Forums

Mathworks Ebook Predictive Maintenance Using MatLab.

Started by Fred Bloggs September 16, 2021
This could get scary. I hope they validate their machine constantly, on the side of safety that is.. 

https://www.mathworks.com/content/dam/mathworks/ebook/gated/predictive-maintenance-ebook-all-chapters.pdf
On Thu, 16 Sep 2021 15:40:34 -0700 (PDT), Fred Bloggs
<bloggs.fredbloggs.fred@gmail.com> wrote:

>This could get scary. I hope they validate their machine constantly, on the side of safety that is.. > >https://www.mathworks.com/content/dam/mathworks/ebook/gated/predictive-maintenance-ebook-all-chapters.pdf
Jet engines and gearboxes have an ODM oil debris monitor system that looks for small metallic particles in the lube oil, to see how much the gadget is tearing itself apart. We make an oil debris monitor simulator! Pretty obscure. People used to glue particles onto a string and drag it through the ODM sensor. -- If a man will begin with certainties, he shall end with doubts, but if he will be content to begin with doubts he shall end in certainties. Francis Bacon
On Thursday, September 16, 2021 at 7:02:58 PM UTC-4, John Larkin wrote:
> On Thu, 16 Sep 2021 15:40:34 -0700 (PDT), Fred Bloggs > <bloggs.fred...@gmail.com> wrote: > > >This could get scary. I hope they validate their machine constantly, on the side of safety that is.. > > > >https://www.mathworks.com/content/dam/mathworks/ebook/gated/predictive-maintenance-ebook-all-chapters.pdf > Jet engines and gearboxes have an ODM oil debris monitor system that > looks for small metallic particles in the lube oil, to see how much > the gadget is tearing itself apart. > > We make an oil debris monitor simulator! Pretty obscure. People used > to glue particles onto a string and drag it through the ODM sensor.
How does that work? Seems there's a LOT of theoretical background here that the average programmer couldn't possibly know about. Pretty soon we'll all be instrumented out with bio sensors. They'll send us a text alert when RUL is getting close.
> > -- > > If a man will begin with certainties, he shall end with doubts, > but if he will be content to begin with doubts he shall end in certainties. > Francis Bacon
On Fri, 17 Sep 2021 09:47:22 -0700 (PDT), Fred Bloggs
<bloggs.fredbloggs.fred@gmail.com> wrote:

>On Thursday, September 16, 2021 at 7:02:58 PM UTC-4, John Larkin wrote: >> On Thu, 16 Sep 2021 15:40:34 -0700 (PDT), Fred Bloggs >> <bloggs.fred...@gmail.com> wrote: >> >> >This could get scary. I hope they validate their machine constantly, on the side of safety that is.. >> > >> >https://www.mathworks.com/content/dam/mathworks/ebook/gated/predictive-maintenance-ebook-all-chapters.pdf >> Jet engines and gearboxes have an ODM oil debris monitor system that >> looks for small metallic particles in the lube oil, to see how much >> the gadget is tearing itself apart. >> >> We make an oil debris monitor simulator! Pretty obscure. People used >> to glue particles onto a string and drag it through the ODM sensor. > >How does that work?
Basically three coils outside a pipe https://tinyurl.com/yed9q7jq with AC on the outers and sensing on the inner. The sensor coil output is zero unless a ferrous or non-ferrous metal particle upsets the field symmetry. The sizes and metal types can be distinguished, which hints at what's being ground up. Gastops makes the sensors. We make a simulator box for testing the aircraft health monitor systems. Basically an ARB and some coupling and modulator stuff. -- If a man will begin with certainties, he shall end with doubts, but if he will be content to begin with doubts he shall end in certainties. Francis Bacon
On 17/09/2021 17:47, Fred Bloggs wrote:
> On Thursday, September 16, 2021 at 7:02:58 PM UTC-4, John Larkin wrote: >> On Thu, 16 Sep 2021 15:40:34 -0700 (PDT), Fred Bloggs >> <bloggs.fred...@gmail.com> wrote: >> >>> This could get scary. I hope they validate their machine constantly, on the side of safety that is.. >>> >>> https://www.mathworks.com/content/dam/mathworks/ebook/gated/predictive-maintenance-ebook-all-chapters.pdf >> Jet engines and gearboxes have an ODM oil debris monitor system that >> looks for small metallic particles in the lube oil, to see how much >> the gadget is tearing itself apart. >> >> We make an oil debris monitor simulator! Pretty obscure. People used >> to glue particles onto a string and drag it through the ODM sensor. > > How does that work? > > Seems there's a LOT of theoretical background here that the average programmer couldn't possibly know about.
How to do it optimally has been known for around a couple of decades. The old traditional way was to try and avoid all critical in service failures by replacing still working parts before they reached MTBF. We christened it causative maintenance since it was pot luck whether or not the damn mainframe would come back up again after a manufacturers "preventative" maintenance. Everyone knew it was a scheme to harvest money from the customer rather than prevent failures but it was another decade or two before the mathematical basis was formally proved. Snag was a lot of components have a burn in infant mortality that seriously affects their apparent MTBF so you end up provoking a burn in failure by replacing a perfectly good part with a decent residual life expectancy with a brand new one that promptly fails. Great news if you are a service company for big kit but very bad news for the customer. Exactly how many unexpected in service failures you can tolerate is a very important part of doing the sums. Also what hot spares you need to hold and where and who will do the repair at a moments notice. When these sorts of risk calculations go wrong they go spectacularly and horribly wrong. Think famous hedge funds with Nobel prize winners that believed that they could never fail to make money. (and with bottomless pits of money they would have been right) -- Regards, Martin Brown
On a sunny day (Sat, 18 Sep 2021 10:25:44 +0100) it happened Martin Brown
<'''newspam'''@nonad.co.uk> wrote in <si4bao$1rp4$2@gioia.aioe.org>:

