RoofTop Scale testing

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

Andy_RR

Well-Known Member
Joined
Sep 29, 2009
Messages
233
Location
Melbourne, Australia
I disagree that it's beyond the homebuilder. I am an engineer, but self-taught in CFD using OpenFOAM - which is free. I have run several projects with it on my laptop (quad-core i7 with 16GB memory - not optimal, but good for about 1e6 cells) and generated this data. Is it accurate? I dunno. Does it show realistic looking stuff? Sure! Can anybody do it? Perhaps not, but the barriers are lower than ever these days and compared to building a model, dealing with scale effects, instrumenting it usefully and regressing all the data, I think it's much better value for effort!

In this case, I was looking at laminar separation bubble effects on this circa-1m wingspan model. I was expecting to see them and, if you squint, you can make out what's going on with a combination of the boundary layer flow and the bulk flow.separation bubble2.jpg

I iterated the spoiler/elevon geometry about seven or eight times during the process and discovered that spoilers on the upper surface produce better results than most other geometry I tried, although they are limited by only upwards travel, obviously. Here we are at a calculated Vy showing the downwash it generates (maybe!) .

Vy3.jpg
 

lr27

Well-Known Member
Joined
Nov 3, 2007
Messages
3,822
Are all those lines streamlines? If so, are we looking at a lot of spanwise flow?

If you fly a model at a constant speed, just as dew is starting to form, condensation will form and show you where the bubble is. Only useful if the Reynolds number is the same, but it's nice to see physical confirmation of the concept. I don't know how perfect conditions have to be, but I have seen it once. The bubble was absent behind rough spots on the leading edge.
 

Andy_RR

Well-Known Member
Joined
Sep 29, 2009
Messages
233
Location
Melbourne, Australia
Yes. The first pic is an estimated stall condition hence the high reverse spanwise flow (typical of swept wings because of the foil pressure distribution and the sweep are amplified at high AoA) I think the color coding on the boundary layer plot is either pressure or velocity - much the same thing, kind of
 

cblink.007

Well-Known Member
Joined
Jul 7, 2014
Messages
206
Location
Texas, USA
I disagree that it's beyond the homebuilder. I am an engineer, but self-taught in CFD using OpenFOAM - which is free. I have run several projects with it on my laptop (quad-core i7 with 16GB memory - not optimal, but good for about 1e6 cells) and generated this data. Is it accurate? I dunno. Does it show realistic looking stuff? Sure! Can anybody do it? Perhaps not, but the barriers are lower than ever these days and compared to building a model, dealing with scale effects, instrumenting it usefully and regressing all the data, I think it's much better value for effort!

In this case, I was looking at laminar separation bubble effects on this circa-1m wingspan model. I was expecting to see them and, if you squint, you can make out what's going on with a combination of the boundary layer flow and the bulk flow.View attachment 95527

I iterated the spoiler/elevon geometry about seven or eight times during the process and discovered that spoilers on the upper surface produce better results than most other geometry I tried, although they are limited by only upwards travel, obviously. Here we are at a calculated Vy showing the downwash it generates (maybe!) .

View attachment 95528
Having not tried out OpenFOAM yet, was it user friendly? I'd love to give it a shot! Looks like you generated some great products!
 

Andy_RR

Well-Known Member
Joined
Sep 29, 2009
Messages
233
Location
Melbourne, Australia
Having not tried out OpenFOAM yet, was it user friendly? I'd love to give it a shot! Looks like you generated some great products!
No, in its basic form, it's not. Not at all! There's no denying a steep learning curve is involved but you get what you pay for. Alternatively at least a couple of companies have re-skinned it and added all the GUI fruit to make it much easier to navigate. Simscale is one place where you can sign up and play for free. Airshaper is another. Both of them are cloud-based solutions so you can buy processing power too, to save smoking your laptop at home. I'm sure there are others too.

I do my work on a Ubuntu VM so I think I could port this to a cloud service somewhere to buy more FLOPS if I needed to.

