Acceleration Enrichment / Multiple Injection Events

[revilla]

An interesting observation whilst looking at the mapping on MEMS3 (in this case VVC).

I was looking at the acceleration enrichment (extra fuelling on rapid throttle opening). I had a scope on Injector 1.

With the engine running 1000rpm (well to be precise, I was fooling the ECU into thinking it was running at 960rpm with the crank and cam sensor signal generator I put together to allow me to swing the throttle without the engine revving itself to pieces) on a closed throttle, this is a trace of the injector voltage:

With a steady open throttle, I expected more fuelling, and here is the trace:

As expected the injector pulses are wider, so more fuel injected. Then I looked at what happens as the throttle opens rapdily. Naively I expected an even wider injection pulse, but this is what actually happens:

There are now FIVE injector events per cycle. The "main" injector pulse is in line with the current throttle position, and all of the enrichment comes from four separate additional injector events per cycle. The enrichment events are equally spaced, with the main event falling between the enrichment events.

I guess this must be to get a better fuel distribution and atomisation in the still relatively slow moving air until the engine picks up when you stamp on the throttle.

Is this normal for other ECUs? It does look like fairly sophisticated injector control to me, but maybe all ECUs do it. I've not looked at anything else.

[oilyhands]

Multiple injector events aren't uncommon, a friend who calibrates Diesel engines mentioned that in some cases there can be as many as 17 injector events per cycle. Pretty sure the Emerald adds transient acceleration fuelling by extending injector duration rather than additional events.

Oily

[revilla]

Thanks Dave. I kind of expected it to work like the Emerald. The four extra enrichment events are spaced equally around the 720-degree engine cycle so in effect the injectors are batch-fired for the enrichment component and sequentially-fired for the main injection pulse.

[StevehS3]

Please can I ask what the advantage is of the multiple injection events technique? More precise fuelling or timing or improved spray pattern...?

[Roger King]

My best guess is that this is an emissions strategy designed to reduce excess HC emissions which are always, and inevitably, a problem with transient enrichment - but I'm more than willing to be schooled about my errors by those better informed.

[Jonathan Kay]

Does it make easier in practice to add in the extra fuel for acceleration in a new mode while leaving the old mode running?

Can that fuel system maintain injection at the right pressure more consistently with several smaller squirts rather than a few big ones? Does the feed pressure drop at all under maximum rates of injection? (I think that's a different point from the characteristics of the mixture as its flow changes.)

Jonathan

[StevehS3]

Jonathan, I was thinking the same as your second point - a short recovery time between squirts.

[revilla]

I think the fuel pump is specified to deliver a lot more file volume flow rate at rated pressure than the injectors can consume. There's a fairly heavy flow in the return line from the regulator. You could also argue that from a transient flow point of view, firing the injectors one at once would be less demanding than batch firing them together, but I doubt the difference is relevant. The ECU runs a fairly powerful MC68336 processor, in terms of architecture it's more like a small PC processor than a typical microcontroller, with virtual memory addressing, supervisor modes etc. I doubt it would be difficult to support either mode on the hardware, so it must have been done out of choice for a good reason. I think the emissions and better fuel atomisation and combustion (which are really two sides of the same thing as it's the unburned fuel which forms the emissions) sound like the most likely explanation.

[StevehS3]

Thank you Andrew. Your work and the collective knowledge base does give me confidence to stay with my K series 7.

[rj]

I am certain it has to do with emissions.

[Dr Slotter]

As Oily says, multiple injection events are common in diesels and are used, for example, to burn off the soot produced by the preceding injection/combustion event.

[Englishmaninwales]

Thank you for your work Andrew, as always very interesting. 

[Petethediesel]

Multiple injection is, as previously stated, very common. Yes it can/does improve emissions but it also broadens the cylinder pressure graph and lowers its peak. This can increase total power and improve NVH.

[rj]

but it also broadens the cylinder pressure graph and lowers its peak.

I very much doubt so in an engine where the fuel is injected before the valve. It is pretty hard to get anything in the combustion chamber mid-cycle :-p

[Petethediesel]

my reference is to diesels. but clearly cylinder pressure build is post valve closure.

[rj]

I thought so, and the multiple injection diesels was well described by VW when they introduced the PD engines.

[StevehS3]

Would the ECU switch to open loop mode during times of acceleration enrichment, thus preventing the lambda sensor attempting to compensate and lean the mixture? This made me wonder if there are other circumstances such as high load conditions during which it runs rich and if so whether these conditions are hard-coded in the ECU firmware.

[revilla]

Yes there's a Target AFR table, by MAP and RPM. It only runs closed-loop at sites where this is 14.7, as that's all the lambda sensor can target, and even then only under stable conditions. You can see here it is mapped to go rich at high RPM and high MAP.

There's another table which models the air flow. So the ECU knows at a given RPM and MAP how much air will be flowing and the target AFR so it calculates fuelling from the two. 

I've tried tinkering with the cells that are set to 14:7, even if you only change them slightly it goes open-loop at those sites, so the Target AFR of 14.7 defines exactly where it will and won't try to run closed-loop when it can.

[StevehS3]

I naively assumed that a factory standard ECU would always target 14.7 when the engine is hot. I also assumed it would use throttle position vs RPM but the actual MAP is better. It is fascinating to see the table. It looks a great ECU.

 

[revilla]

As far as I can see only normally uses the throttle position sensor for two things; to detect closed throttle to activate idle strategy, and to detect rapid throttle movements for the purpose of enrichment etc. There is a table of expected MAP by RPM and TPS which enables it to continue running when the MAP sensor fails, in which case it is effectively using the TPS for load. If you pull the MAP sensor connector off while it is running it stumbles for a second or two then decides that the readings are implausible and substitutes defaults from the table and settles down again.