The Hidden Cost of Knowledge Loss: When Your PLC Expert Retires

When your S5 expert retires, they take 20–30 years of undocumented knowledge with them: why certain timer values were chosen, which interlocks were added after near-misses, what must never be changed. This knowledge cannot be recovered from the code alone. It disappears permanently. The cost is not visible on the day they leave — it shows up months later as longer downtime, more scrap, and eventually a failure that no one can diagnose.

The Knowledge That Lives Only in One Person's Head#

PLC programs contain two layers of information:

Layer 1: What the code does. This can be reconstructed by reading the program — with effort. An experienced programmer can trace the logic, identify inputs and outputs, and understand the sequence. Tools like PLCcheck Pro can accelerate this significantly.

Layer 2: Why the code does it. This cannot be reconstructed from the code. It exists only in the memory of the person who wrote it or maintained it over the years:

• "Timer T5 is set to 4.5 seconds because at 4.0 seconds, the hydraulic cylinder sometimes doesn't fully retract in winter when the oil is cold."
• "The interlock on M 50.3 was added after the incident in 2009 when the conveyor started while someone was inside the machine."
• "DB10 is never used at runtime. It contains calibration values from the original manufacturer. If you delete it, the analog scaling breaks."
• "The sequence in OB1 must run in exactly this order. If you move Network 12 before Network 8, the dosing valve opens before the mixer is running."
• "E 4.7 is connected to the emergency stop, but it's active-low because the original wiring was done wrong in 1997 and nobody wanted to re-wire during production."

Every plant has dozens of these invisible facts. They are the difference between "the machine runs" and "the machine runs correctly." None of them appear in any printout, any cross-reference, any backup.

What Actually Happens After They Leave#

Month 1–3: Nothing Visible#

The machine keeps running. The program has not changed. Everyone assumes the situation is under control.

Month 4–8: Small Problems Accumulate#

A sensor fails and the machine stops. The replacement technician restarts it, but production quality drops slightly. Nobody connects the two events because nobody knows that the failed sensor was part of a quality-monitoring circuit that the retired expert had added manually.

A timer runs out during a cold start. The technician increases it "to be safe." It works, but now the cycle time is 6% slower. Nobody notices because nobody knows what the original value was or why.

Month 9–18: A Real Failure#

A production change requires a program modification. The new programmer opens the S5 code and sees undocumented AWL with cryptic addresses. They make the change, test the basic function, and put the machine back into production.

Three weeks later, a batch fails quality inspection. The root cause: the program modification accidentally bypassed a check that compared the actual dosing amount against the setpoint. The check was in a subroutine that the programmer did not know existed, because it was called conditionally through a jump instruction that only activates when a specific product recipe is selected.

The investigation takes 2 weeks. The scrap costs €40,000. The fix takes 2 hours once someone understands the code.

Year 2+: Strategic Paralysis#

The plant avoids making any changes to the machine. When production requirements change, workarounds are implemented outside the PLC — manual switches, additional sensors wired to separate relays, Excel-based tracking instead of automated counting. The machine becomes increasingly disconnected from the rest of the production system.

Eventually, management decides the machine "must be replaced" — not because the hardware is worn out, but because nobody dares to touch the software anymore.

Calculating the Cost#

The direct cost of knowledge loss is difficult to measure because it shows up as other problems. But we can estimate:

Conservative estimate for a medium-sized production machine: €30,000–100,000 over 3 years after the expert leaves. Most of this cost is invisible because it is booked under "maintenance," "quality," or "downtime" — not "knowledge loss."

What You Can Do Before They Leave#

Start Now: The Knowledge Extraction Process#

Step 1: Identify critical systems. Which machines are controlled by PLCs that only one person understands? Rank them by production impact.

Step 2: Schedule walk-throughs. Book 2–4 hours per machine with the expert. Not at their desk — at the machine. Walk through the process physically. Record the session (audio is enough).

Step 3: Ask "why" questions, not "what" questions. "What does Network 15 do?" gives you a code description. "Why is Network 15 there? What happens without it?" gives you the knowledge that matters.

Step 4: Document the undocumented rules. Every machine has rules that exist only in people's heads:

• What to check after a cold start
• What to never change
• What to do when a specific alarm appears
• Which spare parts are critical and where to find them
• Which quirks the machine has and how to work around them

Step 5: Create program documentation. At minimum: a symbol table with meaningful names, network comments, and an I/O assignment list. See our documentation guide for the complete process.

Step 6: Have the successor work alongside the expert. Not for a one-day handover. For weeks or months, ideally, sharing the on-call rotation and handling maintenance together. Knowledge transfer happens through shared experience, not through documents alone.

Use PLCcheck Pro for Accelerated Knowledge Capture#

PLCcheck Pro can extract Layer 1 knowledge (what the code does) automatically:

• Upload the S5/S7 program and get plain-language explanations
• Every block documented with inputs, outputs, and purpose
• Timer and counter values identified and explained
• Cross-references generated automatically

This frees the expert's time for Layer 2 knowledge — the "why" that only they know. Instead of spending hours explaining what each network does, the expert can focus on explaining the context, the history, and the hidden requirements.

Try PLCcheck Pro →

The Demographic Reality#

This is not a problem that will solve itself. In Germany, the average age of experienced automation technicians is rising steadily. Engineers who learned S5 programming in the 1980s and 1990s are reaching retirement age now. STEP 5 ran on DOS — the last generation that used it professionally is leaving the workforce.

Meanwhile, the machines they programmed still run. Some of them will run for another 10–20 years — if someone can maintain the software.

The window for knowledge transfer is closing. Every month you wait, the available expertise shrinks. The time to act is before the farewell party, not after.

Frequently Asked Questions#

How long does it take to capture one expert's knowledge?

Plan 2–4 hours per machine for a structured walk-through. For a plant with 10 PLC-controlled machines, budget 3–5 full days of the expert's time. This does not capture everything, but it captures the critical 20% of knowledge that prevents 80% of post-retirement problems.

What if the expert has already left?

Focus on documenting the code itself first. PLCcheck Pro can explain the logic. Then interview operators and maintenance technicians — they know the machine's behavior even if they do not understand the code. Reconstruct the "why" from operational experience. It is slower and incomplete, but better than nothing.

Should I hire the retired expert as a consultant?

If possible, yes — but with a clear scope and timeline. A consulting agreement for 10–20 days over 6 months gives you access to their knowledge during the critical transition period. Pay well — their knowledge is worth far more than their daily rate suggests.

---

Maintained by PLCcheck.ai. Last update: March 2026. Not affiliated with Siemens AG.