Maintainability only shows itself after something has already gone wrong, which makes it the easiest of the five properties to specify carelessly and the hardest to recover once the hardware exists. This chapter takes eight systems, one per industry, and works the maintainability question on each: when the thing fails, what has to happen before it works again, which technique measures that honestly, and where the clock actually goes.
The eight systems are the same in every concept topic of this row. The aircraft electrical system worked here for its repair times is the one worked in reliability for its failure rates, in availability for its dispatch, in safety for its catastrophic failure condition and in testability for its built-in test coverage. Read a column and you learn one property across eight industries; read a row and you watch one machine reveal its five RAMS faces. Every page hands you sideways to the same machine's next property. All numbers are illustrative, invented for teaching and chosen to make the arithmetic legible; none is field data from a real product.
The technique is a choice, not a default
The most useful thing this chapter does is refuse to apply one model everywhere. A mean repair time is the right summary almost nowhere: a hard window wants a percentile, a bimodal population wants two numbers, a shared workshop wants a queue, and a hot-swappable array wants no repair-time term at all. Each page names its technique, justifies it, substitutes the numbers, then says what a different technique would have given and why it was rejected.
| Industry | System | Technique this system calls for |
|---|---|---|
| Defence and aerospace | Transport aircraft AC electrical power | Elemental task-time build-up, rolled up by failure rate, with an ambiguity mixture |
| Railway | Wayside level-crossing controller | Task build-up in an additive downtime model, the repair time scored as a percentile against the possession |
| Space systems | Satellite attitude control | A three-stage restoration timeline, no repair at all, and a maintainability function that never reaches one |
| Automotive | Electric vehicle brake system | Lognormal fitted to dealer work-order data, converted into bay-scheduling percentiles |
| Energy and resources | Gas compressor train | A two-component mixture, because the weighted roll-up describes no job on the train |
| Medical devices | Infusion pump fleet | A queueing model of the single biomedical department, with Little's law on the full loop |
| Electronics and high tech | Storage array controller pair | Detection and dispatch, because hot swap removed repair from the availability equation |
| Nuclear | Emergency diesel generator train | A lognormal inverted against a regulatory percentile, with dose as a second currency |
What the eight rows have in common
Read down the column and one finding repeats everywhere: the active repair time was the smallest term in the answer. The aircraft's 1.5-hour removal sits inside a 27.5-hour outstation downtime. The crossing's 2.5-hour repair sits in a possession cycle whose true mean, once window overruns are priced, is 13.1 hours rather than the 9.5 quoted. The satellite has no repair and spends four of its seven restoration hours deciding. The vehicle's 1.2 hours is 3.6% of a 33.6-hour off-road event, the pump's bench time, queue included, is 8% of a loop measured in days, the array's repair time is absent from its availability arithmetic entirely, and the diesel's work is bounded by a ceiling and priced in dose.
The second finding is that the mean was almost never the right summary. Six of the eight pages turned on a percentile, a mixture or a queue: 36% of nuclear jobs would breach a licence limit under a 72-hour MTTR, 15% of crossing repairs miss their window, a 92.6-hour compressor mean describes an event that never happens, and a normal fit to 41,000 brake work orders assigns a tenth of its probability to repairs finishing before they start.
The third is that the decisive lever differed every time, which is why eight rows exist rather than one. Failure-rate weighting named the aircraft's offender; access windows dominated the crossing; decision quality governed the satellite; variance and diagnostic resolution drove the vehicle's bay plan; calendar placement was the compressor's whole answer; arrival rate and float carried the pump fleet; swap architecture deleted the array's repair term; and an unstated percentile hid a factor of four in the diesel's target. None is derivable from a repair-time table.
That is the argument for reading the other four columns. A repair time correct in isolation can still describe a property nobody experiences, and what the operator feels is assembled from the rates in reliability, the downtime accounting in availability, the consequence classes in safety and the coverage in testability.