So I'll say this in response to those talking about the effect of convection pressures in this case.
TL;DR - It's negligible.
For those of you slightly more concerned, I'll list out much of the important information as well as some of the summaries regarding the calculations I made. If there is enough interest, I'll post all pages of calculations and corrections.
NOTES: The following was done using rudimentary thermodynamics and physics. As such, the actual figures will be... Off... for lack of a better term. However, these issues will be addressed in the "Corrections" section, that would address the differences between this and the CFD versions of the calculations.
First, the assumptions.
The fans used are the Noctua NF-P12-1300 Redux. These were chosen as a good middle of the line (and not brown) pressure optimized quiet fans. The specs as given are: 1.68 mmH2O static pressure and 0.0513 m^3/s airflow. The internal volume of the case for the purpose of calculations was 10L, or 0.010m^3. Additionally, an 8700k and RTX 2080's power consumption figures were used for the purposes of DeltaTAir, with 95W for the 8700k and 225W for the RTX 2080 (as that is what I will be putting into the case). Other component power dissipation is neglected as this is already a higher TDP than what should be reasonably seen for long duration usage.
The assumptions for actually determining methodology are as follows: DeltaKE=DeltaPE=0, the air is an ideal gas with a constant specific heat (due to low temp difference), the case has nothing in it and no radiators (this gives the greatest % difference, and is just simplest for calculating [If someone wants to give me the static pressure and airflow volume/velocity and area of fans post-radiator, I will be more than happy to revise some things]), ideal fan performance, sea level air pressure, ambient temp of 21C/~70F, and steady state operation. The internal of the case was basically treated as a solid body of constant density, like a membrane holding the air inside of it.
With that said, let's get into calculating.
Airflow and air density at 21C gave a mass flow of 0.06156 kg/s.
Using energy conservation, a DeltaTAir of 8.5089C was given.
For simplicity, I calculated the buoyant force of a body of 10L with the given DeltaTAir between 21C and 30C of 0.03924N, which is the equivalent of distributing 4g across the top of the case. Converting from Newtons to Pascals, then to mmH2O gives 0.0972mmH2O, with a pressure difference remaining of 1.5828 mmH2O.
After iteration, the total efficiency is ~91.5% in terms of pressure, not airflow, which is, by far, the more important measurement.
(NOTE: this is still a pretty unrealistic number for reasons I'll address, not to mention being solid body force analysis)
However, this is not the CFD solution, so I will explain some of the differences:
The uniform air body was treated as a uniform body, greatly increasing effect of convection forces.
Does not account for actual internal volume with components, which would decrease effect of convection currents.
Does not account for actual fan performance curve, which could increase or decrease effect of convection currents depending on the curve.
Assumes fans as solid bodies.
Fans assumed to be in free-stream. Though accounting for radiators and internal solid bodies will greatly decrease air velocity and pressure, it will also greatly decrease the influence of convection forces due to the nature of fluids.
I'm sure that I made a few small errors in places, but overall I would expect the fans to operate with 5% of each other when inverted vs. non-inverted, which amounts to 1-2 degrees Celsius.
I might be able to do CFD calculations through Solid Edge ST10, but that's free time I don't have right now to learn something I'm not entirely familiar with. I'll see what I can do starting next Thursday.
