[TRNSYS-users] Natural convection in atrium

[Arnaud Candaele via TRNSYS-users <trnsys-users@lists.onebuilding.org>]

Dear TRNSYS Users,

I'd like to model natural convection in an Atrium. I've thus a single thermal zone, divided in 3 (vertically stacked) airnodes.

Now, I know I'm supposed to use coupling between these airnodes to mimic natural convection. But I'd like to have the opportunity to open or close the top of my atrium, depending on the temperature in the highest airnode of the atrium. For that, I'd like to use TRNFLOW. I was thinking that I'd simply connect my 3 airnodes with some very large ducts, and then connect the highest airnode to an external node, in order to model the opening/closure of the top of my atrium.

If I do this TRNFLOW approach, without coupling the airnodes (since the TRNFLOW approach is supposed to replace this coupling), I got an error message for the 2 "virtual ducts" in my atrium, saying "thermal airnode - no adjacent wall/window or (regime) coupling of both airnodes are defined".

So then I activate the coupling, but with constant zero flow rate, to avoid this message. It then crashes during the run (some message involving the inversion of a singular matrix).

My question is: how are TRNFLOW and the coupling of airnodes related? What happens if I specify a coupling on one hand, and on the same time define these virtual ducts to connect my airnodes in the atrium?

And also, what would be the most elegant way to model this stratification, with the possibility of opening the top of my atrium? Should I use 3 separate zones instead of 3 airnodes in a single zone?

Thanks a lot for any suggestion!

Arnaud.