Would adding an exhaust-gas turbine generator be worth doing?

The asymmetric expansion of the Atkinson cycle is intended to do the same thing as an exhaust turbine.  It leaves less expansion energy in the gas.

I can't help but suspect that this is all waiting for variable geometric compression ratios.  At low power, the engine operates on the Atkinson cycle at high geometric CR.  For high power the geometric CR is lowered, the volumetric efficiency is increased and a turbocharger with a motor-generator is brought into play.  The motor-generator is used to spool up the turbo to raise boost pressure quickly, then recover excess exhaust energy for the crankshaft instead of dumping it with a waste gate.  This would allow high peak power from a very small engine, cutting weight.  The smaller engine would operate at higher average load, increasing efficiency and cutting idling losses even without start-stop.

Combining Atkinson cycle with turbocharging is nothing new. It is called the Miller cycle, or perhaps more correct, Miller system. It is a way to recover power density when an Atkinson cycle is used. With fully variable valve timing the effective CR can be varied just by closing the inlet valve early (EIVC) – or late (LIVC). Thus, the complexity of geometrically varying the CR can be avoided (there is no really good solution for this). This is also well-known. To me, it is surprising that all these features combined are not yet used. One practical problem is that single-stage turbocharging cannot provide sufficient charge pressure to achieve similar power and torque as a conventional turbocharged engine (if that high power density is necessary could of course be debated). There is less energy left in the exhaust (for the turbine) due to the high expansion ratio in the cylinder and higher charge pressure would be needed, if EIVC (or LIVC) is used, as in an Atkinson cycle. The power needed by the compressor cannot be supplied by the turbine, i.e. we have an impossible match. One way of overcoming this problem would be to use two-stage turbocharging with both intercooler and aftercooler. The intercooler (between the compressors) reduces the work for the compressors, so that the total compression and expansion work in the compressors and turbines respectively could match. Pioneering work with Miller system already more than 25 years ago on heavy-duty engines used this concept. However, two-stage turbocharging is not really something that is readily available for engines of this size. Yet, this is now common for heavy-duty diesel engines. When twin-turbos are used on light-duty diesel engines it is more of a sequential turbocharging, not “fully” two-stage, so that the small two-stage phase is just limited to the cross-over area between the small and the large turbocharger. Furthermore, the smallest engine (as I know of) with twin turbos is the 1.6-liter Renault diesel. Scaling down the technology to a 1.3-liter engine is probably not impossible but not that easy and it is definitely not readily available today.

I acknowledge the fact that Toyota is trying to pursue this route. An Atkinson engine has less power density than a conventional engine and must be scaled up to compensate for this. This reduces efficiency at part load, as also Toyota notes and which I have mentioned many times on this site. Torque dropped from 104 to 96 Nm. Due to the measures implemented, they raised the torque to 105 Nm. However, in an apples-to-apples comparison, these measures could also be applied on a conventional engine. Perhaps the gain would not have been quite as high as in this case (from 96 to 105 Nm) but my point is that one should not be misled by how the baseline was set for comparison. In addition, 105 Nm from a 1.3-liter engine is not really state-of-the-art. Another point worth noting is the increase in CR by two units. There are engines on the market with higher CR than 11.5:1. In summary, I recognize that the maximum efficiency of 38% is quite high but in reality, the “Atkinson-effect” is relatively small and the real gain in a “fair” apples-to-apples comparison would most likely be relatively small. Atkinson engines are best utilized in hybrid systems.

@E-P & Nick
Indeed, the variable geometric CR is already being done. That's what give the extra torque and power at the high rpm end. So, Atkinson-cycle is used at low to moderate rpm for high efficiency and Otto cycle is used at high rpm to maximize power to permit engine downsizing. To mitigate the risk of detonation at CR of 13.5, high rpm is used as well as increase cooling around the exhaust valve and port to lower the temperature of the hot spots.

This engine is also great for full hybrids (HEV) because it permits more engine downsizing, down to a 3 or even 2 cylinder engine, thereby reducing weight and cost, but also allowing more space under the hood for perhaps 1/2 if not all of the hybrid battery pack, thereby allowing more luggage space. This will make HEV's closer to ICEV in term of cost, weight, and luggage space and will increase market share of HEV's.

