Cumulative data on S65 engine failures.

[regular guy]

Originally Posted by swartzentruber

Since S/C'd engines are showing signs of failure due to detonation, it doesn't seem like it should be that much of a stretch that cars could be failing in lower octane states for the same reason.

Detonation can be caused by too high cylinder pressure -- which is exactly what a supercharger increases with boost. So to me, connecting detonation between NA and FI and saying if it occurs on FI then it can also occur on NA would be the wrong assumption to make.

Here's what's lacking for me in this discussion of detonation. Are we ready to admit that BMW screwed up the ECU programming if they can't handle 91 octane US gas. Yet in spite of five years to fix it...they couldn't? Are we ready to say they're that incompetent? I don't believe that, which is why I don't believe detonation on NA engines is a likely cause.

[signes]

Originally Posted by swartzentruber

I am fully aware that due to elevation, there is less of a need for the higher octane. But still, 91 octane is 91 octane. I'm not going to discount that fact that it represents another failure in a state with 91 octane just because in general there is less of a need for higher octane with elevation. I can think of a few possibilities: Maybe:
a) A colorado car doesn't need 93, but 91 is still too low (i.e. needs 92)
b) I don't really know CO, but I assume they sell 91 statewide, and I assume not the whole state is highly elevated. Perhaps the owner did most of their driving in lower elevation.
c) As RG suggests ealier, perhaps the issue isn't detonation, but something else somehow correlated with the lower octane gas

91 octane is the default premium for the state, very few locations carry 93 or 94. However, the comment "91 octane is 91 octane" is nonsense. It is not a matter of less need for higher octane, it is the physics behind air pressure/density here and combustion and the potential for detonation.

As far as altitude, the major population centers where anybody would be likely to have an M3 are Colo Springs, Denver and Fort Collins, all of which are at altitudes of a mile or higher. Maybe it was in Aspen, higher still. I'm sure SFP can chime in and say where the owner was from.

[swartzentruber]

Originally Posted by regular guy

Here's what's lacking for me in this discussion of detonation. Are we ready to admit that BMW screwed up the ECU programming if they can't handle 91 octane US gas. Yet in spite of five years to fix it...they couldn't? Are we ready to say they're that incompetent? I don't believe that, which is why I don't believe detonation on NA engines is a likely cause.

Why would BMW even be paying attemption? Overall failure rates are still very low for non S/C cars. I think most manufacturers would be pretty happy with a .5 - 1% failure rate of a high performance engine. Saying there's a possible connection between octane and failure does not equate to "BMW screwed up their ECU programming".

Lets put it this way. Suppose someone told you that with statistical significance, we know that drinking increases your risk of dying from a particular disease from 1 in a million to 2 in a million -- a 100% increase in risk. Would you stop drinking? Probably not -- we are still talking pretty low probability of death. It doesn't in any way diminish alcohols value in this case as a powerful explanatory value, but it also doesn't put alcohol on the level of cyanide. Now lets suppose someone tells you that drinking increases your risk of dying from another disease from 1/10 to 1/8, a 25% increase in risk. Would you stop? Probably, I know I would (or at least cut back). In statistics, knowing something might be explanatory is useful, but not that valuable without understanding the context, and particularly without having some idea of whether it's also causative. It is definitely also possible, as you suggest, that octane is correlated but not causative, because people in the lower octane states also do other things that are bad for engines.

An finally, I'm also not sure why the conclusion that lower octane is related to the failures necessary equates to BMW screwup. I have a somewhat strong suspicion that many of the CA failures might have also been tuned, so it could very well be the case that BMW got the ECU programming right for 91 octane, but the aftermarket tunes have not.

[catpat8000]

Originally Posted by swartzentruber

An finally, I'm also not sure why the conclusion that lower octane is related to the failures necessary equates to BMW screwup. I have a somewhat strong suspicion that many of the CA failures might have also been tuned, so it could very well be the case that BMW got the ECU programming right for 91 octane, but the aftermarket tunes have not.

It certainly is an interesting situation.

RG has asserted over and over that you cannot run 12:1 compression with 91 octane fuel unless your knock sensing technology is state of the art and he's right. Further, it seems extremely unlikely to me that BMW's knock sensing technology is the source of these engine failures. In fact, I'd be surprised if the BMW technology couldn't handle lower than 91 octane. Why would they limit it? If I were designing the SW, I would require that if the system couldn't ensure there wouldn't be damage, it wouldn't even allow the engine to run. That should be pretty easy, especially since the throttle is controlled by the computer. It would be easy to simply not allow a load on the engine if the SW believed damage would occur.

