Ten past six on a Tuesday morning: the line is down on an alarm nobody has ever seen, and the only person who truly knew the plant retired last March. Rounds of phone calls, favours called in, and eventually everything points to one damaged part. The spare can still be found — but only refurbished, and in Germany. Two options: twelve weeks for delivery with the line at a standstill, or a ten-hour drive, a 50% premium and two stressed, unhappy employees. The original supplier, when asked, washed their hands of it: that version is no longer supported, and the only solution is to buy the new model or install a new line.

What does unplanned downtime really cost?

According to ABB’s “Value of Reliability” survey of more than 3,200 maintenance decision-makers, the cross-industry median is around $125,000 per hour, and two out of three companies suffer at least one unplanned outage per month (here). Siemens’ “True Cost of Downtime 2024” report estimates that for the world’s 500 largest companies, unplanned downtime is worth 11% of annual revenues (here). None of this comes out of nowhere: obsolescence sends no polite warnings — it builds up in silence, one discontinued component at a time.

The symptoms of obsolescence: what to look for

Obsolescence is a cumulative disease: what matters is not the single symptom, but how many of them pile up, and how fast. This is not a hunch or an impression: according to the same Siemens report quoted above, the average time to bring a plant back into production after a stoppage went from 49 to 81 minutes in five years (2023 vs 2019) — not because failures got worse, but because spare parts and skills got harder to find. Below, we line up the symptoms: the causes behind that number. The good news is that, if you know what to look for, a plant speaks long before it stops — you just need to keep count of how many symptoms have piled up.

The hardware – the plant’s skeleton

Spare parts getting slower and more expensive. Lead times stretching with every order, availability sliding from “in stock” to “on order” and finally to “repairable” or “refurbished only”.
Why it matters: the spares market thins out before the component dies — price and delivery time are the most honest thermometer of your machine fleet’s health.
Components declared end-of-life. PLCs, drives, I/O cards, operator panels with an official end-of-life notice from the manufacturer.
Why it matters: that notice is the only official warning you will ever receive. It is often missed or underestimated; track it and put it on a calendar instead — you will hardly be reminded a second time.
The layered electrical cabinet. Twenty years of accumulated changes: added terminals, “temporary” jumpers turned permanent, drawings never updated.
Why it matters: every undocumented change stretches the diagnosis of the next failure — the cabinet stays readable only to those who were there when it was touched.

The software – the nervous system

Line PCs running unsupported operating systems. The Windows XP or 7 PC that “works, so nobody touches it”, that nobody has rebooted in years.
Why it matters: it is not just a cybersecurity issue — above all, it is a matter of irreproducibility: if that disk dies tonight, what do you replace it with tomorrow morning? And if you fall within the NIS2 perimeter, it is a regulatory issue too: we covered it in our article “The European Maze“.
The supervisor nobody dares — or knows how — to touch. The SCADA or plant software written by a supplier that no longer exists, or frozen on “a version no longer supported”.
Why it matters: untouchable software freezes the process with it; every new production need becomes a manual workaround or an extra application layer — and workarounds and layers pile up like jumpers in the cabinet.
Missing or misaligned source code. The sources cannot be found, or do not match what actually runs on the machines.
Why it matters: without sources, every change is a rewrite, and every rewrite is a project. Aligned source code is the first spare part to keep in stock.
Device backups never tested. They exist — probably. A full restore has never been attempted by anyone.
Why it matters: an untested backup is not a strategy, it is a hope — and it gets put to the test precisely in the one moment when you cannot afford to find out it does not work.

The people – the mind and the memory

The only person who truly knows the plant is about to leave. Retiring, moving to another company — or leaving the supplier’s service department, quietly restructured.
Why it matters: undocumented knowledge is a component with no redundancy. The most expensive single point of failure is rarely inside an electrical cabinet.
The original supplier slows down. Response times stretching, “legacy” price lists rising at every renewal, and every technical conversation converging on the same conclusion: the new model, the new line, the new software.
Why it matters: when your plant drops out of the core business of whoever built it, their interest in keeping it alive drops too — and the signals are commercial long before they are technical.
The coverage that is no longer there. A service contract expired and never renewed, coverage trimmed because “nothing has ever broken anyway”.
Why it matters: being uncovered is almost always a decision nobody actually made — and you find out on the Tuesday morning this article opens with.

History snippets

The B-52 bomber was born over a weekend. On Thursday, 21 October 1948, three Boeing engineers (George Schairer, Art Carlsen and Vaughn Blumenthal) presented at Wright-Patterson, Dayton, yet another turboprop version of the new strategic bomber. Colonel Pete Warden, head of bomber development for the US Air Force, rejected it and asked for a jet design. Joined by Ed Wells, Boeing's vice president of engineering, and by two colleagues who happened to be in town on other business, the engineers shut themselves in the Hotel Van Cleve: on Monday morning they delivered a 33-page proposal for an eight-engine, swept-wing bomber, together with a balsa-wood model carved by Schairer over the weekend. "Now we have an airplane," Warden said. "This is the B-52."

