AFR and knock

[klatinn]

Hi

Here's a log from one of our ST-12 customers (Tracy Pedigo), who graciously send us this data. I combined multiple log files into one with multiple sessions.

I had never seen this AFR signature before in this context. I thought it interresting, that's why I am posting it here. The interresting part is that on this engine the knocking cylinders cause a specific AFR signature. I suspect the abnormal combustion during the knock events cause part of the mixture to not burn at all, causing excess O2 in the exhaust.

This is off a 632 ci Ford drag racing engine with one 1150 CFM Holley 4 barrel. The engine came from the builder with 36 deg. of ignition timing. The engine was designed to run with a 400 HP shot of NOS. Engine was dynoed on a DTS Dyno equipped with an Innovate ST-12. Ignition timing was adjusted anywhere from 28 to 36 degrees. Engine made 850 HP at 36 degrees Ignition timing on VP C16 fuel no NOS.

The engine did not knock audibly, but was not making the expected power.
The first session in the log (named "Detonation 36 deg Advance") is with C16 fuel and 36 deg. advance. Notice cylinder number 1 and 2 (purple and black) showing lean spikes, but not the characteristic sharp spikes of ign. misses, but a lean area with overlaying wild excursions. These cylinders are knocking, badly, as was determined later.

The last session (named "Session Normal 28 deg") was on C16 with ign. advance at 28 degrees. NO changes in fueling at all. In that run the engine made 970 hp (no NOS), a 120 hp gain compared to 36 advance.
Because the engine owner insisted that he needs more timing to make more power, different things were tried. The second session is with 32 deg timing and cam retarded 4 deg on Sunoco 118. The lean excursions on Cyl1 and 2 look almost the same as in the first session if you overlay the two. The third session was with 32 deg. and Sunoco 118 and cam reset to spec. It looks normal but had less power.
In the end the engine was dialed in with 28 deg. advance. With NOS it made 1450 hp.

As an additional note. The actual drag runs were not logged and the engine blew up the first time out, even though it ran perfect on the dyno. The owner did not believe that datalogging and testing on the track is neccessary because the engine was dialed in on the dyno already.

Regards,
Klaus

[Tuner]

Greetings Klaus,

This is from a WWII era M.I.T. book: “Carbon in the exhaust. When detonation occurs in an engine equipped with short exhaust stacks, puffs of gray smoke can usually be seen if a light background is present. At night these puffs are seen as bright yellow flashes, in contrast to the normal exhaust flame, which is light blue. It is thought that the combustion taking place in the detonating part of the charge is chemically somewhat different from normal burning, and free carbon is one of the products. Free carbon is not produced in normal combustion unless the mixture is excessively rich.”

I’ve seen the gray puffs while riding in a boat with open exhaust when I know the engine was detonating. I don’t have to jump very far to a conclusion that if free carbon is coming out of a knocking cycle, free oxygen must be also. I wonder if the detonation wave bouncing off the walls of the chamber and the valves could be a low enough pressure that if the valve is being bounced off the seat a sip of fresh air could get in? I see more soot in the intake of engines that have been detonating. More likely speculation is because a detonating cycle’s combustion is complete much sooner than normal and rejects more heat to the chamber walls, the lower temperature (hence pressure) at the start of blow-down (when the exhaust valve opens) changes the gas flow dynamics during overlap such that more exhaust gets back up the intake and more intake charge gets in the exhaust stream to be seen by the WB02 as lean. That would, if it were the case, dilute and pre-heat the charge for the next cycle, which would then burn less efficiently than a “pure” charge, is more likely to knock and on it goes.

SAE paper 983026 specifically addresses engine knock as a concern in motor-sports. It makes reference to differing combustion chemistry.

[klatinn]

Hi Tuner,

The free carbon reference is interresting and would certainly explain the excess O2.
I think the lower blowdown pressure because of thermal transfer to the cylinder and head during knock is not to blame for the soot in the intake. After all, lower blowdown pressure is also present at part throttle, then even accompanied by partial vacuum in the intake. I rather think that the larger amount of free carbon will condensate on the relatively cold intake walls during overlap, while regular exhaust gas will leave little or no deposits.
For me the look of the knocking cylinder AFR trace compared to a trace with ign. misses is interresting and something to remember. Maybe a few previously posted logs that were suspected of misses had actually knock events?

