SimScale CAE Forum

'Front to back' bypass channel for drag reduction (air dumping from high to low pressure zone)

A bit of a preface:

I’m a dabbling ultracyclist and, ahem, ‘bicycle inventor’, not an engineer per se, so I could use some sanity checks.
I’m desining a fully faired ‘utilitatrian’ recumbent bicycle - not for setting records, but for ultracycling events (200+ km, and I really want to attend 1200km PBP next year) and just having fun.
I have very severe size constraints however - it must fit into a cargo elevator 2200mm by 1400mm with about 1m wide door, so a velomobile is out of a question and simply slapping together an airofoil shape to fit my (considerable, unfortunately) girth is not an option as well, and making a very blunt nose and severely cammed tail negate much of streamlining benefits, so I need to get creative

I’ve read in an article on bicycle aerodynamcs that channeling air from high-pressure zone to low-pressure zones by a hose should result in a net benefit… so theoretically. making ‘scoop’ in front of the bike fairing, running the air thought a frame tube (2" id) and dumping it out into separation zone in cammed tail should result in less of an aero penalty. (This concept is also used in ‘boundary air blowing’, where such air channeling into a zone of separation in an airfoil is supposed to eliminate it entirely).

I’ve also read that about an experiment in aero tube with a hole in a sphere center that resulted in half the drag… unfortunately, there were no concerete numbers or even desription of exprimental conditions (Hole size? Re?)

So, I’ve tried getting my fairing model and running a tube throught it:

I’ve made a few experiments varying size of scoop, ‘exahaust’, center pipe width… got slightly MORE drag in all cases compared to a simple kamm tail :(.

I’ve tried to replicate a ‘sphere with a hole’ setup and indeed got a bit less drag, but much less than halfing it!

So, what gives? I don’t want to waste my efforts to make my bicycle slower, but I’m not sure that using ‘default settings’ is an optiomal way to solve that case, yet I have no idea where even to begin to start changing the conditions.

Can anyone provide some pointers?

1 Like

Hi @OTsyganov, this is actually timed well as I have done alot of work recently validating drag results, here is a good write up for a NACA 0012, indeed you are correct, default numerics are not sufficient for drag:

'NACA 0012 Verification' simulation project by dlynch

There is a heading in the write up about ‘Numerics’ read that first, also I have found best results for K-omega SST when Y+ approaches 1. I reckon both of these will solve your problem.


Edit: Also see how it is effected by mesh finness, in the independence study.

1 Like

“here is a heading in the write up about ‘Numerics’ read that first”
Erm, I’m not sure I understood what you mean by that (my English is not perfect). Can you provide a link?
I’m trying to wrap my mind around Y+ concept for now… it is concerned with optimal mesh density around body for proper boundary layer resolution?

Actually, it could be my English that was bad here :slight_smile: what I meant was that in the link provided, there is a section named ‘numerics’ where there are details about the specific numerics you may have to change to get good drag values.

The Y+ is a dimensionless distance however, we usually refer to Y+ in CFD as the dimensionless distance of the first cell in a layer between it and the boundary. There are requirements for each turbulence model for a first cell Y+, in short, this has to be obeyed to get correct wall modelling. You will need to run a simulation, with the Y+ result control enable, and visualise the Y+ on the aerodynamic bodies surface. Then you can see what the first cell Y+ is like. If too high simply make sure that the mesh layers are finner.

Please let me know if I can explain any more, it’s quite complex.


Yea, to say the least! I seems I need a few years of education to understand underlying mathematics… I guess so far I’ll make do with doing what I was told, I do not intend to become an expert in aerodynamics, my goals are more pedestrian (in spite of me being a cyclist, ehehe) :slight_smile:
I realise that I’m mostly wasting your time, but I’d be very grateful for help.

I see that I should change " Bounded Gauss Upwind to Gauss linearUpwind… all instances of it in divergence section? Which ‘Gauss linearUpwind’, there are a few variances of those?

Unfortunately, trying to understand their exact function runs into a wall of numbers and I, while am more of a ‘tech’ guy by trade, I was never any good at math and prefer biology and philosphy to be frank…

Also, should I manipulate initial conditions as well?

@OTsyganov, actually, it would be much better if we could provide software and resources that cater for everyone to understand, slowly the industry is getting there but for now, it can be a bit painful for beginners, I was there a few years ago, and these forums taught me most of what I needed, so your doing the right thing!

Specifically the divergence scheme for velocity. leave all else as default, if in doubt, there is a project link above that post, you could open it and just duplicate the numerical schemes for say the 10 degrees setup, that would give a good starting point. The joys of having public projects are that you can simply copy a reliable setup and use it as a starting point., particularly validated ones.

You shouldn’t need to, it is well advised but if you have ‘initialise with the potential flow’ checked in the simulation control, you actually don’t need to worry about initial conditions (it is initialised with results from another more basic code).

Keep the questions coming, it helps us understand where our knowledge base lacks.


Edit: Sorry for the confusion, I used the wrong account.