Hey all,

Here is the excel file I have been working on. It is not finished and I am
still sorting out a couple errors here and there but it is getting there.
Eventually the thing will be all correct and allow you to change the
variables and do all sorts of neat stuff. I just got laid off recently from
Delphi (note: F*%k the American auto industry) and have been busy with
finding a new job so I will get back on it when I get the chance. Paul, note
that the rotor equation never nets a constant radius. This data while
representative of the the points cannot be used in a machine shop
environment. This is where my questions of Solidworks 2009 came from as you
would need to graph the sketch of the equation and not a spline of a number
of points. I checked with NX3 platform and SW2009 and it looks like they are
too simplistic to do what pro-e does. Who has pro-e that we can test out? I
can gpo more into it at depth but will only do so if there are serious
questions about it. Here are a couple things that will show you whats up
with what I mean. I am working diligently to get this worked out because if
my plan for a lightweight all steel
brazed rotor works out then I can move to the next step which is increasing
the capacity of the engine and and having the weight of the rotor as the
cast iron version for the larger capacity engine be the same as the
13b/renesis. 8500rpm to 10000 rpm limit on a 1 liter rotor. That sound
intersting to anybody?

http://www.profilesmagazine.com/p33/marsalek.html

http://esoxrepublic.com/blog/category/geometry/splines/

http://www.adras.com/SW2009-Equation-Driven-Curves.t8302-83.html

Rob Woods

If I am understanding  you correctly you are saying the face of the rotor is
 a not a circular arc?

Paul Lamar


 Paul,

The statement I made before is true.You can see by looking at the graphical
output on my program that the curvature in somewhat flat in the middle
where the combustion chambers are in comparison to where the apex seals
are where the equation starts to take a swooping turn which increases the
curvature. There isn't a solid radius arc anywhere in that equation. Kenichi
Yamamoto wrote the book with all the derived and integrated formulas that
net the equations of the inner housing and rotor. If it was a simple as
drawing three circles and trimming the outside stuff I think it would have
been written as such. Any attempt to look at it so simplistically without
actually taking the time to work out what he provided us is a fools quest to
put it lightly. So sagitta has nothing to do with this. This all comes back
to what I have been speaking about. I can set up 1080 points or many times
that number and import that into solidworks and run a spline through it all.
A curvature comb placed on it will show all sorts of weirdness going on.
Getting an accurate cad profile is priority for this since any custom
housings or rotors need to be spot on for any number or reasons. There are a
couple programs that can do this and with some resourcing I can get the
profiles into Solidworks.This board is one of the resources I have and hope
that someone can work something out with this. Or someone who is much
better/smarter than I in programming Macros I can install into Solidworks. I
have one so far but havent verified the accuracy yet.

One thing that I think might be a constant arc is the side seals. They are
not an offset of the rotor profile which makes me wonder why for one and
also because how they might manufacture them. Constant section large
diameter that gets cut into three pieces maybe? Is that the reason for not
being a offset of the rotor surface?

Rob Woods

Rob Woods

You are absolutely right. My mistake. I have no doubt what you say is true.
I happen to know however the rotors have been machined three at a time on a
turn table. I have a picture around here someplace of the machine. I think
it was a Curtiss Wright machine. This is what fooled me.

Obviously Mazda is not doing it this way unless there is something hiding
behind the turn table. I did not see the machine used when I was in the
factory but Mazda would not let me take pictures of that part of the plant
anyway. It is not an NC machine as the plant was built in 1969.

I laid a rotor on a piece of paper and traced the face curve. An arc with a
9.15 radius comes pretty close. Here is a scan of the paper. My 715 mm (28
inches) radius number was way wrong. I am not sure why. The arc radius is
more like 9.15 inches. This is something everybody can try. This also proves
your contention.

Here is the fly in the ointment. All these engines require a depression in
the face of the rotor to form the combustion chamber. What only matters is
the exact path of the apex seals. I contend a pure triangular rotor with
straight sides would work. The compression ratio would be too low but the
engine would run. I contend an arc of some radius with a combustion chamber
depression would be close enough.

Interestedly enough here is a patent on a CW rotor face that did not come
close to an arc.

BTW Ansdale published the math in his book "The Wankel RC Engine"
in
1969.

