This is an account of the research and development of an experimental
high-power, high-torque efficient and balanced 2-stroke Engine.
The goal is to follow prior research in this field, in areas which are *not* part of the standard "proven" Otto-Cycle or Wankel Engine design, and to find out if the designs actually work and provide the benefits that they claim.
An additional goal is to keep the engine as simple as possible. As an unskilled machinist and amateur enthusiast, the build should be kept to within the capabilities of unskilled machinists and amateur enthusiasts. Any special parts should therefore be kept to a minimum and be easy to make by sending off DXF files to a CNC machine shop, or available as off-the-shelf parts.
The primary inspiration for this engine comes from the Bourke Engine.
Russ Bourke's design was done at a time when it was fully recognised that
the Otto Cycle was a massive compromise. He was living at a critical
point in world history where influential innovation was still just about
possible, and where numerous novel engine designs, as people transitioned
from steam to hydrocarbon-based engines, were still within living memory.
Sadly, however, just as Bourke made each breakthrough, events conspired
against him, including Pearl Harbour occurring days after signing a major
contract to supply the U.S. Army. By the 1960s, he had given up,
and wrote a documentary instead. The engines which had been seized by
corrupt individuals in 1952 were discovered over thirty years later, renovated
and shown to be in perfect working order.
Whilst numerous individuals have replicated or near-replicated the Bourke design, none have made any significant mechanical improvements on it or used its best aspects and combined it with other advances since. So, this is what will be done, here. The most significant technological advance over the Bourke design's scotch yoke has to be that of the Revetec counter-rotating cams, which are both beautiful, elegant and easy to manufacture (albeit horrendously complex to design).
The advantage of a scotch yoke compared to a crank is very very clear: increased dwell time at TDC, and no side-wall loading. The opposing pistons are effectively joined together as a single unit, via a shuttle. However, even with Bourke's triple-sleeve bearing invention helping out there is still some side-loading on the shuttle assembly around 90 degrees to TDC and BDC, resulting in at least one patent (in 1974) showing that people did attempt to improve the original Bourke design. By contrast, however, a pair of counter-rotating cams act in tandem with a scissor-action that completely eliminates all side-wall loading. There is literally no movement except in the direction of the piston shuttle at all.
So in many ways it would be foolish to not see what happens when the best aspects of the best available engine technology are combined together into one design. Especially given that Revetec's designs are particularly efficient already, as were Bourke's.
In this illustration, it can be seen that the Bourke design is a 2-stroke ported "opposing" design. Air intake is seen in purple. Water in light blue. Scotch yoke, crank and pistons (effectively a single assembly) are shown in green. Combustion and exhaust are shown in red. Just as in a standard 2-stroke (and a steam engine) ports line up to suck in the air-fuel mixture in from one set of ports and allow the exhaust gases to escape from a second set. A third set of ports is there to transfer the air-fuel mixture from the underside chamber into the compression chamber.
This shows a trilobe which has been calculated from a 20mm diameter bearing rolling in a triple-sine-wave radius with a maximum throw of 60mm and a minimum throw of 30mm. John Rowe's U.S. Patent 5605938 Figure 9A is a good example of what happens when the size of the rolling bearing is increased to become a significant fraction of the maximum diameter of the cams: the cams become much more triangular.
Later variants can take into account the falling pressure against the piston, compensating by increasing the angle of descent of the cam's face - but for now, a relatively simple 50 line python program has been written which would place the bearing into a sine wave pattern, assuming that the output shaft has constant angular velocity.
In the original analysis, because there are twin cams and four CAM Roller
bearings in contact at all times, it was believed that on the opposite
side of the explosion, the Trilobe CAMs would be pushed in the opposite
direction, thus counter-balancing the "twist" that would be imparted.
Unfortunately, this was completely wrong.
The solution is to use triple cams, similar to multi-bladed scissors.
The addition of top and bottom trilobe CAMs sandwiching a middle one
that spins in the opposite direction results in a balanced load on the
CAM Rollers, and stops any potential twisting of the CAM shuttles.
Here, although the third set of CAM Roller bearings have not yet been added, the yellow Trilobes rotate together in the exact same direction, whilst the light-red one rotates in the opposite direction. Now the shuttles will not twist about, which would cause significant wear on both the Trilobes and the CAM Rollers over time.
Originally the plan was to go with a 45mm displacement and a 50mm diameter, to give around a 70cc capacity. However after some consideration, this really is a bit big, especially as the top is to be held with four TR8 3D printer trapezoidal lead screws. These have to take the full force of the explosive compression. All of the dimensions now need adjusting which is a pain, but it is all parametric.
On the TODO list: find a libre licensed Engine Controller, full kit including spark plug. Fuel Injection option for later would be really nice.
