Applied aero for velomobile design, analysis, experiment.

V

viznz2

Beiträge
110
Firstly, lets blow in a new age of enlightenment where only knowledge is useful. Facts, methodologies, deeply considered insights, all good. Emotional reactions, dismissive comments, presenting something that is particular/peculiar or half understood as universal principal, all bad.
1662293285399.jpeg
My son announcing the new beginning. Looking like he missed the call sheet for day 1 on the Gladiator movie set.

So please, only replies that fit the title. If we keep the thread clean, it could be a good place to share ideas, reading, references and research. If someone has a strong negative thesis such as aero being non useful, they should start their own thread.

It may help if I state that I am interested in opportunities inherent in the aero and I'm not aiming at any particular category of velomobile. It may be that the result is only useful for high speed or racing "velomobiles"....

Separately, I'm hoping that Michael (mrue) will start a CFD version of this thread. CFD is such a rabbit hole (as in Alice in Wonderland) that it needs it's own place.

And so, the fun begins..(trumpet noises...)
Can we learn something from the drag values of the sailplane fuselage..?

For now, ignore the long tail boom. It's normally included, so it will add an increment to drag.

The modern sailplane fuselage, the pod, has a fairly simple shape that takes the boundary layer (BL) through a single cycle from laminar, to a transition zone to turbulent, then a pressure recovery zone, before arriving on the tail boom. This simple unitary pod shape can shift the transition point (laminar/turblent) back. It has very low drag values, maybe close to Cd=0.03 at its best (reference area is body section).

Should be noted that the wing roots fair into the rear of the sailplane fuselage pod and do increase drag. Minimized in later iterations, but still there.

If we ignore for the moment the proximity to the ground. Can the velomobile body package the rider, his pedaling legs and the wheels, give the rider comfort, freedom to move, see, enjoy and stay cool....but aim in the direction of these low drag values by managing the pressure distribution to maximize laminar flow.

This is a very open question and some will not like it. Many are committed to the idea of shrink wrapping the surfaces to achieve low frontal area. Even on an intuitive level one should see that this is problematic for the boundary layer.

Back to the sailplane fuselage drag. I looked for something recent. This is an interesting reference and I tried to be careful extracting the data.
"Design of a Flapped Laminar Airfoil for High Performance Sailplane"
Krzysztof Kubrynski...
1662293622594.png
Graph shows Cd contributions at varying CL. I thought this might be data from analysis on the LS8 or ASW27 or the Diana. He's studied the Diana before so I went with that.

I took the fuse Cd values graphically from the Table 1 graph. Error may be 5% on the smallest values.

Assume rho=1,225 kg/m^3
Assume graph is for the Diana sailplane, at AUW=empty+90kg pilot, so 2720N. Diana wing area Sw=8.16m^2
Assume all Cd contributions are non-dimensionalised with reference area =Sw (Diana)
I'm calling Cda the drag coefficient non-dimensionalised with ref area =As, the fuse section at max thickness, approx 0.346m^2.
1662293734861.gif
Dynamic pressure q=1/2 rho V^2
drag D=Cd 1/2 rho V^2 Sw

D/q=Cd Sw=Cda As=cwA

So, easy enough now to compare the drag of the sailplane fuselage to the velomobiles and to the theoretical minimums.

I may post some drawings from generic shapes that will illustrate some of the ideas above, and might help velomobile aero solutions. We'll see.

Gregg...
 
Zuletzt bearbeitet:
Um hier mal weiter zu kommen, ein paar Gedanken:

  • ein vollständiges Velomobilmodell im Detail durchzurechnen ist extrem aufwendig, weil man sehr feine Gitterauflösungen benötigt.
  • Dies würde Zugang zu modernen Supercomputern erfordern
  • Sogar in der KFZ-Industrie bzw in den Dissertationen, die gemeinsam mit der Industrie entstehen, werden regelmäßig gezielt Teilaspekte untersucht.
  • Zum Entwickeln eines Verständnisses für die speziellen Problembereiche eines VM ist das auch sehr sinnvoll.
Daher würde ich folgendes vorschlagen:

  • Untersuchung der Umströmung am geschlossenen Radkasten
  • Untersuchung der Umströmung am offenen Radkasten
  • Optimierung der Formen am jeweils generischen Modell, also z.B. Spoiler vor dem Rad, Optimierung der Ein- und Auslassbereiche vor allem am offenen Rad.
Im Bereich des Hecks kann es bei den bisherigen Ansätzen nur um Feintuning gehen. Da sehe ich beim aktuellen Stand ehrlich gesagt wenig, was man mit CFD noch erreichen kann.
Ausnahmen - eher unkonventionelle Ansätze:

  • Im Buch von Hucho (Aerodynamik der stumpfen Körper) wird eher beiläufig erwähnt, dass elliptische Querschnitte am Heck u.U nicht ungünstiger sein könnten als spitz zulaufende. Das beruht aber auf alten Strömunguntersuchungen. Es wäre sicherlich spannend, sowas mal mit VM-Typischen Formen zu untersuchen, z.B. ausgehend von den Querschnitten von Kopf- drinnen und Kopf-draußen VMs hinter der Fahrerposition. Wenn das wirklich zutrifft, könnte das für Alltagsfahrzeuge, aber auch für vierrädrige VMs neue Perspektiven eröffnen.
 
Zuletzt bearbeitet:
I think for the actual normal VM closed wheel models like Alpha 9M, Milan, Bülk, Snoek etc. there it gives also aerodynamic improvements else, for example with "noses" before and behind the wheels, also included more in the shape of the floor, more bulbous shape of the floor, maybe longer VM tail behind the rear wheel, other shape of the hood with more inclined windows, better exhaust air way along the surface of VM,..
 
But resolving all that with CFD is illusionary, especially in the context of full scale vehicle models. You will immediately understand that if You find a mathematical formulation for such details and then a sufficiently fine discretisation.

And maybe You should have a look into

a) the literature dealing with the fundamentals of CFD, e.g., Chung, Finite Elemente in der Strömungsmechanik (that's an old textbook)
b) read the literature i had referred to in my threads collecting results on that stuff, e.g. the thread on areonynamics of the wheel boxes. Especially the dissertations and especially the sections on methodology. Yes, some even mention the computational effort.

Maybe that might prevent You from such dreams.
If You have limited ressources - and we all have - and even more limited experience - which most of us inlcluding me also have (but my experience is large enough to let me understand that) - You should stick to simpler cases.

BTW: my list above has sufficient material for a few master thesis or even PhD thesis.
 
beate schrieb:
To go further, a few thoughts:

  • Calculating a complete velomobile model in detail is extremely time-consuming because very fine grid resolution is required.
  • This would require access to modern supercomputers
  • Even in the automotive industry or in the dissertations that are produced jointly with the industry, partial aspects are regularly examined in a targeted manner.
  • This is also very useful for developing an understanding of the specific problem areas of a VM.
Therefore I would suggest the following:
  • Investigation of the flow around the closed wheel housing
  • Investigation of the flow around the open wheel housing
  • Optimization of the shapes on the respective generic model, eg spoiler in front of the wheel, optimization of the inlet and outlet areas, especially on the open wheel.
In the rear area,....I don't see much that can be achieved with CFD at the moment.
Exceptions - rather unconventional approaches:
Zum Vergrößern anklicken....

Straight away we veer into the full CFD topic :) . It's a subject that needs its own thread, especially if we debate the feasibility of it, the available computing power, software or the skills required. I was hoping for Michael (mrue) to start that. He has some interest, and may do when there is time.

I'm interested in your ideas above and agree with most of them. On the issue of feasibility, hardware, skills we need the opinion of a current practitioner using open source code or a FLUENT student licence on a workstation or good PC.

Things have changed. Twenty years ago the little Octane workstation at the FLUENT seminar was about USD30K here (<8Gb RAM and 600MHz CPU, dual CPU option). Now one can can hide just behind the wave front of expensive hardware. My HP Z800 work station, dual 3GHz CPUs, 12 threads, 96GB DDR3 RAM cost about USD1000 three years ago. I was excited, then I read that the basic FLUENT licence would limit me to 1 core in one CPU.

Now I'm going to zip it (silencio) on that issue until we have an appropriate thread.

I'm interested in these threads you refer to. How do I find them? The forum search tools are not good for me yet.
"...literature i had referred to in my threads collecting results on that stuff (CFD ?), e.g. the thread on areonynamics of the wheel boxes. Especially the dissertations and especially the sections on methodology. ..."

Thanks for any links or suggestions on how to find them,

Gregg.
 
First regarding the computational power: i am aware that the meanwhile old UltraSparc1 workstation (that former equivalent to the Octane) i still have lying around in a corner is computationlly out of date ...

i am here on an older PC with 2 cores and 4GB Ram, and i have a slightly larger but similarly old machine available. Both running Devuan Linux and therefore openfoam in the distro.

