Client at a steelworks in northern Slovakia, March 2025. Old flat rolling line for stainless steel, in operation since 1987, definitively shut down. Our task: full decommissioning with reuse of ~35% of components (motors, gearboxes, electrical) as spare parts for a sister line in Hungary. Planned duration: **28 calendar days.** Real duration: **42 days.** This is a breakdown of why, including everything we would do differently today.
Tagging — Class 1 Gen 2 RFID vs hand-written labels
The first decision was how to track the 2,800+ components going into the spare-parts warehouse. Hand labels (waterproof polyester, A6, laser-printed) would have cost ~€0.40 each, but warehouse tracking would have stayed manual — operator reads the number, clicks in the WMS. With 2,800 parts and a typical 2–3% write-off error rate, that means 60–80 parts logged badly that nobody finds six months later.
Choice: **EPC Class 1 Gen 2 UHF RFID tags** (Impinj Monza R6-P chip), polyamide + epoxy encapsulation for thermal tolerance up to 200 °C.
What works, what doesn't
- **Read range on a bare part** (motor, gearbox): 4–6 m with an Impinj Speedway R420 reader + circular polarised antenna. Fine.
- **Read range through a metal cover**: ~0 m. Metal absorbs RF energy entirely. In practice that meant tags MUST be off metal surfaces — either on a sticker with a **foam spacer** (3M VHB 4926 or TagSurance UHF on-metal pad, 5 mm thick) or entirely off-metal on a hanging plastic label.
- **Read range through oil and grease**: almost unchanged. Oil is RF-transparent.
- **Read range through water**: catastrophic. Wet components after washing had to dry 2–4 hours before tag reading. That added a logistics layer we hadn't planned for.
Tagging cost (real)
- RFID tags (on-metal, IP67): **€1.10–1.80 each** × 2,800 = **€3,500.**
- Stationary reader (Impinj Speedway + 2× antenna): **€2,200.** Handheld reader (Zebra MC3300xR gun): **€2,400.**
- Software integration into the SAP MM module (custom ABAP wrapper around the RFID API): **€8–12k engineering.**
- Tag application on a component (cleaning + bonding + check): **3–7 min/piece** × 2,800 = ~250 hours × €25 = **€6,250.**
**Total RFID infrastructure: ~€22–25k.**
With hand labels it would have been ~€3–5k. But RFID paid back inside 8 months thanks to faster warehouse stocktakes — instead of 3 days × 4 people of manual count, now 4 hours × 2 people with a walkthrough scan.
Photo documentation — the protocol nobody writes
The second stumbling block: photo documentation. The client originally wanted "a few photos for the record." We arrived with a protocol — and still expanded it twice during the project.
Final per-component protocol
1. **Overall view before disassembly** from 4 cardinal directions + top view. **5 photos minimum.** 2. **Detail of the nameplate** (Schild) with visible serial number, legible. If the plate is corroded/illegible, write the data next to it in chalk + take another photo. 3. **Detail of the mounting interface** before disconnection (terminal block, drive shaft, hose, anchoring). At least 2–3 photos depending on interface complexity. 4. **Photo before lifting** with visible lifting points and crane attachment. 5. **Photo after disassembly** of the component on a clean surface with a scale object in frame (1 m aluminium rod or folding measure in shot). 6. **GPS metadata** in EXIF on every photo — automatic via Garmin GLO 2 paired with iPhone/Android. At the sister-line install in Hungary that metadata saved 3–4 days of "where did this sit in the old hall."
**Minimum 8 photos per component.** 2,800 components × 8 = ~22,400 photos. That's 3–4 hours a day for a dedicated photo documenter who does nothing else. We actually had 2 people full-time on documentation.
Software workflow
- Capture: **Adobe Lightroom Mobile** + custom preset (no RAW, JPEG fine, GPS on, watermark with tag ID).
- Sync: cloud (Lightroom Mobile → our own S3 bucket via Lambda hook).
- Tagging: after upload each photo is matched to the RFID tag ID via a QR code in frame (on a flexible board the photo documenter holds during the shot). **A QR code in every frame is non-negotiable.** Without it you lose 30–40% of matching.
- Sorting and archive: into a **Synology DSM Photos** server, with metadata in Postgres.
Real cost of the photo documentation bundle for the project: **€18–22k** (2 people × 42 days + software + storage). In the first deck the client budgeted €3–5k. That's the difference that hit him — but six months later, when the sister setup started in Hungary, photo documentation saved an estimated **€80–120k** in engineering hours of reconstruction from the unknown.
14 days of overrun — anatomy
The plan was 28 days. The reality was 42. Here's the breakdown of where the days went:
Days 1–3: Discovery and surprises (+2 days)
Planned scope: 280 hours of the team mapping the real state. Surprises:
- **Machine 4 (straightening stand)** was not in the drawings but stood on the line. Nobody remembered when or why it had been added (~1996 by weld style). Documentation lost. **+8 hours** for extra survey + photos.
- **Cable runs in the floor.** The client said "we have drawings." They did. From 1987. 60% of cables in floor ducts didn't actually run as the drawings showed. **+24 hours** for re-mapping via metallic cable locator and GPR (Ground Penetrating Radar) — €4,200 extra to rent GPR.
Days 4–7: Safety isolation (on plan)
LOTO procedures, decommissioning, isolation of media (steam, water, air, hydraulics), draining of hydraulic circuits (4,500 l mineral oil → certified disposal, **€8,500**), pneumatic depressurisation. Nothing was lost here.
