Machine History & Reliability Record

What this tool does

CM8-279 keeps the life history of a machine. One line per intervention, from the day it was commissioned to the day it is sold or scrapped: every breakdown, every planned repair, every service, inspection, modification, calibration and overhaul, with the hours it cost and the money it cost.

A card like this exists to answer three questions that nobody can answer from memory:

• Is this machine getting worse? Not "does it feel unreliable" — is the interval between failures shortening? The breakdowns-by-month chart and the MTBF figure answer that from the record rather than from the last bad week.
• What keeps failing? Not what failed last, but what has failed more than once. The failure-mode chart is the whole point of the exercise: a fault that recurs is a fault with a cause you have not yet removed.
• Should we still be repairing it? Lifetime maintenance spend, set against what a replacement costs and how much life is left, turns an argument into a comparison.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so your asset history, contractor costs and reliability figures stay on the machine you are using.

Recording discipline — including the five-minute fixes

A history card is only worth what people put into it, and the failure mode of every maintenance record ever kept is the same: the big jobs get written up and the small ones do not. Somebody clears a sensor fault in four minutes on the night shift and says nothing, because it was nothing. Six months later the same sensor has been cleared eleven times, nobody knows that, and it is still fitted.

The small entries carry the pattern. A twenty-thousand rebuild tells you what happened once. Eleven four-minute resets tell you what is wrong. Log the reset. It takes longer to decide whether to log it than to log it.

Practical rules that keep a card usable:

• One row per visit to the machine, not per day and not per work order. If two separate faults were fixed on the same shift, that is two rows with two failure modes.
• Write the description as symptom, finding, action. "Spindle stopped on a temperature alarm; front bearing found seized and blued; bearing set and seal replaced, spindle rebuilt and re-tramed." Someone three years from now who has never seen the machine can use that. "Fixed spindle" helps nobody, including the person who wrote it.
• Separate downtime hours from repair hours. Downtime is production time lost. Repair hours are hands-on labour. A service done on a Saturday is four repair hours and zero downtime; a breakdown that waited two shifts for a part may be sixteen downtime hours against three repair hours. Adding the two together produces a number that means nothing, so the tool keeps them apart everywhere.
• Record the cost even when it is internal labour at an assumed rate. A blank cost column reads as free.
• Keep the machine name spelled the same way. "CNC-1", "CNC1" and "cnc 1" are three machines to any grouping, here or in a spreadsheet.

Failure modes as a controlled vocabulary

The failure-mode field is a small free-text box, and it is the one field where free text will quietly ruin your analysis. "Bearing seizure", "seized bearing", "bearing failed", "spindle bearing" and "brg seized" are one fault written five ways, and a frequency chart will show five faults that happened once each instead of one fault that happened five times — which is precisely the signal you were looking for. Agree a short list of terms for your equipment before anyone starts typing, keep it somewhere the technicians can see it, and write the mode the same way every single time. Twelve to twenty terms is usually enough for a whole shop: bearing seizure, seal failure, hose failure, control board failure, sensor fault, motor winding failure, drive belt failure, hydraulic leak, pneumatic leak, corrosion, wear beyond tolerance, blockage, software or parameter fault, operator damage, and a plain "not yet established" for the ones you genuinely have not diagnosed. Name the component and how it failed, not the symptom the operator saw — "machine stopped" is not a failure mode. The tool matches modes case-insensitively and ignores surrounding spaces, which forgives a capital letter but nothing else.

Condition monitoring — vibration readings, oil analysis, thermal images, current signatures — belongs in this card as inspection rows, because a trend of readings taken before a failure is the only thing that ever turns a history card into a warning.

MTBF and MTTR, honestly

Two figures get quoted about reliability more than any others, and both are simple enough to state in full:

MTBF — mean time between failures — is shown in days for the featured machine, and only when that machine has at least two recorded breakdowns. With one breakdown there is no interval to measure, so the tool prints an em dash rather than inventing one. Note what the formula does and does not use: it measures calendar days from first failure to last, not running hours, and it takes no account of how hard the machine was worked between them. A machine that sat idle for four months will flatter its own MTBF.

MTTR — mean time to repair — is the average hands-on repair hours per breakdown, taken across every machine in the current filter so that the average has as many events behind it as possible. It deliberately excludes waiting time for parts and for a free technician, because those are supply and staffing problems rather than repair problems. If you want the full stop-to-start figure, read the downtime hours instead; the gap between the two is your logistics.

The reliability summary shows MTBF and MTTR per machine so that you can see how thin the underlying counts are. Read the breakdown column first and the mean second. If the count is one or two, the mean is decoration.

