15 captured repliesPartial discussion capture · Source accounts, not a confirmed diagnosis
THIS DISCUSSION / SOURCE-DERIVED OVERVIEW
Discussion summary
YOUR RESEARCHIMS data collection update - 132 cars (Page 3)
Reported or suspected cause
current sample size is not big enough
Suggestion or failed attempt
No successful repair reported for this finding.
0
Identified cause
1
Suspected cause
0
Reported fixes
Read evidence and context
Explicit resolution reports before identified causes, performed actions and suggestions. Within each level, matching component labels precede direct query mentions and inherited thread context; successful versus unsuccessful discussion counts, owner confirmations and independent discussion support determine order. This is not a probability of success.
Discussions that suspect a cause
1
0 original-poster resolution reports. Counts can overlap; they are not unique people or a success rate.
0 discussions have direct query context; 0 report a fix in that context. Other evidence may inherit the discussion’s subject.
Suspected cause
However, I suspect the current sample size is not big enough to be conclusive one way or the other.
Leading evidence: 'current sample size is not big enough' (no explicit attributable resolution; this is a reported action or suggestion). Across 1 matching discussions, the most frequently mentioned possible cause is 'current sample size is not big enough' (1 discussions; 0 report identifying it and 1 suggest it). The most frequently discussed action is 'Replace bearing due' (1 discussions). For that action, 0 discussions report a resolution and 0 report an unsuccessful attempt. No action has an explicit, attributable resolution report under the current rules. These are forum reports and suggestions, not a confirmed diagnosis.
Source reports, not a diagnosis or a success probability. Counts refer to discussions and can overlap.
Explicit resolution reports before identified causes, performed actions and suggestions. Within each level, matching component labels precede direct query mentions and inherited thread context; successful versus unsuccessful discussion counts, owner confirmations and independent discussion support determine order. This is not a probability of success.
Discussions that contain suggestions
1
0 original-poster resolution reports. Counts can overlap; they are not unique people or a success rate.
0 discussions have direct query context; 0 report a fix in that context. Other evidence may inherit the discussion’s subject.
Suggestion
Will be interesting to see how many have changed the bearing due to paranoia rather than need?
Explicit resolution reports before identified causes, performed actions and suggestions. Within each level, matching component labels precede direct query mentions and inherited thread context; successful versus unsuccessful discussion counts, owner confirmations and independent discussion support determine order. This is not a probability of success.
Discussions that contain suggestions
1
0 original-poster resolution reports. Counts can overlap; they are not unique people or a success rate.
0 discussions have direct query context; 0 report a fix in that context. Other evidence may inherit the discussion’s subject.
Suggestion
You could look at it the other way and it might be interesting to see if any bearings that have been changed consequently failed?
Counts describe captured discussions, not unique people or a diagnosis. Author identities are not reliably stored. Exact duplicate evidence is counted once across matching discussions; one discussion counts at most once per action/state. Multiple repairs reported together remain a combination. Suggestions and failed attempts do not establish a fix. Matches cover whole discussions and can include related faults, model variants and incomplete threads.
15 eligible posts · 0 unverified posts · 1 discussions match the subject in their title or opening post.
911uk: 1 discussions
0 evidence records excluded for query context · 0 excluded by outcome qualification · 0 adjusted by outcome qualification.
0 acknowledgements have no verified action link; they do not establish a repair result.
Generation: 20261027T005001Z · outcome-qualification-6. Deterministic extraction; no AI model inference. How we use sources
Just been looking at the data collected so far and thrown a couple of graphs together for folk to look at.
Here's a couple of bar charts showing which engines and models have sustained a failure:
Total number of failures - 9.8%
of which 3.4 - 0% and 3.6 - 13.3%
of which all X51 - 0%, C2 - 16.6% and C4S - 14.81%
Looking at the 6 failures reported I've done a graph displaying their mileage over time and 3 points - registered date, Failure date and data inputted date.
Notice how 5 of the 6 failures were all registered within a 12 month period and out of those, 3 of them failed within a few weeks of each other 9 years later.
Last reported failure was July 2014 and then June 2013 before that.
I'll do any updates in this post rather than the sticky so it doesn't get too messy in that post.
Once again, thank you to everybody involved. :thumbs:
Thanks for doing the analysis Alex. I'm still considering getting into an early car before prices begin to rise.
I suspect the % failure rate may be higher than the true norm for reasons discussed in the original thread.
Hopefully more data will continue to be added as this should increase the accuracy.
The breakdown between models is really interesting. Seems that some of the theories (such as early 3.4's are less prone to this kind of failure) are borne out in the sample.
Are you thinking of doing a level-headed bore score survey once this is completed? Might throw up some interesting patterns too? You can never have too much information and nice to see these issues discussed in a sensible way.
Very interesting results.. so most seem to be at quite low mileage? My new engine is coming upto 40k odd now. Fingers, toes, legs and arms crossed :lol:
Difficult to prove that one, though. You could look at it the other way and it might be interesting to see if any bearings that have been changed consequently failed? If not (or the failure rate is significantly lower), it's difficult to say that changing is necessarily a bad thing, if the result is a lower failure rate and less paranoia? However, I suspect the current sample size is not big enough to be conclusive one way or the other.
This is a source reference. Confirm exact model applicability in the original material.
