[esp-r] Simulation breaks in combined plant and air flow network

[Christian Anker Hviid]

Hello ESP-r community
 
I am doing combined simulations of plant and air flow network. I am only using the stack and windpressure as driving forces and controlling the flow with a flow corrector and a proportional control law. The simulation often breaks with the error:
 
cmaslv: forward reduction phase
forward error: zero diagonal in row xx
 
STOP LUDCMS: singular matrix! Statement executed
 
I am guessing the error arises when two things coincide, the controller has closed the damper too much and driving pressure is too small, i.e. the mass flow is close to zero. I have not encountered the same error for pure mfs simulations. 
 
Has any had the same problems and is there a remedy that does not include altering climate data? (increasing wind speed actually does help). Switching matrix solution methods from gaussian to LU decomposition does not help.
 
 
 
Best regards
 
Christian Anker Hviid
Industrial PhD candidate
 
DTU Byg
Technical University of Denmark
Brovej, building 118
DK-2800 Lyngby
Phone DTU: +45 4525 1886
Phone Alectia: +45 8819 1365
Mail: cah at byg.dtu.dk