Regards,
Klaus

[Tuner]

Hi Klaus,

I know this is strechin’, but: At WOT the dynamics of the gas activity are greatest. The largest and hottest amount of mass is involved. The tuning of the exhaust is optimized for “normal” circumstances. When the pulse tuning is changed because the temperature and pressure are suddenly different (knocking) the pipe would have to be a different length to accommodate that and still keep the gas activity during overlap the same as it is normally. The gas exchange during overlap would have to be different in knocking and non-knocking operation.

Regards

[klatinn]

Originally Posted by Tuner

At WOT the dynamics of the gas activity are greatest. The largest and hottest amount of mass is involved. The tuning of the exhaust is optimized for “normal” circumstances. When the pulse tuning is changed because the temperature and pressure are suddenly different (knocking) the pipe would have to be a different length to accommodate that and still keep the gas activity during overlap the same as it is normally. The gas exchange during overlap would have to be different in knocking and non-knocking operation.

Yes, I agree with all that. But the resonance tuning of intake and exhaust is optimized for a specific RPM. Different RPMs will cause also a phase shift between these oscillations and the engine cycle. Even though at a specific RPM with knock the oscillation cycles and engine are out of phase compared to regular combustion, that, or a similar phase shift would also occur at normal operation but at a different engine operating state. Wouldn't you agree?

Regards,
Klaus

[Tuner]

Yes. I was trying to associate the greater amount of free carbon in the knocking combustion products with noticing it in the ports of knocking engines.

[Tuner]

Hi Klaus,

So here I am whittling away on a QJet and the little voice in my head quiets down enough to listen to what you are trying to point out. The subject you’re trying to get across to us isn’t the detonation but the effect of it as seen in the WBO2 data, right? Sorry. Now you’ve got me wondering if some of the lean excursions in 471 Magnum’s data last May and June, that wouldn’t respond to changes of the jetting and accelerator pump like I would have expected, weren’t knock related. He was having trouble with knock when the weather got warm.

Regards

[klatinn]

Hi Tuner,

Yes, that was what I was trying to get across. I didn't know knock can have such a large effect on AFR data when seen on a cylinder by cylinder basis. When using the WB in the collector the effects will be of course diminished as the lean gas will be diluted by other cylinders normal gas. This dilution would make it harder to see. But when you see sudden lean areas that are not typical spikes from misses and can't be explained by carb/injection events, high speed knock is something to consider. This is especially true for engines with factory knock sensors, as most factory ECUs don't listen to knock sensors above ~4k RPM anyway because engine noise masks the signal too much.

Regards,
Klaus

[automotivebreat]

Hi,
Very interesting subject, I personally don’t get involved with NOS but I have an obsession with combustion and the combustion characteristics of different performance applications so I get involved with these types of discussions.

On another forum a discussion turned to high HP NOS engines, some of the tuning problems, solutions and explanations for these solutions. One engine builder in the discussion described a trend to opening squish clearances very wide to control high cylinder pressures encountered with NOS. Another member, Rob considered a "combustion expert" gives his thoughts in the quote below.

Originally Posted by Rob

I *think* what might be going on in the nitrous motors nowadays is that peak cylinder pressures are getting extreme. This should be no surprise. Using the increased squish distance to lower turbulence, and thus flame velocity, would be one way to slow the rate of pressure rise.

Lets use my current academic research as an example. Here is what happens when combustion durations become extremely small:

http://www.ualberta.ca/~rlupul/pressure.pdf

With the high pressure rise rates, the combustion becomes more violent, and eventually 'knocks'. The knock shown here is a bit of a seperate phenomena, that I won't get into, but I already had this pdf made, so here it is

With a fast combustion, the maximum pressure of the cylinder is very high, compared with a slower combustion. Keeping this maximum cylinder pressure under control, in order to keep head gaskets in the motor say, could be the reason the increased squish helps things out. Since these motors are not detonation limited, in general, engine builders should be able to get away with this.

In the example currently being discussed the engine was detonation limited. My thinking is that the high burn rate and resulting cylinder pressure spikes encountered with detonation, NOS, supercharging or too much ignition lead put the flame front very near the edges of the squish regions with the piston still very close to TDC. In this the situation, the squish region could act as a quench zone, the mixture in the end zones of the quench area remain inaccessible to the flame front. If indeed this is happening, the recorded spikes may be related to unburned mixture in the end zone.

[coocoo67]

The engine wan not detonation limited since it made more power on 28 deg than 36 degrees.
Ignition timing is different on any engines. The tuner should have stopped advancing timing when you see no more increase in hp/torque then go down 1 deg for safety.

[Bobalos]

Originally Posted by coocoo67

The engine wan not detonation limited since it made more power on 28 deg than 36 degrees.
Ignition timing is different on any engines. The tuner should have stopped advancing timing when you see no more increase in hp/torque then go down 1 deg for safety.

