NDT

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by Eric Chadwick:
<B>The common theme also seems to be engine braking (or possibly more correctly, breaking ) under high loads. Is it likely that the damage was already done while running at high revs or is it the slowing down itself that kills the bearings - can slowing down gently after a high speed run help reduce the strain?
[/quote]

Peak load on the bearing during the four strokes of the cycle is at TDC on the exhaust stroke
there are the same inertial forces at TDC/BDC on all strokes, but on the compression stroke there's gas pressure reducing the overall load on the bearing (i.e.inertial load trying to stretch the rod, gas load trying to compress it).
Inertial loads at high engine speeds mare HUGE compared to the gas loads.
When you back off the throttle at high speed THE OVERALL LOAD ON THE BEARING WILL INCREASE - as there's no gas load acting against the inertial loads
sorry for SHOUTING but it's counterintuitive (but true)
I've got a spreadsheet somewhere which shows all this stuff.
I've also got a load of pressure diagrams for bearing films (I used to work in engine design at Cosworth), but I just haven't got round to doing the sums for the Impreza engine.
Basically it all comes down to rpm - the best thing you could do to increase Impreza engine life is to use less revs - i.e. don't use excessive engine braking.
Inertial load goes up with the square of the rpm (i.e. at 7050rpm the inertial load is TWICE that at 5000rpm.)

sorry - haven't got time to edit this into a coherent story......but may do do later.

Bob - I'll post the details of my bigend failure later too......

Nick

Adam M

NDT,

Thanks for that.

Makes complete sense. can't believe I didnt think of it before.

I now dont feel so bad about flooring the clutch whenever I let go of the accelerator.

Have conciously tried not to do, but it is automatic to me, and I can't get out of the bad habit.

So which is worse, engine braking or riding the clutch?

NDT

Alex
Actually I agree with how Adam interpreted it - declutch as soon as you come off the power.
Nick

Stelios

Just 2 let u know FYI.
*NO* no3 failures reported by the main Subaru service which i suspect services 80% of all imprezas in Athens. Suprisingly 2 no 4 failures.

jwhitton

All,

You can still get the Castrol 10w60 it is just that Halford et al do not seem to stock
it.

You can get it in the UK from PartCo.
Mobil 1 Motorsport can be had of 25 quid for 4 litres if you shop around or get it from Halfords with the 10 quid of 50 quid spent deal.

jon

AlexM

Hi NDT,

Ok - accepted that less time at mega revs on overrun will minimise the loading on the bearing.

It doesn't make sense to dip the clutch every time you're not on the throttle though - better to restrict your usage of very high rpms a bit. It can also be dangerous in the wrong circumstances, as Moray says.

Did you get a look at your engine while it was in bits? Did it show signs of having spun the bearing shells as has been suggested in this thread? Was it rebuilt with oem rods and bolts or something a bit better?

Cheers,

A.

Adam M

Alex,
In my coasting post, I wasnt entirely descriptive as I considered it a passing comment regarding my bad habits.

I was referring to being on track, or slowing down after motorway runs to the traffic lights on slip roads.

I rarely ride the clutch from 70+ mph, unless I am coming to a standstill or am certain of the road layout (eg track - the one time I went)

Moray, it is not illegal, if it is, how can it be proven?

The highway code states that you should always be in a gear from which you can accelerate, and under this interpretation you should slow down through the gears (which I try to do - even if I am riding the clutch).
I am not sure if they draw a line between being in a gear, and having that gear engaged.

Either way, I do know that what I do is frowned upon by driving instructors.

PaulMc

My UK MY97 gets serviced in 2 weeks. I am going to get the dealer to put in Castrol RS 10-60. Can anyone tell me where I can get it from if Halfords are not selling it anymore?

Cheers and very informative thread.

MorayMackenzie

Adam,

Re: "Moray, it is not illegal, if it is, how can it be proven?"

I'd call this covering all bases... what do Lawyers call it?

Have you just checked in the 'code or are you just going from memory too? I will have a look at a copy sometime and get back to you.

