Question about EFI cold start AFR

General engine tech -- Drag Racing to Circle Track

Moderator: Team

User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Question about EFI cold start AFR

Post by ptuomov »

I got some very valuable comments in the thread about cruise ignition advance, so figured out I’d ask a similarly practical question on a different topic:

What’s the right AFR for cold starts? Not talking about the cranking injector pulse width (no issue there) but the AFR after engine is idling but very cold. This is in Northeast US so the car needs to theoretically starting idle stably at below 0F. More practically, I’d want to start it with stable idle and minimum hydrocarbon smells on 32F morning when the ground has frozen overnight. What AFRs should I target at each coolant temperature?

For reference, the car warmed up idles around lambda 0.95-1 open loop and it would idle 0.97-1.03 closed loop if I’d let the oxygen sensor control the fueling.

But what’s the ideal cold idle AFR for say 30F coolant temperature?

Port injected, 4-valve head, 5-liter V8, 8.3:1 or so compression, 775 rpm hot idle rpm, 8000 rpm redline rpm.
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
David Redszus
Guru
Guru
Posts: 9633
Joined: Tue Nov 27, 2007 9:27 am
Location: Chicago
Contact:

Re: Question about EFI cold start AFR

Post by David Redszus »

The fuel enrichment should be just enough to compensate for the poor fuel evaporation due to cold temperature.
That would depend on the fuel distillation curve and ambient temperature. Fuel evaporation will lower inlet air
making fuel even harder to evaporate.

The more critical factor is how much air is necessary to overcome increased friction due to cold oil.
If we know the power required to overcome friction, we could then calculate the fuel needed to
produce the required power.
User avatar
Rick!
Expert
Expert
Posts: 531
Joined: Fri Jan 01, 2016 4:13 pm
Location:
Contact:

Re: Question about EFI cold start AFR

Post by Rick! »

You need extra air along with extra fuel.

One way to work it out backwards is to tap a hole pre-cat in a later model fuel injected car and watch PW, CO, HC and CO2 until it goes into closed loop.
Do some calcs to swag CO into an AFR and calc some more to get your cold target PW (or increased fuel pressure WRT coolant temp).
The more modern way would be to add an O2 to the system, pre-cat, and directly measure AFR, easy peasy.
IACs help with the idle speed, or back in the old days, the auxiliary air valve worked.

Or going back even further, remember choke pull-offs?
0.130" gap or so once the engine starts and comes off in 3-5 minutes.
Those ran power mixture type AFRs to create driveability until the heated air intake took over.

YMMV.
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

David Redszus wrote: Mon Nov 02, 2020 5:51 pm The fuel enrichment should be just enough to compensate for the poor fuel evaporation due to cold temperature.
That would depend on the fuel distillation curve and ambient temperature. Fuel evaporation will lower inlet air
making fuel even harder to evaporate.

The more critical factor is how much air is necessary to overcome increased friction due to cold oil.
If we know the power required to overcome friction, we could then calculate the fuel needed to
produce the required power.
I can measure and control air flow reasonably well. The engine has two MAF sensors and an idle adjustment controller. The idle adjustment controller logic has some very simple "feed-forward" or open loop parameters that first map the engine coolant temperature to a target idle rpm value and then a look-up table that gives a rough guess of the IAC duty cycle that should produce that target idle rpm. This rough guess is the base IAC duty cycle. Then there's proportional-integral-derivative controller that controls deviation of actual rpm from the target rpm by adding to or subtracting from the base IAC duty cycle value the P-I-D values.

It's not a perfect setup, even if its pretty good. One imperfection is that the mapping that guesses the base IAC that would produce a certain rpm doesn't depend on the engine coolant temperature. Also, one needs to take into account that the same amount of air in a cold engine will produce more power with rich mixture than with stoichiometric mixture, simply because the vaporization is incomplete and thus the burn also. One implication of this is that I have to decide and tune the idle warm up enrichment before I finish "tuning" the idle adjustment controller control algorithm.

