Severe error: Unanticipated combination of heating and cooling loads

edit

I have a large model with complex geomerty. Part of the geometry is imported from gbXML to DesignBuilder. The model without any HVAC systems, runs fine. Then when I started to put the systems in, I got this error first:

 ** Severe  ** DualSetPointWithDeadBand: Unanticipated combination of heating and cooling loads - report to EnergyPlus Development Team
   **   ~~~   ** occurs in Zone=1XBDXHEGYALLATH:SZULOBOX1X1 During Warmup & Sizing, Environment=SUMMER 
DESIGN DAY IN UNTITLED (01-06:07-06) JUL, at Simulation time=07/15 00:00 - 00:10
   **   ~~~   ** LoadToHeatingSetPoint=NAN, LoadToCoolingSetPoint=NAN
   **   ~~~   ** Zone Heating Set-point=18.00
   **   ~~~   ** Zone Cooling Set-point=100.00
   **   ~~~   ** Zone TempDepZnLd=NAN
   **   ~~~   ** Zone TempIndZnLd=NAN
   **   ~~~   ** Zone ThermostatSetPoint=0.00
   **  Fatal  ** Program terminates due to above conditions.

Then I looked here: https://github.com/NREL/EnergyPlus/is...

Probably my zones are malformed somehow. I tried to divide zones to smaller ones, change the geometry in the zone in question, no luck. I removed the HVAC systems from this zone, the error came up in an other zone.

Then I put IdealLoadsAir systems in every zone. This way the error came up like this:

** Severe  ** Temperature (low) out of bounds [-102.77] for zone="0XBDXMAJOMKIFUT:MAJOMKIFUTX0", for surface="0XBDXMAJOMKIFUT:MAJOMKIFUTX0_CEILING_1_1_2_1_2_0_HOLE"
   **   ~~~   **  During Warmup & Sizing, Environment=SUMMER DESIGN DAY IN UNTITLED (01-01:31-01) JUL, at Simulation time=07/15 12:30 - 12:31
   **   ~~~   ** Zone="0XBDXMAJOMKIFUT:MAJOMKIFUTX0", Diagnostic Details:
   **   ~~~   ** ...Internal Heat Gain [5.000] W/m2
   **   ~~~   ** ...Airflow Network Simulation: Nominal Infiltration/Ventilation/Mixing not available.
   **   ~~~   ** ...Zone is part of HVAC controlled system.

So at least I found, the problematic surface. Removed the hole (it's a horizontal door) but the error moved on to an other zone.

Then I found this: https://github.com/NREL/EnergyPlus/pu...

And my model has the same characteristics. I used displacement ventilation room air model, and it had horizontal doors in it. I changed the room air model to Mixed. Still does not work.

I thought I am close to the source of the problem, but looks like I'm looking at the wrong places. Any suggestions?

EDIT:

I managed to reproduce the error on a smaller dummy. It has simpler geometry, everything is convex and so on. The model is two zones on top of each other, and a hole (horizontal door) between them. The upper zone is made out of glass, almost entirely transparent.

C:\fakepath\dummy.PNG

And I found that the simulation runs fine in the winter, but as it's getting warmer, the super cold surface temperature error comes up again at theese horizontal door surfaces. If I recall correctly, the solar heat gain is being put in the zone floor if I use Full exterior solar distribution.

Is it possible, that this extreme high solar heat gain is someow drives wild the horizoltal door? It's made of IRTMaterial. Maybe it has wery little mass, and extreme temperatures may occour because it can heat up swiftly.

I ran the same model, with the same settings, I just brought down the transparency of the glass in the upper zone ... (more)