PI : DI Ratio Discussion

[wparsons]

@jamesm (banned, don't wait for a reply) stated he was able to find an extra ~10whp in the most knock prone areas just by increasing the DI %age on 91/94 octane. On e85 it isn't as big of a difference (or a difference at all) because of how knock resistant e85 is.

[arghx7]

Originally Posted by Turdinator

First of all, great post. Thank you for your input. I am trying to get my head around all the info here and the other threads linked earlier. One thing I was curious about is, what exactly does "better combustion stability" mean?

You'd think an engine makes some amount of torque but that's constantly jumping up and down. It's usually worst at cold start, when pulling spark to reduce knock, and part load where it's being choked down with internal EGR for fuel economy. The ugly zigzags you see on a dynojet sheet are sometimes noise but there really are torque fluctuations. If they're bad enough you'll get hesitations, surging, and misfires.

If one were to manually control an engine on an engine dyno, holding speed with loading, one might see all sorts of fluctuations in torque and combustion pressure. Then one's software could calculate an instability % which corresponds to these fluctuations in torque or combustion pressure. It might look like this, with the bigger changes corresponding to same change in some parameter you are tuning...



so is the torque really say 100 lb/ft or is it 100 +/- 5% , or 10%? Is it bad enough that a driver could feel it, or even bad enough that you start getting misfire?

The thing that's relevant to the debate is this: DI can cause poor combustion stability if it's not mixing correctly. However, it relieves knock, and that means more advanced spark and with the right heads probably faster combustion speed.

More advanced combustion due to knock relief improves combustion stability. That's a + for DI.
Poorer mixing due to injection timing and mixing time hurts combustion stability. That's a - for DI.
More retarded combustion due to spark retard for knock hurts combustion stability. That's a - for PFI.
Better mixing due to vaporizing the fuel in the intake port and having more time improves combustion stability. That's a + for PFI.

Originally Posted by wparsons

@jamesm (banned, don't wait for a reply) stated he was able to find an extra ~10whp in the most knock prone areas just by increasing the DI %age on 91/94 octane. On e85 it isn't as big of a difference (or a difference at all) because of how knock resistant e85 is.

I can't even keep up with this forum and who's in and or who's out. It's like a middle school clique sometimes.

For E85 it's going to depend on if/how boosted the engine is and how retarded the spark/combustion was on normal E10.

[Xero-Limit]

I have grown to LOVE the split DI/PI system on this car. It is more work--yes, but in practice it works incredibly well. Out of all the things we can tweak on this car to make power, the DI/PI ratios and timings are the least touched. The engineers knew what they were doing here, and any tuner who actually spent time trying out the different variations empirically, will be using something very similar to OEM.

The reality here is that DI works BEST by injecting in a very narrow window, just as @arghx7 outlined. Too early and you're blowing out the exhaust, too late and you're still injecting when the combustion event goes off. Another factor is that you don't want to inject too far away from the spark, that's why it isn't just 370 across the board.

We've seen some tuners swear by DI only, and others use 50/50 split or even full on PI in really high HP applications. I don't think either one is a good solution.

Think of it this way: DI fuel delivery will always be the optimal method, UNTIL you make the mixture too unstable or wash the cylinders. Wash is unlikely, but once you approach 7ms injection times (250+whp with e85) you begin to lose some of that benefit. The mixture becomes less stable. Between about 300whp you cross over the threshold where 35% PI has no negative effect. 20% PI on OEM tunes? We have found absolutely no repeatable HP gains, as in 0, like none-what-so-ever. Despite changes in timing, leaner or richer, more or less aggressive cam timing, or different injection timings...I think one car actually made 1hp over 20% PI mix...which is within the margin of error between runs.

So in reality, there's just no reason to overstress the DI system with long IPWs. Take advantage of the dual injection, max out the DI to about 6 ms (this way in sub freezing temperatures you have headroom) and then run port at no more than 80% duty cycle, or about 15ms.

Also, as a comparison the DI equipped FA20 in the 15 WRX...we can't run half the timing on those! DI only we hit the knock threshold at a much lower HP using DI only. Can't explain it and it's still a very new system, but it made me wish for the BRZ setup having tuned it.

[arghx7]

Here's an exercise. Go into your DI firing angle map and set the whole map to 260. Take some datalogs and see how it runs.

[arghx7]

Originally Posted by moto-mike

Also, as a comparison the DI equipped FA20 in the 15 WRX...we can't run half the timing on those! DI only we hit the knock threshold at a much lower HP using DI only. Can't explain it and it's still a very new system, but it made me wish for the BRZ setup having tuned it.

You can't really see this accurately without drilling the heads for combustion transducers, but keep in mind the difference between spark timing, combustion phasing, and combustion speed.

I'll tell you exactly why you can't put much timing into a FA20DIT besides some situational knock tendency. It burns WAAAAY faster with the high tumble port and combustion chamber shape. Put it into a flow bench and you'll see it flows like crap compares to a BRZ engine.

