HVAC simulation gets stuck / Best Practice?

[h2rob]

I am modeling a staircase with two apartment doors on each floor. Enclosed is a clipped section of the geometry. The fire compartment is on the second floor on the right.
I modeled the leakage through the door with a "complex" HVAC, to account for the change in leakage under heat exposure. The inlets and outlets of the HVAC for the apartments on the left are included in the domain but in different meshes. The apartments on the right are not modeled; here the HVAC is connected to AMBIENT. There are also localized leakages applied to the staircase.

The doors are modeled with four VENT per side. The System is constructed like (example for one VENT):

&VENT ID='1_NE_l_1_TR', SURF_ID='HVAC', XB=4.7,4.7,1.3,2.3,1.5,2.0/  located in MESH1
&VENT ID='1_NE_l_1_NE', SURF_ID='HVAC', XB=4.9,4.9,1.3,2.3,1.5,2.0/  located in MESH2

&HVAC ID='N1_1_NE_l_1', TYPE_ID='NODE', VENT_ID='1_NE_l_1_TR', DUCT_ID='D1_1_NE_l_1'/
&HVAC ID='N2_1_NE_l_1', TYPE_ID='NODE', DUCT_ID='D1_1_NE_l_1','D2_1_NE_l_1','D3_1_NE_l_1','D4_1_NE_l_1','D5_1_NE_l_1','D6_1_NE_l_1','D7_1_NE_l_1',XYZ=4.2,0,.1/
&HVAC ID='N3_1_NE_l_1', TYPE_ID='NODE', DUCT_ID='D2_1_NE_l_1','D3_1_NE_l_1','D4_1_NE_l_1','D5_1_NE_l_1','D6_1_NE_l_1','D7_1_NE_l_1','D8_1_NE_l_1',XYZ=4.2,0,.1/
&HVAC ID='N4_1_NE_l_1', TYPE_ID='NODE', VENT_ID='1_NE_l_1_NE', DUCT_ID='D8_1_NE_l_1'/
&HVAC ID='D1_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N1_1_NE_l_1','N2_1_NE_l_1', AREA=5E-2,LOSS=1,1,LENGTH=0.01/
&HVAC ID='D2_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=2.0E-4,LOSS=1,1,LENGTH=0.01/
&HVAC ID='D3_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=2.0E-4,LOSS=1,1,DAMPER=T,CTRL_ID='1_NE_l_2',LENGTH=0.01/
&HVAC ID='D4_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=4.0E-4,LOSS=1,1,DAMPER=T,CTRL_ID='1_NE_l_3',LENGTH=0.01/
&HVAC ID='D5_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=8.0E-4,LOSS=1,1,DAMPER=T,CTRL_ID='1_NE_l_4',LENGTH=0.01/
&HVAC ID='D6_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=1.6E-3,LOSS=1,1,DAMPER=T,CTRL_ID='1_NE_l_5',LENGTH=0.01/
&HVAC ID='D7_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N2_1_NE_l_1', AREA=3.2E-3,LOSS=1,1,DAMPER=T,CTRL_ID='1_NE_l_6',LENGTH=0.01/
&HVAC ID='D8_1_NE_l_1', TYPE_ID='DUCT', NODE_ID='N3_1_NE_l_1','N4_1_NE_l_1', AREA=5E-2,LOSS=1,1,LENGTH=0.01/

With increase in HRR the CPU-time increases and the simulation gets stuck at around 400 s. The volume flow measured in DUCT 1 looks good and fits well to the pressure in front of the door.

I tried the following, but the problem is not solved yet:

• Moving the VENTs for HVAC at least one cell into the mesh, so its not position on the boundary
• &PRES VELOCITY_TOLERANCE=0.005, MAX_PRESSURE_ITERATIONS=50/ (according to the .out-file there are about 15 iterations)
• SUSPEND_PRESSURE_ITERATIONS=T
• Removing the HVAC which are connected to AMBIENT

Right now I am trying to:

• Use only one mesh for the staircase, one for the apartments and one for the fire compartment
• Reduce the complexity of the file to trace the problem
• I just saw that XYZ on the NODE-line is not correct, running again with correct positions

In the meanwhile: are there any best practice routines regarding HVAC I did not use yet?

[drjfloyd]

What do you mean by the simulation gets stuck? Does it stop making any forward progress or just very small time steps?
What version of FDS are you running?
Have you tried just having one duct at whatever the area is at 400 s? That might help point to whether or not it is an HVAC issue.

[h2rob]

The time steps are very small all of the time (somewhat around 0.xE-2), but at some point it is not moving to the next time step.
Version 6.10.0 (EDIT; I will try 6.10.1)
Yes, it seems to be working if the area is not changed within the simulation. Without any CTRL I am at 900 s right now. Looking at the devc_ctrl_log of the unsuccessful runs the simulation stops at the time step the area is increased.
I tried HVAC_PRESS_RELAX, but neither 0.9 nor 0.5 is helping.
This reduced file shows the same behaviour: HVAC_test.txt

[h2rob]

Switching to 6.10.1 solved the problem

[h2rob]

Update: Using FDS 6.10.1 worked for the reduced file. For the large file the simulation got stuck again. Instead of 8 ducts with 3 nodes I simplified the setup using only one duct and controlling the LOSS to account for a change in ductarea. It seems to be working now.