I agree with Coocoo67. Klaus, do you remember Bruce Plecan? He used to say "you dont run as much timing as you can, you run as much timing as you need & there is a BIG difference between the two".

Very cool thread. I sure wish I could run 9 O2's on my hot rod, but that would be a bit much I think. Maybe I might add another 1 or 2 down the road.

Bob

[klatinn]

"you dont run as much timing as you can, you run as much timing as you need & there is a BIG difference between the two".

That's exactly right. During the second seminar at PRI Tracy was there as well and told the story of a sprint car racer. This guy had an ign. retard device, activated by a button on the steering wheel. It retarded ign. by 6 degrees. It's purpose was to reduce power by a small amount when he looses traction. When Tracy put that engine on the dyno it turned out the engine was making MORE power when retarded 6 degrees, the exact opposite of what the device was supposed to do.

Regards,
Klaus

[shrinker]

Hello to all. The issue here I think we need to discuss is why does the LM-1 interpret an assumed detonation as lean spikes of AFR. I thought the LM-1 was able to workout the AFR irrespective of combustion or not. The chambered sections of the sensor do not need combustion to work but do require sufficient heat, could that be the problem?
Also why does the Lean area extend for approx 5 combustion events. Does this mean the problem is not accurately recorded because of cooling the sensor or is it a probem with the tune that results in a cyclic combustive event? Well at least that is what I have gathered from my research, am I wrong?
Also I have seen our race engine partially melt 3 pistons going down the track and it blew grey smoke the same as tuner has described, I thought that this grey smoke was most likely piston aluminium. I have seen other cars blow black smoke they are usually turbo injected dont ever see carbs doing it that bad. I think it indicates the differences between vaporization time of the 2 systems, it shows up in these high power engines. Our engine makes 900hp 412 ci small block chev
Thanks for a great forum.

[klatinn]

Hi,

I thought the LM-1 was able to workout the AFR irrespective of combustion or not. The chambered sections of the sensor do not need combustion to work but do require sufficient heat, could that be the problem?

No, that is not the problem. The LM-1 or LC-1 even show the correct Lambda of a Lambda lab gas mix at room temp.

As Tuner indicated, and confirmed by the SAE papers, during knock part of the carbon in the fuel does not parttake in the combustion. That creates the grey smoke. Unless there is substantial engine damage, molten aluminum particles would hardly make it through the muffler to the tailpipe.
Excess carbon particles during knock means the air-O2 intended to burn them is not used. Any O2 in the exhaust make a wideband read lean. A wideband works by measuring excess O2 (lean of stoich). Rich of stoich there should be no O2 in the exhaust at all (by definition all is used up at stoich and richer). Instead the wideband measures how much additional O2 is needed to catalytically burn the partially combusted exh. gas (CO, HC, H2 ...) to stoich. That then is a measure for the fuel air equivalence ratio FAER = (1/Lambda).
Excess O2 in the gas does not require the WB sensor to add any via it's pump cell and therefore reads lean.

The reason the lean spikes (even during a single ign. miss at high RPM) can span multiple combustion events in the log is the way the LC-1/LM-1 log.

A regular (analog feedback, not LC-1/LM-1) wideband works is by measuring the fuel/air equivalence ratio with the wideband's pump current. This feedback loop by neccessity has low pass filters. Without them the measurements would have wild over/underswings and can even oscillate uncontrollably. These filters have the effect of averaging the readings. During an ign. miss no fuel is burned and measured Lambda would be infinite, or, consequently, the measured fuel/air equivalence ratio would be 0 as all the original O2 from the air is still there.
If that wideband for example has the equivalent of averaging 50 measurements for a logging interval, at Lambda 0.80 (FAER = 1.25) with one miss, it averages 49 samples of 1.25 FAER and one sample at 0. The result after calculating the average FAER and converting to Lambda is (50 / (49*1.25)) = 0.816, hardly a noticable difference.

The LC-1/LM-1 use a different measurement principle that does not involve low pass filters. The LC-1 measures up to 500 times/second (LM-1 up to 250)and calculates Lambda for each. The resulting Lambda values are then averaged. So for 50 measurements with one miss (infinite Lambda), the averaged and logged result is still infinite. This is specifically done that way so that abnormal combustion events show clearly up in the logs.

Regards,
Klaus

[shrinker]

Thanks a lot Klaus. That explains a lot of observations Ive made in the past. I appreciate that. The method you used for logging, now that you explain it is good. Some of us just have to know this stuff or we cant die happy!