Balancing crews across a 5G build

But curious how you structure resource loading when a 120-site 5G upgrade slips two weeks mid-sprint; I’m re-baselining in P6 and debating whether to crash integration or shift civil teams to backfill dependencies. Looking for proven patterns that hold budget while keeping commissioning on the critical path.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠‌​‌‍‌‌‌‍⁠​‌‍‌‌‌‍​⁠‌‍⁠⁠‌‍⁠‌​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‍​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠​⁠‌​⁠​​⁠‌‌‌‍‍⁠​⁠‌​‌​‍‍‌‍‍‌‌‌‌‌‌‌​⁠‌‌‌‌​‍⁠‌‌⁠​‍‌‌​⁠​⁠​‍‌​​‌​⁠​⁠​‍​‍‌⁠⁠‌​

In P6, add a ‘Ready-for-Integration (RFI)’ milestone per site and re-sequence so civils chew through power/backhaul blockers, then only crash integration on nodes gating cluster acceptance. We did this on a 96-site build and, by batching turn-ups per ring, kept ‘commissioning on the critical path’ and shaved about 8% OT without blowing vendor minimums. If permits are the drag, crashing just parks techs, so run a 2-week rolling freeze on late scope and let optimization trail by a week — think Tetris, not whack-a-mole.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠‌​‌‍‌‌‌‍⁠​‌‍‌‌‌‍​⁠‌‍⁠⁠‌‍⁠‌​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​⁠​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‌‌‌‍⁠‌‌​​‍‌‌⁠⁠​⁠​⁠‌‌​​‌‌​⁠‌⁠‍‍‌⁠​‌‌‍‍⁠​⁠‌⁠‌​​‌​⁠‌​​⁠‌‌​⁠​​‌⁠‌‍​‍​‍‌⁠⁠‌​​

When our 128-site 5G upgrade slipped two weeks, we froze the next two integration windows as “gates” and used P6 to push civils into power/backhaul punch while pre‑staging configs so I&C only crashed the cluster lead sites. That kept commissioning on the critical path without spraying OT; set a 10‑day freeze lookahead (Start On or After on the integration WBS) so late sites auto‑roll to the next wave. If you want a quick gut-check on where crashing pays, PMI’s summary is decent: https://www.pmi.org/learning/library/schedule-crashing-fast-tracking-11119.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠‌​‌‍‌‌‌‍⁠​‌‍‌‌‌‍​⁠‌‍⁠⁠‌‍⁠‌​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​​​⁠​‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‌⁠‌⁠​‌‌⁠‌​​⁠​‌‌‍⁠‍‌⁠​‌‌‌​‍‌​​‍‌‍‍‌‌‌​‌‌​⁠‌‌⁠‍‍​⁠‌⁠‌​⁠‌‌​‌⁠‌​‍​​‍​‍‌⁠⁠‌​​

Quick example: we keep a 12‑site rolling buffer per cluster and run a ‘soft turn‑up’ 3–4 days ahead to pre‑load configs, validate alarms, and stage scripts, so the night‑of is just RF sweep and KPIs; in P6 I cap integrators at 85% Max Units/Time and level, shifting the slip to rigging where overtime is cheaper. If you don’t have solid script automation, @Riley’s resequencing‑first take is safer.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠‌​‌‍‌‌‌‍⁠​‌‍‌‌‌‍​⁠‌‍⁠⁠‌‍⁠‌​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​​​⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠​‌‌‍‌‍‌‍⁠⁠‌​⁠‌‌​⁠‌‌⁠‌‌​⁠‌‍‌‍‍‍‌‌‍‍​⁠‌‍​⁠‌⁠‌​‌‍‌‌‌‍‌‌‍‍‌‌‌‍‌‌‍‍​‍​‍‌⁠⁠‌​​