Oh, and Suicide is probably still illegal, but I don't remember hearing of a case where they managed to prosecute a successful suicide "victim".

Moray



[This message has been edited by MorayMackenzie (edited 13 July 2001).]

Adam M

suicide is not illegal in this country, not the same in the states I dont think, but then I am not a criminal lawyer - or any kind of lawyer for that matter.

and yes it was just from memory!

AlexM

Let's get back on topic...

I don't think that this is really the issue. To back off or not is irrelevant really.. I would want to feel confident that the engine will stay together provided that certain operational constraints are observed e.g. standard rpm limits not exceeded, servicing as per the specified schedule etc. This doesn't seem to be the case.

Until the cause can be more positively identified, no mitigating actions can really be proposed and it is this which is disturbing us all .

So... why does extended high rpm / high load use make a failure more likely than an instantaneous high rpm / load situation?.

Take your pick from:

loss of oil viscosity over time
Incorrect servicing procedures
Incorrect oil grade / specification
reduction in oil pressures at high engine temps
bearing shell material failure
rod bolts too weak
insufficient bearing area / bearing overload
temperature limits exceeded for oil or bearing shells.
.
.
.
other problems or a combination of the above.


Cheers,

A.


[This message has been edited by AlexM (edited 13 July 2001).]

Hoppy

Thread of the year, this one

Good news! You can get Castrol 10-60. I've seen it in a couple of small outlets but thought I'd check with Halfords as they're doing £10 off £50 worth of oil.

The assistant denied knowledge of it but checked with Castrol and, yes it's still a Castol line. Even better, the price at Halfords is £20.72 for 4l. DING-BL00DY- DONG

I've ordered two 16l cases - should take two weeks as not an official Halfords line and therefore needs special purchase authorisation. I'll post this info on a new thread in case people miss it.

Andrew - enjoyed your write up. Get down to Halfords!

Adam - dipping the clutch when lifting off at max speed/revs. Given what's been said about internal loads, this makes perfect sense. Could be the cure we're searching for!? I'm sure it's frowned upon as a driving technique, but then doing 150mph is not terribly popular with the law, either. What would they prefer, dipping the clutch or a thrown rod spilling it's guts? Best dip the clutch then anyway if you're going to avoid a mega accident.

And contravening the highway code is not illegal in itself, but you could get done for dangerous driving, or without due care and attention and generally being a very naughty boy.

Now all we need is a summary of today's posts from Bob 'The Man' Rawle.

Richard.

WREXY

Paul,

You probably missed it, as can happen and does happen to me alot, three posts above your post, Jon has said that Castrol RS 10W/60 is still available from PartCo. I'm happy to say that my dealer still stocks it and will continue to stock it here in Greece. I bought 5 litres of it today.

WREXY.

WREXY

Hoppy,

I'm a slow typer so while you were typing I was typing and didn't see your post till after I sent mine. You've covered the availabity of Castrol quite well. For 4 litres it cost me 30 GBP here, so 20.72 is very cheap.

WREXY.

AlexM

Sorry - I don't think dipping the clutch will really make any difference .

The bearing has already essentially failed - there are same inertial loads on the exhaust stroke whether you're on the throttle or not. All you are doing is slightly reducing the amount of time spent in the high rpm 'danger zone'

[This message has been edited by AlexM (edited 13 July 2001).]

NDT

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:<HR>Originally posted by AlexM:
<B>Sorry - I don't think dipping the clutch will really make any difference .

The bearing has already essentially failed - there are same inertial loads on the exhaust stroke whether you're on the throttle or not. All you are doing is slightly reducing the amount of time spent in the high rpm 'danger zone'

[This message has been edited by AlexM (edited 13 July 2001).][/quote]

OK here's my theory (based on some engineering knowledge, but feel free to blow it away)
initial pick up on the bearing (beyond normal wear, whatever that is) may be a single incident.
this one small bit of damage then escalates ie small bit of damage to bearing results in breaking down of the oil film in that area, so the damaged area gets large.

