A Commissioning Window That Cannot Hold: 2,400 Links, Two Testers, Two Weeks
A regional bank is enabling a 2,400-link single-mode hall. The plan: contractor mobilizes in week 2, 96 preterminated trunks with a 10-week lead time are ordered in week 1, cabinets arrive week 3, and the commissioning team is booked for weeks 7–8 only. Owner acceptance is 100% OLTS plus 10% OTDR; the manufacturer's extended warranty additionally requires 100% end-face inspection with records. Two certified testers can complete 120 duplex links per day on OLTS; OTDR runs about 20 minutes per link; end-face inspection and documentation takes about 2 minutes per end face. In week 4 a change order arrives for 240 additional links, and the business asks to pull go-live forward by two weeks for a regulatory deadline. The contract is fixed price with a change-order clause requiring written approval before work proceeds.
1. Do the throughput math and determine whether the booked windows can absorb the scope, including the change.
Read the worked answer
Work in tech-days, not calendar days. Original scope: 2,400 links at 120 links/day for a two-person crew is 40 tech-days of OLTS. OTDR at 10% is 240 links × 20 min = 80 hours = 10 tech-days. Inspection is 4,800 end faces × 2 min = 160 hours = 20 tech-days. Total ≈ 70 tech-days.
A two-week window at 5 working days is 10 days. Two testers supply 20 tech-days. The plan was infeasible before the change order arrived - the shortfall is roughly 50 tech-days, or five testers short. Add the 240 links: 44 tech-days OLTS, 11 OTDR, 22 inspection, about 77 tech-days, requiring eight testers working the full 10 days with no rework allowance. Pulling go-live forward two weeks removes the window entirely.
The conclusion is not 'work harder.' The commissioning plan must be re-sequenced so testing follows cabling in a rolling fashion, and the tester count must be resourced to the arithmetic, not to the original assumption.
2. Present your recovery options with cost, schedule, and risk for each.
Read the worked answer
Option 1, add testers: four certified testers plus two helpers for inspection and documentation roughly halve elapsed time, but only if trunks are terminated and labeled ahead of them; otherwise testers idle and the cost is wasted. Risk is medium - tester availability and calibration lead time.
Option 2, rolling commissioning: test row by row as trunks are installed from week 4 onward, using the weeks 7–8 booking only for retest and closeout. This is the lowest-cost recovery and the one I would push, but it requires the contractor to deliver complete, labeled, cleanable rows early and it exposes any termination defects earlier.
Option 3, sequence-breaking alternatives: defer the 240-link change to a post-go-live phase, or accept partial go-live with a documented punch list. Both are business decisions, not technical ones, and both must be priced.
Lead time is the hard constraint: 10 weeks from a week-1 order is week 11, already past a go-live pulled two weeks earlier. No testing strategy fixes that - expedite, split the order, or use field-terminated assemblies for the critical rows.
3. What do you refuse, and how do you document it?
Read the worked answer
Refuse to compress acceptance testing: no sampling in place of 100% OLTS, no dropping end-face inspection, no unidirectional OTDR substituting for bidirectional where the warranty requires it. Those items are the record that supports the 25-year warranty and the basis for rejecting a defect claim later; trading them for four days converts a schedule problem into a liability.
Refuse to start the change order before written approval, and refuse to absorb the 240 links into the fixed-price scope. Issue a written impact statement: added tech-days, added tester cost, revised completion date, and the effect on the regulatory milestone. Flag the lead-time conflict explicitly, naming which date is at risk and which decision the business must make.
Finally, refuse to accept 'the tester is conservative' or any verbal assurance about marginal links. Every disputed link gets a calibrated reference, a documented method, and a recorded result before it is closed.