How to Design the Complex Shapes of a Rotary Engine

[LtIvan]

Allo, this is my first post here.
I must say this forum has everything not just Physics but Mathematics and Engineering, all my favorite things.

Introduction:
I would like to start with an introduction of a 'rotary engine' for those who don't know.
Unlike the common boxer piston engine with vertical motion going to rotating by the use of a crank and reciprocal other pistons doing the same, the rotary or Wankel, (named by the first to innovate it), it works of a curved faced triangle rotating in an oval shape housing. Each segment of what the triangle leaves is for the three stage cycle; inlet, ignition and exhaust. Here is a link with a gif, (at 'Design').
https://en.wikipedia.org/wiki/Wankel_engine

Summary:
Now when looking at the design of the rotary engine, you notice the triangular rotor moves on an eccentric shaft, (or sometimes known as E-shaft), and most significantly it is curved of each of the faces. It moves in a similar crank motion as the piston engine does but also rotates at the same time. The oval shape housing it self is not in fact an oval but an epitrchoid.
This strange design of the rotary engine is very much important for the engine to work as for the special movement for the rotor and the correct housing for it to fit.

Now this is my question. How would you design such a shape as in both the epitrchoid chamber and the Reuleaux triangle, because it seems to me they are very complicated and need some special attention to design them.

I am a graphics student so I might want to draw it on a graphics board with a T-square.

Conclusion:
So if anyone who knows how to draw these complicated shapes and tell me how is fine.

Cyallo

 

[Nidum]

A very quick search on Google produced large amounts of good information .

Just Google 'Wankel engine geometry' and 'geometry of cycloids and trochoids' .

Drawing these shapes is much less difficult now than when the engine was first designed . Some CAD systems can generate these shapes easily and do animations of the resulting motion ,

The Wankel is only one of a large family of possible 'rotary piston engines' . Whilst there have been many proposed designs over the years most of them have rapidly disappeared .

There was some interest a few years ago in designing a Wankel engine which ran more like a gas turbine than a piston engine . Conceptually promising but never heard any more .

 

[LtIvan]

I found this PDF. You do have to buy it to download it but if you sign up you can read for a free trial.

It was somewhat helpful for the housing but very vague on the rotor shape. It did gave a formula but that's it.
Although I haven't read through it extensively though.

Thanks for the advise.

 

[Nidum]

http://foxed.ca/rx7manual/manuals/REbyKenichiYamamoto-1981.pdf
http://www.animatedengines.com/wankel.html
http://scot.tk/re/Trochoids/Trochoids.htm
https://en.wikipedia.org/wiki/Epitrochoid

 

[LtIvan]

I think I most likely don't have the right equipment to design such a shape to do splines and have the technique of a cog going around another.
I found this tutorial to draw a rotary engine on Autodesk Inventor.


But thanks Nidum for your input.

 

[David Macclain]

I would probably make two circles, one for the inner radius and one for the outer. I'd then make a construction line from the center of the outer circle to the edge of the inner circle, place a mid-point on it. Circular pattern this mid-point around N times about the center of the outer circle. I would then circular pattern those patterned points about the inner circle N * inner_radius/outer_radius times. This should give a full set (and more) of points to attach a closed spline to.

 

[dean barry]

CAD is recommended
Use a known crank radius : rotor centre to tip radius ratio of 1:7 (mazda)
Draw a vertical line up from your crank rotation center = to crank radius (1 unit)
Extend this line by 7 units to represent distance from rotor centre to rotor tip.
Rotate the crank line say 15 degrees clockwise, but only rotate the rotor 1/3 of this amount (in this case 5 degrees)
(remember they are still connected)
Mark the point where the tip of the rotor is
Repeat rotation increments until you have created 1/4 of the casing profile (270 degrees of crank rotation), join the dots and then you can mirror the image twice to complete the full casing.
I recommend you do this on a CAD, I believe there are line smoothing algorithms to smooth the created series of lines
(you have to amalgamate all the individual lines into what's known as a polyline before applying the smoothing algorithm)

 

[dean barry]

Here's the sketch to go with post 7.