I believe it's also true that tuners really have no idea how their tunes might affect detonation/knock/preignition in various kinds of corner cases. If it works on their dynos in LA during 80 degree weather, it must work well in every scenario, right? I personally think you'd have to be crazy to get your car tuned IF you were also worried about the potential of losing your engine. The tuner is going outside the factory derived settings and without *any* solid engineering to determine whether exceeding these factory settings has a long term detriment.

But I think this data is also suggesting that rod bearing engine failures are not singley correlated with tight bearing clearances. For one thing, the failure rate is way, way too low. For another, there shouldn't be giant disparities between states, although I admit we may still just have noisy data.

I think one of the most important things to continue to do is to try to find and track additional engine failures. That's how we can all contribute. The more data we have their, the more likely we'll be able to draw firm conclusions.

Pat

[catpat8000]

Originally Posted by US///M3

Most dont advetise having a tune when they end up with a blown engine for obvious reasons.

You are right and I am wrong. SFP clarified - his data does not exclude tuned cars (it does exclude SC cars). We simply don't know if the cars were tuned or not.

Pat

[swamp2]

Originally Posted by regular guy

Here's what's lacking for me in this discussion of detonation. Are we ready to admit that BMW screwed up the ECU programming if they can't handle 91 octane US gas. Yet in spite of five years to fix it...they couldn't? Are we ready to say they're that incompetent? I don't believe that, which is why I don't believe detonation on NA engines is a likely cause.

Hate to beat the dead horse but you are willing to accept that BMW screwed up the oil and bearing clearance specification. Or perhaps the slightly less harsh opinion of BMW M engineering that this is happening only in cases of unfavorable tolerance stack up.

[swamp2]

0.5% is the worst case number I calculated (back of the envelope, given the reported failures along with an estimate to account for those owners not present here on this forum).

Originally Posted by swartzentruber

Lets put it this way. Suppose someone told you that with statistical significance, we know that drinking increases your risk of dying from a particular disease from 1 in a million to 2 in a million -- a 100% increase in risk. Would you stop drinking? Probably not -- we are still talking pretty low probability of death. It doesn't in any way diminish alcohols value in this case as a powerful explanatory value, but it also doesn't put alcohol on the level of cyanide. Now lets suppose someone tells you that drinking increases your risk of dying from another disease from 1/10 to 1/8, a 25% increase in risk. Would you stop? Probably, I know I would (or at least cut back).

Excellent analogy.

This just is not the E46 M3 bearing case.

[regular guy]

Originally Posted by swamp2

Hate to beat the dead horse but you are willing to accept that BMW screwed up the oil and bearing clearance specification. Or perhaps the slightly less harsh opinion of BMW M engineering that this is happening only in cases of unfavorable tolerance stack up.

I'll beat the dead horse too!

I'm more inclined to call it a misguided design decision, instead of a screw up. And of course I'm more willing to give a pass to the ECU programmers because I design similar closed-loop software for a living and I have more faith in what those guys are doing.

But yes, I think it's mostly in the cases of tolerance stack up, not an endemic problem. But to the guy who said 1% is tolerable, I would disagree and be horrified by that number. I think 0.1% is tolerable. But admittedly, I'm pulling that out of thin air based on my own industry.

[swamp2]

Some of our forum tuners have chimed in on this topic. I think they could help us out a lot more.

• How many tunes are sold in CA vs. other states? I bet it is higher in CA on a per car ownership basis.
• What are they really tweaking? Is the timing advance riding an already fine razor edge WRT octane?
• I'm sure no admissions of guilt will be made here, but I firmly agree with the point that tuners ARE almost never engineers (at least not in a relevant domain) and likely don't understand some of these subtleties as well as they should.
• SFP - perhaps you can expand your work reaching out to those with failures and get some private, non owner identifiable data about who did and did not have a tune?

[swamp2]

Originally Posted by regular guy

I'm more inclined to call it a misguided design decision, instead of a screw up. And of course I'm more willing to give a pass to the ECU programmers because I design similar closed-loop software for a living and I have more faith in what those guys are doing.

Wow, you've got to be kidding, you can't be that biased for such a poor reason.

