Author Topic: How to define unreliability.  (Read 2041 times)

peteo48

  • Approved Member
  • *
  • Posts: 2689
  • Country: gb
  • I have entered the Jazz Age
  • Fuel economy:
  • My Honda: 2021 Honda Jazz Mk4 1.5 i-MMD EX
How to define unreliability.
« on: May 23, 2018, 11:16:05 AM »
The issue around Deflation Warning Systems or Tyre Pressure Monitoring Systems got me thinking about how you would define unreliability. When I had my Golf there was often heated debate around the many issues that plagued the Mk5 Golf - especially the diesel. One such issue was failure of the ABS pump. VW enthusiasts would jump up and down when somebody complained about this saying "Well mine is OK so you are talking rubbish."

The nearest definition I can come up with is the measure used by the drug companies when describing side effects of any given drug. They are grouped into common, uncommon, rare and very rare. A side effect is common if it affects one in ten of people taking the pill. In short 90% won't experience this problem.

We have a number of engineering bods on the forum and I wondered if there is an engineering definition of "unreliable" for any given piece of technology?

On TPMS systems I don't have any figures of failure but the anecdotal evidence is pretty convincing as shown by the number of entries on the internet and the growth of technology to defeat them. I have also been struck by the number of acquaintances who have had issues. For this reason I deem TPMS systems to be unreliable. The fact that a majority do not have issues does not mean they are reliable.

ColinB

  • Approved Member
  • *
  • Posts: 1162
  • Country: gb
  • My Honda: 2015 Jazz 1.3 SE manual in Milano Red
Re: How to define unreliability.
« Reply #1 on: May 23, 2018, 01:23:36 PM »
There’s another thread about “reliability” that’s worth a browse:
https://clubjazz.org/forum/index.php?topic=9589.0
Some comments in there about what we mean by “reliability”, eg some contributors felt that only defects that actually immobilised a car would be reason enough to call the car unreliable. Some surveys use data from warranty companies, ie if an owner has felt strongly enough about something to get it fixed under warranty then that gets factored into some quantitative measure of “reliability” (doesn’t work for manufacturer’s warranties though, they don’t publish that data, nor for defects that aren’t covered by the warranty).

Personally I think there’s a missing piece of the puzzle, which is the methodology for translating a list of defects into a measure of reliability. That would involve making assumptions about the severity of any defect, whether it affected safety, performance, environment, and any number of other things. That’s going to be very subjective. It’s probably possible to base something around your drug analogy but I doubt there’s a universally-agreed way of doing that.

Regarding your TPMS issue, it’s undoubtedly annoying for you and I wouldn’t belittle that. But it only occurs intermittently, it doesn’t prevent you driving the car (after manually checking the tyres, of course), and it can be reset easily. Should that brand the car as “unreliable” ?

Jocko

  • Approved Member
  • *
  • Posts: 9356
  • Country: scotland
  • Fuel economy:
  • My Honda: Died from rust.
Re: How to define unreliability.
« Reply #2 on: May 23, 2018, 03:37:42 PM »
We have a number of engineering bods on the forum and I wondered if there is an engineering definition of "unreliable" for any given piece of technology?
We used to measure Mean Time Between Failures. MTBF would be measured in Hours, Miles, Cycles, or whatever was appropriate for the piece of kit.

peteo48

  • Topic Starter
  • Approved Member
  • *
  • Posts: 2689
  • Country: gb
  • I have entered the Jazz Age
  • Fuel economy:
  • My Honda: 2021 Honda Jazz Mk4 1.5 i-MMD EX
Re: How to define unreliability.
« Reply #3 on: May 23, 2018, 05:34:42 PM »


Personally I think there’s a missing piece of the puzzle, which is the methodology for translating a list of defects into a measure of reliability. That would involve making assumptions about the severity of any defect, whether it affected safety, performance, environment, and any number of other things. That’s going to be very subjective. It’s probably possible to base something around your drug analogy but I doubt there’s a universally-agreed way of doing that.

Regarding your TPMS issue, it’s undoubtedly annoying for you and I wouldn’t belittle that. But it only occurs intermittently, it doesn’t prevent you driving the car (after manually checking the tyres, of course), and it can be reset easily. Should that brand the car as “unreliable” ?