If the initial pick up was caused by the peak load then reducing the peak load will help avoid the overall failure.
The worst danger zone will be the period at high revs with no gas pressure, ie. just after you lift off, effectively in engine braking.
Dipping the clutch reduces this 'danger zone' completely.
Obviously if the bearing had alredy failed....then it would make no earthly difference.
But all other things being equal (yes I know they never are) it would reduce the paek load, and take you that bit further away from danger.

but I agree, it's not terribly practical, and if the underlying cause of the problem were to be sorted, then it wouldn't be necessary.

Nick

NDT

if someone can send me a full cycle pressure trace for the Impreza, then I'll do the sums on bearing loadings - it's all set up on a spreadsheet, I just don't have the pressure data.
Then if the gas pressure idea makes bugger all difference I'll freely admit it (and post the results)

Nick

MorayMackenzie

Hoppy,

Declutching at 150mph will destabilise the car and destabilising a car that's travelling 220 feet every second isn't a good thing(tm).

Moray

[This message has been edited by MorayMackenzie (edited 13 July 2001).]

Bob Rawle

Comment has already been made that the boxer engine inherantly has smaller bearing areas than other engine configurations. The Subaru crank is rated as a "toughy" but, bearing width is limited due to the engine's physical size. NDT's post is in line with my own past experience (non-Subaru) where decelleration has caused me to blow an engine.
In thinking about it that is what started me on the driving technique that I have used for so long which is to "roll" the throttle and not snap it or "prod" it either on accelleration or decelleration. The bearing loading is, as NDT points out, completely different on lift off and, when allied to the stress of high revs high load could certainly breakdown the oil film, if bolt stretch is added into the equation and bearing clearance has increased then the oil film strength in the bearing will decrease as well, this would make the bearing even more likely to pick up. Hot spots have been discussed early on in the thread, occasional pickup is not necessarily an issue as it is "worked off" under more normal conditions. Under high rev high load conditions with sudden lift off then the pick up will translate into bigger things, don't forget that if the bearing spins (ie moves in its housing not literally spinning) then the oil feed hole can be obscured and the bearing will then disintegrate very quickly.

If the oil in the engine is of "more average" grade or not suitable for the turbo heat then the higher temperatures induced will definately cause it to be less able to maintain its film strength, under those circumstances the bearing will pick up. A high quality oil will possibly withstand the event better thus saving the engine.

Declutching ... as Moray says that is a very dangerous thing to do ... anyone ever been caught by lift off oversteer, sooner or later the car will be in a situation where declutching will de-stabilise it so much someone will have a big accident. Roll off the throttle from high revs instead, balance the engine against the car and minimise the stress that way. Don't drive hard at high revs and use the engine to brake on its own, if you need to help it with the brakes. That way the car will stay loaded up and under control. Some out there will have their own ways that will be their normal style of driving but any driving instructor should advise against coasting ... especially at high speed. Remember the 4 wheel drive system relies on the engine driving the wheels to work properly, thats why you should never lift off but only accellerate if the car starts to drift away.

Dropping the engine straight to idle from a high rev/high load situation is also a bad thing as the oil pressure drops with revs but temps will rocket due to heat soak from the turbo, so what happens when you go again ? ... just a thought.

Hoppy, didn't realise that Halfords would special order, the Castrol seems to be capable of standing the pace, I also had a look at the Motul site after Jerry's post re the 300V, that also seems a good bet, I've emailed them for a comment as to suitability.

Todays mails re failures are all showing the same pattern, recent service, car driven hard, lift off then bearing failed. No no. 3 bearings today, no.2 and no.4 instead.

SecretAgentMan

We've got a nasty thing here in Sweden...the dealers get paid a given amount for an oil change...then it's up to them to choose whichever oil to put in our beloved scoobs.

Don't know if it's the same in England.

My local dealer insisted that I should use QUaker semi 5-30 in my car... Guess what...that oil happens to be dirt cheap...