Also measured a broken snowmobile cylinder, to check out the piston diameter and ring gap spacing. Bill also informs me that the inside of the cylinder needs crosshatch scoring, so that oil can get into the grooves. This could also work well for the spring loaded cylinder in between the two pistons. To explain: unlike in an ordinary 2 stroke where the piston has skirts which cut off air intake, a second (lower) piston head is to be used, which of course means that in between there must be no air allowed. A full seal is needed, so imagine a normal split piston ring but make it a full 40mm high.
This led to the realisation that, actually, the 3D CAD work would be easier than previously imagined (and, after 5 part time days is 90% complete). The entire casework is being designed from 5mm lasercut steel, with wood style "slots" and bolts to hold sandwiches together. The only exception is the Cylinder Liner which will need holes drilled then bungs welded on, as wellas other small pipes (some as spacers).
The trilobe CAMs have been reduced to a sine wave, for now, and double ring SKF CAM roller bearings located.
Holding the trilobes is slightly problematic as they have to be
perfectly aligned. The holder is again a pipe plus a lasercut sheet,
welded together, then 6 grub screws balancing it out and the 3 bolts
holding it onto the trilobes. This has to go inside the shuttles, so
ends up making the shuttles slightly larger than anticipated. However
it is better than welding the shafts directly to the trilobes.
Bill, who has done materials science, has advised me that to get the trilobes hardened, what you do is oversize them, case-harden them by rolling, then have them re-machined to actual size. I also want to get them nickel-plated. Therefore it will be quite important not to mess them up, hence the bolt-on shaft holders.
The Cylinder liner, mechanically, is a pain. Stainless Steel is a nuisance
to drill: Bill informs me that it is so hard that drills have to be run
at only 200 RPM, with plenty of oil. A 3mm hole can take an hour. We
have eleven 8mm holes per liner to drill.
The reason is because the cylinder is only around 50mm in diameter, which is quite small. Larger bung holes would result in the piston top being exposed for a considerable part of the stroke.
We initially considered just going with slots (so that displacement is greater), however that would require the construction of boxes that would need welding on the outside. Perhaps this is worth considering, although cutting slots in 3mm thick Stainless steel without damage is not very convenient.
The variable compression end, the current arrangement (which is a pragmatic
one) is to use TR8 lead screws and brass TR8 nuts commonly used on 3D
printers. Four of them, one in each corner, will give stability. There
is currently a ring missing (cyan) which creates a sealed chamber that
can be water-cooled, with the spark plug going into the centre. The four
TR8 nuts will each have a sprocket bolted on, and a chain goes round all
four and can be driven to move the entire assembly (cyan) in and out, to
alter the compression ratio.
The venturi still needs to be sourced and designed. This is basically a "paint sprayer" tube arrangement, where fuel is drip fed into a connecting tube between the transfer ports (a standars 2 stroke arrangement). The venturi accelerates airflow, causing the fuel to vaporise, just like a paint sprayer.
Also under consideration, is the inclusion of US Patent US1339176A, from 1921, by Leonard Dyer. Leonard came up with the 6-stroke design, which apparently, on its own, is 40% more fuel efficient than a 4 stroke, due to the hugely reduced internal temperature as well as clear air due to in effect two scavenging and exhaust cycles. This design being a 2 stroke, with no valves, will need a little thought.
Here is a picture of the gear arrangement for the trilobe cams:
The problem is this: right in the middle of the picture, it can be seen that the shafts going through the trilobe cams are floating in mid-air. Also, as you can see from the image (below) when there are 6 cylinders, there's no real easy way "in" to support the ends of those two shafts, because there are 3 sets of shuttles in the way. Sure, it would be possible to bring in a bar or two (or three) offset at 30 degrees (giving 12 spokes) but that's getting complicated, and also supporting those bars might get in the way of the gear shafts going through the whole thing. or, damn it, the gears have to be made large enough (read: expensive) to reach all the way to the outer edge of the hexagonal box that joins everything together.
So what the proposed solution is, is to make a cylinder that contains 4
bearings, separated by two spacer cylinders. Two of the bearings will
be at the ends of the outer cylinder, and two of the bearings will be
right next to each other. The basic idea is to fit everything together by:
* Dropping in one gear+shaft+cam into the bearing in the centre of the hexagonal box
* Dropping the 3 piston shuttles on top
* Slotting the cylinder-thingy over the shaft attached to the first trilobe cam
* Dropping the 2nd gear+shaft+cam into the cylinder-thingy
* Putting the top hexagonal lid on.
This would mean that the two shafts would effectively be one unit, but just each half rotating in opposite directions, hopefully keeping the whole lot steady, assuming that the 4 bearings are jammed tight in the cylinder. If that turns out not to be stable enough (too much force on the cams causing the two shafts to go off-centre) then it's always possible to look at squeezing in some support bars, attached directly to the cylinder.
The crank-case is to be made from layers of 12mm laser-cut steel, in order to make it low-cost and easy to acquire and assemble. Laser-cutting does up to 0.2mm accuracy. Accuracy is not a big concern, here, with the possible exception of the hole for the piston rod. Even here: if oil-soaked leather is to be used for the seals, it's not a big issue, because actually the leather needs to permit a small amount of oil to drip into the piston chamber. Here are the plates - been designed to fit the 70mm big-bore cylinders:
The bolt-holes are 8mm width, and are continuations of the holes through the cylinder head. The top-left plate is the one that acts as an air-intake. After this has been tested, it will be interesting to see if the concept can be extended to a complete cylinder, using a steel liner to keep them all together.