So I divide my combustion into 4 basic parameters:

1) the time of spark, which you're all familiar with
2) the burn delay in crank angle degrees between 0-10% burn
3) the location of 50% burn - this is the combustion phasing!
4) the bulk burn duration - crank angle between 10-90% burn

with a modern DI turbo engines you'll see high tumble ports. This makes the 10-90% very fast and helps mixing, but they flow pretty bad.

So if I fire the spark at 20 degrees BTDC on a slow burning combustion system it may still burn 50% of the mixture at 8 degrees ATDC (considered to be MBT). If I fire the spark at 5 degrees BTDC on a fast burning combustion chamber it may still burn burn at 8 degree ATDC.

The difference is the combustion speed and the spark, not the phasing.

The reason why an FA20DIT doesn't take a lot of spark is in part because it doesn't need a lot of spark. It burns fast, but the heads don't flow like a BRZ's head. If they didn't have the fast burning head, it would run like crap at part load (poor combustion stability), have high CO and O2 emissions, and probably knock a lot more at full load.

[arghx7]

Let me point out that combustion stability perception is kind of relative. When you see a V8 camming, it's having poor combustion stability at idle and low load. It still sounds cool though, even if it's burning poorly.

Likewise when you drive down the road in a modern engine you will have higher instability at part throttle due to dilution with AVCS induced overlap. However it's not unstable enough to notice usually, and you are getting better fuel economy.

[Kodename47]

Bump for any more tried and tested ratios etc?


I have to say that I prefer a bit of extra control at the top end of the rev range, rather than the same at 5.2k and over. I have tendency to ramp the DI in more gently.


The other thing that has occurred to me is that depending on AFR and induction, that actual percentages will vary as to what's ideal. An example is that a FI car with a lower % PI could still run more PI volume than a NA car with a higher %.

[mid_life_crisis]

Isn't the shorter length of time offset somewhat by the violence of the expansion of fuel under that much pressure being released into an open space? Wouldn't that help with the mixing?

I haven't seen it put as plainly as this (maybe it was stated in technical terms and I missed it) but if you boost this motor you need to add more fuel. If the DI is effectively maxxed out due to restrictions of the mechanical pump, you can only increase the amount of fuel coming through PI. In that case, a boosted engine has to have the ratio shifted a bit towards PI compared to stock, doesn't it?

[Kodename47]

Originally Posted by mid_life_crisis

I haven't seen it put as plainly as this (maybe it was stated in technical terms and I missed it) but if you boost this motor you need to add more fuel. If the DI is effectively maxxed out due to restrictions of the mechanical pump, you can only increase the amount of fuel coming through PI. In that case, a boosted engine has to have the ratio shifted a bit towards PI compared to stock, doesn't it?

As @moto-mike said, its not so much the percentage as the IPW or injection time. On my SC car I run a lower percent PI than stock yet my PI injection time is higher than stock. Think of it like this, 20% of the fuel at 12:1 AFR will be using the injectors less than 20% for 11:1 AFR. Now, let's put aside boost and needing both sets of injectors to fulfill the actual fuelling requirements, if we assume that a certain amount of PI may be required to give a homogeneous mixture, does that sit as a percentage of the mix or as a certain volume of fuel? Again, I understand that this may be over simplifying

Having gone an re-read all the old posts on here, there must be a reason why running a 20:80 split was chosen. I have seen little benefit of leaning heavily on the DI on a NA car. I have taken Mike's advice though on my car, running up to 6ms DI and then having it pegged there until redline seems to.work nicely.

[mid_life_crisis]

Originally Posted by Kodename47

As @moto-mike said, its not so much the percentage as the IPW or injection time. On my SC car I run a lower percent PI than stock yet my PI injection time is higher than stock. Think of it like this, 20% of the fuel at 12:1 AFR will be using the injectors less than 20% for 11:1 AFR. Now, let's put aside boost and needing both sets of injectors to fulfill the actual fuelling requirements, if we assume that a certain amount of PI may be required to give a homogeneous mixture, does that sit as a percentage of the mix or as a certain volume of fuel? Again, I understand that this may be over simplifying.
Having gone an re-read all the old posts on here, there must be a reason why running a 20:80 split was chosen. I have seen little benefit of leaning heavily on the DI on a NA car. I have taken Mike's advice though on my car, running up to 6ms DI and then having it pegged there until redline seems to.work nicely.

