Technical help - combustion and C/R

[Barry Hart]

Having been impressed with the technical standard some of the comments following our posting about our 3 litre 944 turbo, and being at the stage to start another more powerful version - there are a couple of technical issues I would like to know a little more about and think somone on here may be able to offer the answers. the point has been discussed about the speed of the burn of molecules in the cylinder which has led me to think about another observation I cannot explain.

It was over 40 years ago that I was studying thermodynamics and heat engines (as we called them then) and I vaguely remember a couple of things that now interest me but not enough to answer the problem.

It relates to the power that results from burning petrol in an engine at different compression pressures and the resulting temperatures.

I seem to remember various different types of engine efficiency formula for example for - isothermal or adiabatic expansion - etc but one in particular had a formula with an integer to the power of something that linked it to the C/R.

We also used to calculate the power resulting from burning (and the emissions) from the chemical analysis of the fuel - but you know 40 odd years has made it very misty and I don't really need to go over it all again - just seek some answers or explanations.

The reason I am interested is because in all the work I have done on both two strokes and 4 strokes - it seems to me that there is a difference in the results of burning petrol that is not linear with the compression pressure before ignition (or if you like effectively the C/R).

I know for example that petrol vapour from a carburettor will burn at atmospheric pressure - but very weakly (as aged 10 I loosened the head of a villiers JDL engine that was on my dad's bench and to see what happened - how it all worked - to see the ignition that was previously hidden inside the engine - I rotated the magneto and caught the bench alight from the blue flame that spread outwards from the gap. Even more stupid was my quick answer to douse the flames with the bucket of water next to me - yes the one I had been using to wash off the petrol I had cleaned the engine parts with before re-assembling it - woof the whole shed lit up and thank goodness for the black black out curtains used to darken the windows in war time - that they still had in the corner of the shed that I managed to put the fire out with that had now spread all round the shed - and they never found out either!).

It seems from testing engines on a dyno that as the compression pressure increases - so of course does the temperature and the resulting burn starts off creating relatively low burn pressures - not very powerful. Then - as the pressure and temperature is increased there is a point reached when there is a sudden jump in the output to a new level on a graph - as if some critical point in the temperature and pressure has been reached and the resulting burn is suddenly much better. After this is seems that the improvement continues to be either liear or perhaps more exponantial than linear - up to the point of detonation.

I have proven this with two strokes from measuring air flow and making extremely high compression heads that transform an engine unable to pull the proverbial skin off a rice pudding to a reasonably powerfull one when running well outside its designed areas of good air flow dynamics, resonant exhaust frequencies etc. Just raising the C/R at this area of the engines performance transformed the resulting power. I suspect it applies to four strokes because again when the compression pressures are low (like a 911 3.2 with worn rings, the power is low until it runs in the area of the camshaft coming to life, but re-ringed - all the bottom end returns) it is just as if the compression pressures before were just below the point at which some huge leap in output takes place. Similarly (and more to the point of origin of the question) with turbos - although the flow from the compressor reaches a point where there is suddenly a rapid increase of delivery and turbo speed - creating a circle of higher exhaust pressure and higher and higher compression pressures - never the less the graphs of output are less severe that the rapid rise in torque resulting from - it seems - some critical point reached after which the burn fficiency is greatly increased.

I know (well I think I do) that too high a compression pressure can make the molecules of petrol become larger and start to precipitate out of the vapour and require some squish design to improve the mixing (hence squish bands) but it is at the lower end of this cycle I am more interested in the performance.

I just cannot finds anything on the Internet to help - so if anyone can throw some light on this - preferably what I would like to see is a graph plotting the rise in pressure after ignition of a normal fuel mixture as the compression pressure (and therefore also the temperature) rises - to see if it is a straight line or as I suspect either two straight lines with a point that jumps bigger or two curves with similar characteristics.

The answer could assist me in deciding how to build this engine we hope to answer all requirements at both ends of th e rev range - but if you can help please keep it simple!

Thanks

Baz

 

[DivineE]

Are these the formulas you were reffering to?

http://opensource.zyba.com/reference/engineering/thermodynamics/internal_combustion_engine_cycles.php

Also speaking from total ignorance but I was once told by a man on a dyno that BP quote their fuel will be most efficient for torque at something like 12/1 and most efficient for hp at something like 13/1 C/R's? Leads me to thing that it may be to do with the fuel and how it reacts?

