[ncolafrau1@gmail.com]

Hello everyone,

I am trying to simulate the air conditioning of a data center. For this task I am using Simcenter Flov-nt (I am not using the full name for spam issues).
The IT room is air conditioned by 8 AHUs, simulated with 8 fixed flow inlets. This room features 10 rows of IT racks, divided into 5 groups of 2 rows each. Each IT rack contains a set of fans, that extract cool air from the internal ambient (cold aisles) and delivers it to confined ones (hot aisles). One hot aisle collects the hot air of 2 rows of rack. The hot air is then conveyed in a false ceiling located above the IT room through a rectangular continuous stack. From there, the air is expelled through the front.
An opening at the ground floor disposes of the excess air, entering through the inlets and used to create a positive pressurization in the cold aisles.

I have modeled the main parts of the systems as following:
• AHU supplies: Fixed flow rate ports; Inlets
• Racks: Racks, with a given flow rate and heat rejection
• Outlet: Free opening with air resistance
• Ground floor opening: Free opening with air resistance
• Walls: Non conductive walls

For this model I have severe problems of convergence.
After many trials (you can see the simulation resets in the attached residual plot), done tweaking various parameters such as:
• Fan Relaxation
• False time step
• Termination Residuals
• Initial iterations (I tried for both pressure and temperature)
The normalized residuals (especially X, Y, Z velocities) keep remaining very high.
Do you have any general/particular suggestions to give me? Thanks a lot in advance!