Every tenant in the city was operating. None of them had agreed to stop. The project brief said “utilities corridor”; the actual job was fourteen kilometres of negotiation with a trench attached.
The record drawings were fiction
Thirty-year-old as-builts showed services that were not there and omitted services that were. We spent four months and 400-odd trial holes building a survey we could actually trust, backed by ground-penetrating radar across the full alignment.
That looked like a slow start. It bought us a project with three utility strikes instead of the thirty-plus that the original drawings were quietly pointing us toward.
Sabkha, water and sheet piles
Below about four metres the ground turns to sabkha — salt flat, saturated, happy to flow into any hole you dig. Every deep run was sheet-piled and dewatered continuously, with discharge quality monitored because the water comes out saline enough to kill landscaping.
We ran two dewatering systems in parallel on the north reach after losing a pump on a Thursday night. Redundancy on dewatering is not a luxury; a flooded trench is a week of re-excavation and a compaction argument.
Access as a deliverable
We wrote a rule into the method statement: no business loses vehicle access for more than 48 consecutive hours. That single constraint drove crew count, plate-bridge inventory and the leapfrog sequence across the south reach.
It cost more in temporary works than a continuous trench would have. It also meant we finished without a single tenant escalation reaching the authority, which was the outcome the client actually cared about.
Twenty-eight live tie-ins
Each connection to a live network was a booked night window with a written rollback plan and the authority’s operator on site. Twenty-seven went to plan. The twenty-eighth found a valve that would not isolate; we backed out, reinstated, and returned nine days later with a different isolation strategy rather than improvising at 03:00.