My next CFD goal is to get moving mesh geometry working so as to simulate propellers properly and analyse the effects of prop wash but that is quite a tall order!
 

sming

Well-Known Member
Joined
Apr 10, 2019
Messages
63
No, in its basic form, it's not. Not at all!
You don't say! I'm a software engineer, and it was definitely not easy (weelll, mainly because I don't know what I'm doing on the physics side, but still). You need to be comfortable with linux, the command line, editing text files for configuration and creating bash scripts to run everything. If you started computing in the 80's, nothing new ;)

FreeCAD offer a cool OpenFoam solver though, it helps a lot! If you can model something in FreeCAD that is ;)
 

cblink.007

Well-Known Member
Joined
Jul 7, 2014
Messages
206
Location
Texas, USA
No, in its basic form, it's not. Not at all! There's no denying a steep learning curve is involved but you get what you pay for. Alternatively at least a couple of companies have re-skinned it and added all the GUI fruit to make it much easier to navigate. Simscale is one place where you can sign up and play for free. Airshaper is another. Both of them are cloud-based solutions so you can buy processing power too, to save smoking your laptop at home. I'm sure there are others too.

I do my work on a Ubuntu VM so I think I could port this to a cloud service somewhere to buy more FLOPS if I needed to.

My next CFD goal is to get moving mesh geometry working so as to simulate propellers properly and analyse the effects of prop wash but that is quite a tall order!
That is pretty cool; keep up the good work! We here use a combination of AutoCAD, an in-house modified version of XFLR5 (modified for higher Re) and RDS. Just recently went from AutoCAD to Solidworks. We prefer to do dynamically scaled models to validate what we see on computer. I have always wondered "how long" until CFD is so accurate that wind tunnels and scale models become obsolete! Lord knows it will sure save some time & $$!- Your prediction?
 

sming

Well-Known Member
Joined
Apr 10, 2019
Messages
63
There is an amazing talk by the Jonker brothers explaining how they designed, built and tested a world championship winning glider in something like one year, using CAD and CFD. For half the price of state of the art german competitor! It's quite mind blowing... Let me see if I can find it (it was posted somewhere here by autoreply I think)

edit: can't find it!!
Some slides but it's not the same: Hall2C Tuesday13h00 - Uys Jonker
Some screenshot of the slides a remembered: Pilot's Post - SSSA Second Annual Conference and AGM - Orient 2019
An interesting paper but unreadable, other links dead: Theoretical question
Another good post : Advice from anyone running open foam CFD for aerodynamics ?
 
Last edited:

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
916
Location
madrid
i wonder how much wind tunnel balances cost (Wind Tunnel Balances) I found 4 companies selling them , but any of them answered my price request. Anybody got a guestimate?
In my particular benchmark scale testing i have an additional handicap, due to the fact that vortex lift is not affected by scaling, but the benchmark model, that relies purely on normal lift is severly affected by reynolds scale effects. In fact i run all the scale parameters and a 2m span model should weight 26 kg in order to match the scale weight effect. How many hp should a 26kg-2m span model require?
 

cblink.007

Well-Known Member
Joined
Jul 7, 2014
Messages
206
Location
Texas, USA
How many hp should a 26kg-2m span model require?
26kg seems alot, even by dynamic scaling math. What density altitude/gross weight are you trying to simulate?

We are doing a 2.5m model of our design, and our weights are going nowhere near 26kg...maybe 19...if we are wanting to simulate a 10000 foot density altitude at our max gross design weight.

For our aircraft, we are using a brushless motor:


The motor and prop combination actually match our power scaling requirements spot-on!
 

rotax618

Well-Known Member
Joined
Oct 31, 2005
Messages
824
Location
Evans Head Australia
For low aspect ratio wings (AR <3) the airfoil has very little to do with stability and lift at high Alpha, the plan form has a far greater effect on low speed flight at high alpha, a flat plate will provide just about the same lift/drag as any given airfoil at AOA > 17deg, the planform shape will effect the stability of the vortex which will form on the upper surface.
 

