most obvious question would be: 10m/s inlet with a 5cm geometry - quite some speed, is that intended? The next question would be how you came up with the values for k and omega? Did you use some correlation / hand calculation? That could be one reason for the instability and the next place to look would be the numerical schemes.
@gholami has got more experience with CHT. Babak, anything obvious to you?