Originally Posted by regular guy

But yes, I think it's mostly in the cases of tolerance stack up, not an endemic problem. But to the guy who said 1% is tolerable, I would disagree and be horrified by that number. I think 0.1% is tolerable. But admittedly, I'm pulling that out of thin air based on my own industry.

1% pretty high also in my mind. Hard data (not any of your anecdotal evidence from phone calls with dealers) puts the number at about 0.1%. Again we can get up to about 0.5% with a fairly liberal approach to account for failures not reported on this forum (using number of forum members claiming ownership of an M3 vs. total number of US cars).

[US///M3]

Imo it's a combination of things, running low octane gas, tunes( especially those raising the redline), poor design....like what happened in Chernobyl.

[SenorFunkyPants]

Originally Posted by signes

I'm sure SFP can chime in and say where the owner was from.

All I was given was Colorado.

[SenorFunkyPants]

Originally Posted by swamp2

[*]SFP - perhaps you can expand your work reaching out to those with failures and get some private, non owner identifiable data about who did and did not have a tune?

I'll have another go..

[swartzentruber]

Originally Posted by regular guy

I'm more inclined to call it a misguided design decision, instead of a screw up. And of course I'm more willing to give a pass to the ECU programmers because I design similar closed-loop software for a living and I have more faith in what those guys are doing.

I work in software too, and actually take the opposite perspective, that know how easy it is to overlook the "corner cases" as referred to earlier, it isn't a stretch for me to imagine some cases the programmers did not anticipate and account for. State of the art hardware makes it even more likely that something could have been programmed wrong for certain rare combinations of environmental conditions, as there'd be less time to work out the kinks in the field. I like to think back to the original batman (with Michael Keaton), where the joker was poisoning people, but only if you used certain combinations of products. Hair spray was fine, but hair spray and deodorant killed you. Maybe 91 octane by itself is fine, but 91 octane with slightly over spec levels of ethanol causes issues.

Another item that has surprisingly never came up is what if any impact the later software revisions made. I really don't know what all got changed in 240e, but I certain notice some behaviors are changed, including warmup. Maybe BMW did discover some issues and fix them, but didn't think the failure level was high enough to warrant pushing 240e out for everyone. I only mention this because their seems to be somewhat of an attitude that if it's an ECU issue, why hasn't it been fixed, and my thought is it possibly has been, but they just didn't want to announce it to the world and issue a recall, because to-date the number of failures has been pretty low.

[SenorFunkyPants]

Originally Posted by catpat8000

You are right and I am wrong. SFP clarified - his data does not exclude tuned cars (it does exclude SC cars). We simply don't know if the cars were tuned or not.

Pat

There is actually only one car in the chart (previous page) that stated a tune, the rest all stated no.
As part of the research I read most all of each owners posts to see if there was mention of a tune or other mods.

[regular guy]

Originally Posted by SenorFunkyPants

There is actually only one car in the chart (previous page) that stated a tune, the rest all stated no.
As part of the research I read most all of each owners posts to see if there was mention of a tune or other mods.

I feel for ya...I have to do the same detective work when searching for Dyno Database entries.

[Petros]

If detonation or low octane or bad software is the cause, is that also the case with the S85 in the M5? Are they also getting bearing failure mostly in California? Do they also have crappy software that can't prevent knock? This car has been produced since 2005. So from the first M5 in 2005 to the last M3 in 2013, a run of 9 years, BMW has been unable to prevent their engines from detonating? Is this scenario even remotely believable?????

[Ezio]

Originally Posted by Petros

If detonation or low octane or bad software is the cause, is that also the case with the S85 in the M5? Are they also getting bearing failure mostly in California? Do they also have crappy software that can't prevent knock? This car has been produced since 2005. So from the first M5 in 2005 to the last M3 in 2013, a run of 9 years, BMW has been unable to prevent their engines from detonating? Is this scenario even remotely believable?????

because cally has the most M3s

[SenorFunkyPants]

Originally Posted by Petros

If detonation or low octane or bad software is the cause, is that also the case with the S85 in the M5? Are they also getting bearing failure mostly in California? Do they also have crappy software that can't prevent knock? This car has been produced since 2005. So from the first M5 in 2005 to the last M3 in 2013, a run of 9 years, BMW has been unable to prevent their engines from detonating? Is this scenario even remotely believable?????

