www.recumbents.com - Streamliner Fairing Construction and Design
www.recumbents.com
www.recumbents.com
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
 All Forums
 www.recumbents.com
 Technical / Bike Building
 Streamliner Fairing Construction and Design
 New Topic  Reply to Topic
 Printer Friendly
Previous Page
Author Previous Topic Topic Next Topic
Page: of 10

warren
human power expert

USA
6454 Posts

Posted - 04/19/2019 :  15:20:02  Show Profile  Visit warren's Homepage  Reply with Quote
Well, there's this at a high level: http://www.recumbents.com/wisil/Plans/how_to_build_a_streamliner.htm
And this: http://www.recumbents.com/wisil/barracuda/barracudafairingdesign2.htm which shows how I arrived at a shape fairly similar to Matthew's current design.
Go to Top of Page

Jerry
human power supergeek

USA
1459 Posts

Posted - 04/19/2019 :  21:13:11  Show Profile  Reply with Quote
And this: http://www.recumbents.com/wisil/costin/swift/default.htm .
Go to Top of Page

jjackstone
recumbent enthusiast

USA
268 Posts

Posted - 04/19/2019 :  21:40:06  Show Profile  Reply with Quote
quote:
Originally posted by Toecutter

quote:
Originally posted by warren

That's not an airfoil. Fairing designers don't use airfoil designs any more because they are designed for lift, not minimun CdA.

quote:

...
Do you have any links to design algorithms or math one can use to arrive at a very low CdA shape for their application without having to throw down money on software?


Shooky did some software a while back. Not sure how it ended up.

http://www.recumbents.com/forums/topic.asp?TOPIC_ID=5972

[quote]I really like the shape Matthew has come up with and would like to make a main body shell of similar purpose, except for an open-wheeled trike application(the application being the lowest possible CdA for a given set of design constrains). Being that the way the air interacts with objects is complicated and that our use cases, ergonomics and storage needs differ, I can't simply expect to copy his shape and get good results.


Look at the Velotilt. It's a delta design but could probably be modded for a tadpole.



[quote]I'm very curious how one goes about designing a shape and arriving at what he has gotten. For all the information this site has, I can find precious little info on the process of designing a shape to meet one's needs.



What do you mean by designing? Just determining what you want to accomplish, doing the actual build, or doing the math behind the shape?



JJ
Go to Top of Page

Toecutter
New Member

USA
57 Posts

Posted - 04/19/2019 :  23:42:04  Show Profile  Reply with Quote
Thanks for the links. I'd been through some of them already before I asked my question. While helpful, they don't tell me WHY the front, side profile, and tail are shaped precisely the way they are. They don't go into the actual technical details of what the airflow is doing as the shape moves through it. They don't explain why the slope of the tail changes or why it changes in the spot that it does, or why it is the length that it is, or why the top/back of the side profile tapers at the angle it does. I know those things have a tremendous impact on the Cd of the vehicle because if you get something wrong, unwanted turbulence is generated.

The link to the Barracuda was interesting, but the pressure distribution java applet didn't work(Server not found).

I also found that my next design is already using some of the techniques in those links. For instance, Sean had some "latex knee pops" to accommodate his pedal stroke clearances, and I'm doing the same thing to accommodate my steering bars on my trike to allow a slightly narrower shell(620mm vs the 650mm of my last one; I could go narrower, perhaps 450mm with a rack and pinion with steering wheel or with T-bars but I need a bunch of custom parts for that and more money, so that's something for the 3rd design iteration).

quote:
Originally posted by jjackstone

What do you mean by designing? Just determining what you want to accomplish, doing the actual build, or doing the math behind the shape?



I want to be able to go through the math to make sure my shape is as good as I can get it for my application before finishing an actual physical build. Or failing that, at least have a free program that can simulate various shapes and spit out Cd values and/or pressure vs section values that I can compare with other designs.

As for shooky's program, I'm using a linux operating system so unfortunately it doesn't work on my computer.

Edited by - Toecutter on 04/20/2019 01:20:28
Go to Top of Page

Matthew Martin
New Member

USA
62 Posts

Posted - 04/20/2019 :  08:38:03  Show Profile  Reply with Quote
I can only speak for how I designed. I am not a Phd in aerodynamics, and my design reflects that. I knew I was going to have to design for pressure distribution, because sst transition modeling is very difficult to get accurate results. designing for pressure distribution is faster as well.