 

[appletonn]

I cannot help there Barry, but just wanted to say welcome to the forum again and that I loved the shed anecdote! []

Looking forward to following your 3l developments over the coming months.

 

[Barry Hart]

Thanks for that DivineE it really takes me back a lot.

Those are the type of formula but they refer more to the results of differences in the actual geometric compression ration in an engine that is working perfectly and providing typical answers if it was all running perfectly. It is more theoretical than based on practical results.

What I would like to see is say if you had a container filled with fuel (petrol and air at the right ratio) and compressed it to a specific pressure (not with a crankshaft turning just say by pumping in more of the the fuel to the required pressure and closing the inlet valve you pumped it in through and burning it while measuring the resulting pressure somehow (perhaps with a surface strain gauge).

This would then be repeated for different pressures and what I want to see is how the resulting pressure after the ignition varies with the pressure reached before ignition in graphical form to see if there is a sort of critical compression pressure below which the output is very poor and another pressure after which it is notably better, and what shape that graph would be.

This is quite different to the engine compression ratio because it is only that ramped up by the volumetric efficiency at the revs in question that enables the pressure at the point of ignition to be calculated. But if the torque (proportional to bmep) was say particularly low at low revs - this could be partly due to the poor cylinder filling but could also be because despite the cylinder filling being only a little below the previously good output figure it may also be just below the threshold where the critical pressure changes after which decent power can be delivered.

That was my problem with the two strokes - that they didn't produce poor power at low revs because they were breathing inefficiently - the amount of air they trapped was not significantly less than a regular graph would show improving as the revs rose - yet the output changed dramatically at one point in that rising efficiency cycle.

In other words the "tuning" which everyone at the time blamed the poor bottom end on and thought it meant that the engine simply was not breathing in hardly any air - was not the case - the engine was breathing in good quantites of air but the output after compressing it was extremely low until a sudden rapid rise was noted.

It may be that while I understand what I am asking for - I may not be able to explain or describe it well enough for others to grasp the critical difference in the information I was looking for.

The formula are useful as they remind me that the efficiency is related to the compression ratio (which I thought it did) but even though the results follow a graph - it is a smooth curve and it is that that my practical experience doubts is the actual case.

Baz

 

[sulzeruk]

Baz, as I am led to understand with 2 stroke engines, the pulse from the exhaust is critical to stop gases flowing out of the exhaust port when there is a greater overlap when the transfer ports are still open. The exhaust design will also play a large part in where this occurs over the speed range. The Japanese motorcycle engineers (particularly Yamaha) developed the 2 stroke expansion design (aided by knowledge from MZs Walter Kaaden) in the 1960s. They proved that changes to the resonance of the exhaust greatly affects where the power will occur in the engine. The size of the exhaust port opening and also at the point that it was revealed by the piston also had a massive impact on power delivery. The various power valve systems that were developed in the 1980s certainly were quite effective in improving bottom and mid range torque. These basically lowered the exhaust port thus reducing the overlap time with the transfers open. At higher rpm, the exhaust opened earlier and gave a longer duration as well as closing later.
I think this also applies to 4 stroke engines as well as with higher output, longer valve opening and valve overlap are required to get as much air in as possible as this is pretty much the limiting factor in most engines. Back pressure form the exhaust also plays a roll in keeping the air in the cylinder rather than blowing it out due to the overlap. As you say, volumetric effiency of the engine also plays a massive roll in what happens.
If you were to try out measuring various burning pressures with fuel, I am sure you would quickly come to a ceiling where detonation would occur (unless using Blugas!). What ratio would you also use for testing, stoichiometric rate or leaner or richer than this???
As always, the higher the CR, the better in any vehicle, my old 2 stroke race bike ran about 15 to 1 CR. Was using 50/50 Avgas / Super Unleaded in it and it never pinked once. This also revved to 14,000 rpm, not bad for a road based 125 engine! Also was tractable from about 6,000 but never really came alive until 9,000 rpm with a wide power band of about 3,800 rpm. We found squish was also crucial to promote a clean and even burn over the piston crown, also to prevent detonation too with no unburnt pockets of fuel.
What will you be using for your next project, 8 or 16 valve head? I would think that the efficiences of a 16 V head would be the one to use as you have the bonus of low CR, good flow and CCs for bottom end torque and then greater volumetric effieciency with boost and a big single stage turbo. Has anyone every tried twin turboing a 944?
Sorry for rambling a bit but as you say, it is sometimes hard to get down in a post what you want to say in your head!
Alasdair Cowan

 

[Neil Haughey]

Baz I have to hand here, Advanced Engine Technology by Heinz Heisler. I have had this book a while but I imagine it is still available (you probably have a copy somewhere anyways).