Planned versus unplanned spend

The donut splits maintenance money by type of intervention, and the ratio it shows is the single best indicator of maintenance maturity available from a record this simple. Unplanned spend is the breakdown slice: money spent because a machine stopped without asking. Everything else — preventive maintenance, inspection, planned corrective repair, calibration, overhaul — is money you chose to spend at a time you chose.

Organisations that have their maintenance under control spend most of their maintenance money on planned work. Those still fighting fires spend most of it on breakdowns, and pay a premium for the privilege: emergency callouts, expedited parts, overtime, and repairs done under pressure that do not last. The unplanned share column in the reliability summary gives you the ratio per machine.

Do not chase the ratio for its own sake. Two things can move it without anything improving: doing more preventive work than the equipment needs, and quietly reclassifying breakdowns as planned repairs. The question worth asking each quarter is whether the unplanned share is falling while the breakdown count is falling. One without the other means the money moved, not the reliability.

Repair or replace

This is the conversation the card enables, and the honest answer is that no formula settles it. Enter a replacement cost on the Settings tab and the life card foot expresses lifetime maintenance spend as a share of it — a machine that has consumed half its replacement value in repairs is worth a serious look, particularly if most of that went in the last two years. But the share is a prompt, not a decision. Set it beside the things this tool cannot know: how many years of useful life are realistically left, what the downtime actually costs you in lost orders rather than in labour, whether the failures are ageing or a fixable application problem, whether a new machine would be faster or need fewer people, what the old one would fetch, how long a replacement takes to deliver and commission, and whether the capital exists at all. A machine with a rising repair bill and a fixable recurring failure is usually worth fixing properly. A machine with a rising repair bill and five different failure modes is usually wearing out. The card tells you which of those you have.

Where this sits alongside your other records

Three records overlap here and each answers a different question. The Machine Downtime & Loss Log costs production loss — stops, causes, minutes lost, units not made and what that was worth — and is the right place for the money that lost output represents, which this card deliberately excludes. The Preventive Maintenance Schedule looks forward: what is due, on what interval, by whom, and what is overdue. This card looks backward across the whole life of the asset and is the only one of the three that spans commissioning to disposal.

They feed each other. A recurring failure mode found here should produce a new task on the preventive schedule; a big stop costed in the downtime log should appear here as a breakdown row with its failure mode. Keeping all three does not mean typing everything three times — it means each record holds the field the others do not need.

FAQ

Should preventive maintenance be logged here as well as on the schedule? Yes. Without the planned work in the history you cannot see the planned-versus-unplanned ratio, and you cannot tell whether a run of good reliability followed a service or happened despite one.

What counts as a breakdown rather than a repair? A breakdown stopped production when you did not intend it to. A repair is corrective work you scheduled — the fault was known, the machine was taken out at a time that suited. The distinction drives every unplanned figure in the tool, so make it consistently. If a fault was found on a Friday and fixed on the Monday shutdown, that is a planned repair.

Why is MTBF in days rather than running hours? Because most workshops can reconstruct dates from records and cannot reconstruct running hours. If you do have reliable hour meters, hours are the better basis and you should calculate MTBF that way separately — say which basis you used whenever you quote the figure.

What if a machine is moved, rebuilt or given a new number? Keep the history under one name and note the change in the description of a modification row. Splitting the record at a rename destroys the only thing the card is for.

How far back should I go when starting? Start from today and add what you can reconstruct from work orders, invoices and the store's parts issues. Do not wait for a complete history; a card that starts thin and is kept honestly is more useful within a year than a perfect reconstruction that nobody maintains.

Should contractor invoices go in the cost fields? Yes — put the parts on the invoice into parts cost and the labour into labour cost, and tick the external contractor box. Contractor work is often the largest line in a machine's life and leaving it out makes internal machines look expensive by comparison.

Saving your work

Interventions, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. A machine history is a long-lived record, so keep a dated export somewhere backed up rather than trusting a browser for a decade. Export CSV gives you the register for spreadsheet work and includes every filtered record, not only those drawn on screen. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

The arithmetic here is deliberately plain and the tool does it faithfully. Everything that matters sits underneath it: whether every intervention was recorded, whether the failure modes were written consistently, whether downtime and repair hours were separated properly, and whether internal labour was costed at all. A card missing the small jobs will show good reliability and a low cost, and both will be wrong.

Reliability figures built from a handful of events are indicative, not statistical, and a mean drawn from two or three intervals should never be presented as a prediction. The repeating pattern in the failure modes is almost always worth more than either mean. Costs are the direct parts and labour you enter and exclude the value of lost production. This is a record-keeping and analysis aid, not engineering advice, not a maintenance strategy, and not a basis on its own for a capital decision.