The fuel is E10 93 octane pump gas. So what ball park lambda for, say, 32F coolant temperature at idle?
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
turbo camino
Expert
Expert
Posts: 674
Joined: Sun Aug 10, 2008 11:51 am
Location:

Re: Question about EFI cold start AFR

Post by turbo camino »

DON'T PANIC
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

Rick! wrote: Mon Nov 02, 2020 6:45 pm One way to work it out backwards is to tap a hole pre-cat in a later model fuel injected car and watch PW, CO, HC and CO2 until it goes into closed loop.
Do some calcs to swag CO into an AFR and calc some more to get your cold target PW (or increased fuel pressure WRT coolant temp).
The more modern way would be to add an O2 to the system, pre-cat, and directly measure AFR, easy peasy.
IACs help with the idle speed, or back in the old days, the auxiliary air valve worked.
I can measure the oxygen demand with a lambda sensor, at least in principle. This however doesn't tell me where the mixture burned. It may have burned in the combustion chamber while making power, too late in the exhaust stroke, in the exhaust manifold, or in the wideband sensor. Short of going to dyno, how do I measure whether the mixture burned at the right time? That is, how do I measure the idle combustion efficiency (in my garage, not in a lab facility)?

Given the limitations of the control algorithm, one potential practical approach would be this. Set the warm engine idle to be stoichiometric. Try to set the warm up enrichment such that the same idle adjustment controller duty cycle will result in the same engine rpm regardless of the coolant temperature. in other words, if possible, simply add enough fuel to the cold engine to burn a greater fraction of the same amount of oxygen than hot engine does to overcome the extra friction from cold oil. If the hot engine would burn perfectly efficiently every air molecule with stoichiometric mixture, this would not be possible. However, even the hot engine probably doesn't fully burn the stoichiometric mixture early enough, so it may be possible to enrich the cold engine to make more gross power (just enough to overcome the additional friction) from the same amount of air. It's a thought.
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

This less theoretical than search for extraterrestrial life. I've been driving the car every day here for the last weeks and we've had one snow storm and a bunch of cold days.
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
peejay
Guru
Guru
Posts: 1946
Joined: Tue Aug 10, 2010 9:16 pm
Location:

Re: Question about EFI cold start AFR

Post by peejay »

Correct is "what it needs". This is going to depend a lot on your fuel injector spray pattern, injection timing, port geometry, injectior location and geometry...

My engine actually likes the same air/fuel ratio dead cold as hot, according to the sensor. However I still need 20-30% enrichment to get there!
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

peejay wrote: Tue Nov 03, 2020 6:23 am Correct is "what it needs". This is going to depend a lot on your fuel injector spray pattern, injection timing, port geometry, injectior location and geometry...

My engine actually likes the same air/fuel ratio dead cold as hot, according to the sensor. However I still need 20-30% enrichment to get there!
So that's a sensor error? Shouldn't we just disregard the wideband readings until the sensor is correctly heated, exhaust piping is hot enough to boil all condensation water, and therefore engine not too far from the operating temperature?

Is the 20-30% relative to lambda one or some other benchmark? I assume that you're measuring air mass with a MAF and can make a reasonably intelligent guess about the actual air-fuel ratio that you're feeding to the engine, hence the 20-30% estimate regardless of what the sensor says?

I'm trying to think of how to measure whether the engine is getting "what it needs". How do you measure whether the engine is getting what it needs?
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
peejay
Guru
Guru
Posts: 1946
Joined: Tue Aug 10, 2010 9:16 pm
Location:

Re: Question about EFI cold start AFR

Post by peejay »

If it feels like it is lean stumbling, it's not getting enough fuel. Seat of pants tuning, no precision needed.

No direct airflow measurement, just a "calculated guess" speed density system. I am thinking the enrichment is simply due to the engine having higher VE when cold because the entire intake tract and inside of the engine are cold. Which, given that the temperatures of intake tract and inside of engine are mostly a function of coolant temps, using coolant temps to correct fueling actually is the appropriate scenario. Oil temp also has an effect on VE, but given that higher oil temps would result in lower VE and therefore overfueling, which will cool the engine, I see this as a self correcting issue with no need to add more complexity.

The OEMs go to closed loop, stoich mixure, as soon as the oxygen sensor (well, UEGO nowadays) is heated and reporting for duty.
User avatar
Rick!
Expert
Expert
Posts: 531
Joined: Fri Jan 01, 2016 4:13 pm
Location:
Contact:

Re: Question about EFI cold start AFR

Post by Rick! »

ptuomov wrote: Mon Nov 02, 2020 10:13 pm I can measure the oxygen demand with a lambda sensor, at least in principle. This however doesn't tell me where the mixture burned. It may have burned in the combustion chamber while making power, too late in the exhaust stroke, in the exhaust manifold, or in the wideband sensor. Short of going to dyno, how do I measure whether the mixture burned at the right time? That is, how do I measure the idle combustion efficiency (in my garage, not in a lab facility)?