> >How to do it optimally has been known for around a couple of decades. >The old traditional way was to try and avoid all critical in service >failures by replacing still working parts before they reached MTBF. > >We christened it causative maintenance since it was pot luck whether or >not the damn mainframe would come back up again after a manufacturers >"preventative" maintenance.
In my TV days we would run shifts and if something failed fix it ASAP there was also some redundancy in equipment so you could divert operations. One day there was some genius who decided to reduce interruptions in the broadcasts to make a 'preventive maintenance' group, and to work those engineers went,,, Basically with Ampex quadruplex video recorders that meant taking one apart and putting it back together again. The usual things we already did on schedule, like replacing recording heads etc. Disassembling and putting back together solid state electronics does not really help, 'if it ain't broken do not fix it; Anyways one machine started to produce little black spots every now and then in the picture, It was sidelined, and several engineers had a go. One evening I had the late night shift, not much to do, everything worked, so I decided to have a look at that machine, Took the scope cart, Tek, could detect strong impulses everywhere, was midnight and we went off air (in those days), so I wrote in the book for the next shift 'seen pulses everywhere, looks like a shield is lose'. My boss, it was his turn next morning, he read that remark and spend the whole day taking the thing apart, no luck. Next day it was my shift, took all circuit diagrams, scope and went for it. The pulses reminded me of some tacho signal used, found the pulse generator and followed its output via a coax cable that send it to an other unit, PL259 connector, was sort of lose, screwed it back on tight, problem fixed. 10 minutes.... Went back with all the books and scope card and said: Hey guys, let's go get some coffee. All the faces there in disbelief.. OK, I said I will show you and reproduce it,.. So how did that cable come lose? The preventive maintenance group had forgotten to tighten that PL259 connector.
On Saturday, September 18, 2021 at 11:23:16 AM UTC-7, Jan Panteltje wrote:
> On a sunny day (Sat, 18 Sep 2021 10:25:44 +0100) it happened Martin Brown > <'''newspam'''@nonad.co.uk> wrote in <si4bao$1rp4$2...@gioia.aioe.org>: > > > >How to do it optimally has been known for around a couple of decades. > >The old traditional way was to try and avoid all critical in service > >failures by replacing still working parts before they reached MTBF. > > > >We christened it causative maintenance since it was pot luck whether or > >not the damn mainframe would come back up again after a manufacturers > >"preventative" maintenance. > In my TV days we would run shifts and if something failed fix it ASAP > there was also some redundancy in equipment so you could divert operations. > One day there was some genius who decided to reduce interruptions in > the broadcasts to make a 'preventive maintenance' group, and to work those engineers went,,, > Basically with Ampex quadruplex video recorders that meant taking one apart and putting it back together again. > The usual things we already did on schedule, like replacing recording heads etc. > Disassembling and putting back together solid state electronics does not really help, > 'if it ain't broken do not fix it; > > Anyways one machine started to produce little black spots every now and then in the picture, > It was sidelined, and several engineers had a go. > One evening I had the late night shift, not much to do, > everything worked, so I decided to have a look at that machine, > Took the scope cart, Tek, could detect strong impulses everywhere, was midnight and we went off air (in those days), > so I wrote in the book for the next shift 'seen pulses everywhere, looks like a shield is lose'. > My boss, it was his turn next morning, he read that remark and spend the whole day taking the thing apart, no luck. > Next day it was my shift, took all circuit diagrams, scope and went for it. > The pulses reminded me of some tacho signal used, > found the pulse generator and followed its output via a coax cable that send it to an other unit, > PL259 connector, was sort of lose, screwed it back on tight, problem fixed. > 10 minutes.... Went back with all the books and scope card and said: Hey guys, let's go get some coffee. > All the faces there in disbelief.. OK, I said I will show you and reproduce it,.. > So how did that cable come lose? The preventive maintenance group had forgotten to tighten that PL259 connector.
A lot of this theory depends upon the deterministic nature of systems, namely machines behave in predictable ways that we can detect if we are smart enough. But what if this is not true? What if systems have an unpredictable component that we don't know about? This has happened, for example, in jet engines, which are inspected very regularly with sophisticated instruments. This happened to United flight 232, a DC-10 which had a catastrophic engine failure (https://en.wikipedia.org/wiki/United_Airlines_Flight_232). The cause of the failure was a metallurgical anomaly in a stage 1 fan disk. Chaos theory covers this type of situation. If you are blind folded, or walking in a fog, you will be OK most of the time. But, in rare circumstances, you might walk off o, f a cliff. All of the data you could collect prior to the catastrophe will probably indicate that there is nothing to worry about. But that would be wrong, just as it was for Flight 232. Sometimes you just have to shrug and say "It's better than nothing" which it might be. While flying I am dealing with similar probabilities. Most of the time it works out fine, but sometimes it doesn't. A couple of years ago a good friend of mine, who was a very experienced pilot, died doing a routine flight maneuver that he had done tens of thousands of times before. Chaos theory caught up to him.
On a sunny day (Sat, 18 Sep 2021 23:28:04 -0700 (PDT)) it happened Flyguy
<soar2morrow@yahoo.com> wrote in
<ce283d15-f727-4535-874c-d374ae80b55fn@googlegroups.com>:

>On Saturday, September 18, 2021 at 11:23:16 AM UTC-7, Jan Panteltje wrote: >> On a sunny day (Sat, 18 Sep 2021 10:25:44 +0100) it happened Martin Brown >> ><'''newspam'''@nonad.co.uk> wrote in <si4bao$1rp4$2...@gioia.aioe.org>: >> > >> >How to do it optimally has been known for around a couple of decades. >> >The old traditional way was to try and avoid all critical in service >> >failures by replacing still working parts before they reached MTBF. >> > >> >We christened it causative maintenance since it was pot luck whether or >> >>not the damn mainframe would come back up again after a manufacturers >> >"preventative" maintenance. >> In my TV days we would run shifts and if something failed fix it ASAP >> there was also some redundancy in equipment so you could divert operations. >> >One day there was some genius who decided to reduce interruptions in >> the broadcasts to make a 'preventive maintenance' group, and to work those >engineers went,,, >> Basically with Ampex quadruplex video recorders that meant taking one apart >and putting it back together again. >> The usual things we already did on schedule, like replacing recording heads >etc. >> Disassembling and putting back together solid state electronics does not really >help, >> 'if it ain't broken do not fix it; >> >> Anyways one machine started to produce little black spots every now and then >in the picture, >> It was sidelined, and several engineers had a go. >> One evening I had the late night shift, not much to do, >> everything worked, so I decided to have a look at that machine, >> Took the scope cart, Tek, could detect strong impulses everywhere, was midnight >and we went off air (in those days), >> so I wrote in the book for the next shift 'seen pulses everywhere, looks like >a shield is lose'. >> My boss, it was his turn next morning, he read that remark and spend the whole >day taking the thing apart, no luck. >> Next day it was my shift, took all circuit diagrams, scope and went for it. >> >The pulses reminded me of some tacho signal used, >> found the pulse generator and followed its output via a coax cable that send >it to an other unit, >> PL259 connector, was sort of lose, screwed it back on tight, problem fixed. >> >10 minutes.... Went back with all the books and scope card and said: Hey guys, >let's go get some coffee. >> All the faces there in disbelief.. OK, I said I will show you and reproduce >it,.. >> So how did that cable come lose? The preventive maintenance group had forgotten >to tighten that PL259 connector. > >A lot of this theory depends upon the deterministic nature of systems, namely >machines behave in predictable ways that we can detect if we are smart enough. >But what if this is not true? What if systems have an unpredictable component >that we don't know about? This has happened, for example, in jet engines, >which are inspected very regularly with sophisticated instruments. >This happened to United flight 232, a DC-10 which had a catastrophic engine >failure (https://en.wikipedia.org/wiki/United_Airlines_Flight_232). The cause >of the failure was a metallurgical anomaly in a stage 1 fan disk. > >Chaos theory covers this type of situation. If you are blind folded, or walking >in a fog, you will be OK most of the time. But, in rare circumstances, >you might walk off o, f a cliff. All of the data you could collect prior to >the catastrophe will probably indicate that there is nothing to worry about. >But that would be wrong, just as it was for Flight 232. Sometimes you just >have to shrug and say "It's better than nothing" which it might be. > >While flying I am dealing with similar probabilities. Most of the time it works >out fine, but sometimes it doesn't. A couple of years ago a good friend >of mine, who was a very experienced pilot, died doing a routine flight maneuver >that he had done tens of thousands of times before. Chaos theory caught >up to him.