/

It'd be really interesting to know what make/weight of oils have been in the engines that has blown...and when it was changed last.

I'm sticking with the silly expensive Motul.

[This message has been edited by SecretAgentMan (edited 13 July 2001).]

JamesS

....just to follow up on previous post; I recently had a 15K service at large main dealer.

Dealer had no knowledge of Mob 1 5W-40 or that 0-40 was not to be used.....supplied my own!

Big Bear

Bob
The Motul V300 oil is used in the world rally car so I was told by a rep of the company . There name is on the front bumper of the car . I now always use this oil .
Dave

darren johnson

ive been reading up on this thread for the last week as my crank (bearing picked up on no 3 cylinder) last sunday i,m just asuming every one is jumping on the band wagon blameing high revs etc for this problem mine picked up at tick over stuck in traffic. i had been givning the car some stick about 15 mins earlier i also checked oil water levels before journey ( ironic) as i intended to give the car a good blast i use castrol magnetex as my local halfords store recomend but also use synthetic slick 50. as ive done in every motor ive ever owned but i was under the impression that this and other oil aditives coat the metal parts of the engine ie reducing wear ( as any one ever seen the demo at car shows etc when they roll two bearings to gether with oil and then with friction reducer then dry) impresive! my car is uk 97 with 60k on the clockand i change oil at every 3k hoping this would never happen hope all this makes sence as been up since 4am and also had a few bacardi,s well i am missing my scoob after all its been nearly a week !"

darren johnson

Sorry i also forgot to add i,m running a vf 24 turbo on my car which is obviously bigger than standard so in my opinion perhaps heat soak is were i should be pointing the finger as i said mine went in traffic. only my own opinion DAZ

WREXY

As I said earlier I got my Castrol RS 10W/60 today. The dealer also showed me a 1 litre carton of oil with subaru on it. He said Subaru Greece were now bringing the subaru labelled oil. It was fully synthetic and rated at 10W/40. Anyone else seen this? He said it was also more expensive than the Castrol RS. I paid 30.00GBP for the Castrol, as it's expensive here. For myself however I'm sticking to my Castrol RS.


Bob brought back memories when he mentioned the way to roll the throttle and I second his advice.
When I did my advanced driving course back in OZ I was taught to balance the power to the car by distributing the power evenly with the accelerator. I remember the words "squeeze on the throttle," " don't snap it hard to the floor," same with the braking, "squeeze the brake pedal don't stomp on it." " Squeeze off the throttle." I was told by the instructor that this was the way to preserve the car, for long and short races and also for the road. "How do you think those 600hp engines last all day going hard at the Barthust 1000 touring car race?" "If they'd be driven like you lot think they should be driven, they wouldn't last 5 laps," said the instructor.

WREXY.

Bob Rawle

Darren, your car still fits the pattern, hard blast then bearing fails "after", don't feel bad about the Bacardis ... well justified.
Sorry to say that IMHO Magnatec is not up to it, Slick 50 .... modifiers are also not the thing to do, the oil you use has been designed to work as it is, additives will/can reduce their effectiveness significantly.

For what its worth I used to do the same but stopped several years ago when I became more "educated". That was when I drove normally aspirated cars.

The VF24 would not, in itself, have contributed, yes the car would have been running with more air but the oe ecu would/should have been able to cope with that. If it had been a VF23/22 on the other hand. In fact the VF24 would have reduced charge temperatures for the same boost.

WREXY

Darren,

Sorry to hear that m8. Changing oil every 3000 miles and this happens. I'm shocked. Were you doing high speeds or just high revs in lower gears before you were in traffic?

WREXY.

WREXY

Bobs just answered my question. Thanks Bob.

WREXY.

RobJenks

I have uncovered some science behind the design of bearings in GIEK -Technical Formulae if any members want the relevant information let me know and I will dispatch

pat

Goddam this is a long thread, just read all six pages of it (sad, I know, but what's a guy to do when he can't sleep due to a chest infection? )

OK there are very many points made, some of which I will try to address, in spearate "sections"...