Originally I started by eyeballing the shape. I tried to make it look like what looked fast. I also did not have access to and real CFd at the time. I also did this method for the side view. as I learned more my design philosophy changed. first I realized I needed consistency in the topology of my model that played the biggest part in maintaining a consistent pressure gradient.
Before:
After

So at this point I had a rough idea of what I needed. I needed a shape that had a high pressure at the nose (that's a given), and the pressure progressively gets lower as you go further down, where the lowest pressure is around the widest point. The after the lowest pressure, begin to gradually but quickly bring the pressure back up at the tail. An idea know as statford seperation criterion (pressure recovery). Now if you look this up you will find all the mathematical definitions for it. Most all the math you will find is how to calculate the likely hood of laminar boundary separation. Currently too complicated for me, But by solving for pressure distribution I could get good enough results.

this still doe not answer your question. "how do I determine the shape?" Well Once I determined what I wanted, I did a lot of guess and check. I began to look at fast designs with new eyes. I began looking for similarities among top bikes. the velox bikes were a nice example because I was able to observe iterative design changes in regards to the shape.

What I noticed was most of the top profiles of the bike were more of less the same. The most noticeable difference came from the side profile. But then after a dozen or so test models, I saw the trend. The were basing the side profile of a shape that had the desired pressure distribution. That can look more like Eta or more like the VeloxX bike (not the delft bike). Both have a slope to the highest point then drop off. But this still did not help me design my shape. The latest velox (9) seems to combine these two different designs.

So what I did initially was find my top profile. I started with a Naca-67-018 aerofoil just to start. I ran a stimulation of the profile revolved around the center axis to see the pressure distribution. ( I did not run a naca 67-018, I ran a Naca 66-018) The mesh was very rough and so are the results, but they showed to be close enough for my needs.
Now for the side view. I realized an aerofoil would fit over the body like so.

I chose these aerofoils because they were both 6 series, which are designed for laminar flow.
This was a starting place and this is how I made this model
Ignore the stuff hanging out its irrelevant to my current stuff.

Now this worked ok but as shown above my topology was not great and my pressure distribution was very rough ( see post on 11/21/18) . So I tried again using the same foil and this time sculpting the bike of the body in Blender(open-source 3d modeling software) then bringing it into Fusion 360. This allowed me to have more control points while also maintaining control of the without too much mess. this go me here:
Better but still not great. So I tried another foil ( only for the side). this got me here:
You can see a much more consistent gradient. Now these are only the best models from each series I tried. The mesh settings for the simulations were near identical.

So this was good but still not great. I liked the design but something in me knew it had to change somehow ( it was too long). So I contacted Calvin Moes for the first time just to see if he would offer any advice. And he did. He said I should make the bike shorter and have a more aggressive transition in the back. So I made these changes. This time only using the the top aerofoil as reference. this allowed me to achieve a lower pressure and the widest point, while also having the gradient travel for more then 70% of the body. Calvin had told me that it is possible(perfect world) that laminar flow could be achieved all the way until the pressure begins to increase. So this got me to my current and final:


Note how the rear is completely different than a standard Naca 67-018. This was done though several iterations.

So in summary the top teams use shapes based on a specific pressure distribution(I could be wrong), in which they determine using different mathematical models. What I had to do was use shaped available to me that had a close resemblance to what I wanted and modify it iteratively, in order to achieve the desired pressure distribution. sorry for the long post I hope it helps.
Go to Top of Page

Matthew Martin
New Member

USA
62 Posts

Posted - 04/20/2019 :  10:58:09  Show Profile  Reply with Quote
I should also note ignore all aforementioned drag numbers I say this because none of the simulations were done in a sst transition model, which simulates both laminar and turbulent air. in fully turbulent conditions my bike should have a Cda of about .0245. That's with no laminar flow, in reality it should be much lower. I am still learning, and as I learn more I should clarify my mistakes.
Go to Top of Page

Toecutter
New Member

USA
57 Posts

Posted - 04/20/2019 :  12:53:34  Show Profile  Reply with Quote
I appreciate you for taking the time to describe your process. Very informative.