After a long night, and tracking down both LibreCAD and sdxf.py it was possible to cut out OpenSCAD from the equation and go directly to a DXF file. Also, rather than use an O(N-cubed) loop, an O(N-squared) loop was used instead, alongside solving a quadratic equation by substituting y=mx+c into the equation of a circle - y-squared + x-squared = r-squared. This gave a pair of points where the line (projection of the cam face for a particular angle) intersected the bearing:
m = y/x a = ((m*m)+1) b = 2*(-xN-(m*yN)) c = ((xN*xN)+(yN*yN)-(radius*radius)) b2 = (b*b) ac4 = 4*a*c if b2 < ac4: return a1 = math.sqrt(b2 - ac4) x1 = (-b + a1) / (2*a) x2 = (-b - a1) / (2*a) y1 = x1*m y2 = x2*mBy covering in incremental steps all positions of the bearing, the cam face could be "shaved" in a similar fashion to CNC milling.
packrat from CNCZone recommended the use of a HEI Module which will need evaluation. thank you!
The piston and cylinders arrived, and a bit of thinking done on how best
to ignite the fuel. One way would be to use spark plugs (14mm thread)
then ignition coils and an ignition rotor with a magnetic-triggered relay.
However: a much simpler way, and one which is more in line with the end
design, is to use dieselling. 14mm off-the-shelf glow plugs for scooter
cylinders however simply don't exist. Instead, an accidental search revealed
14mm to 1/4-32mm adaptors.
The use of these adaptors would allow a standard R/C model glow-plug to be
used. Whether this will produce enough heat to ignite the fuel for such a
large engine is going to be a matter of experimentation.
Cams: the CNC company in Middlesex had already tried making a plastic version of the cams, from a DXF file sent to them last week. Apparently it worked great, but the DXF file generated by OpenSCAD had absolutely no scale, and it was made up of "points", not lines.
Gears: hpcgears.com is a company that many people contacted when "Robot Wars" was popular. They have an extremely competent technical team, and they are evaluating the drawings, to advise of the best gears for the job. More information on Monday.
Carburetor: Steve, of Recycle Scooters, advised the other day on the parts needed to get ignition going, and, although it strictly is not needed for the final design, a decision to go with a carburetor for the first prototype has been made. So, Steve is looking out for a simple 125cc 2-stroke Carburetor that can do very lean mixtures.
Crank case: now that the cylinder top is here, a crank case needs to be designed around it. Using full 3D CNC machining for something like this feels somewhat inappropriate, so instead a series of "slices" that can be made with laser cutting will be made. They will either be sandwiched, using long bolts, or simply welded. A simple design with 3 15mm layers should suffice.
Cam followers: with a bit of messing about (i.e. inverting how they're put in) these 22x12x10 KR22B "budget" cam followers should do the trick, although this is still up in the air. These 10x22x6mm W619002Z bearings can apparently do up to 36,000 RPM which might be more appropriate.
After contacting enthusiasts on the
who have been enormously helpful, a decision was made to make the first
engine out of off-the-shelf 2-stroke cylinders. One person recommended
motorcycle cylinders, however after some calculations which pointed at using
small cylinders, it was decided that 2-stroke Scooter cylinders might prove
more appropriate. As it turns out, 2-stroke "Big Bore" Cylinder Kits are
available on ebay, brand new, for £32 + £15 shipping! This is a fairly
major breakthrough, not just because it simplifies the build process, but
also because anyone else also wishing to experiment with engine designs can do
so, knowing that the hardest part - machining a cylinder and matching piston -
is entirely taken care of.
The next most difficult part also turns out to be easier than it first seems. One google search for "CNC Machinist Services UK" turned up a no-nonsense guy in Middlesex named "Gary", who was amused but not put off by an email with a hand-drawn illustration of a trilobe cam. One phone call later and I had the crucial information I needed: a one-off pair of 120mm diameter cams made out of mild steel 8mm thick could be machined for £50 each - all that Gary needed was a DXF file.
Ordinarily, most people who are not Software Engineers would go and find some ultra-expensive CAD/CAM software, and either spend a fortune on it or rip it off, then spend months wondering how the hell to use it. However, as a Software (Libre) advocate, there is a cat in hell's chance of that ever happening, so I dusted off some python code I had written 8 months ago which uses PyOpenSCAD, fired up OpenSCAD, and, 120 lines of python code and 2 hours later, thanks to the help of the people on the openscad IRC channel, I had a DXF file which Gary should see when he checks his messages in the morning.
Total required budget so far, covering the two major most complex components: £194. stunning.
|Moto Story||70mm Big Bore Cylinder Kit|