ECU
---

The Subaru ECU (be it JECS as in the early cars, Hitachi as in some of the later cars or Denso as in the MY01/01) derives it fuelling control using data from the Mass Air Flow sensor, so obviously if that gets contaminated things aren't gonna be looking best clever... the way it reacts on boost is model year specific, but they ALL go open loop on boost, even the MY01 (it seems to have a pretty trick primary Oxygen sensor, just as an aside).

This means that on cruise, the effects of replacing any part which could affect volumetric efficiency (and let's make no mistake here, that is all that an air filter / exhaust does... it improves VE) will be compensated for twofold: firstly there will be more air, and this will be reflected by the MAF sensor, and even if it were not, the lambda control on cruise uses the ONLY point on a normal lambda sensor's output curve which is COMPLETELY independant of temperature, the "knee" where the oxygen partial pressure in the exhaust gas swings violently many orders of manitude as you go from lean of stoichiometric to rich. It "rocks" the mixture across this point, ensuring an average of stoich, which is a requirement of 3 way catalytic converters such as those fitted to Impreza.

When running on boost, the lambda doesn't come into it, you DO NOT want to rock the mixtureabout stoich on boost, this will run VERY hot (but interestingly this AFR is actually less likely to det on its own, it's just the horresdous temps that it generates that causes problems.... if you could keep it cool somehow, it would actually be quite stable!). So it runs on open loop control; depending on model year it will either use MAF, MAP or both to determine fuelling.

These cars almost invariably show rich on the rollers, BUT this does NOT mean that they will be the same on the road. I have an air charge temperature sensor in my intake which I use to trim the ignition timing to control det as the intake heats up (I back off the timing, but this raises EGT... it's a trade-off, but fortunately the Scoobysport / Haywards & Scoot equal length stainless steel headers have had an earthshattering effect on gas temps going into the turbo, dropping this up to 150 degrees C!!!).

I normally ask rolling road operators to moderate the load impedance to keep intake temps "sensible"... but still I see temps over 70 degrees C on the rollers. I NEVER EVER get anywhere even remotely close to that on the road. So, charge air density on the rollers is artificially low... just because a car runs rich on the rollers DOES NOT mean that it will be fine on the road. To be sure, fit a wide band lambda sensor to it (see below) and measure it on the road.

After market ECUs such as the Link, MoTeC and Pectel (just listing those I have personal experience of) generally are not setup to use closed loop lambda. I personally DO run closed loop lambda on cruise with lambda targets about 1.20, this makes it run quite hot but is quite friendly to the old fuel economy (theorectically the best brake specific fuel consumption is achieved at lambda of 1.05, but normal lambda sensor are very crude for anything other than the transition from lean to rich, I find 1.2 works well, it MAY in reality be 1.05, just can't trust these sensors! )

On boost I also run open loop, the temperature dependance of the transfer curve is such that you can't trust the output.... not much anyway It IS possible to do compensation with the Pectel (it can determine EGT either from its EGT input or from the Lambda sensor heater current). This is a Good Thing (TM) becuase it means that if you have a sensor you CAN trust (such as an NTK UEGO, or Bosch LSU4.7, both of which are wide band) then you can run closed loop on boost, and because of the extremely rapid response time of these sensors, as long as the PID lambda control is correctly configured would preclude any possibility of running lean..... this would be useful for fuel pressure related problems; ie if the fuel pressure were to fall off at the top end a "normal" after market ECU setup, as well as the OEM would just lean out, whereas this would react by increasing injector duty within a few milliseconds (ie less than the time it takes one cylinder to fire). This control strategy is in use in competition engines and has been for a long time.... it's just that traditionally people didn't much enjoy buying a new UKP 500 sensor every 50 hours of driving.... not an issue if the most important thing in the world is winning and cost is a secondary consideration but not ideal for road users.... this has all changed with the Bosch LSU4.7 which will provide reliable service for a reputed 100,000 miles, even with leaded petrol!! Needless to say I'm quite interested in these devices

Another stragety is to fit a fuel pressure sensor. Again, a good aftermarket ECU will allow injector duty compensation based on fuel pressure... this way any reduction in fuel pressure is immediately compensated for (it's a feedforward rather than a feedback control system).