Based on what you have done, I think I've been heading in he right direction with my design all along in spite of my anxiety that I could be designing it to the contrary, except I don't have CFD software to figure out what the air is doing around the shape. I have a drawing of my current in-progress coroplast trike body(doesn't include windshield/roof/turtledeck yet) that I am going to scan when I get the chance and ask people to critique it.

Months ago I made a spreadsheet to model a NACA 66-018 to see how it would fit me and my trike(the data was obtained from www.airfoiltools.com), but because of the minimum width imposed by my steering bars, I determined it would be too long of a body to work for my needs(I want a full tail to a point, and not have to cut the shape into a Kamm design). I really need a new steering system so I can get a width down to around 450mm in order to keep the length of the body reasonable(it needs to be under 2.8m for practical reasons). I have many similar spreadsheets, which allow me to test foot boxes, shoulder width, storage areas, knee bumps, ect to see if the shape will meet my needs without even having to make a 3D model for each and every one.

Instead, I settled on a shape described in my latest post in this topic(4/20/2019; 12:55:06):

http://recumbents.com/forums/topic.asp?whichpage=1&TOPIC_ID=7064

When it is finally built, tuft testing will certainly be in order.

I read your entire thread when I discovered it last year and can relate to the passion of designing something at a young age, having designed my first electric car conversion back in high school. I'm glad to see that you've made a lot of progress from that very first rendering until today. You've got the wooden jig ready to go and the design pretty much settled on, so a finished and working design isn't too far off, relatively speaking. I bet it's going to haul ass too.

My interest in HPVs stems from wanting to make electric vehicles more efficient and higher in performance. I think cars are far more wasteful than they should be, and the sort of holy grail of automotive efficiency is a street legal car designed like a streamlined HPV. What better way to test that efficiency than having to first pedal it! And think of what kind of performance it could have if it only needed 2 horsepower to do 70 mph on the highway, had the occupant(s) as most of the vehicle weight, yet had 200+ electric horsepower on tap with all wheel drive! This design philosophy would also keep costs down by reducing the number of batteries needed for a given range. Imagine getting a 250 mile range at highway speeds on only 5 kWh of batteries. Such efficiency would also make running it on solar panels a viable proposition, at minimum helping to extend range and possibly rendering range a non-issue as long as there's sunlight if there's enough area for panels.

Since I'm using a recumbent trike as a platform and building a coroplast body over it(don't have the resources to custom build an actual car from the ground up yet), I'm going to start much more modest than 200+ horsepower, maybe 8-10 horsepower instead, with perhaps a 1-1.5 kWh pack, and human power being able to add motive force at all usable speed ranges. Even with such a small amount of power(compared to a car; quite monstrous by ebike standards), it is not out of the realm of possibility that such a vehicle could be faster to accelerate than most cars on the road, and it would be a truly off grid vehicle that wouldn't be reliant on a fuel source since it could always be pedaled with the motor off if need be.

Edited by - Toecutter on 04/20/2019 13:24:38
Go to Top of Page

Matthew Martin
New Member

USA
62 Posts

Posted - 04/20/2019 :  13:23:58  Show Profile  Reply with Quote
starting with a Naca 67-018 may be a better option if you are trying to build a laminar flow body (a human fits better).


As for Cfd Simscale currently offers a free community plan. Only thing is your designs have to be public and you are limited in problem size. it is good for getting a rough idea of pressure distribution, though for more refined solutions you would need to purchase a plan.

Go to Top of Page

Toecutter
New Member

USA
57 Posts

Posted - 04/20/2019 :  13:35:32  Show Profile  Reply with Quote
quote:
Originally posted by Matthew Martin

starting with a Naca 67-018 may be a better option if you are trying to build a laminar flow body (a human fits better).


I found the S1016 to be an even better fit than the NACA 67-018. The only problem is that for any of those shapes to work to their best potential and remain practical(excess shell length increases crosswind vulnerability and makes storage/transport difficult or impossible), I need to keep the width down, and that requires getting rid of the steering bars and rigging up a rack and pinion, tiller, T-bar, or other steering so that my shoulders and Q-factor of my pedals become the new limiting factors regarding width. My shoulders are 420mm width with a thick leather jacket on, and with my current crank/pedals I need 410mm width for my feet, so that would require a shape of roughly 450mm width at its widest point to accommodate both without messing the streamlining up.

quote:
As for Cfd Simscale currently offers a free community plan. Only thing is your designs have to be public and you are limited in problem size. it is good for getting a rough idea of pressure distribution, though for more refined solutions you would need to purchase a plan.