The rod bearing wear patterns are quite similar in S85 engines.
The ionic sensing anti knock system of the S85 was redesigned for the S65.
The Vanos system of the S85 was redesigned for the S65.
The lubrication system was redesigned for the S65.
There actually weren't that many recorded S85 engine failures. I did list them at one time (its in the bearing thread) but didn't put as much effort in as there there were so few to work with. The majority though did state an engine tune. Only 1 S85 was a Euro failure at ~143,000 miles.

Out of 30,000 owners how many do you think would on occasion have used low (<91) octane fuel - because they didn't know better, by mistake, to save a few $ or simply a lease car owner who couldn't care less?

The ECU may well be perfectly adapted to run 91 octane fuel but it won't handle 87 octane at all well.
Its a simple application of Murphy's law - in Euro land you can't buy <91 octane fuel so no one can put it in their cars - though Euro car owners still conspire to put diesel in instead!

[swamp2]

Originally Posted by Ezio

because cally has the most M3s

Perhaps you've missed much of the other discussion...

When making various attempts to look at the failure rate (not NUMBER of failures) on a country and state by state basis, CA comes out with a much higher rate. There is no obvious reason for the rates to be significantly different.

[Ezio]

Originally Posted by swamp2

Perhaps you've missed much of the other discussion...

When making various attempts to look at the failure rate (not NUMBER of failures) on a country and state by state basis, CA comes out with a much higher rate. There is no obvious reason for the rates to be significantly different.

perhaps

[MrStinky]

Just want to add something in favor of tuners on here. I have logged data with and without a popular tune on here and I also know from asking all major tuners leave timing targets largely unchanged. Under perfect conditions the timing targets were increased by 2 degrees in latter half of power curve but the vast majority of time our ecu pulls timing constantly that it never gets close to hitting new targets. No tuner forces ignition timing like an open loop scenario and since all operation is closed loop, its impossible to have too much timing on these cars if you keep the knock system in tact which all tuners do. It is almost impossible to hit stock timing of 32 degrees.

Also a/f is really unchanged or barely. In my view power increase often happens in mid range and is more a result of vanos timing overlap adjustment and timing in midrange but again since closed loop occurs if the conditionsxarent right it won't hut stock timing let alone higher values. Vanos for all intents and purpose provides a mild camswap without changing hardware. You can change durations and overlap like a performance can would. There is a limit but anyone who has done a small can improvement in older car sees 15-20 HP which is what most M3 tunes produce.

Imo BMW leaves room on the table for cam /vanos tuning is stock settings are emissions driven to some degree in terms of overlap and duration. However changing these really don't out any risk of detonation, esp since gains cone mid range to high rpm which at 4500 plus rpm, things already moving fast enough where preignitiin is unlikely and this no detonation.

Anyone who has logged sews ecu is constantly pulling timing almost always. Only in sub 70 degree weather with newer spark plugs and 93 plus octane have I ever it max timing targets.

So in my view it only makes sense detonation would be problem with FI where detonation can occur for reasons other than preignition due to blower creating crazy high cylinder pressures vs NA where detonation exclusively comes from preignition which other than lugging motor, doesn't happen when engines spin to 8400

One note also you can't sat 12:1 cr can't run 91-93 octane although I agree you can't quite hit max timing but you have too look at 2 important things. First is effective dynamic CR which varies drastically between different static 12:1 cr engines. Second is rpm operation of engine because at 5k plus preignition simply has no time to occur and that's why spinning ahigh cr engine to high rpm allows for high cr as no time exists for ignition of unburned fuel by heat or hotspots.

By more aggressively setting cam timing analogous to putting in larger cam, the dynamic cr is even smaller because as the intake valve remains open longer, it will allow fas escape even past BDC and thus no true compression occurs at a 12:1. variable car is often 7-8:1 for a static cr of 12::1. So tuners can " lower" the dynamic cr by increasing cam durations and thus can often hit higher ignition targets mid range than stock and is why most increase comes from mud range. At higher rpms this effect is less relevant because intake air under higher ROMs is being sucked in with more vacuum pressure and thus air cannot exit back out of cylinder out intake valve when incoming air pressure is higher at high rpm.

My point, a static cr of 12:1 is quite meaningless and until someone accounts for dynamic ratio including intake valve duration, intake and cylinder design from swirling to cooling, and exhaust valve gas flow which allows even more expulsion of gases and more intake vaccum and higher tolerance of higher cr without gas reversing flow leaving intake valve ( is cat less cars tolerate higher CR due to this more efficient expulsion and secondary I'm proved intake. Vacuum effect on dynamic car

12:1. Is meaningless in my view as any way to dictate detonation risk