The prototype flew on 15 April 1952; service entry came in June 1955. Between 1952 and 1962, 744 were built: the 76 still in service all belong to the final series, the B-52H, which left the lines between 1960 and 1962. The youngest aircraft in the fleet is over sixty years old.

In those sixty years the aircraft has changed avionics, radar and communication systems several times. The latest upgrade fits a radar derived from the F/A-18 fighters', and in September 2021 the US Air Force selected the Rolls-Royce F130 engines — a contract worth up to $2.6 billion — to replace the eight Pratt & Whitney TF33s designed in the late 1950s. The structure, meanwhile, is still the one that left the lines in the 1960s.

Current plans keep it flying into the 2050s of this century: close to a full century of service. It has already outlived the bombers designed to replace it, starting with the B-58 Hustler, retired in 1970 — and according to current plans it will outlive the B-1 and B-2 as well. The Hotel Van Cleve, on the other hand, was demolished in 1969: the hotel is gone, and the aircraft designed in its rooms is still flying.

The lesson for anyone running a plant is the same: structures live on timescales of decades, electronics and software on timescales of years. It is almost never the mechanics that call time — it is the systems ageing around them, while the structure is still sound.

The diagnosis: maintenance, retrofit or revamping?

The first question to answer is not “how old is the plant?” but “where do the symptoms concentrate?”. If you kept count in the previous section, the answer may have surprised you: almost everything clusters in the nervous system and the memory, little or nothing in the skeleton. That is no accident — it is the rule: mechanics wear out, and wear is managed with preventive and routine maintenance; electronics and software, instead, expire — they drop out of the market, and over that expiry date you have no control. It is the B-52 lesson: the structure lasts a century; it is the avionics that need updating. The diagnosis exists to establish one thing only: how much of your plant is structure that is still “young”, and how much is avionics that has reached the end of its life.

Each diagnosis has its own therapies; we group them into four families, in increasing order of depth and complexity.

Targeted extraordinary maintenance. Replace the single critical component with a current equivalent, or build a strategic stock — the “last time buy” when the manufacturer announces end of production.
When it is the right answer: isolated symptoms, or a single family of devices affected, while the rest of the plant is in good health. One limit, in the case of the strategic stock: it treats the symptom, it does not move the prognosis. Buying the last available parts postpones the replacement deadline — it does not cancel it.

Targeted retrofit. Renew the nervous system while leaving the skeleton untouched: PLC migration to current platforms, new operator panels, replacement or virtualisation of line PCs, network upgrades, a restyling of management and supervision software. The plant keeps doing exactly what it did before, but with components that have a market again — spare parts available, and people who know them.
When it is the right answer: symptoms concentrated in electronics and software, with a mechanical structure that still has years ahead. It is the intervention with the best ratio of cost to years of life gained: you work on a fraction of the plant, and extend the life of the whole.

Deep revamping. Here you are not replacing like for like: you are redesigning the way the plant thinks. A new supervisor, a new control architecture, capabilities that were not there before, integration with factory systems. In short: new logic and new functions.
When it is the right answer: symptoms across several parts of the plant, plus process needs that have changed in the meantime. One caution: when the intervention touches control or safety functions, you enter the territory of substantial modification, with precise consequences for CE marking and liability. Retrofit and revamping are not the same thing even for the lawmaker: we covered the regulatory side — the grey areas and the warning bells — in our article “The European Maze“: in those cases, the two articles should be read together.

Replacement. Let’s be honest: sometimes the supplier proposing a new line, a new warehouse or a new model is right. It happens when the skeleton itself is failing, when the required capacity is beyond what the structure can give, when the process has changed to the point that the plant answers a question that no longer exists. A serious diagnosis includes this outcome too — otherwise it is not a diagnosis, it is a sale.

How to choose, in practice

Four questions to ask, in order:

What does one hour of downtime on this line cost? The numbers in the opening are averages; the real figure, measured on your plant, is worth more.
How many years does the mechanical structure really have left?
How much market is left for the critical components — spare parts and skills included?
Will the intervention touch safety functions, i.e. reopen the regulatory perimeter?

These questions deserve measured answers, not rough guesses — and the answers, once truly measured, position the plant almost by themselves. How to measure them is a topic we will get to shortly.

One final warning, about what may look like a smart, money-saving strategy: choosing the therapy one failure at a time. In the short term it seems to save a great deal; overall, it often turns out to be the most expensive option of all. Three emergency interventions in two years frequently cost more than one planned retrofit, and they impose further hidden costs (unplanned line stops, extra recovery shifts, penalties on missed deliveries or production, in the worst cases lost customers and lost orders) — with the difference that nobody has ever priced them, lined them up and called them by their name: unplanned economic damage.

That is exactly what the diagnosis is for: it lets you decide once — calmly, with the numbers in hand — the strategy for the years to come, instead of at every stoppage, urgently, in the dark, with no way of estimating the real cost as it happens.