 

[dean barry]

note the deliberate mistake, I used 30/10 degrees on the drawing and 15/5 degrees in the description, either will suffice, one is more refined than the other.

 

[LtIvan]

Allo
I have just gotten back.

I read carefully through your instructions Dean and got through to this (fig.1).
I use a crank rotor ratio of 10mm:70mm and did the 30/10 degrees.
I was then able to draw a full template of a housing (fig.2).
I ended up drawing two for the outside and inside of the design. I then finished up with coolant gaps and holes for holders. This is mainly supposed to be a basic design nothing to serious. I didn't go that much into the full detail of things because I was mainly focusing on getting the shape right.

It may look a bit rigid around the edges of the eptrichoids mainly because I wanted to blend those corners out.

Thanks for the input.

I guess the rotor would be next anyone got any tips for that.
perhaps draw an equilateral triangle touching the edges of the housing with one face facing the side then doing an arch so the vertex touches the edge on that one side, then repeating the radius again on the other two sides? haven't tried it out yet.

 

[Rx7man]

Very interesting... I've wondered about this for some time!

 

[LtIvan]

Allo

Housing:
I have done full orthographic drawing this time with a lot more detail and dimensions.
(fig.1, fig.2, fig.3).

Rotor:
I think I have found away to draw the rotor and I will have a drawing done.

 

[dean barry]

Looking good, I like the drawings, can you get access to CAD ? your local college may be able to help, CAD is hard at first to learn the basics, but fantastic once you get the basics, Microsoft AutoCAD is recommended (its an industry standard) 2D and 3D (surface and solid modelling) are part of the package and it takes the sheer hard work out of technical drawing.
Good luck

 

[Rx7man]

Microsoft Autocad? I think that's Autodesk Autocad.

Might want to check out Solidworks or Autodesk Inventor too

Autocad is a steep learning curve to get it to do what you want it to do, especially if you start trying to do things in 3D.. in 2D its a lot simpler.

This is the bumper for my truck, It took me months to get the design just right, and then I got the 3/16th aluminum plate CNC plasma cut and welded it all together

The finished product.. with a day of welding and a day of grinding

There's no way I could have made it fit and finish as nicely doing it by hand

 

[LtIvan]

Allo

CAD:

Looking good, I like the drawings, can you get access to CAD ? your local college may be able to help, CAD is hard at first to learn the basics, but fantastic once you get the basics, Microsoft AutoCAD is recommended (its an industry standard) 2D and 3D (surface and solid modelling) are part of the package and it takes the sheer hard work out of technical drawing.
Good luck

I have work on Autodesk AutoCad and Autodesk Inventor on my school computer labs for awhile and I am quite familiar at Inventor. However for the moment at this stage I do not have access to CAD since it is the holidays, (this task was to keep my drafting skills occupied). I will though be getting a laptop soon which would be able to do AutoDesk Inventor and AutoCad.
Here are some examples of what I have done (fig1, fig4).

Rotor:
(fig2, fig3).
As for the rotor it look good. I just drawn an elevation and depression of 30° at the bit where the shape dims inwards a little. These should help create an equilateral triangle. Then just draw an arch from that same point. You should find that the radius is the same length as the sides on the triangle. This arch should touch the the side on the housing.
Even though I am confident this is (approximately) right, the side might not be a perfect arch. In the video on how to make it on Inventor, the poster made an ellipse to make the side.

 

[Rx7man]

OK, this prompted me to try it out in CAD... I used 15 degree eccentric shaft increments, the eccentric shaft is off-center by 1 unit, the distance from the center of the eccentric base circle to the tip of the rotor is 7 units

As suggested, I drew and rotated lines and then connected them together, mirroring twice to get the full housing... that was the easy part.