FWIW, the Pectel T6 (same price bracket as the MoTeC M48Pro) used to have a knock sensor option; sadly this has now been withdrawn, since the mezzannine board cost as much as the ECU itself. It did, however, use a digital signal processor to identify knock on a per-cylinder basis, and adjust ignition timing to compensate. It's nice to see this kind of feature making it into mainline Subaru engine management (as in the MY01 Denso ECU)... but Bosch have been doing this for years, but with a little less "horsepower" in the CPU front.... Just as a complete p155take, the T10 ECU has up to 10 knock sensor inputs and a pair of DSPS.... but the pricetag is quite astronomical... you could buy an Impreza for the cost of one of these!

Of course, any advanced after market ECU will allow a per-cylinder trim to be applied with regard to mapped fuelling and ignition timing (ie not a feedback function, just "simply" pulling a bit of fuel out of No. 4 and put a bit more into No. 2.... I have been doing this for a while now).

Rev limiters are a Bad Thing (TM), they are about as mechanically sympathetic as a nuclear bomb.... no matter what strategy is used, on a turbocharged car they won't do your engine any favours at all.... keep away from them. The fact that most people with KnockLinks observe a BIG RED when hitting the limit should be warning enough! To just expan a little on the subject, a limiter must rob the engine of torque. There are two ways that this can be done.... 1) get rif of fuel or 2) get rid of the spark... you could do both of course. Option 2) is VERY VERY dangerous on a turbocharged engine, IMVHO. If the spark is removed, then for the first few cycles, fuelmix will just get blown through the exngine.... this will then light up in the very hot manifold. At this point, even if you has a wastegate the size of a dustbin it would be unlikely to prevent the turbo from overspooling.... you're effectively shoving the best part of one megawatt of heat energy into it.... pooor little thing. So you have NO boost control... so the boost builds relentlessly and all of a sudden even though you have no spark, fuelmix lights up in the cylinders in a very detty way. Not good for rods or bearings!

The alterantive is to cut fuel. This is a bit kinder, but also risky. The injectors wet the manifold when running. So if you shut them down there will still be fuel, so for one or two cycles the cylinder will run pretty lean, before it shuts down good and proper. Conversely when the injector is fired up again it will have to wet the manifold again, so the first few cycles will be lean unless the ECU can extend the first pulse (a wetting-pulse.... some advanced ECUs can do this). But it's still not ideal...

Rev limiters are there to stop the engine machanically flying apart due to over-rev, they are NOT there for use as a shift indicator, OK?

With regard to the fuel rails, this is the sort of thing that might be addressed by mapping... there are a few ideas that cross my mind. I find it unlikely that the OEM fuel pump cannot provide sufficient fuel. I do find it more plausible that there are restrictions in the fuel rails which could cause an appreciable difference in fuel pressure between injectors. Certainly a "fatter" rail would help here, and considering the cost and time taken to do the Gp N mod (my car has been running like this for almost a year now, admittedly due to necessity rather than design) it certainly seems like a worth while "insurance". If I get time I may try to measure the flow... I have a spare manifold or two, and a spare loom... just gotta rig it up to summat to pulse the injectors at 80% (they don't like any more, fuelling becomes inaccurate.... also they down't like to be held open more than 16ms, but the JECS drives them up to about 20ms, wonder if that is related? not lilely, but interesting lontheless).

I think that pretty much covers the ECU functions that could contribute / prevent bearing related failures....


OIL
---

Not all oils are the same, and care needs to be taken when choosing one. In general I'de advise getting the best base stock that sensible money can buy. Examples of decent roadcar oils include (but are not limited to) Silkolene, RedLine, Amsoil. Interestingly there was a long thread on the MR2 mailing list where a lot of research had been done and a few interesting points raised regarding multigrade oils and why, in some ways, it's actually better NOT to use them!

