I don't expect Diesel to over take gasoline. The gas engine still has a long way to go. It's just not as efficient as a diesel engine. Like I said, keep a watch for Honda, which happens to be the cleanest and largest engine manufacturer in the world.
Honda has done it before: the 1974 CVCC engine was truly revolutionary, and proved that emmission rules could be met by a standard gasoline engine. A truly remarkable engine.
I'm not convinced by diesel, however.
The gas engine still has a long way to go. It's just not as efficient as a diesel engine. The present design of gasoline engines, even in the complete absense of frictional losses, would be less than 50% efficient; in any engine where the expansion ratio is no greater than the expansion ratio, more than half the energy of the fuel will either be lost as friction or waste heat to the engine block, or will else go out the tailpipe.
There are many techniques that could overcome the theoretical efficiency limitations of the Otto Cycle engine. I wonder which ones will end up being practical? A few I like:
- Use some sort of energy-harnessing device (e.g. turbine) in place of the throttle, and either include a butterfly valve for use in wide-open-throttle situations or design the device for two-quadrant operation (e.g. so it can convert airflow into electricity, or convert electricity into airflow) so it can act as either a throttle or a supercharger. The purpose of a normal throttle is to waste energy; any energy captured from the process would be 100% "free".
- Use delayed intake closure instead of throttling except when running the engine at the very bottom of its power curve (when delayed closure alone would not suffice). When throttled to 90% atmospheric pressure, an engine must expend energy drawing vacuum through its entire downward stroke; it only gets 10% of that energy back at the start of the upward stroke. By contrast, an engine using delayed intake valve closure wouldn't have to waste any energy in deliberate frictional losses, though incidental frictional losses would still exist (and could pose real issues).
- When operating in non-peak-power conditions, use the two center cylinders together as a compounding cylinder for the other two cylinders (they would together serve one end cylinder on one stroke, and the other end cylinder on the other). A Freeper wrote a white paper on that one; sounds like a neat concept.
I don't know what all the technical difficulties are with such approaches, but there certainly seems to be room for improvement.