[edited]
One of the major problems with Wankel Rotary Engines, is that due to the
fact that one side of the housing is hot, and the other side is (relatively)
cool, there is uneven thermal expansion.
This uneven thermal expansion causes distortion, which causes stress on
the housing, and uneven
wear on the rotor seals and on the interior of the
housing. It's also a big cause of inefficiency, since gas will leak past worn
seals, which reduces the effective compression ratio.
This idea is to use two separate rotors, and two separate housings. One
rotor and housing is used for compressing the air or air/fuel mix, and the
other acts as the actual engine. The number of power strokes per rotation
is the same as if the two "engines" were separate, however, all of the
power strokes occur in only one of the two housings.
The compressor has two input ports, one of which is where a normal
Wankel engine might have it's input port, and the other diagonally
opposite it. There are also two "crossover" ports, one where a normal
Wankel has it's exhaust port, the other opposite it. To simplify engine
geometry, the crossover ports are both on one face of the housing, not on
the edge. It may be necessary to have check valves over the crossover
ports, letting gas out but not in, but I'm not quite certain about this.
The main engine's housing has two intake ports, both on the face of the
housing, and both connected (via short, sturdy pipes) to the crossover
ports of the compressor. Naturally, there are also two exhaust ports --
these are probably on the face of the housing, not on the edge, but I'm
not sure how important that is.
The rotors share an axle, and are slightly out of phase from one another,
so that each compression event completes shortly after the main engine's
intake/combustion event. The amount of phase difference determines
the system's compression ratio -- a smaller phase difference results in a
greater compression ratio.
The main engine has two spark plugs, located near their intake ports.
Each plug fires just after the nearby intake port is uncovered. If the
fuel/air mix comes in through the port faster than the flame speed, the
flame cannot propagate backwards into the crossover tube, but will
instead be basically stationary.
With combustion occurring during the overlap of the end of the
compression event in one housing and the start of the expansion event in
the other, it is nearly isochoric, which makes this similar to an ideal Otto
cycle.
A stationary flame front should allow for complete combustion of the
fuel/air mix, eliminating unburnt hydrocarbons from the exhaust. Also,
there's no possibility, at all, of engine knock.
After the compressor's crossover port is covered by the rotor, or the
check valve is closed by gas pressure, the main engine's rotor continues to
expand it's heated gases, and then exhausts them.
For greatest efficiency, the main engine should have a slightly larger
displacement than the compressor, so that the working gas's pressure
equals atmospheric pressure just before the exhaust port is uncovered.
This optional feature would make the cycle more similar to an Atkinson
cycle. Besides being more fuel efficient, it be quieter.
Theoretically, the intake ports for the compressor could have a variable
geometry, eliminating the need for a throttle valve, and thus eliminating
throttling loss. In reality, if this were possible, Mazda would already be
using it in their rotary engines. Eh, whatever.
If gasoline direct injection is done, with one injector (or set of injectors)
placed in each crossover pipe, it should be easy to produce a stratified
fuel charge. This would allow us to have a relatively rich mixture near
the spark plug when it fires, but a lean or stochiometric mix everywhere
else. This should also completely eliminate any chance of preignition. It
also protects the injectors from the heat of combustion, which happens in
the main motor, not in the crossover pipes.
Cooling is simpler than in a regular Wankel engine, since the compressor
will be fairly cool, and the main motor fairly hot. In contrast, in a regular
Wankel engine, half the housing is cool, and the other half hot.
Basically, the engine would have all the positive features of a Wankel
engine, but with greater efficiency, and a cleaner, quieter exhaust. The
mechanical complexity, and such things as torque, power, and redline,
would be about the same as that of a pair of Wankel engines. Sealing
could probably be simplified, since the housings expand fairly evenly as
they warm up, unlike an ordinary Wankel engine which expands
unevenly... a Mazda rotary engine has about a hundred seal-related parts,
to compensate for the uneven expansion.
The compression ratio could probably be made rather higher than that of
a regular Wankel engine (since this engine is highly resistant to pre-
ignition and knocking), which would boost thermodynamic efficiency, and
probably boost torque, power, and overall efficiency. On the other hand,
increasing CR increases friction and mechanical inefficiency... testing
would need to be done to determine the best ratio.
I've read that ordinary Wankel engines are sensitive to back-pressure, and
don't take well to being turbo-boosted. I'm not sure why that is, but it
would be a good idea to see if this engine would benefit from a turbo.
The Scuderi Split Cycle Engine [link] is prior art. And of course, so is the
Wankel engine [link]! :)