Sure, unexpected things happen, try to keep a TV station (we had 3 channels) without interrupts running. Or a studio production with many cameras recorders, audio, artists waiting, freaked out director... the show must go on... You cannot sell minutes of black . It is bad practice to - as a zombie - take things apart and put it back together for the fun of it. There _is_ of course an area for preventive maintenance, but that must be done with brains. I don't give a shit about chaos theory, what COUNTS is that you know the stuff to component level and can fix it. Planes have fallen apart due to metal fatigue... pressurization and de-pressurization stress, birds... suicidal pilots, bad programming of the computahs, pilot errors, bombs, what not. address those issues and you do something . I do not like pilots that just walk into a plane 'I am the Pilot' without walking around it and having a good look at everything. Now take my experience in the F100 Super Sabre, well no reason to repeat it here, posted about that long time ago... So many lives at stake... ;-)
On 18/09/2021 19:22, Jan Panteltje wrote:
> On a sunny day (Sat, 18 Sep 2021 10:25:44 +0100) it happened Martin Brown > <'''newspam'''@nonad.co.uk> wrote in <si4bao$1rp4$2@gioia.aioe.org>: > >> >> How to do it optimally has been known for around a couple of decades. >> The old traditional way was to try and avoid all critical in service >> failures by replacing still working parts before they reached MTBF. >> >> We christened it causative maintenance since it was pot luck whether or >> not the damn mainframe would come back up again after a manufacturers >> "preventative" maintenance. > > In my TV days we would run shifts and if something failed fix it ASAP > there was also some redundancy in equipment so you could divert operations.
The magic smoke came out of our local (10th highest structure in the UK) TV mast last month (literally right out of the top). 1M people without a TDTV signal and no convincing estimate for a repair time either. https://www.bbc.co.uk/news/uk-england-tees-58163612 https://www.arqiva.com/news-views/news/update-on-incident-at-bilsdale-mast It is very probably scrap now. A new smaller 80m temporary mast is to be erected nearby when they overcome the legal objections of the farmer. Meanwhile many of my neighbours have no TV signal (I'm on satellite). -- Regards, Martin Brown
On 19/09/2021 08:03, Jan Panteltje wrote:
> On a sunny day (Sat, 18 Sep 2021 23:28:04 -0700 (PDT)) it happened Flyguy > <soar2morrow@yahoo.com> wrote in
>>> So how did that cable come lose? The preventive maintenance group had forgotten >> to tighten that PL259 connector. >> >> A lot of this theory depends upon the deterministic nature of systems, namely >> machines behave in predictable ways that we can detect if we are smart enough. >> But what if this is not true? What if systems have an unpredictable component >> that we don't know about? This has happened, for example, in jet engines, >> which are inspected very regularly with sophisticated instruments. >> This happened to United flight 232, a DC-10 which had a catastrophic engine >> failure (https://en.wikipedia.org/wiki/United_Airlines_Flight_232). The cause >> of the failure was a metallurgical anomaly in a stage 1 fan disk. >> >> Chaos theory covers this type of situation. If you are blind folded, or walking >> in a fog, you will be OK most of the time. But, in rare circumstances, >> you might walk off o, f a cliff. All of the data you could collect prior to >> the catastrophe will probably indicate that there is nothing to worry about. >> But that would be wrong, just as it was for Flight 232. Sometimes you just >> have to shrug and say "It's better than nothing" which it might be.
In this sort of risk management it is invariably the "Unknown unknowns" that get you. Known ones have all been carefully modelled and remediated. The one remaining serious gotcha that you didn't even know was there hasn't even been considered until it actually happens IRL. Rumsfeld was widely laughed at for making a reference to this subtle distinction but he was actually spot on. -- Regards, Martin Brown