Good point and I accept that. The issue doesn't mean I have an unreliable car - just a component which, as you say, is more nuisance value than anything.

culzean

  • Approved Member
  • *
  • Posts: 8017
  • Country: england
Re: How to define unreliability.
« Reply #4 on: May 23, 2018, 05:53:14 PM »
Mean time between failures is, as Jocko says what industry uses. The accepted truth is that the more complex a piece of equipment the more often it will be expected to break down (if you factor in the MTBF for each component and sub-assembly).  There is a graph of failure rates called a ' bathtub curve' which shows higher failure rates early in the life of equipment followed by a flat line during its normal life and then a rise in failures towards the end of its life as things wear out' .   https://en.wikipedia.org/wiki/Bathtub_curve

Military equipment and aerospace (satellites etc.) are normally subject to a rigorous 'burn in period' where they are subject to loads above their normal running design spec to weed out any defective parts which would fail early (infant mortality) as much as humanly possible.  I did read years ago that Honda ran engines flat out for 24 hours straight and if anything failed it was analysed and strengthened if needed and then another 24 hour run and so on until the engineers were satisfied.  When I worked in a design office we made drive train test rigs for auto companies which would fit the whole engine and drivetrain on, the wheel were loaded up by DC motors and the whole rig was fed with 'road-load data' (a favourite one was a circuit of the Nurbergring) which would replicate the drive as many times as they liked, with robot working the accelerator, clutch and gearchange. The rigs ran 24/7 and were locked in fireproof cells and the control room windows were armoured glass in case anything disintegrated.

It is only fair to grade problems from 'annoyance' to 'leaving you stranded' and also by cost of repair. Otherwise a lot of annoyances can make something look worse than a fewer number of absolute expensive breakdowns that left people stranded.

It is common in industry to grade breakdowns by severity, and as Jocko said earlier engineer them out if possible to prevent the same thing happening again. Industry also uses 'root cause analysis' to make sure the failed part is identified properly and not just lumped into 'equipment failure'.
« Last Edit: May 23, 2018, 07:06:10 PM by culzean »
Some people will only consider you an expert if they agree with your point of view or advice,  when you give them advice they don't like they consider you an idiot

Jocko

  • Approved Member
  • *
  • Posts: 9356
  • Country: scotland
  • Fuel economy:
  • My Honda: Died from rust.
Re: How to define unreliability.
« Reply #5 on: May 23, 2018, 06:04:45 PM »
When I worked for Lexmark, we did a super course on Root Cause Analysis. Unfortunately, unless we had a very expensive failure, we got little chance to use it, as production just wanted it "fixed", then on to something else.

VicW

  • Approved Member
  • *
  • Posts: 1441
  • Country: england
  • My Honda: 07 Plate Civic 1.8 i-Shift.
Re: How to define unreliability.
« Reply #6 on: May 23, 2018, 07:21:29 PM »
Mean time between failures is, as Jocko says what industry uses. The accepted truth is that the more complex a piece of equipment the more often it will be expected to break down (if you factor in the MTBF for each component and sub-assembly).

I spent some time in the RAF and on one particular aircraft, a complicated interceptor, an MTBF of four hours for the main radar was normal. It has to be appreciated,however, that the ratio of technicians to aircraft in any one aspect of expertise was about one to two. The working day was about twenty hours split into two shifts.

Vic.

« Last Edit: May 23, 2018, 07:25:46 PM by VicW »

culzean

  • Approved Member
  • *
  • Posts: 8017
  • Country: england
Re: How to define unreliability.
« Reply #7 on: November 10, 2018, 03:26:11 PM »
Shock horror my sister-in-laws 16 year old Honda Civic just failed an MOT ( never happened before - never even an advisory on the sheet ) - the cause ? a steering ball joint problem because the gaiter got split and let water in.  The only things replaced on it in 16 years are pads, discs ( but not drums or brake shoes ), a battery and an exhaust box + cam belts. Never been serviced by a garage,  just given oil and filter changes by my brother,  done over 90,000 miles.

Now that is what I call reliability.............
« Last Edit: November 10, 2018, 03:30:23 PM by culzean »
Some people will only consider you an expert if they agree with your point of view or advice,  when you give them advice they don't like they consider you an idiot

peteo48

  • Topic Starter
  • Approved Member
  • *
  • Posts: 2689
  • Country: gb
  • I have entered the Jazz Age
  • Fuel economy:
  • My Honda: 2021 Honda Jazz Mk4 1.5 i-MMD EX
Re: How to define unreliability.
« Reply #8 on: November 10, 2018, 05:09:55 PM »
Absolutely. Impressive stuff. There are loads of old Hondas on the road round our neighbourhood. Civics and Jazzes mainly with a few Accords.

Tags:
 

anything
Back to top