In general, though, the wider the disparity in hot / cold viscosity equivalent index, the worse an oil will be to degradation over time.... ie a 10W30 will last longer than a 10W60 of the same "quality". Of course it's all down to the base stocks used, but as a general rule of thumb, it was a case of choosing an oil that's right for the prevailing weather conditions. Let's make no mistake here, the Toyota 3S-GTE tuning scene is lightyears ahead of Subaru.... these guys are getting, with relative "ease" in excess of 700BHP from a 2 litre turbocharged lump and you'll appreciate that in such a highly stressed engine, choice of oil is very important! So, use decent oil, and replace it regularly (some of the Toyota guys will dump it after 1k road miles :eek ...

Oil flow to the no. 2 and 3 big ends comes from the No. 3 main bearing, but I do not believe that the feeder to this is any larger than to the no. 1 and no. 5 mains (which feed no. 1 and no. 4 big ends respectively). The reason that the main bearings have "recesses" is to allow oil to flow from the block, round the main journal and up the crank to the big ends even when the holes are not lined up, ensuring a continuous flow regardless of crank orientation.

The crank design was changed in 1999 when the EJ20K was replaced by the EJ205/EJ207 Phase II engine. One of the design revisions was to move the thrust bearing from No. 3 main to No.5 main, placing it closer to the source of the thrust (ie the clutch / gearbox), which should reduce any crank distortion due to microscopic compression caused by end load. I don't think that this has helped with the No. 3 as we are seeing Phase II engines letting go as well, BUT that could be due to the Phase II being much closer to the limit to start with, and that if it were mapped more conservatively it may actually stand a better chance than a Phase I. Phase 2.5 as I affectionately call the MY01 engine (it is VERY different from a normal Phase II) shares the same internals and hence also the same advantage of the thrust bearing being in a more favourable location. Only time will tell whether an individual cylinder trim is gonna save these engines...

An interesting thought had crossed my mind; perhaps the oil film breakdown could be due to either a) crank case blowby causing dilution of the oil or b) crank case blowby causing heating of the crank. It is interesting to observe that the main crank case ventilation orifice is, you guessed it, next to No. 3 cylinder (there are breathers on the heads too, but only ONE on the crank case). Perhaps just a coincidence? Perhaps more to it?

An inalienable truth has already been identified earlier in the thread, that the oil film on No.3 *MUST* be breaking down, since this is the ONLY reason that it would pick up and consequently fail. The trouble is just to identify the reason that, in particular, No. 3 breaks down (or breaks down ahead of all the others, in the absence of any other fault condition which might cause a film breakdown such as a blown head gasket).

The film shear strength of a high quality motor oil is staggering if it's within its operating range. It can break due to a few reasons.... 1) it is diluted, 2) foreign matter is introduced, 3) it is operating outside its design parameters (too hot) or 4) it is being subjected to such high loads that the film shear strength is being exceeded.

1) is possible from (for example) excessively rich running and blow-by... this can happen with older cars where the lambda sensor was in the headers and routinely got fried

2) crud being flushed through filters or introduction of water into the oil due to a cylinder head failure could alter the surface tension sufficiently to allow the film to collapse.

3) As Nito has discovered, high speed cruise creates very high temperatures. This is quite expected, since the engine is running closed loop under these conditions (ie pretty mcuh stoich) with no extra fuel to cool the thing down....

4) Detonation (well, detonation isn't really the right word for it, a proper detonation in the cylinder would literally polverise the piston) causes huge pressure spikes in the cylinder which will travel down the 'rod and will have to be absorbed by the crank; this may be sufficient to break the oil film. Indeed I have seen a dead engine which showed pickup EXACTLY where you would expect it due to this effect, and virtually none elsewhere on the big end.... scary! It is also possible that high oil temps coupled with a high speed trailing throtle could cause similar problems. "Odd" isn't it that most failures on track have happened on a training throttle?