I think perhaps the section 4.2 Spark ignition combustion process, is partly what you are asking for although it doesn't have the equations. Briefly they describe 3 phases; ignition + early flame development, flame propogation and then after burning period. The critical bit to our question is the first phase. This period tends to be constant in time but is dependant on; Temp of the spark flame, nature of the fuel, Temp and pressure of the charge, thoroughness of mixing of the charge, strength of the charge i.e. air/fuel ratio.

They give a couple of graphs, one shows the reaction time getting shorter from roughly 0.005s for a spark flame temperature of 2K celsius down to 0.0005s for a temp of 2.5K celsius.

They give another graph for the air/fuel ratio that plots temp & delay time, best is around 15:1 to 16:1 for minimal delay time and maximum temp.

The second phase is supposedly almost linearly shorter with respect to engine revs, as it depends on turbulence (i.e. is more or less a constant amount of crank angle), so I may be wrong but it seems the first phase is a big limiting factor as to high fast one can push a spark ignition engine. The worst case of 5/1000th's of a second is not that great, an engine makes 2 strokes for each revolution so 200 hundred strokes per second is equivalent to 6K rpm (100 revs per second), so even the best case on that graph at 6K rpm accounts for 10% of the time required to make the power stroke. The worst case would take up the entire power stroke so clearly wouldn't work.

Its late however so please check my mental arithmetic.
Cheers
Neil

 

[Barry Hart]

Alasdair, thanks for that "" I cannot tell you how pleased I am with your response because you have inadvertently confirmed everything I said about the problem and also why I seek the answer "" but without knowing it.

I assumed that in the last 35 years the understanding of two strokes had caught up with (or passed) where I was then "" but still the same issues remain misunderstood.

How you stated things is how others understood it when I tested things 35 years ago to show a different explanation and hence my question about burn pressures.

The notion that the exhaust pulse is so critical (as I said) imagines that it has no affect until say in your case 9000 rpm and strong from 10,200 ending at 14,000 as if the "tuning" has a small range it worked in.

What I found out I can best explain by the following. Imagine your own engine is on a dyno and you test it at all the different revs to find a power graph.

Imagine that you fit a 20/1 C/R cylinder head and do the test again. You will suddenly find very good bmep/torque and resulting power similar to what you had from 9,000 revs, from (say for argument sake) 6,000 revs. If power was all to do with resonance of the exhaust "" how would changing the C/R influence this so much? It would then seize at say (again for argument sake) 10,000 rpm as then the C/R will be too high at those revs with better cylinder filling and detonation would occur.

Similarly if you fitted a 25/1 C/R head a similar result would occur from say 3,000 rpm and it would seize at say 7,500 rpm.

So if you did nothing else but fitted a variable compression head and reduced the C/R as the revs rose to the C/R you use in the upper rev band (to stop it seizing) you would end up with a fantastic engine that would allow you to alter the gear ratios for wider rev drops in lower gears and closer ones in high gears and overall improve the rear wheel torque (as a result of those gear ratios) and end up with a much faster motorcycle.

Some of this is because the higher C/R increased the exhaust temperature and therefore extended the range (not by much though) but most of it is simply because the amount if fuel trapped is usually lower at those lower revs and so the resulting compression pressure before ignition is much lower and ineffective until a "critical" pressure is reached "" which by altering the C/R at lower revs you will have raised.

Similarly with power valves. About 3 years after I left the motorcycle industry "" Honda brought their latest new power valve cylinder to Armstrong to show it off and were rather stunned when a director went down to the dyno room (no longer is use) and in a cupboard pulled out the same design (covered in dust) that I had made and tested 5 years beforehand (and these engines are in build again for classic motor cycle racing events and I am presently involved as a consultant with the manufacturer "Rave motor sport" - who had found some of those power valve cylinders in the parts they took overas and the alternate firing engine I previously mentioned in another posting).