Main problem to start: defining the geometry of the bodies. My CAD "experience" is limited to using Xfig :-)

And i will not have any time for such things before christmas.

Here a few links to some of the aerodynamics discussions during the history of this forum:

the wheelboxes:

Aerodynamik der Radkästen

Das Thema der Strömungverbesserung an den Radkästen taucht ja immer mal wieder auf, vor allem natürlich die Frage, wie man diese Problemzone verbessern kann. Klar, Deckel drauf und sich freuen ... ist aber nicht immer erwünscht. Und bei Rekordfahrzeugen wohl auch nicht mehr ausreichend. Es gibt...
www.velomobilforum.de

Aerodynamik - neueste Erkenntnisse!

AW: Aerodynamik - neueste Erkenntnisse! Aoredynamik ist nichts das richtige Gebiet, um mit Halbwissen auf einen grünen Zweig zu kommen (hab ich gehört). Entweder richtig Ahnung haben oder richtig messen. Stimmt schon. Ich würde auch gerne verschiedene Versionen testen. Das Problem ist bloß...
www.velomobilforum.de

Aerodynamik bei Seitenwind für Vortrieb

In der letzten Zeit gab es etliche Diskussionen dazu - z.B. im Quest-Starkwind-Thread und bei Alphaseven. Ich denke es ist an der Zeit dies in einem eigenen Thread grundsätzlich zu diskutieren. Es ist klar, dass wir bei Luftströmung direkt von vorn einen möglichst niedrigen Luftwiderstand haben...
www.velomobilforum.de

Aerodynamik-Thread

Liebe Mit-Interessierte, dieser Thread ist dazu gedacht, Infos zum Thema Aerodynamik am Fahrrad, insbesondere im Hinblick auf Liegeräder, Trikes und Velomobile, zu sammeln. Zunächst der Klassiker: Kreuzotter bzw. Velomobil-Version von Kreuzotter Noch ein anderer Rechner: Human Powered...
www.velomobilforum.de

VM - Aerodynamik: Warum 2 Räder vorne und 1 hinten?

Seit einigen Tagen surfe ich im Internet auf allen möglichen Seiten zum Thema Velomobil. Ich bin noch nie eines gefahren, es ist aber ein Traum für mich, den ich so bald als möglich verwirklichen will. Ich bin zwar noch absoluter Anfänger :D, vielleicht darf ich aber trotzdem schon eine Frage...
www.velomobilforum.de
 
beate schrieb:
Main problem to start: defining the geometry of the bodies. My CAD "experience" is limited to using Xfig:)
Zum Vergrößern anklicken....

So you are already using OpenFoam? Someone else may help with the surface model.
Some ideas come about the priority subjects for flow analysis, but I will wait.

Long ago, when I could have made the leap into Rhino, I didn't. Surface models prepping for CMARC-12 3D panel code were made with the adjunct tool Loftsman. There was too much else to learn. Finite element (structures, composites) was a big bite. All those tools were on 32 bit machines, and the last one just died.

Thanks for the links. Will save me a lot of time. If others come to mind, this thread might be a good place to share them, at least untill the CFD thread starts.

greg.
 
There is a reason for writing a good introduction in journal papers: It forces you to come to grips with what has been done before. And I feel the spirit of this thread was to do just that: gather what has been done in the field of glider aerodynamics that would apply to VMs.

Jumping straight to CFD (while tempting, I dabbled with OpenFOAM myself) is rarely productive without understanding what is already known. In the best case you reproduce something that somebody already did years ago, in the worst case you mess up and build your theories on false assumptions.

So ... if we really want to try a collective effort to (theoretically) optimize velomobile aerodynamics ... the first step is to gather information (as in read academic and white papers) on ground-based vehicle aerodynamics. (Ideally the gathered information can then be collected for others in a review paper, but writing good papers is hard).
 
Sailplanes are my own obsession so I naturally gravitate there, but for this thread I was thinking of a more general scope, a hub where people can share ideas and anything useful.

Gregg.
 
CFd and windtunnel experiments both work by filtering out lots of unknown variables. If a factor disturbs things to much, simply leave it out. Why bother about the windbehavior close to the ground. They produce reproducable results as a result of a very clean proces. Problem is that the real world is not so clean and straight forward. Some of the results as shown in the velomobile windkanal test, are correct, while others may be correctly measured, but do not match with reality, because of mistakes, wrong assumptions, over simplifieing, made in the proces leading to the results. it takes a good designer Like Daniel Fenn or Eggert Bulk to figure out whats fact and what's irrelevant.

