Mailing List flyrotary@lancaironline.net Message #35832
From: cbeazley <cbeazley@innovista.net>
Subject: Re: [FlyRotary] Re: Pinched ducts was : [FlyRotary] Re: cowl openings for water radiators
Date: Wed, 28 Feb 2007 22:16:39 -0500
To: <flyrotary@lancaironline.net>

 
On 2/27/07, Bob White <rlwhite@comcast.net> wrote:
I haven't said much about ducts as I don't have any experience except
that my first installation doesn't work very well.  They have been on
my mind quite a bit lately though as I'm going to have to redo the
system.  My take from K&N was similar to yours Thomas.  But maybe Ed
has made an advancement.

I'm thinking about putting the rad and oil cooler horizontally next to
each other with the wedge duct as shown on Paul's typical cooling
layout picture.  Except I think my radiator/oil cooler will be closer
to the front than shown.  So i was just going to run the wedge opening
straight out to just behind the prop.

I'm also not fond of the exit as shown, and plan on something similar
to the wedge on the bottom but with more room.

Anyone want to speculate on my chance of success.

Bob W.
 
Bob,
 
I give you a very high chance of success.  The wedge inlet is almost as efficient as the K&N and is a heck of a lot easier to install (much of the reason I chose it for my latest iteration). 
 
Paul's website shows the wedge being less efficient as an exit, but I had a modified wedge exit and it worked fine..
 

On Tue, 27 Feb 2007 13:52:26 -0500
"Thomas y Reina Jakits" < rijakits@cwpanama.net> wrote:

> Hmm,
>
> if I remember right from reading what I have from K&N - if you do not have the optimal lenght available you start to cut it from the intake - take radiator and apply the optimum duct, measure from the radiator towards the intake whatever distance you have available and cut the duct there....
>
 
Thomas, I agree with you and Ed on this point.  That is the optimal way for pressure recovery and airflow, but will be high drag if not all that cooling is needed.
 

> Dave,
> are you a Navy-flyer or a medic?? :)
>
 
Ex Navy Flight Surgeon.


>
>   Hummm, Dave, perhaps my understanding of what it takes to keep the boundary layer attached to the duct wall is flawed.
>
>   From what I believe I understood regarding airflow in a duct is that the pressure recovery both aids and hinders the boundary layer's attachment to the duct wall.  The pressure build up (area of slower molecules) tend to push and keep the boundary layer pushed against the wall of the duct as it curves out - at the same time it is slowing down the boundary layer.  So its the point of separation is (at least in part) contingent on how much speed the boundary layer has enabling it to push how far into the pressure recovery area - before it ultimately separates.  The further the better is my understanding.
 
I agree with everything except the pressure causing the separation.  The pressure aids in boundary layer adhesion except for possibly some small effects on viscosity and Reynolds number.

>
>     My understanding is that in a duct - it is the recovery pressure  which  builds in the expansion area just  before the core.  This "high" pressure area  will "push" back on the boundary layer causing it slow and eventually  to separate from the wall.  .  However, if you keep the boundary layer speed up it pushes further into the pressure recovery area following the duct curve before the "back pressure" slows it enough to cause it to separate.
>
 
I still don't get why slowing the boundary layer would make it want to separate.  This is a laminar vs. turbulent issue.  Slower is better for laminar every time.

>   Also  the speed of a molecule in all random directions is much, much higher than the component imparted by the airspeed - about 1100 ft/sec at sea level as I recall compared to about 40 ft/sec in the duct.  
 
True, but it is still subject to fluid flow dynamics and the speed of the air (airplane) does make a difference.
 

So my interpretation is that (at least in a duct) its the back pressure of the recovered pressure that causes the separation - not necessary the curve of the duct alone although that certainly contributes to the pressure recovery. 
 
Not buying it yet..

 That being said,its clear that  the three factors  (duct curve, expansion area and separation)   really go hand in hand. 
 
Agree totally

 The greater the curve the more pressure recovery occurs and the greater the tendency for separation.   The higher the velocity of the boundary layer the further it can penetrate into the higher pressure area before being slowed and separation occurs.
>
 
greater curve = more separation = LESS pressure recovery

>   There is NO doubt that having a longer duct would improve the situation.  However given I only had 3 -6" my take  was that speeding up the air (and boundary layer energy) would ensure it penetrated deeper into the bell shape before the pressure recovery caused separation.  But, as I have often stated - I could be completely wrong about what I think I understand.
 
You don't want boundary layer to penetrate, you want laminar flow to penetrate.  Boundary layer is turbulent and slow (yielding less pressure recovery).  The central laminar high velocity is what you are trying to keep as long as possible.  It has to terminate sometime before going through the rad.  So the idea is to minimize separation and boundary thickness as long as possible, keeping the high velocity non-boundary air.

>
>     You are after-all the Navy flyer and I know they cram a lot of areo into Navy pilot's heads.  Me- I'm a electrical engineer, so what  I know about aerodynamics is what I have read (and think I understand).
 
I'm not claiming to be educated on the subject.  And Navy flight school was certainly no help.  They are still teaching that lift is derived from air flowing faster over the top of the wing.  I get my gestalt from my undergraduate degree, Atmospheric Sciences (way too much fluid dynamics and stuff that I was not that good at anyway). 
 
More importantly though, I think there is no real way to test WHY one duct works and one does not, you can only test WHICH ones work.  The theories are just conjecture on anyone's part.  Measurements of say, pressures at every point in the duct, will of course help refine the theories but they are still just theories - no matter where you hear them.
 
That being said, someone must have hit upon a theory that is closest to accurate (mine appears to be second best...  according to anyone I happen to be discussing the issue with at the time  :-)
 
And someone else must have hit upon the best duct design given the conditions required to cool a rotary via the stock inlet of an RV-6, and that appears to be you!
Hi Ed/Dave;

The pinched duct obviously works :)
One possibility is that the pinched neck section is tripping the laminar flow to turbulent at or shortly after the neck-down.
This should energize the boundary layer, reduce laminar bubbles and delay separation.  Separated flow is much worse than turbulent.
Note that some high performance sailplanes use turbulent airfoils sections near the fuselage to avoid separation in the wing-fuselage junction.

2D airfoil packages such as xfoil are good for one surface, not two or more divergent (3D) surfaces, then you are into the realm of CFD or expensive packages.
    http://web.mit.edu/drela/Public/web/xfoil/

The slight angle on the rad should give the air more distance to make the bend, possibly piling up in the area farthest away from the inlet.
 
If someone suspects they might have separated flow on an existing installation you might consider trying turbolator, zig-zap or
even hockey tape to try to trip the flow early.  Much like the vortex generators added after the fact to wings.

Ed,
what are the rad dimensions? thickness? type of rad?
any ducting on the back of the rad? size of the exit area?
temps for some power/speed settings?
I am also curious if the area farthest away from the inlet is working, particularly in the corners.
Have you considered filling in parts of the furthest areas as test?

Another guess.
Thanks
Cary
Subscribe (FEED) Subscribe (DIGEST) Subscribe (INDEX) Unsubscribe Mail to Listmaster