FWIW, after a prolonged boost run I do NOT trail the throttle, I dip the clutch and blip the throttle to allow the revs to rapidly fall to about 2000 then keep them about there; this helps keep oil pressure up..... interestingly the Pectel has function whereby one connects an oil pressure sensor and then if the oil pressure falls below a preset limit if the revs are above a preset threshold it shuts the engine down.... this could be an engine-saver!

Although not strictly oil related, the use of lead-indium bearings helps insofar as them being quite "soft", so any pickup will not be so much a a gouge as a small "smear" which will easily disappear again... they ain't cheap and they won't prevent it from going bang eventually but they can save you once or twice...

Another point to note is that high EGTs will NOT raise the oil temp due to the turbo getting hot, because the percentage of oil flowing through the turbo is very small indeed. It is beyond question that high EGT WILL raise oil temps, but this perhaps more to do with the oil return from the heads / oil coming off the back of the pistons (which has either been sprayed there by squiters or has been "flung" there by the crank). Plus, the turbo's core is water cooled, so its temperature will not rise as drastically as one might first expect.


Water
-----

Interestingly noone has mentioned water / the cooling system as a possible cause.... (apart from Mark, but only in connection with possible consequence of an FMIC on oil / water temps) close inspection of the cooling system reveals two interesting snippets of info....

1) No. 3 is furthest away from the supply of cool water (the pump)

2) The flow path round No. 3 is not favourable, ie the flow of water will take an easier route if it can.

Looking at disassembled engines shows a residue (perhaps scale, perhaps "baked" on aintifreeze - unlikely) around the outside of No. 3 especially round the area closest to the turbo. Hmmm.... coincidence? The turbo should NOT effect the engine since Aluminium is an excellent conductor of heat, but the combination of (apparently, one would really need to measure it to confirm) poor water flow in close proximity to a source of extreme heat (I used to see up to 950 degrees C in the turbo!) is interesting, if nothing else.


Aerodynamics
------------

The Impreza was not designed to sustain high speed crusing, or at least not in its Japanese form. Apparently, and I have not measured this yet, at high speed airlflow through the engine bay / scoop is adversely affected. Perhaps there is a cooling problem at very high speed; this is almost confirmed by Nito's readings on a high speed cruise (although a lot of that has already been explained). Still, an interesting thought nontheless. Ideally more data is required before it would be possible to draw anything resembling a semi-sensible conclusion....


Having said all of that, I have driven my car with its original engine (STi II) off the clock for fairly large distances and have not suffered a big end failure yet (I have suffered a rod failure but the reasons are quite weird - hello Moray! and a piston ring land failure). Thia was without any additional oil cooler (I had the cooler fitted but not plumbed in at the time). It was however using good oil, and proper filling procedure

I do not believe that there is a single commmon set of circumstances that have caused all the recored failure.... I think that there is probably a common theme to a a fair number, but there will also be "odd ones out".


Engine geometry / configuration
-------------------------------

Well, short of taking the thing to bits and rebuilding it, there isn't a lot that can be done. FWIW, the ratio of the rod length to the stroke, known as the conrod ratio (funnily enough) is actually pretty close to "perfect" in the EJ20 engine, and still "acceptable" in the EJ25 with it's 2mm greater crank throw (to give a 4mm increase in displacement).

The configuration does present some interesting challenges w.r.t housing the bearings, but I do no believe that they are fundamentally too narrow. Although the acceleration (and hence force) experienced at TDC is high, it's not horrific... probably significantly lower than in a high revving motorbike engine, or something like an F1 engine (admittedly the latter does just have to last the race!). If it were simply a case of bearings being too narrow then one would expect a more even distribution of failures among the big ends 1 through 4....

This is not to say, though that there is not some other mechanical influence which is causing No 3 in particular to let go... I'm sure that there is a mechanical influence involved, just not entirely which one / how (yet)....


And finally
--------------

Bob, I wish you luck in your information
gathering excercise, you have a tall task ahead of you, but it would be good to finally find the common thread to at least one "group" of failures....

Damn this post has got long!

Cheers,

Pat.