The Expertise Exodus
In the next five years, half of industrial Britain will retire. The knowledge walking out the door was never written down.
Welcome back to The Ledger - a weekly briefing of what’s happening inside complex systems around industrial decarbonisation. Let’s get straight into Issue #11.
THE OPENING ENTRY
A chemicals complex in North West England. The control room shows 47 alarms. The night operator ignores 46 of them.
“That one,” he points to a temperature reading, “that’s the only real problem.”
He’s right. I ask how he knows.
“Twenty-eight years of Thursdays.”
He retires in March. His replacement starts Monday. The training manual is 400 pages long and contains nothing about those 47 alarms.
Across industrial Britain, the same exodus is underway. The people who actually know how things work are leaving. And they’re taking something that was never documented: the difference between how systems should run and how they actually run.
That gap? It’s worth about 30% of your energy consumption.
FIELD REPORTS
1. The Whisperer
Heavy industry plant, Scotland. Brian can hear when the turbine’s running rich. Not from instrumentation. From the sound it makes three floors down.
“Bit throaty today,” he mentions, walking past. Adjusts the fuel trim. Saves 200kg of natural gas per hour.
The control system says everything’s within parameters. Brian’s ears say different. He’s been right every time for two decades.
Brian retires next year. They’re replacing him with someone who trusts the instruments.
Annual cost of losing Brian’s ears: £280,000
2. The Weather Prophet
Food processing facility, Midlands. Maggie starts the boilers two hours early when she “feels the cold coming.”
No weather app. No forecast. Just something about the pressure.
She’s been right 94% of the time. IT installed a £50,000 predictive analytics system to replace her intuition. It’s been right 71% of the time.
Maggie retires in eighteen months. The system that doesn’t work as well as Maggie costs £8,000 annually to license.
3. The Sequence Keeper
Pharmaceutical plant, North West. Derek knows the undocumented startup sequence. Not the one in the manual. The one that actually works without tripping the ammonia detection system.
“You bring up Unit 3 first, but only to 40%. Then Unit 1 to full. Wait seventeen minutes - not fifteen, not twenty - then Unit 3 to full, then Unit 2. Do it any other way, you’ll trip the alarms and lose four hours.”
This isn’t written anywhere. Derek worked it out in 2001. He’s taught three people. Two have already retired.
Every startup without Derek costs an extra 4.7 tonnes of steam. At 300 startups per year, that’s 1,410 tonnes of steam that exist because we never wrote down Derek’s sequence.
4. The Valve Genealogist
Chemical plant, Yorkshire. Terry knows which valves leak, which stick, which lie about their position.
“That one says it’s closed but it’s always 5% open. That one needs a tap with a spanner at 3 o’clock. Those two were swapped in 1998 but the drawings were never updated.”
The digital twin they’re building doesn’t know any of this. It assumes the engineering drawings tell the truth. Terry knows the official blueprints are fiction - the real plant diverged from the drawings twenty years ago.
When Terry leaves, they’ll spend six months wondering why the mass balance never adds up. The consultants will charge £400,000 to rediscover what Terry knew for free.
5. The Load Predictor
Decades old mill, Wales. Graham knows what production will really need tomorrow. Not what the schedule says - what will actually happen.
“Hywel’s order will be late, so we’ll run Aled’s first. That means different alloy, so preheat to 1,180 not 1,220. Save ninety minutes of gas.”
He’s reading between the lines of emails, knowing which customers always delay, which production manager over-promises, which supplier ships short.
The AI scheduling system they bought doesn’t read emotions. Graham does.
Graham’s intuition saves 15% on energy costs. The AI system that will replace him costs £2.4 million.
WHAT THE LEDGER REVEALS
Every industrial site runs on two operating systems:
The Documented System: Procedures, manuals, training guides, digital twins
The Tribal System: What Maggie knows, what Derek discovered, what Brian hears
The tribal system is 30-40% more efficient. And it’s about to disappear.
We’ve spent decades building Industry 4.0 digital systems to capture 5% efficiency gains. Meanwhile, we’re about to lose 30% efficiency as the people who actually know how things work walk out the door.
The bitter irony? We’re so busy installing sensors that we forgot to interview the sensors that are retiring.
THE KNOWLEDGE THAT’S LEAVING
It’s not just operational knowledge. It’s the archaeology of bodges:
Why that heat exchanger runs backwards (because it works better)
Which instruments lie (most of them)
Where the real constraints are (never where design says)
What the alarms actually mean (rarely what they say)
How to read the plant’s moods (everything has a personality)
This knowledge emerged over decades. It can’t be documented in a rush. It won’t fit in a database. It lives in stories, in muscle memory, in the corner of someone’s eye.
THE UNIVERSAL TRUTH
This isn’t just about factories. It’s every organisation losing the people who know:
Which rules to break
Which processes to ignore
Which metrics are meaningless
Which shortcuts work
Which official fixes made things worse
The efficiency isn’t in the procedures. It’s in knowing when to ignore them.
THE OPERATOR’S PLAYBOOK
Map your dependency risk. List every person over 55. Write down what only they know. Be terrified by the result.
Create shadow shifts. Pair every Brian with someone for six months. Not training - just shadowing. Knowledge transfers through presence, not PowerPoints.
Document the undocumented. Not procedures - stories. “Remember when the compressor failed and Terry fixed it with a hammer?” That story contains more knowledge than any manual.
Value intuition in your digital systems. If Maggie says it’ll snow, override the algorithm. Build systems that can learn from humans, not replace them.
Pay for knowledge transfer. Offer retiring experts consulting rates to stay three days a month. Cheaper than rediscovering what they knew.
THE LEDGER LINE
We’re about to lose more industrial efficiency through retirement than we’ll gain through digitalisation.
The race isn’t to install AI before the old systems fail. It’s to download wisdom before the wise leave.
Because when a Maggie, a Terry or a Brian retires, you don’t lose an employee. You lose a library that was never written.
Thanks for reading, and please share The Ledger if you have found it useful or insightful in any way.
Here’s to what’s possible
Dom