Given the limitations of the control algorithm, one potential practical approach would be this. Set the warm engine idle to be stoichiometric. Try to set the warm up enrichment such that the same idle adjustment controller duty cycle will result in the same engine rpm regardless of the coolant temperature. in other words, if possible, simply add enough fuel to the cold engine to burn a greater fraction of the same amount of oxygen than hot engine does to overcome the extra friction from cold oil. If the hot engine would burn perfectly efficiently every air molecule with stoichiometric mixture, this would not be possible. However, even the hot engine probably doesn't fully burn the stoichiometric mixture early enough, so it may be possible to enrich the cold engine to make more gross power (just enough to overcome the additional friction) from the same amount of air. It's a thought.
Couple of basic questions:
  1. Do you know the base fuel map for the engine management system?
  2. Do you have the base timing map for the engine management system?
  3. Can you modify either relative to coolant temp and IAT?
  4. Do you have electronic control over fuel pressure? This would be the easiest way to increase fuel - think "warm-up regulator."
  5. Do you have data logging capability?
  6. Why do you think just altering the fuel for cold running will raise the engine rpm up adequately? You talk about oxygen needs - the engine needs a boost in power when cold to overcome oil friction and somewhat lower "suck" due to cylinder leakage being higher when cold, hence a higher idle speed setting to increase airflow to accompany the extra fuel.
I think you're overthinking this.
Every vehicle I ever worked on had a means of raising idle speed, a means of enriching fuel mixture and usually a means of increasing base timing and altering the timing curve during cold engine operation.
Nowadays, if you notice, somewhere between 32F-40F IAT is where these changes are employed. Gas, diesel, they both seem to have the same effect which is a slightly higher, smooth idle, good driveability and minimal emissions.
If you have only one knob to turn, turn it a bunch and see what happens. Then make sure to run the engine a proper amount of time at operating coolant and oil temps and put it away until the next morning and try another change.
Cold driveability is making the right changes and using your calibrated butt to interpret the feedback.
You'll get it, just unwrap yourself around the "theoretical engine" axle and apply some basics. :)
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

Rick! wrote: Tue Nov 03, 2020 7:43 am Couple of basic questions:
  1. Do you know the base fuel map for the engine management system?
    YES, I CAN ACCESS AND MODIFY IT.
  2. Do you have the base timing map for the engine management system?
    YES, I CAN ACCESS AND MODIFY IT.
  3. Can you modify either relative to coolant temp and IAT?
    YES, BASED ON COOLANT; NO, BASED ON IAT.
  4. Do you have electronic control over fuel pressure? This would be the easiest way to increase fuel - think "warm-up regulator."
    NO, IT'S A MECHANICAL BYPASS REGULATOR SYSTEM THAT CONTROLS THE PRESSURE DIFFERENTIAL BETWEEN THE FUEL RAIL AND INTAKE MANIFOLD TO BE APPROXIMATELY CONSTANT.
  5. Do you have data logging capability?
    BASIC.
  6. Why do you think just altering the fuel for cold running will raise the engine rpm up adequately? You talk about oxygen needs - the engine needs a boost in power when cold to overcome oil friction and somewhat lower "suck" due to cylinder leakage being higher when cold, hence a higher idle speed setting to increase airflow to accompany the extra fuel.
BECAUSE WHEN I FIX THE IDLE ADJUSTMENT CONTROLLER TO A FIXED VALUE AT OPERATING TEMPERATURE, INCREASING FUELING FROM SAY LAMBDA 1 TO LAMBDA 0.95 HAS A SMALL BUT MEANINGFUL IMPACT INCREASING THE ENGINE RPM. I EXPECT ADDITIONAL FUEL TO BE EVEN MORE BENEFICIAL WITH A COLD ENGINE.
Rick! wrote: Tue Nov 03, 2020 7:43 am I think you're overthinking this.
Every vehicle I ever worked on had a means of raising idle speed, a means of enriching fuel mixture and usually a means of increasing base timing and altering the timing curve during cold engine operation.
Nowadays, if you notice, somewhere between 32F-40F IAT is where these changes are employed. Gas, diesel, they both seem to have the same effect which is a slightly higher, smooth idle, good driveability and minimal emissions.
If you have only one knob to turn, turn it a bunch and see what happens. Then make sure to run the engine a proper amount of time at operating coolant and oil temps and put it away until the next morning and try another change.
Cold driveability is making the right changes and using your calibrated butt to interpret the feedback.
You'll get it, just unwrap yourself around the "theoretical engine" axle and apply some basics. :)
I don't trust my butt to calibrate machines. All that my butt can tell me is that something is seriously wrong when something is seriously wrong. It can't tell the difference between good and better.