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
916
Location
madrid
26 kg comes out of the formula from the NASA book on scaling. I applied the formula at sea level. The original has 4,5m and weights 300kg mtow.
That brushless motor looks like a great tip. I guess in the second phase of dçynamic scale testing i prefer and electric to an combustion engine. Will review your tip.
After asking Verhees he confirmed that the Verhees doesnt at all rely on vortex lift. his argument was that the plane once in approach vortex lift regime wont be able to power up enough to avoid landing and retake a non vortex lift regime.
His comment is a good point.
Regarding my experiment, that resembles strongly the 1947 paper on the DM1 glider, is that i wanted to benchmark the vortex test against the verhees model. However this one will be treated unfairly due to a different lift regime. Most of my thesis has been confirmed by the papers i reviewed recently. Still comparing a sclae vortex lift design against a non vortex lift will show misleading design.
 
Last edited:

Steve C

Well-Known Member
Joined
Oct 6, 2008
Messages
78
Location
Lodi, CA
An old friend once told me about flying the F-106. Here's an interceptor that can hit mach 2 in a very short time from takeoff, yet he said if you entered high alpha mode, full burner would not arrest the sink rate. He said it remained very controllable with rudder and elevator. The ailerons created such adverse yaw they were worthless and the only way to recover was to stick the nose down.
 

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
916
Location
madrid
Would such a feature (stickiness of high alpha mode at low altitude) be a "no go" for you in a delta travel machine? I admit i dont have experience in planes besides cessnas, but personally i just remember a few times I aborted landings due to high crosswinds.
 
Last edited:

Steve C

Well-Known Member
Joined
Oct 6, 2008
Messages
78
Location
Lodi, CA
I think it's certainly something to avoid until it's tested at a high altitude. Perhaps deltas with low wing loadings will recover very quickly.

It would be extremely helpful to have an aoa indicator.
 

Bill-Higdon

Well-Known Member
Joined
Feb 6, 2011
Messages
594
Location
Salem, Oregon, USA
i wonder how much wind tunnel balances cost (Wind Tunnel Balances) I found 4 companies selling them , but any of them answered my price request. Anybody got a guestimate?
In my particular benchmark scale testing i have an additional handicap, due to the fact that vortex lift is not affected by scaling, but the benchmark model, that relies purely on normal lift is severly affected by reynolds scale effects. In fact i run all the scale parameters and a 2m span model should weight 26 kg in order to match the scale weight effect. How many hp should a 26kg-2m span model require?
I'll try & get some pictures of some home made balances with strain gauges attached to them this week end. They're on EAA 292's demo wind tunnel
 

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
916
Location
madrid
I would be very grateful for that. I have something in my head with linear carriages. But i am pretty lost on the strain gauges to use.
The beauty of sting balances is that they read 6 moments. I guess that inducing a progressive increase in AoA until stall, may allow reading a differential loss of lift on the wing halves and also give an inside into stall behaviour. A rotational strain gauge could be feasible without too much added complexity.
Alternatively that could be acchieved by hanging the plane on three points. One on each swept back wingtip and one on the nose and reading lift and drag on the three points..
One of the issues i want to test is to create a second vortex at 60% of semispan (or reinforcing tha appex vortex at that point) added to the vortex at the appex. Such a vortex should have effects on the stall beahaviour.
 
Last edited:

berridos

Well-Known Member
Joined
Oct 10, 2009
Messages
916
Location
madrid
Would appreciate if someone could give me an opinion or answer on a specific doubt.
If for example i am testing a model at a high aoa, lets say 25º based on vortex lift, do i have to adjust the ailerons so that, a this constant aoa the pitching moment becomes 0 and overcomes the positive dynamic stability?
Would that be the proper measurement procedure for lift om a rooftop device?
 
2
Top