I can see that. Increasing the PI pulse length would increase the amount of fuel per injector cycle. I wish I had a better understanding of how the ECU tries to meet the percentage target. With all those other tables to reference, what exactly is it doing by way of reading what table and then physically activating injectors, cycling pump valves, etc. What does it take to satisfy it that the target percentage was met? If you change a pulse length, does it take that into account? Times like this you really want the programming code.
From what I've read, you want some fuel hitting the back of the intake valves. This not only helps keep them clean, but I bet it has a cooling effect as well. In a high rpm, high compression motor, you need to consider the health of the valves. This could explain the PI in the upper rpms when the DI can handle it.
As far as the benefit of DI in an NA motor, again, high revs, high compression = high heat. The DI has a definite cooling effect, which not only allows for more aggressive tuning, but could also contribute to the long term health of things like valves and piston tops. They don't spray the tops of the pistons just to create a particular combustion pattern. Although that is extremely important, it also has to have a cooling effect on the piston top. It may be small, but every little bit helps.

[arghx7]

Originally Posted by mid_life_crisis

Isn't the shorter length of time offset somewhat by the violence of the expansion of fuel under that much pressure being released into an open space? Wouldn't that help with the mixing

Yes, if the rail pressure is high enough, and the spray targeting is appropriate for the injection timing. Raising the rail pressure increases parasitic losses though. Even good spray targeting has a range of injection timing where it won't significantly impinge on the cylinder wall on piston top.

Originally Posted by Kodename47

Having gone an re-read all the old posts on here, there must be a reason why running a 20:80 split was chosen. I have seen little benefit of leaning heavily on the DI on a NA car. I have taken Mike's advice though on my car, running up to 6ms DI and then having it pegged there until redline seems to.work nicely.

Remember that knock is a bigger problem at low speed and high load, and that's where the most DI percentage is being used. At low speed there is more time for the gases at the end of the combustion chamber to auto-ignite.

Originally Posted by mid_life_crisis

I can see that. Increasing the PI pulse length would increase the amount of fuel per injector cycle. I wish I had a better understanding of how the ECU tries to meet the percentage target. With all those other tables to reference, what exactly is it doing by way of reading what table and then physically activating injectors, cycling pump valves, etc. What does it take to satisfy it that the target percentage was met? If you change a pulse length, does it take that into account?

Solenoid injectors have a linear and non-linear range of operation. On modern PFI with saturated injectors (voltage controlled) it's simpler because the current and voltage control are simple compared to the old "peak and hold" injectors from the 80s which used a burst of current to open those clunky 80s needle valves and then a lower current to keep it going. The old peak and hold control lives on in solenoid direct injection, but it's more complex (see below).

So basically on modern PFI there's a bunch of stuff in there to translate pulsewidth into mass flow, based on the characteristics of the injector and on the environmental conditions (air temperature, water temperature, etc).

The direct injection is more complicated because they are run with peak-and-hold current control and booster voltage. They also are much trickier to control at low pulsewidths (typically under 1 millisecond). You have to have an injector flow vs pulsewidth model in the code with a bunch of look up tables populated by bench tests and engine dynos for the most part. The basic intellectual property is owned by the supplier and then adapted to the application.



With a lower flow DI injector, you flow less and don't need low pulsewidths so much. You spend less time in the "ballistic" non linear range, where the valve opens and closes like you are lobbing a Hail Mary pass at the end of the Super Bowl.



So when the DI injector is being driven, you have 4 basic phases:

1) booster voltage phase -- a bunch of current and voltage applied to open the injector quickly

2) high current holding phase - current dithers at a high level as the injector reaches full open

3) lower current holding phase - the majority of the on the time when long pulsewidths are used. the split between high and low current hold has effects on injection noise and behavior of the spray pattern, especially with respect to emissions

4) closing phase - sort of the inverse of the boosting phase

So the ECU signal is the pulsewidth, yes, but there are all these separate sub-operations going on within that pulsewidth that are in separate modules of code with a bunch of lookup tables and settings controlling them. The fuel mass model can calculate flow taking into account all this complexity.

[mid_life_crisis]

It sounds more and more like unless you have ridiculous research resources at your disposal you should leave the DI portion alone and mess with the PI and the percentage between the two. Unless there are areas that are well known to be optimized for, say, clean burning or economy and the way to change them to be optimized for power is well established.

[Phoenix*]

Just wanted to say...
This has been a refreshing thread to read, since I agree with the technical views of both arghx7 and moto-mike!

We've developed several custom maps for the BRZ and FT86, for both pump gas and ethanol. When I get a gap, I shall share some do's and don'ts.

[mid_life_crisis]

Okay, this is coming from left field, but this thread plus another thread where a guy talked about running 100% ethanol got me thinking. For the guys running boosted engines and really pushing it, would it be advantageous to have complete control over how much ethanol is mixed and when? If you completely isolated the PI and DI systems; separate tanks, pumps, and lines, you could fill the PI with 100% ethanol, and using these tables have precise control over how much ethanol you are adding to the fuel. This would theoretically give you the ability to really fine tune the system for engines pushing the envelope. Gasoline burns with more energy, but the ethanol raises effective octane. I find myself doubting it would be worth the trouble, but it is a somewhat interesting thought experiment.