If u build a realy good velomobile with open wheelwells, it can be realy fast. As soon as you experience how a realy fast Velomobile acts at speeds over say 60 km/h, other factors start to get important. The handling of a velomobile with open wheelwells, is much more sensitive to sidewinds, compared to the same velomobile with Pants, covering the wheels entirely.

There is a relativly small improvement between a current good velomobile with open wheelwells, and the same, with closed wells. About 5km/h at 50 km/h i recall. No matter how it's done an open moving wheel will never be as good as a perfectly aero shaped surface. So the maximum achievable result of the proces will not be beyond the improvement made by using a closed well. Likely significantly less.

With the M9 it is shown that u do not need an open wheelwell to have a Velomobile with acceptable turning radius and the abilitiy to use wider tyres or get the front wheel off easy.

Yes i know lots of marginal gains also add up. But if u focus to much on 1 detail, u start to lose sight on the bigger picture. If u want to climb up the steep side of a cliff, while we all know there is an easy path from the other side, go ahead. It may be an enjoyable proces.
 
tieflieger schrieb:
CFd and windtunnel experiments both work by filtering out lots of unknown variables. If a factor disturbs things to much, simply leave it out. Why bother about the windbehavior close to the ground.
Zum Vergrößern anklicken....

Serious windtunnel experiments (the guys with a lot of money) actually model the moving ground by having a huuuuge treadmill below the vehicle:


I assume that serious CFD simulation does include a moving ground, too.
 
There is a lot more about that than a moving ground: rotation of the wheels. And that also involves a fine discretisation of the time and really fine grids in the wheel area. Which means huge systems of equations. That means massive computational effort.

Please stop dreaming and start with simpler and realistic scenarios.
 
beate schrieb:
There is a lot more about that than a moving ground: rotation of the wheels. And that also involves a fine discretisation of the time and really fine grids in the wheel area. Which means huge systems of equations. That means massive computational effort.
Zum Vergrößern anklicken....

Exactly, which is why car manufactures and racing teams still rely on wind tunnel and real world testing. But the computational effort actually got manageable, you can do basic CFD analysis on a computer that people have at home these days.

beate schrieb:
fine grids in the wheel area
Zum Vergrößern anklicken....

I mean ... the major benefit of encapsulated wheels is that they don't interact too much with the surrounding air ... I don't see why we should deviate from that design. And then we are left with a relatively simple shape compared to F1 or road cars.


My major gripe with all of this is that, except for empirical and "gut" knowledge, we (and by that I mean ... I) know very little about how the velomobile floor interacts with the ground. Considering that "downforce from ground-effect" is a pretty hot topic right now and does not seem to affect drag too much, I think that it would be interesting to investigate different floor designs for VMs.
 
Zuletzt bearbeitet:
How do you handle "blowby" out or into the holes in the wheel housing or the manhole? This extra flow definitively influences the flow on the outside.
To me it makes no sense to ignore this if you will find an improvement.

Maybe it is useful to make this areas airtight and put a simple Controllabel valve to it. So you can test the influence of this airflow in rolling tests.
 
Riese schrieb:
put a simple Controllabel valve to it.
Zum Vergrößern anklicken....
In the simulation? Or in real tests? :unsure:

For the simulation, probably the best approach would be to check results from others first (as I wrote before). The university solar racing teams have vehicle designs that are similar to velomobiles, with encapsulated wheels. I would assume that somebody must have investigated the interaction of the slim slit wheel openings with the outside. From that I would derive a simplified pressure model that simulates that behavior. (Effectively you would define geometry that "plugs" the holes, but has properties that creates or destroys "air" depending on some rules ... not sure if that is what @Riese alludes to).
 
fho schrieb:
I would assume that somebody must have investigated the interaction of the slim slit wheel openings with the outside
Zum Vergrößern anklicken....
Let us know if you found something.


fho schrieb:
In the simulation? Or in real tests? :unsure:
Zum Vergrößern anklicken....
I was thinking of real tests. But the problem with real tests is, that tiny temperature differences or changes in wind often produce bigger errors as the changes you wanted to test. So maybe my first thoughts were too superficial.
 
Oben Neue Beiträge
Du musst dich einloggen oder registrieren, um hier zu antworten.
Zurück
Oben Unten