Paul Lamar


Paul,

Looking at that patent he is showing a knife edge apex seal that will
never work. Interesting shape but I would always stick to the math. All
of the work in Kenichi book is all dependent on each other as far as
calcs go so I would mess with anything. One thing to note on my
spreadsheet is that I will eventually replace the constants(all of the
2/3, 9/2 and such)in the equations with variables so we can visualize
lobe and housing number bigger that two and 3. It is a work in progress
and it is important to understand everything in and out before solid
design decisions are made.

Rob Woods

Thanks for doing all this work Rob. I am sure it will be useful when it
comes time to machine some rotors.

I see what you are saying but I think the guy drawing the patent just got
lazy and did not draw an arc on the apex seal. The apex seal had nothing to
do with the patent anyway. If you want the complete patent I have it in a
pdf.

I found that dwg on how to machinerotors faces in the 1969 Kenichi Yamamoto
book on page 119 and pagwe 120. Here they are.

There is one other thing going on and that is crevasses that store unburned
hydrocarbons are not good. The side seals should be out as far as possible.
I think the side seal slot is an arc on the current rotors so an arc along
side an arc would work best. What Mazda is doing now is filling that gap
with Teflon which must be pretty expensive.

I am still thinking about how to best machine this thing out of one piece
and get some cooling fins inside for oil cooling. Steve Brazil's two piece
idea weldingthem together is the best idea so far. The face of the rotor
must be supported by ribs the same as the bearing support. Rotor faces have
been known to collapse. It behooves us to use the stock phasing gear and all
the stock side seals if possible. I am open to suggestions. The Curtiss
Wright method is a bit heavy. Perhaps we should not go for the lightest
configuration at first but develop the optimum aluminnum rotor one step at a
time.

Paul Lamar



Paul,

I read that portion of the book too. Seems like they just rough mill
them then final grind and lap for accuracy. Then again this book was
written in 1969 when cnc equipment sucked so I bet it is much different
today. Also by there manufacturing methods back then lead to much
variance in the rotors as written about in the compression ration of the
rotors in that turbo 2 design paper you sent out. So two point
approximations and such are worthless these days.

I think aluminum is the wrong direction to go. Talk of any of the exotic
alloys that it would take to make it work well will never happen and we
all know it in terms of cost,saftey,etc. With aluminums greater
expansion rate you would have to increase the clearances which means in
start up and part throttle the thing is going to blowby a bunch.
Aluminum does have a fatigue life and is just to damn soft to take any
kind of oops factor either from assembly or coughing seals. Titanium is
way to prohibitive after re-reading the nasa paper on it useage. Special
coatings on the seal grooves and needing chemical machining to take off
the out layers after heat treating is just unrealistic. Like I said
before Steve B. gave me an idea to do a furnace brazed sheetmetal rotor
after he mentioned they did it on the side housing with good results.
Welding is good but brazing is better since you would be able to get all
the fins/vanes/whathaveyou to join instead of just the outer edges where
the weld can go. A multi-piece steel rotor brazed I think is the
direction we should concentrate on. The only thing to beat steel in
strength to weight, modulus to weight,etc is carbon fibers and titaniums
which are not usable. If you went the extra mile and made the rotors out
of 4340 Hardened to RC 48-50 the yield is 180kpsi and ultimate is
230kpsi. Not much is going to touch that. 1020 or 4130 is more realistic
but with steel you have a ton of choices.

Rob Woods

The problem is wall stiffness. To get it down to the weight of aluminum the walls would have to be one third as thick as cast iron. About the thinnest one can cast in iron is about 3/16 of an inch. That means most of the walls and webs would be 1/16 thick. I don't think that is going to be stiff enough for the rotor faces.

What is the NASA paper you are referring to?. Can you upload a copy?

If Ted wants to do Titanium let him try.

BTW Rob please set your left margin to zero so it does not screw up the forwarding.

Paul Lamar

As far as the ringing is concerned I am not to familiar with the ins and
outs but the connection of the ringing would be pathed through the roll
pins of the gear, journal bearings,etc so I would assume that would take
care of it. I am unaware of any system in an engine that doesnt benefit
from stiffness.  I am against titanium for the reasons i listed in the
reply i just sent.

Rob Woods

It is local panel vibration due to the shock of combustion.
The roll pin loads would definitely be different due to the reduced
mass.

Paul Lamar