Fuel metering in Atkinson cycle applications

[rationull]

I've been curious about this for some time but not until today did I think to myself "I wonder if BITOG has the answer?"

I want to talk about Atkinson cycle applications with multi-point fuel injection -- no direct injection, no batch fire (i.e. assuming one fuel injector per cylinder firing into the intake manifold timed specifically for that cylinder's power stroke). Just your "standard" modern EFI setup. Such applications include the current Toyota HSD cars, as well as the "fuel economy" VTEC mode on the current non-hybrid gasoline Civic.

How does the fuel charge get precisely metered in this type of engine? Obviously the fuel must be injected while the intake valves are open, which means the unburned air/fuel mixture is still floating in the cylinder as the piston upstroke starts, with the intake valve(s) still open. Thus, it seems like some fuel will be pushed back out of the cylinder on the upstroke. It seems like to get a good smooth burn the ECU will have to know how much fuel will be pushed out for a given injector pulse time based on conditions. Further, ideally it would have to take that leftover fuel into account for the next power stroke (i.e. less fuel needs to be injected because some is still floating around in the intake manifold).

Presumably the O2 sensors can help stabilize all this. Is there just some really heavy modeling going on to determine how much fuel to inject based on engine speed, MAP, MAF, and combustion chamber flow dynamics in order to arrive at the ignition event with the right amount of fuel in the chamber? Or, are the combustion chamber shapes and fuel injector timing somehow worked out so that only/mostly air gets shoved back out into the manifold?

Obviously normal EFI programming is fairly complex but to my mind it seems like the escape of injected fuel back into the intake manifold would add another dimension of complexity.

 

[Shannow]

As long as the fuel is measured with regard to the nett airflow into the engine, and as you say trimmed with O2 sensors, it should be fine.

That being told, I'm curious how they measure the airflow accurately.

When that little bit of air is pumped back out, does the airflow sensor see it as a reversion, or is the manifold designed that it goes into initially feeding the next cylinder in the cycle...which with Atkinson operation must start drawing.

 

[rationull]

Good question. I don't know much about the intake setup on the various Toyota HSD cars but on my Civic at least I can't imagine that the MAF sensor sees any kind of reverse flow from that situation. First, I'm not sure if MAF sensors can measure direction at all, and second, there would have to be a lot of air pumped back out to make it all the way back through the intake manifold. I could see the MAP reading being affected by the push-back, though. Moreover I don't think it can be fed to any other cylinder (if that's what you meant) because of the intake runner structure.

The problem with dismissing it as an airflow measurement problem, as I see it, is that some proportion of fuel will be pushed out of the cylinder with the air. You can easily choose dose of fuel based on airflow but after accounting for the amount that gets pushed back out you're left with some "leftover" fuel for the next cycle which must then be accounted for. Supposing you didn't account for this "leftover" fuel, you'd get a rich power stroke and the O2 sensor could theoretically account for that. If the "leftover" fuel stays approximately the same for each power stroke (because the amount left over from further past power strokes dimishes) then the trim could potentially stabilize the fueling calculation. But that seems rather crude to me.

In short you have to inject some amount more fuel than you actually intend to burn and it seems like there might be some kind of accumulation effect with that. Is this "some amount" derived based on conditions and knowledge of the structure of the engine, or is it maintained based on O2 sensor feedback? (I realize the answer is probably "both"). I may be making this more complicated than it is, though. Wouldn't be the first time. I'm a software guy, not a mechanical or fluid dynamics one

 

[chrome]

The flow inside an engine can now be very well modeled using computational fluid dynamics, accounting for everything from valve overlap to rpm and engine load. The intake charge push back isn't so much a problem as during engine development, the engineers would have modeled that to the n'th degree, and then programmed the fuel vs. load map in the ECU to account for reversal. This will then be tested very throughly during dyno sessions.

 

[oilyriser]

BMW was experimenting with early intake valve closing, so there would be no flow reversion, and there would be better evaporation of fuel, due to the reduced pressure.

 

[Shannow]

oilyriser, I'd oft considered that was the better way of doing it also...for the players to have gone the other way must mean that there's a bigger negative in the pumping loop or something. Maybe the intake valve shape still flows reasonably well forwards rather than backwards, making shutoff more certain the other way.

I still am in awe at fuel stand-off in carburetted engines on the dyno...huge flows into the engine, but enough flow/pressure backwards to make a little cloud.

 

[Titan]

Isn't it just the inertia of the quickly-moving intake gases that allows the flow NOT to reverse even though the piston has begun moving upwards? Quickly moving gas has a lot of intertia, so the relatively slowly moving piston (it moves most slowly at the bottom and top of the stroke) isn't generating sufficient impulse to reject the quickly incoming charge prior to the intake valve closing...even when it's left open during the first moment of the piston's upward movement.

 

[Shannow]

Titan, spot on, that's why big cams work at high revs...and when we muck around with engine programmes, intake closing is the most important event.

The "Atkinson" and Miller cycles hold the intake valve open WAY past BDC for a normal revving engine to push air/fuel back into the intake, making the engine proportionally smaller.

May have a 1.5 litre engine ingesting the air/fuel of a 1 litre, but having an expansion ratio 30% greater.