The soon-to-come solid-state battery that is much more compact and lighter than current NiMh battery used in HEV's, as well as SiC-based inverter, together with this significant engine innovation will make HEV's must-have vehicles! Very significant gain in MPG's, along with increase luggage space AND better handling due to lighter weight of the car, AND better reliability and lower maintenance cost, for purchasing cost comparable to that of a comparable ICEV will dramatically increase EV market penetration.

My experience with the Ford hybrid is that the 2.0 liter Atkinson is also very "buzzy" (I have no idea what RPMs it gets to, as the various dash displays do not include any engine speed or temperature data).

However, if the drivetrain allows the engine to get up to "buzzy" speeds whenever required, would it not be able to generate the mass-flow required to spool up a turbocharger on demand?  Diesel EGTs are quite a bit lower than gasser EGTs, so would not the exhaust energy be sufficient?

Peter_XX comments illustrate perfectly the Atkinson Cycle/Miller System.
Back to Otis question ("... adding an exhaust-gas turbine generator) and possibly correcting the drawbacks that Peter_XX pointed out.
The answer is yes. Recall that Formula 1 uses a electric turbo compounded system (MGU-H) to reduce turbo lag. Both Audi and Honeywell have an motor generated assisted turbocharging systems that should be in production soon.

@E-P,
You're a second person after GasperG who is affected by the "buzziness" of HEV engine. To me, the Prius' engine seems to be smooth and quiet enough that its presence is barely perceptible at any rpm. If the radio (NPR) is on, or if carrying a normal conversation, the engine would not even be heard.

IMHO, Atkinson-cycle engine provides improve efficiency at low to medium engine speeds at high load, the speeds too low for turbocompounder to work efficiently due to low mass flow. Adding a turbocharger to an Atkinson-cycle engine can help increase the power and aid in downsizing further, or to increase maximum power for Autobahn cruising.

An Otto-cycle engine with a turbocompounder to harness extra exhaust energy would not be as efficient at low to moderate rpm even at high load, when the mass flow is still low.

For a given displacement, the Atkinson cycle engine cannot be boosted to the same power level as an Otto-cycle engine for reasons as Peter explained, so it remains a compromise for applications wherein efficiency is a higher priority than maximum power.

Whenever the discussion turns to exhaust energy recovery and the limitations of turbocharging, I keep turning back to an electrically decoupled "turbocharger", more or less the device Aeristech has suggested. Since you aren't forced to compromise the individual characteristics of axially-joined turbine and compressor components, you can optimize a turbine design for the best overall turbine/generator performance, and in turn design an electrically-driven blower that is operated as-needed to provide boost. The "accumulator" exists in the form of an ultracap bank that can provide boost from a dead start with no "lag" experience, and as the turbo-generator comes up to speed it fully powers the blower as well as recharging the caps. Modern switched reluctance technology can already provide the electromagnetics for both devices (without the need for some sort of notional anti-gravity device), with necessary switching speeds no longer a challenge for today's power electronics. E-P you essentially described a form of this with your first post.

One can reasonably posit that a capacitor bank sufficient to also provide creep power (single-digit kW for a few tens of seconds) during engine-off periods. Honestly, with non-exotic engines like the 1.0 liter Eco-Boost producing 100kW/liter today, it is not at all hard to imagine a 1.2 liter 3 or 4-cyl engine in the Camry/Fusion class of microhybrid providing very drivable, under 9 sec, 0-100km/h performance with pleasant NVH characteristics.

Roger, I do think pretty much everything you addressed ("better" batteries, whether solid state or other, SiC inverter technology, and "building in more lightness") are imminent and complementary to the engine improvements mentioned. Batteries still are the best storage medium for non-propulsion needs like A/C, steering, and other hotel loads.

Breaking the 4 liter/100km (60mpg) barrier in the current US "family" sedan in a manner completely transparent to today's driver will happen soon, and without Plug-In option. Add a very reasonable Plug-In storage (Volt-like or better) that permits most driving as EV and offers much more braking recuperation, and we are looking at a very bright set of possibilities indeed, even without a battery Nirvana.

@Engineer-Poet
I have said many times in the past that I do not want your comments but you never seem to respect that. You have taken out three excerpts from my explanation and I just have to say that you are wrong, wrong and wrong. Obviously, you have to read what I wrote thoroughly. I suggest that you read it again, again and again. I will not give a more thorough explanation this time. If you cannot understand what I have written, sorry, I cannot help you.

EP & Peter_XX,

Now now guys, be nice!