Secondly, I drew the line from the uppermost centerpoint of the shaft to the top of the housing, rotated that line ±60 degrees and extended it, then drew a connecting line between where each one crossed the housing, this gives me the 3 tips of the rotor, and the center point.
Next I drew an arc with the centerpoint at a tip of the rotor and a radius equal to the length of the side of the rotor and placed it in the housing.
Incrementally, I rotated the rotor along the same path I used to generate the housing (5/15 degree increments), and whenever any part of the rotor extended past the housing, I'd trim that section off and replace it with the intruding part of the housing..

rinse, repeat, repeat, repeat!

Throw away one half of the remaining arc, mirror it along the centerline of that face, rotate ± 120 degrees to finish the 3 faces...

I couldn't stop there, so I drew in some apex seal grooves, corner seals, side seals and oil control rings.. The rotor bearing has a minimum unit size of 2.

Here's a screenshot.. I worked my way from top left to bottom right.. should give you an idea.. I didn't play with smoothing the lines

 

[LtIvan]

Next I drew an arc with the centerpoint at a tip of the rotor and a radius equal to the length of the side of the rotor and placed it in the housing.

Accept for the first bit on the finding the three points, I did exactly the same in Post 15.
But the CAD models look good.

 

[Rx7man]

Yeah, the posts were 3 minutes apart.. I had already drawn everything.. didn't notice you had written something until I submitted mine :)

It might be possible to get a closer approximation some other way.. perhaps with an elliptic arc?

 

[Rx7man]

Fiddled a little more and made it 3d.. no cooling system or compression divet implemented yet.. who needs that anyhow

 

[LeeK9]

your in luck, i built a model rotary engine as part of my second year project in my Mechanical Eng degree.

it is a difficult shape to machine but i have the two driving formulas which locate the tip of the rotor at any point in the cycle.

here is my solidworks model

i drew my rotor pocket on excel with the equations

 

[LeeK9]

here is a quick video of my model in action

 

[Rx7man]

I just drew up an excel spreadsheet with the formula for finding the tip at any point in the rotation... Once I thought about it, it was really easy actually.

Here's a screenshot with the formulas for each column at the top

First you define the angle increments for the E shaft and rotor in degrees and radians, Then calculate the center point of the eccentric shaft X and Y coordinates, then the relative location of the tip of the rotor to the rotor center, sum them to get the relative location of the tip to the E shaft, and graph them. If you look at the images I posted earlier of my CAD drawing it may be more apparent how it works. You can also play with the rotor ratio and see how it affects the shape of the epitrochoid.

I haven't figured out a way to mathematically draw the rotor

 

[LeeK9]

The formulas for the X and Y co-ord in my model were taken from the Yamamoto book.

Yamamoto, K 1971. Rotary Engine, Toyo Kogyo Co ltd, Tokyo.

X = e.Cos(3(O))+RCos(O)
Y = e.Sin(3(O))+RCos(O)

where
e = eccentricity
R = Rotor radius
O = Angle in radians

the angle is progressed from 0-360 at 0.5 degrees increments which is then converted to radians to be input in the formula above given 720 X and 720 Y points

The rotor in my owl model was somewhat fudged in solidworks to be as small a radius as possible without fouling against the housing, this was not a critical dimension for me as it was just a model for the demonstration of the principle.

here is my full report documenting the full build of the model, for design calculations see page 24 and onwards.

https://www.dropbox.com/s/dld9nqlf044ml7y/Wankel Rotary Design Report 2014 PDF.pdf?dl=0

 

[Rx7man]

Those formulas are pretty much identical to what I drew up, I just separated the steps a little bit and used a ratio for the rotor size instead of a direct value.

 

[LtIvan]

CAD:
I have got AutoCad and made a simple 2D sketch for the housing, (fig.1).
I will have to make a 3D model soon but since I am mostly used to Inventor and relied on it for more in depth 3D than AutoCad I will have to get my self used to the AutoCad.

