Outstanding work as usual mate.
Looking at the spiralling around cylinder number 4... Do you see if there's a pressure drop in the port/trumpet area? Would vanes in the chamber or raising the floor help counter the spiralling?
I learned a fair bit from this process, but there are also a bunch of limitations that I encountered, as well as a few things that relate to Rama's tried and tested experience.
Without bludgeoning people with Excel spreadsheets - which is fun for no one - there are a few rules I learned from reading theory, Rama's experience and from the simulations:
From Theory:
- Until you hit ~ 10 000 RPM, airboxes almost certainly improve engine performance in all cases
- Tapered designs always perform better
- 40mm of clearance from the trumpet to the airbox results in relatively equal mass flow to each cylinder (usually less than 5% variance)
In Rama's experience these points are all true, and since he is a racing engineer I will take it his word for it.
From the simulations:
- I learned that vorticity is the bane of flow rate. If significant vorticity forms around a particular cylinder its performance will drop catastrophically. This makes sense since the air will carry high kinetic energy to move 'out' rather than in (centripetal forces)
- Baffled design massively outperform non baffled
Now that said, Rama isn't keen to experiment with a baffled design and I can understand why. He is keen to use the 'best performing' non-baffled design, since it fits with his experience and the performance difference can be illustrated through the simulation. He is (understandably) concerned about any simplifications made in the simulations that don't reflect reality, as well as it being an extra cost for him to implement and then to test. He also pointed out, that it hadn't been done previously - however I believe that to be more related to the fact that high performance low RPM NA intake design were becoming redundant by the time advanced CFD was available at a price to be justified on anything other than F1. I believe that the model I made was sufficiently advanced enough to warrant physical investigation, but cost is always a prohibiting factor so I can't hold that against him. I might try and investigate it personally though.
To illustrate some of what is going on, the below video is of a standard design Rama uses that sees good performance. I ran a simulation over 24 hours and generated ~500Gb of data for that 17s clip...
https://www.youtube.com/watch?v=eLPiHUe0wx0Basically, what you're seeing is that the air loses momentum (and thus reduced velocity) by the time it reaches cylinder 4 reducing its MAF rate by ~5% compared to cylinder 2. The curved end of the plenum actually generated a small vortice as well, which contributed to the loss.
When you compared it to this video, which pulled in nearly ~10% more air overall:
https://www.youtube.com/watch?v=-WbrzmInXX0The differences are very pronounced. The first thing you should notice is there is a much higher average velocity in the chamber, and much less dead space. Almost the entire volume of the plenum is air in motion. Regarding the vortice at the end of the chamber, you will see a small baffle I placed to break it up - that on its own massively improved the airflow in every design to cylinder 4. However, one major issue is that this flies against the conventional wisdom that 40mm clearance is essential for equalised flow rates, which is why it I don't think it will get made.
For the record the 'optimal' intake plenum without baffles uses a funky shape where the intake point follows a slightly curved path to aim between cylinders 3 and 4 while sitting a bit further away from cylinders 1 and 2. I didn't make an animation of it, but it looks very similar to the model with the baffles. It comes in at a funny angle, but I will have to dig around a bit to find where the model is. When I find it again I will post it up!
I'm keen to make a full intake simulation, though I haven't yet put aside the time to do it!