Days 8–18: Mechanical disassembly (+4 days)
Plan: 11 days. Real: 15.
- **Seized bolts.** After 38 years of operation, the M36 anchor bolts of the rolling stand have a layer of corrosion + thermal seizure. Manual pneumatic pullers don't cope. We needed **Enerpac hydraulic bolt tensioners (up to 90 kNm torque)** + an induction heater to 250 °C to create thermal expansion. Set rental from Hydratec: **€6,200/week**, used 2 weeks. **+€12,400 extra.**
- **Asbestos gaskets.** In the steam circuit we found the original 1987 gaskets — Klingerit Sil C-4400 with asbestos filler. As if cut from a textbook chapter "why decommissioning old industrial lines never has a plan A." **Stop work, mandatory notification to ÚVZ SR**, waiting for approval of the removal procedure with a certified firm (Saneko, EMS Slovakia). **+5 days of pure delay**, **€18–22k** for certified removal of 14 kg of asbestos-containing material. In reality the most expensive 5 days of the whole project.
Days 19–26: Electrical disassembly (+3 days)
Plan: 8 days. Real: 11. Main reasons:
- **Cables without labels.** The cable rack in the main switchboard had original 1987 labels cracked and mostly illegible. Before disconnecting every feeder we had to do **manual tracing via Fluke 2042 cable locator** or via an inductive tester. **+18 hours** for a pair of electricians.
- **One cut cable.** Classic. On day 22 one of the technicians, dismantling a cable tray, cut the main control cable 12×1.5 mm² between the PLC cabinet and the hydraulic power unit **before** the power unit isolation was complete. The cable was part of the emergency STOP circuit for downstream hydraulic equipment, which still had a pressurised accumulator. **No injury**, but an OHS incident → 4-hour stop, documentation, root-cause analysis. From that came the protocol for all future projects: **every cable is first tested with a signalogram (tone generator + audio probe) BEFORE cutting**, even if labelling shows it isolated.
Days 27–30: Haulage and sorting (on plan)
No surprises. Iron to scrap (~410 tonnes × €170/t = **€69,700** revenue), non-ferrous metals (~28 tonnes × €1,800/t = **€50,400**), electronics (separately, ~€4,200), hydraulic oil (assessed as contaminated, **€8,500** disposal, see above).
Days 31–42: Re-export to Hungary + final documentation (+5 days)
- Planned time: 4 days for loading and transport. Real: 6 days, due to the oversized load (rolling stand 12.8 t, width 4.2 m) → need for a **special transport permit** via the National Motorway Company, lead time 3–4 working days. **+2 days.**
- Final documentation (RFID database export, photo library, BOM, material certificates) — planned for 3 days, real 6. **+3 days.** Reason: during disassembly we added ~280 ad-hoc photos and 60 hand sketches of "surprises" that had to be archived and indexed.
Final cost analysis
| Item | Plan | Real | |------|------|------| | Labour (12 people × days × €350 / person / day average with bonuses) | €117,600 | **€176,400** (+50%) | | Enerpac hydraulic tensioners | €0 | **€12,400** | | Asbestos removal | €0 | **€20,000** | | Hydraulic oil disposal | €5,000 | **€8,500** | | GPR + cable locator rental | €1,200 | **€5,400** | | RFID infrastructure | €18,000 | **€23,000** | | Photo documentation | €4,000 | **€20,000** | | Special transport | €18,000 | **€26,000** | | Accommodation for the team (northern SK steelworks, local guesthouses) | €15,000 | **€22,000** (+14 days) | | **Total cost** | **€178,800** | **€313,700** (+75%) | | Revenue from scrap and non-ferrous | €115,000 | **€124,300** | | **Net cost** | **€63,800** | **€189,400** |
The client budgeted net cost ~€60k. Real: €189k. **3.15× over budget.**
What we would do differently today
If you're asking whether the project ended in a loss — it didn't. The client paid under the fixed contract + change orders, and the documentation saved the sister line in Hungary ~€100–150k. But if we were doing a similar project today:
1. **5-day discovery phase instead of 3.** Including GPR scanning from day 1. Pre-project asbestos screening (Z3 sampling, accredited lab ~€800–1,200) — MANDATORY for anything pre-2000. 2. **+30% schedule buffer** off plan for lines > 25 years old. Assume 2–3 undocumented modifications. 3. **RFID protocol including "on-metal" tags priced in** from the start, not as an add-on. 4. **Photo documentation as a separate line item** with a dedicated team of 2 people full-time, not as a side job for fitters. 5. **Cable tracing BEFORE every cut** — without exception, even if labelling shows isolated. 6. **Special transport permits** requested on day 1, not day 28.
The hidden profit of decommissioning
The least documented benefit of decommissioning is not the scrap revenue. It's **diagnostics on sister lines.** The client had a line in Hungary from the same era, still in operation. Our photo documentation + RFID inventory showed 3 components (one main motor, two planetary reducers) that were in such a state on the Slovak line that they indicated imminent failure on the Hungarian line too. Preventive replacement was done — saved an estimated **€280–340k** in unplanned outage. That's not in the deck. But it's the reason serious firms commission decommissioning as part of a predictive-maintenance roadmap, not just as a disposal process.
---
*We write this as a technical partner that has decommissioned 40+ industrial lines over the last 12 years. If you have a decommissioning ahead of you with a budget < €100k, the first consultation (90 minutes) walks through these cost categories that usually don't show in the first quote and gives you a realistic estimate — including asbestos risk assessment and discovery scope.*