The (lack of) Prior Art on Annular (Cylindrical) Flow Dividers

Myself, using old Google Scholar, I previously struggled to find any literature that might hint at the fluid dynamics mechanism I've proposed on this blog, which I tend to call the "friction-buffers". Let me be super clear what this describes:

• Layman:
• A tube in air is rotating, and it is surrounded by cylinders of increasing size. The cylinders rotate at slower speeds, because they are dragged by air movements caused by the tube. The cylinders allow us to spin the tube more easily than if they were not there.
• Kind of fancy:
• Multi-Annulus Taylor–Couette system with freely-moving intermediate cylinders
• Another:
• Segmented annular flow with passive staggered rotation enabled by flow dividers that suppress Taylor Vortices

I want to be abundantly clear about this - I do not want this to be my original idea. That just makes it harder to defend. Over and over again I tell myself that somebody must have come up with this idea before. But if I search I come back empty handed.

Gemini Research Output

That hasn't changed. But what has changed is that we have AI now, so at least I can prove that the AI can't find comparable background literature either. In case anyone was going to ask, although I am a major AI power-user, absolutely none of what's written here is from AI. Here's output from the latest Gemini:

https://gemini.google.com/share/0c90b978750a

Let me highlight the 2 key conclusions

1. It agrees with me. It didn't even take convincing by follow-up prompts, which other AIs have.
2. It can't find background literature either

On the first point, here's an evidentiary quote

The investigation into using intermediate, free-floating concentric cylinders to reduce viscous drag and suppress turbulence in high-speed annular centrifuges confirms the viability of this hydrodynamic control technique as a compelling alternative to vacuum operation.

I want to call out the bold, and really emphasize that this goes way further with the claims than I ever would have. This is probably not useful for Uranium enrichment or any other lab centrifuge applications. If drag matters, you can probably solve it by pulling a vacuum, and if you can reasonably pull a vacuum, that is obviously superior to the method I've proposed. I wouldn't call that "compelling", but if I was in academic I would probably write that anyway... but that's just an academia thing.

In any case, if this could be useful and practical for centrifuges, it's almost guaranteed to work for the much larger and slower case of free-floating tubes in microgravity. This is a full-throated endorsement. All the AIs agree, for whatever that's worth, which isn't much.

And as for (2), I'll just have to show you what it did find.

Rotating Half-Discs Drag Reduction

The best reference by far is:

2021

Reduction of turbulent skin-friction drag by passively rotating discs

https://eprints.whiterose.ac.uk/id/eprint/209042/1/2106.12824v1.pdf

This is abundantly clear that it proposes to have discs on a surface with a fluid moving over it, where half of the disc is kept under a divider. This allows the disc surface to be closer to the velocity of the fluid as opposed to the surface. It is very intuitive how this could work with the discs being rotated passively by the fluid, just like it is intuitive that you could reduce drag by putting a treadmill on it. These are just slightly more practical variations of a passive treadmill (if I am to tell it).

You see half of the discs because the other half is covered

Ok but this is still a bit abstract without a use case. I like this sketch, because it looks like an airplane wing.

Airplane wing maybe, my own interpretation

Could this method improve the fuel economy of an airplane? Yeah, that's totally physically possible. This idea does share many properties of the friction-buffers proposed in this blog.

• Introduce a flow divider (in my case) or a skin friction attacher (their case), which is a "sheet" in all cases
• Allow that sheet to move passively, meaning, it is moved by the flow itself
• There is an expectation that the flow becomes less turbulent, and the energy lost due to viscous forces decreases

So almost-check, check, and check. The other difference we might point out is the entire geometry is different - a surface vs. annular flow.

It looks like this paper also covered the same thing

2013

Turbulent drag reduction through oscillating discs

https://www.researchgate.net/publication/258796176_Turbulent_drag_reduction_through_rotating_discs

However, that was much harder to follow because none of the pictures made it entirely clear what it was showing.

Other Close Misses

This paper describes a passive mechanism for drag reduction in Taylor-Couette flow. Seems promising!

2024

Research on the Sealing Performance of Segmented Annular Seals Based on Fluid–Solid–Thermal Coupling Model

https://discovery.researcher.life/article/research-on-the-sealing-performance-of-segmented-annular-seals-based-on-fluid-solid-thermal-coupling-model/aabc017bc11133f9a41c1f81a3ec4b33

However, that is very clear that it optimized using a groove design. And looking at the pictures further, it might not even be talking about the general topic at all.

Looking further into papers on Taylor-Couette flow is an exercise in madness. One takes two cylinders (tubes) and places them next to each other and has them spin.

I'm tempted to believe that somebody wouldn't write a paper on this friction-divider concept. Let me explain why in 2 scenarios:

• Do not include the effect of oscillations with movement of the cylinders (global stability problem), you've effectively made an undergrad-level problem, not worthy of CFD or of a paper. It's too easy and the effect is obvious
• Include the movement of cylinders, you've added mechanical boundary movement to your CFD at which point you've made a problem that's too hard and give up

So basically, to get progress, I, or you (the reader) need to do it ourselves.

CFD or Experiment

Experiment.

This brings me back to my prior dichotomy. It's either "CFD is useless" or "CFD is impossible". Also, looking towards the specific application, the predictable objection should be a tiny bit more nuanced than "you can't divide the flow in half". Problems you're most likely to hit probably won't occur until you're at a high number of sheets. Not just 1. I'm thinking 4+ sheets to really get some value out of it.

I have some material to write on global oscillations, and maybe it's correct in its approach. But there's no natural confidence in it. Actually doing the experiment, showing the friction-buffers stay in place, and demonstrating a performance gain, would get miles and miles further confirmation that eigenvalues from the coefficients that you came up with on your own. Garbage in, garbage out.