The therapy: why planning beats reacting

The very same intervention (same components, same working hours) costs one figure if scheduled and another if carried out as an emergency. A very similar distinction exists in medicine, and it is a sharp one: elective surgery is prepared — you choose the day, you arrive in theatre with the tests already done; the same operation in an emergency happens at night, with whatever is at hand. The difference in cost and outcome is not in the instruments, but in everything that surrounds the intervention.

For a plant, it is the same. Planning does not eliminate the stoppage — it lets you choose it. Date, duration, crew, spares already on the floor, production built up in advance, cold testing before restart. Downtime you suffer has a cost you discover afterwards, plus a hidden cost still to be quantified; downtime you choose has a cost you fix and sign beforehand. The same idle hours, bought at two different prices.

Besides, in many plants the right window already exists — it is written in the calendar. Every production has its campaigns and its pauses: the August shutdown, the season change, the tail between one campaign and the next. In food, this is even more true: production runs in campaigns, and the distance between intervening in the window and intervening mid-campaign is not measured in hours of downtime — it is measured in orders, shipments and penalties. A retrofit planned into the shutdown costs the intervention; the same retrofit forced by a failure at the peak of a campaign costs the intervention plus the campaign: lost production, missed shipments, penalties.

The Siemens report adds a detail that closely concerns consumer goods producers (FMCG, Fast-Moving Consumer Goods, the category used in the report). The FMCG sector has the lowest hourly downtime cost of all those analysed — around $36,000 per hour — and that is precisely where the trap lies. An hourly cost perceived as “bearable” makes postponing easy; and indeed, according to the same report, FMCG is the only sector where hours lost to unplanned downtime have increased since 2019, while every other sector was reducing them. The sector that pays the least for a single hour of downtime is the one that has stopped reducing those hours: the final bill is not set by the hourly price — it is set by the total.

Then there is a cost that appears on no hourly rate card, and for an SME it is the highest of all: delivery reliability. Suppliers to the big brands are measured on OTIF (On Time, In Full): deliver everything, on time. A stoppage at the wrong moment does not just burn production hours — it calls into question your very status as a supplier. For a company whose contract lives on that rating, the worst stoppage is not the longest one. It is the one that lands in the wrong week.

Which leaves the practical point: the right window cannot be improvised. To use it, you need to know in advance what to do, how long it will take, and what has to be on the floor that day. You need a photograph of the plant, taken before the decision. How that photograph is brought into focus is the subject of the next section.

The check-up: where to start

The photograph is brought into focus with an assessment. The word is overused, so let’s say straight away what it must contain to deserve it: a serious assessment walks back through the three families of symptoms (the hardware, the software, the people) — but with instruments instead of impressions.

On the hardware, you start from the inventory of critical components — make, version, lifecycle status — and dig through the end-of-life notices already issued and the residual spares market. On the software, you assess versions and support of operating systems and line applications, you locate where the source code is kept and verify that it matches what actually runs on the machines, and you check when the last full restore was tested. On the people, you establish who can put their hands on what, inside and outside the company, and how much of that knowledge is written down anywhere. This is the part of the assessment you do not do by looking at machines: you do it by involving the departments and the managers who know the plant up close (maintenance, production, IT, OT), because part of the photograph lives in their heads, not in the documentation.

The result is not a report to be filed away — it is meant to be used, through its three tools:

The photograph of the as-is, very often the first updated as-built the plant has seen in years.
The diagnosis, i.e. the position on the plant’s lifecycle continuum: what is still sound structure, what has reached end of life, and what can wait.
Finally, the plan: interventions in order of priority, anchored to the real windows of your production calendar — suggesting what to do at the next shutdown, what within one to three years, and what merely to keep under observation.

There are two essential requirements without which the check-up loses its value. The first is independence: whoever makes the diagnosis must not have the sale of a new line as their first interest. The second is that the final report stays in your hands: a complete, readable document, usable with whatever supplier you later decide to involve, with intervention priorities and the list of critical points traced out.

And this is where, for transparency, we say how we work. At noname.solutions, every take-over starts like this: an assessment of the as-is before any offer — it is the starting point of our Retrofit and Revamping interventions and the doorway to our Lifecycle Management service, because we only take on systems we know. First the photograph, then the decisions.

Conclusions

The title of this article was a promise: your plant still has years ahead. Now it can be substantiated — and they are almost always more years than the supplier of the new line is willing to grant it. The mechanics, like the B-52’s structure, is rarely the problem: what ages is the electronics and the software working around it — and that is a disease that can be diagnosed and treated. Provided you listen to the symptoms when the plant shows them, not when it screams them: after that, only the most expensive therapy is left — the one decided in an emergency, on the wrong Tuesday morning.

Three questions to ask yourself tomorrow morning

The source code of the latest version running on the machines: do you know where it is?
The last full restore from backup: when was it tested?
If the line stops tonight: who answers, and how fast?

If even one answer made you hesitate, your plant is already talking to you.

“Another flaw in the human character is that everybody wants to build and nobody wants to do maintenance.”
— from “Hocus Pocus” (1990) by Kurt Vonnegut

👉🏻 Tell us about your plant: let’s assess the starting point together.

Your plant still has years ahead: the warning signs that tell you when it’s time for a revamp

Table of contents