While power valves helped charge efficiency "" they did so because they simply reduce the effective time area of the exhaust port when the revs are lower. Since port are open longer at lower revs a large port is open too long and so there is too much overlap and fresh charge is lost. By putting a temporary blockage in the exhaust you simply effectively slow that charge exit speed down to time the overlap more efficiently. As you rightly say 4 strokes are really no different except for fixed ports you have moving valves.
I explained this at work this way. We have a corridor about as wide as a person leading to a door. I said imagine the corridor is full of people standing one behind the other. How many people can I get through the door if I hold the door open for 10 seconds? After the various guesses, we tested it. I opened the door a tiny bit for 8 seconds "" no one could get through, then fully opened for 1 sec and closed. Only 1 person got through. Not fair they shouted "" but it was "open" for 10 seconds I replied. Then I fully opened it in half a second, held it open for for 4 seconds and closed it in half a second and of course more got through in half the time.

I used this to explain that (just as with 2 stroke exhaust ports and 4 stroke valves) it is not so much the area measured physically nor even the time the area is open "" but it is the effective area during the speed at which it opens and closes at different speeds that determines the flow (which you can calculate knowing the basic geometry and angle limits and port areas in two strokes or cam profiles combined with port areas in four strokes - with integration).

This is called the "time area" and for all engines "" once you know the pressure differences - it determines the flow rate amazingly accurately and therefore can predict torque or power output.

The problem is that the larger the time area the more flow (bmep, torque and power) you can achieve at high revs but the less you trap at lower revs when the gas speed is lower and the overlap too long because the area is basically too big at those lower revs.

So the most powerful engine has a high time area port or valve opening and comes up with a way to make it work as if it was a smaller area at lower revs to spread the power band "" using power valves, or variable cam timing, valve lift or variable inlet tracts etc.

My tests showed that you don't really need all this if you instead vary the compression ratio and increase the lower revs cylinder pressure that would have resulted from a lower fill (or lower volumetric efficiency) by raising the compression ratio (and hence cylinder pressure) back to a more acceptable pressure. This therefore implies that there is a sort of critical cylinder pressure you want to keep the engine above to burn the fuel efficiently.

This is why "" after leaving that industry - I expected others to develop variable compression ratio engines and why Turbos are effectively doing just that by instead of varying the time area they vary the pressure difference at different revs to adjust the cylinder filling rates. This critical burn pressure and temperature - is what I am trying to find out more about "" so far with absolutely no results.

The down side of the turbo is that while you can pump more air in the exhaust is at a higher temperature and therefore volume "" so there is more to get out "" in the same time (or preferably thought of as "time area") and so it struggles to get out and there is more back pressure feeding back into the cylinder - so if the overlap is too high it reverses the initial flow. Hence the need for negative overlap which a N/A cam from that era probably wouldn't have.

You do not of course need overlap to fill the cylinder when it is in your hands how much manifold pressure you use to fill the cylinder in the "time area" you have available "" you can therefore compensate for it but in so doing you have to run higher manifold pressures, bigger turbos and pipe areas and again lose out at the lower end of the power band "" which we want to avoid. You don't really need back pressure to maintain cylinder filling if you can charge fill the engine but you do to drive the turbo.

This is why we will be using the 16 valve head for our next project as it has the potential to flow more in and out and to more easily experiment with the overlap and even change it using the variocam solenoid.

Thanks Niel, I think your maths is good but like all other information I have found and been taught "" most things connected to ignition, burning, expansion etc seem almost linear or progressive and my experiences suggest to me that there is something else going on "" and that there is a compression pressure that below it - gives poor burn pressures and above suddenly gives good results - that I cannot confirm.

This actually does not alter our plan or handicap our ideas for the future "" it is just that I was so impressed with the standard of replies I received on this forum that was new to me "" that I wondered if anyone could confirm this issue that has niggled in my head for decades.

Baz

 

[barks944]

Perhaps it is to do with being close to the point of knock? Maybe you get to a point where you get partial burn and partial knock?