 

[Max_Wander]

precisely! in fact the entire concept of the cycle is to exploit those exact flow characteristics.

the fuel metering should be quite easy with an air flow and MAP sensor. when on atkinson cycle, MAP drops dramatically near or equal atmospheric as the throttle is pegged wide open like a diesel, and intake stroke event offset dramatically. however with the intake stroke delayed to snip the valve shut just before TDC compression, the cylinder has on average claimed as much air as equivalent light-throttle or cruising. the MAF reads the actual mass of air moving, and the MAP basically modifies the equillibrium as need be. Also, either can be a reliant sensor in the failure of the other. o2, barometric pressure, temperature, TPS all fine tune the symphony according to the thoughtful engineers' hard work.

atkinson is for fuel efficiency and works only at times when, say cylinder deactivation would; cruising. by metering air with intake valve manipulation instead of the throttle, you can eliminate pumping losses as the creation of manifold vacuum and compression losses.

the miller cycle is when the engine retains it normal power level, as with the use of a compressor and high manifold pressures, while still retaining the pumping efficiency.

 

[rationull]

Originally Posted By: Max_Wander
when on atkinson cycle, MAP drops dramatically near or equal atmospheric ...


I assume you meant that MAP rises dramatically, assuming no forced induction, right? (i.e. high vacuum at low throttle --> low MAP, turns into low vacuum --> high MAP when starting atkinson cycle).

chrome: good point that it would be possible to model the accumulation effects of the flow inversion to the nth degree and take it into account in the fuel tables.

Max: Very informative. For whatever reason it hadn't occurred to me that while the pushing of air/fuel mix back out of the cylinder won't actually reverse flow further upstream, it will still be felt at the MAF sensor as the intake manifold fills with air in the absence of any kind of compressor. So, if the engine is accepting less air (whether because of a throttle restriction or because the cylinder is ejecting part of the charge) the MAF will still give an accurate air flow reading. As you said, the MAP gives us the other side of the picture by showing a pressure closer to atmospheric than it would be if the throttle restriction was the only thing limiting the air charge.

So, with both MAP and MAF you can model this situation "easily" but with only one of them it would be quite hard. And of course (again like you said) the other sensors play the same part they do during non-atkinson cycle operation.

This makes *much* more sense to me now.

 

[artificialist]

During the 1980s, many EFI engines used a Volume Air Sensor which was pushed back by air going in, and had a damper so that any air going in reverse would not be measured. The reason they were phased out was because they created a serious airflow restriction, and they were expensive. Hot wire Mass Air Sensors became cheaper, and speed density systems that only used a MAP sensor were possible thanks to better computer modules.

Another thing used in some EFI systems is "Alpha-N" This is the throttle position, divided by the RPM. This was popular with some race engines because it was simple, however, it is not very accurate at part throttle.

In the exhaust system, there is an O2 sensor that sees what the air fuel ratio is. Most are the basic type that can tell when the AFR is 14.7:1, and when it is not. That isn't useful when dealing with a special high efficiency engine with a leaner than normal AFR, and that also doesn't work with the richer AFR needed at full throttle acceleration. A wide band O2 sensor can detect between 10:1 and 20:1 or so, however, they are very expensive. Perhaps a future engine will use the wide band O2, and it will be running in "Closed loop" but the target AFR will be variable instead of being fixed at 14.7:1.

 

[Max_Wander]

Originally Posted By: rationull
Originally Posted By: Max_Wander
when on atkinson cycle, MAP drops dramatically near or equal atmospheric ...


I assume you meant that MAP rises dramatically, assuming no forced induction, right? (i.e. high vacuum at low throttle --> low MAP, turns into low vacuum --> high MAP when starting atkinson cycle)


yes thats right, sorry I was just thinking of the inches of mercury on the dial *smh*

Originally Posted By: artificialist
During the 1980s, many EFI engines used a Volume Air Sensor which was pushed back by air going in, and had a damper so that any air going in reverse would not be measured. The reason they were phased out was because they created a serious airflow restriction, and they were expensive. Hot wire Mass Air Sensors became cheaper, and speed density systems that only used a MAP sensor were possible thanks to better computer modules.

Another thing used in some EFI systems is "Alpha-N" This is the throttle position, divided by the RPM. This was popular with some race engines because it was simple, however, it is not very accurate at part throttle.

In the exhaust system, there is an O2 sensor that sees what the air fuel ratio is. Most are the basic type that can tell when the AFR is 14.7:1, and when it is not. That isn't useful when dealing with a special high efficiency engine with a leaner than normal AFR, and that also doesn't work with the richer AFR needed at full throttle acceleration. A wide band O2 sensor can detect between 10:1 and 20:1 or so, however, they are very expensive. Perhaps a future engine will use the wide band O2, and it will be running in "Closed loop" but the target AFR will be variable instead of being fixed at 14.7:1.


Most cars now run a wideband o2 sensor, in order to detect total oxygen content in the exhaust. This is the tuning difference that makes an engine "flex fuel". Widebands in this app help maintain a good AF ratio regardless of the fuel's oxygen content (any blend of alcohol and gasoline)