This idea makes use of an uncommon, but *probably* not
novel linkage for converting between rotary and
oscillating motion.
For the sake of discussion, let's assume that our rotating
drive shaft is oriented vertically. Around it's outer
surface, create a groove with a sinusoidal shape, with
two peaks and two troughs. The groove has rotational
symmetry of order 2, around the drive shaft's axis.
Surround the drive shaft with a ring, with two inward-
facing pins. These pins are 180 degrees apart, and fit
into the groove on the drive shaft. The ring is
prevented from rotating using some sort of linear
bearings, and is allowed to move parallel to the drive
shaft.
Rotating the drive shaft will cause the ring to oscillate
perpendicular to it, in a sinusoidal manner.
If you want, you can think of this assembly as an unusual
form of swashplate or cam.
Next, we add a couple of vertical rods, parallel to the
drive shaft, and link them to the ring either rigidly, or
using something like a scotch yoke. The purpose of the
yoke (if present) would be to prevent the ring from
exerting any horizontal force on the rods, and only
allow vertical force.
Each rod has an upwards-facing piston on it's top end,
and a downwards-facing piston on it's bottom end.
For balance (both static and dynamic) we add to our
drive shaft a second groove and ring (below the first
groove and ring), and two more bars, and four more
pistons. The second pair of bars and quartet of pistons
should be 90 degrees away from the first set. The
phase difference should be such that when one ring is
moving up, the other is moving down.
Note that the distance from the first ring to the pistons
above it is shorter than the distance to the pistons
below it. Similarly, the distance from the second ring
to the pistons below it is shorter than the distance to
the pistons above it.
Finally, add a four cylinder engine block above, and
another below. Each block's cylinders should be
arranged in a square around the drive shaft.
If desired, the cylinder heads can use ordinary poppet
valves actuated by an ordinary cam (which might be
located directly on the drive shaft, since the pistons
already make four strokes per rotation), or for less
vibration and higher rpm, we could use rotary valves.
To reduce friction and wear on the grooves, we can
replace each pin with either a sliding block (as we
might do in a scotch yoke), or a pair of wheels (one
facing upward, and the other downward). If wheels are
used, the grooves should be beveled, and have a
varying angle to that bevel, so that each wheel's RPM is
some constant multiple of the drive shaft RPM.
The groove shape doesn't need to be a perfect sine
curve, for the same reason why, in real world internal
combustion engines, scotch yokes are no more efficient
than crankshafts. That is, more time at TDC results in
more heat conduction to the engine block, which
balances out theoretical efficiency improvements. If
you want to, you can give the groove a shape which
mimics the motion of a crank and slider.
Advantages:
I'm positive that this engine can be more compact than
a conventional 8 cylinder, though I'm not sure by how
much.
Higher max RPM, due to balanced moving parts.
More power at any given RPM. Why? Each quarter turn
of the drive shaft results in eight simultaneous strokes,
four up and four down. This results in two power
strokes per quarter turn, and eight per full rotation!
Lower internal losses. In spite of friction which might
occur in the linear bearings (which prevent the rings
from turning) and between the pins and grooves, the
reduction in piston side loading, and elimination of the
wrist pins and crank pins should more than make up for
it.
Reduced piston and cylinder wear (due to lower side
loading), thus less maintenance.
Disadvantages:
It's weird and different, and the automotive industry
doesn't like doing new things.
I don't know whether or not it would be lighter than a
conventional 8 cylinder engine.
At low vehicle speeds, it might be necessary to shut
down the top or bottom half of the engine, for fuel
efficiency.
More cylinders is possible (12, 16, etc.) but fewer is
not. More than 12 is just silly, though.
If the drive shaft is vertical, then half the engine (one
of the two blocks) is upside down, potentially making
maintenance very awkward. OTOH, I only described it
as vertical to simplify the explanation; horizontal would
be more practical for most purposes.
Hmm...
Other advantages and disadvantages would be
appreciated!
Also, if anyone knows the proper name for the linkage I
described, that would also be appreciated. If it's really
and truly new, then would somebody lend me a few
thousand dollars so I can go patent it? ;) The most
similar thing I've seen is here [link], but that's kinda
inside out (both literally, and because on that page, the
cam is stationary and the engine block rotating).