 

[sulzeruk]

Baz, pass on my regards to Richard Tracy next time you see him. I have known Richard for a long time thorugh the love of old Yamahas. I have a small sideline specialising in parts for old 1960s Yamahas and Kawasakis which keeps me busy!
As you have stated, you must use the 16 valve head as it is guaranteed to flow more gases in and out of the engine, 4 valves are always better than 2. It would be interesting to see a comparison with variable valve timing with a 16 valve head.
I am sure that Honda developed an engine that had a variable compression ratio. It was all to do with the free radicals within the engine during the combustion process. I must do a google search and see what I can find.
Alasdair

 

[Neil Haughey]

Hi Baz, the graphs in that book are not linear but its true they have a smooth progressive shape. TBH I think they may be fairly approximate to show effect, its a book aimed at BTEC level vehicle/automotive engineering students rather then a heavy weight academic text. How fuel burns in the cylinder is definitely not a simple problem, a mate of mine was doing a PhD into just this sort of thing at Cranfield some years ago. They pushed the boundaries of what could be done with Computational Fluid Dynamics to better understand model these things. Although I have never worked in the industry my understanding is that the OEM's are still putting a lot of time and money into scientific research in this area.

 

[sawood12]

ORIGINAL: barks944

Perhaps it is to do with being close to the point of knock? Maybe you get to a point where you get partial burn and partial knock?

You really want to avoid knock at all costs. You are ideally looking for a constant pressure expansion of the gasses, this means that the fuel in the air burns slowly over a time releasing it's energy as heat in a 'controlled' way. When knock occurs all the fuel burns instantaneously and gives up its energy as a shock wave rather than heat - and it is heat you are after. All the shock wave does is impact against the cylinder walls and take huge chunks our of it, and the energy impacts the piston crown causing huge short, sharp stresses to transmit down through the piston, con-rod and into the crank.

Ultimately the fuel has a calorific value. That energy is being applied as pressure onto the piston crown and basically you can work out the force on the piston as Force=Pressure x Area. From there on in you are looking at straight mechanics. The tricky bit is understanding the calorific value of the air fuel mixture and how that fuel is burned up over time to maintain the pressure in the cylinder as the piston descends and volume increases. There is no accurate way of controlling how the fuel burns - unless you employ direct injection, so you are really relying on pot luck as to how the fuel molecules distribute themselves and mix in the air.

However the tricky thing in all these instances is not the primary things you can conceive of and anticipate, but those losses that happen that you can't always anticipate or estimate accurately if you do anticipate them. Basically the key thing is getting the fuel to mix evenly through the air such that every fuel molecule can combine with an air molecule prior to ignition so the fuel molecule will actually burn and give off its energy. This is one area piston engines are not very good at. Typically the air fuel mix in a cylinder prior to the spark is about 30% excess air relative to Stoichiometric (to give the fuel the best chance of combining with air molecules), but generally the products of combustion are rich with unburned fuel - as your wideband O2 sensors will confirm, because the fuel doesn't mix very well with the air and you end up with large chunks of fuel that doesn't burn and just carburises in the heat of the combustion process happening around it. This effect is very difficult/impossible to accurately predict and engineers will employ all sorts of Computational Fluid Dynamic models that attempt to model the aerodynamics/fluid dynamics of the air and fuel entering the cylinder and how the fuel might mix into it. Piston engines are not very good at mixing fuel and air and this is exasperated as revs rise.

Ultimately at the detail design phases you start to get into chemistry and physics at the molecular level and incorporating Entropy and other tricky things that are difficult to imagine in the formula's in an attempt to estimate what is happening at a molecular level.

Fascinating stuff, but mind bending.

 

[944Turbo]

I heard there was a company in chessington back in the 70's that built glass cylinder heads so the combustion could be filmed with high speed cameras, I understand they didn't last very long but an interesting approach It was for F1 engine development. Wonder if those films are still around somewhere.
Tony

 

[333pg333]

While in context, can someone explain the difference between knock and pinging? Or is one just the precursor for the other? Pinging being pre ignition, so where does knock occur?

 

[Neil Haughey]

I think in general you mean detonation, which is what Scott is talking about. This will result in a shock wave that can travel at speeds up around 1200 m/s (ISTR), or about 4 times the speed of sound. The term pinking relates to the fact that this shock wave hits the cylinder wall and results in sound waves being transmitted through the block and makes a sound somewhat like a 'ping'. Note that detonation can happen at any time. Pre-ignition is supposedly normally caused by hot spots, however one of the side effects of even slight detonation is that it breaks down any protective veils of gases and film on the cylinder wall, increasing heat transfer and thus making pre-ignition more likely. Engines really are a very tricky balancing act, just from reading the books it seems that almost any one bad thing has knock on effects that create all sorts of secondary problems.