 

[LtIvan]

CAD:
I have done a concept of a housing and rotor, (fig.1, fig.2).
Still trying to get used to the 3D like having trouble trying to understand to extrude cut like how you would in Inventor. I have learned subtracting into models but limits me. Also how would you render a model to make it look more metal like?

 

[Rx7man]

It looks good so far... I really have NO clue as to how to shade things... I know you can set up lighting of all sorts and there's a huge material library, but I never bother with that. There's got to be some books and websites that will help with that though.

 

[LtIvan]

CAD:
Here I have done a details into the coolet jackets, (fig.1, fig.2).
Then I did a much more detailed design for the rotor using three-point-arch instead of a radius one for the edges. I did this because after testing out the rotor shape on the housing the arch does not actually touch the sides all the time so I used three-point-arch. Any suggestions on how to do this with an elliptical arch for I have found it is different to how the person in the video did it on Inventor. Besides the edges of the rotor I have also done more detail with the apexes adding round slots, (by the way what are they for?). I also added arch seal slots and spaces on the side to allow movement. A big change is the combustion cavities or divots on each side. This was done by extruding a block of the base shape out from the rotor and subtracting it, (fig.3)

AutoCad Mechanical:
Also would you recommend me to get AutoCad mechanical? Does it have something similar to inventor where you can assembly individual separate parts?

 

[Rx7man]

I think the "rounded slots" are for the "corner seals", and the arched slot is for "side seals".. Usually there is a pair (one may suffice at least for demonstration purposes) of oil control rings.. you have to keep the oil and combustion separate, the steel seals (Side and corner seals) prevent pressurized gasses from getting into the oil, the apex seals prevent gasses from mixing on different sides of the rotor, and the oil seals prevent oil from getting into the combustion chamber. All the seals have a 'kinked metal' kind of spring behind them to keep them pushed against the housing. Go to www.mazdatrix.com.. I think they have all the parts, with pictures.

The arch of the rotor doesn't really need to be super close the the 'peak' of the epitrochoid.. the critical thing is it can never touch.

I've never used Autocad Mechanical.. I just got 2016 and it's got a few things that nerve me, but that's probably because i haven't fiddled with the settings enough.. What drives me ballistic is when I want to draw a circle and it draws it on the wrong plane!

 

[LtIvan]

I think the "rounded slots" are for the "corner seals", and the arched slot is for "side seals".. Usually there is a pair (one may suffice at least for demonstration purposes) of oil control rings.. you have to keep the oil and combustion separate, the steel seals (Side and corner seals) prevent pressurized gasses from getting into the oil, the apex seals prevent gasses from mixing on different sides of the rotor, and the oil seals prevent oil from getting into the combustion chamber. All the seals have a 'kinked metal' kind of spring behind them to keep them pushed against the housing. Go to www.mazdatrix.com.. I think they have all the parts, with pictures.

The arch of the rotor doesn't really need to be super close the the 'peak' of the epitrochoid.. the critical thing is it can never touch.

I've never used Autocad Mechanical.. I just got 2016 and it's got a few things that nerve me, but that's probably because i haven't fiddled with the settings enough.. What drives me ballistic is when I want to draw a circle and it draws it on the wrong plane!

I knew what the main purpose of the seals were, just not the round corner seals.
So it has nothing for assembling? like in Inventor?
I have encountered such a thing in AutoCad where it draws the circle on the wrong plane when trying to draw the template for the divots. I ended up drawing a box and filleting it, that might provide a temporary solution. Does anyone know how to fix that?

 

[Rx7man]

I know about 1% of Autocad's functions... and that may even be optimistic..

When a circle gets drawn on the wrong plane I usually just 3drotate it back to where it's supposed to be.. .I think it has something to do with dynamic UCS.