Jerry recommended this program to me on another forum. I definitely need the commercial version, but can't afford it at this time. I may try making a model with a free program and trying it out anyway.
Go to Top of Page

jjackstone
recumbent enthusiast

USA
268 Posts

Posted - 04/20/2019 :  15:45:25  Show Profile  Reply with Quote
http://www.recumbents.com/wisil/qfactor/qfactor.htm

JJ
Go to Top of Page

alevand
human power expert

USA
3478 Posts

Posted - 04/24/2019 :  08:59:37  Show Profile  Visit alevand's Homepage  Reply with Quote
Good work, Matthew.


C:
Tony Levand
Go to Top of Page

Speedy
recumbent guru

USA
907 Posts

Posted - 04/24/2019 :  14:46:55  Show Profile  Reply with Quote
Yes ... excellent computer design skills Matthew.

Go to Top of Page

alevand
human power expert

USA
3478 Posts

Posted - 04/26/2019 :  06:14:23  Show Profile  Visit alevand's Homepage  Reply with Quote
Can you give us more detail son the CNC machine shop?

C:
Tony Levand
Go to Top of Page

Matthew Martin
New Member

USA
62 Posts

Posted - 04/26/2019 :  18:19:10  Show Profile  Reply with Quote
Well long story short I got in contact with a family friend who knew a retiree who had access to the CnC router at MacDill air force base. He was willing to cut the slices for me but warned he had limited experience. And after receiving the sad news that he could not machine the plug for me I had to get back on the lookout. I still have a few more places to ask around.
Go to Top of Page

Upright Mike
human power expert

USA
3864 Posts

Posted - 05/02/2019 :  13:48:58  Show Profile  Reply with Quote
Matthew, I thoroughly enjoy reading your straight-forward progression of your design. You are explaining things and going through things that have taken folks like me years to understand (and well not having the software tools to work it either limits someone like me from trying to design a proper streamliner) And to to think you're still in high school, WOW! Kudos and much respect to you!
Go to Top of Page

Upright Mike
human power expert

USA
3864 Posts

Posted - 05/02/2019 :  14:05:55  Show Profile  Reply with Quote
quote:
Originally posted by jjackstone

http://www.recumbents.com/wisil/qfactor/qfactor.htm


Years ago I read and studied everything on this page. I found a simple solution was to purchase 10+ year old DuraAce 7402 double (I think this is the number) cranksets as these had one of the narrowest crankarm shapes to begin with as the arms are not flared out, hence the manufactured spec said they needed like a wide 127 mm BB. BUT, you simply mount them to a 103 or 104 mm square taper bottom bracket (readily available). This produced a Q-factor of I think 132 mm (about the same as the BMX Black widow Odyssey cranks listed, but with a much more solidly built crankset).

The Black widow have a "cheap" removable spider held on by 3 bolts, and also only come with a 110 mm Bolt Circle Diameter. The DuraAce are standard forged Al crankset with a "standard" 130 mm Bolt Circle Diameter - important for more strength if you're putting say a 65 tooth of 80 tooth Chainring available from Dutchbikebits.com, recumbentparts.com or Raptobike.com

Also I ordered shortened pedal spindles as well to bring my feet in even closer, like another 12 mm, all of these parts (old cranks, 103 mm BB, and spindles I got off of eBay.

You can also shorten say a 172.5 mm old DuraAce 7702 crankset down to about 147 mm, which I had help with from Thom Ollinger. This narrow configuration, shortened, modified crankset is what I finally put in my streamliner.

EDIT: Here's an old post about everything I said above. More details and photos http://www.recumbents.com/forums/topic.asp?TOPIC_ID=6003

Edited by - Upright Mike on 05/02/2019 14:09:56
Go to Top of Page
Page: of 10 Previous Topic Topic Next Topic  
Previous Page
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
www.recumbents.com © 2019 www.recumbents.com Go To Top Of Page
Snitz Forums 2000