 

[Barry Hart]

Compression ratio is such a misleading concept because it doesn't tell you what pressure the gas is at before ignition and only really tells you what the pressure in the cylinder would be likely to be if you knew the volumetric efficiency and it is different anyway throughout the rev cycle.

My tests and life's experience tells me that there is indeed a minimum pressure below which the expansion of the fuel is limited and after which it follows a more or less straight line up to detonation (and for those posting recently - this discussion was never about that end of the scale "" detonation - only the lower end).

I am guessing that there is a pressure and temperature below which the droplets of petrol do not come out of being like small rain drops in air and after which the droplet size reduces considerably so aiding much better burning and release of energy.

An engine capable of running at 300 bhp (or 350 lbs ft torque) when running at say 25 to 50 hp is only using 1/6 to 1/12 of the air it will be using at higher revs and so the compression that was say 8.5/1 will only be compressing that air to a very small fraction of the pressure and temperature it will reach at peak output.

The fresh charge velocity and swirl will also be dramatically reduced and it seems to me that something different happens to that petrol vapour at such low pressures. Is it just a mixture of reduced pressure, reduced temperature and reduced mixing that all combine to make a mixture that just doesn't burn as well as it does later when the three combined raise the situation and change the effective mixture. Most of the stuff I have read suggests it is almost linear (ok a slight curve) but not a sudden leap "" which it seems to me to be!

Anyway I don't feel so bad about now knowing the answer since it seems no one else does either (if you do please explain).

I only am interested because I have experienced this many times and a turbo with low C/R pistons and a big turbo running at low revs also exhibits poor performance up to a seemingly critical point and I wondered if it was similar to the problem I previous noticed which raising the C/R solved "" and if I hit the same snag with the next engine it means that a rise in boost pressure at that low rev point should solve it.

Baz

 

[944Turbo]

Also mildly related and suprising is the difference between batch fired and fully sequential injection. I have no experience myself but Rick here went from batch fired to sequential on his 944 Turbo, and I read an article by Dave Walker that also said there is no real benefit in either economy or power which suprised me. Maybe to get the required amount of fuel in you have to start squirting before the valve is open so you end up with a puddle behind the valve anyway.

Oh and welcome Baz, enjoying your input and have respected your work from afar since I bought my first '44 and read the articles on your site, my brother is looking for a 996 or boxster at the moment and i pointed him to your engine rebuild articles as well. Will have to drop in one day when I'm up that way - but then I live 5 mins walk from Jon Mitchells and I've only been in there once this year!

Tony

 

[944Turbo]

And for a completely unworkable idea (probably) - I wonder how big an electric motor you would need to create meaningful boost pressure out of a turbo - Spin it up at low engine rpms electrically until the exhaust energy can take over. So you would need a motor a drive shaft with a release mechanism. I think a shaft driving the cold side through a hard intake pipe. The intake pipe could be curved so the motor would sit outside. It should be possible to drive the shaft in the centre of the turbo. Not sure where you would fit it all on a 944

 

[Neil Haughey]

http://www.turbodyne.com/index.php?Itemid=12&id=17&option=com_content&task=view

Old idea but now has merit if we see more hybrid vehicles with large amount of stored electrical power in batteries.

Another old idea that dates back to 1973 is to use a store of compressed air and fire it as and when required into the inlet of the turbocharger. Surprised actually that the air injection idea AFAIK hasn't been used on a production vehicle as it would probably be much cheaper then VNT's.

http://eprints.hud.ac.uk/3790/

 

[944Turbo]

But that is an electric turbo rather than electric assisted, I was thinking that the standard exhaust driven side would take over once the energy is there, was also wondering if a fan spinning in the exhaust would create some benefits at low rpm. Air injection also makes sense.
Tony

 

[Neil Haughey]

Yup, both good ideas in a hybrid sort of way. Thinking about it years ago ppl thought it was crazy to generate electrical power, store it then use it again as a power boost. The amount of energy lost is pretty severe but I guess it just goes to show how inefficient the internal combustion engine at anything other then within its ideal operating parameters.

Something we haven't mentioned yet that would be great if it could be practically applied to a piston engine is adjustable heat regeneration. Basic idea would be to extract heat from the exhaust and use this to pre-heat the inlet charge when running at low power. Thus increasing thermal efficiency. Then when more power is demanded ideally one could take away the inlet preheat.

Heat rejection is another one. Raise the operating temperature of the cylinder components as far as possible then less power is lost by absorption of heat energy into the cylinder components.