An update from 30 minutes ago:

The starting point for what I did this morning is that, at the operating temperature, the idle is tuned to about lambda 0.95-1.00 and ISV% 50 gives me about 800 rpm idle. When I say ISV 50%, I mean the idle adjustment valve set fixed 50% duty cycle. In practical terms, that's acting like an idle adjustment screw that is fixed into a constant position. The hole thru which air comes thru is now of a fixed size.

Here's what I ended up doing today. The engine block was about 25F to start with this morning, measured with an infrared gun. I overrode the idle adjustment controller to ISV 50%, so now it's just like a dumb idle adjustment screw. The idle was high at over 1000 rpm when engine was cold, when ISV% 50 would get me only about 800 rpm idle at operating temperature. This indicates to me that we are/were overcompensating for temperature with more fuel than necessary. (The ignition advance temperature compensation at these engine temperatures is not an important factor.)

I sat in the car as it warmed up and I pulled fuel from the warmup enrichment map at each cell until I got the idle below 1000 rpm. I had to pull a lot of fuel away from the warmup enrichment map to bring the idle rpm down and if anything it started running smoother although at lower rpm. Tomorrow morning, I'll repeat the procedure and pull more fuel trying to get it idle closer to 900 rpm with ISV% 50.

Does what I am doing make sense?
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
adam728
Member
Member
Posts: 137
Joined: Fri Dec 06, 2013 12:53 pm
Location: Michigan

Re: Question about EFI cold start AFR

Post by adam728 »

So you are holding a constant air flow and tuning speed with fuel only?
User avatar
ptuomov
Guru
Guru
Posts: 3587
Joined: Fri Aug 07, 2009 3:52 am
Location:

Re: Question about EFI cold start AFR

Post by ptuomov »

adam728 wrote: Tue Nov 03, 2020 8:42 am So you are holding a constant air flow and tuning speed with fuel only?
Yes, but as a test of the warmup enrichment only. The idea is to make air mass and ignition timing constant and see how high rpm the engine runs at different temperatures and different warmup enrichment parameters.

When I turn all of the systems back on, it runs at idle based on
(1) base fuel map + warmup enrichment map that determine the fuel per unit of air
(2) idle adjustment control algorithm that determines the units of air based on a feed-forward table and a PID controller on rpm error
(3) Ignition map + temperature compensation map for ignition (although temperature compensation map doesn't do much to ignition at cold temperatures, it's more for very hot temperatures)
Paradigms often shift without the clutch -- https://www.youtube.com/watch?v=cxn-LxwsrnU
https://www.instagram.com/ptuomov/
Put Search Keywords Here
adam728
Member
Member
Posts: 137
Joined: Fri Dec 06, 2013 12:53 pm
Location: Michigan

Re: Question about EFI cold start AFR

Post by adam728 »

I'm trying to picture how a cold engine, with the same airflow and ignition timing as a hot engine, runs 200 rpm faster cold vs hot just by richening things up.

Being very over-generalized, we normally add a few hundred rpm idle speed for a cold engine, feed it fuel that puts it in a happy place (often nearly the same as WOT best power, or 0.80-0.86 Lambda watching a wideband), and keep timing in an area where the engine is happy to idle (one engine might get grumpy going higher than 10 deg BTDC, a different design might be happiest at 22 deg BTDC).

An example would be an engine that idles 800 rpm at 100C engine temp. Might cal it to idle, say, 1000 rpm at 40C, 1200 rpm at 0C, and 1400 rpm at -20C. Fuel qty is tweaked to keep it at 0.85-ish Lambda throughout, and timing targeted at 15 deg BTDC.

If emissions and a catalyst are involved then things change up. It's closed loop ASAP, and timing is generally very late to get heat in the cat, and a very open throttle to make up for those first two items. Lots of production stuff out there will run with the throttle very open and ignition well after TDC on a cold idle.
Post Reply