Regional Hospital Expansion
Fixed opening date drove a 16-week acceleration. Acceleration cost exceeded liquidated damages; the maths had failed at month 18 and nobody flagged it.
- Value
- $142M
- Duration
- 36 months
- Outcome
- Late & over
- TCPI > 1.15 sustained
- RAG stayed green for six cycles
- Risk register stale
- Re-baseline contingency every quarter against the live risk register
- TCPI above 1.10 is a recovery signal, not a forecasting one
- RAG status needs explicit criteria, not feelings
Project context
A 142-bed expansion to an existing regional hospital in the South Island of New Zealand, comprising a new four-storey clinical block, a two-storey link bridge to the existing hospital, a refurbishment of the existing emergency department, and a new central energy centre. Gross floor area was approximately 22,000 m² of new build and 4,500 m² of refurbishment. The contract was an NZS 3910 amended construct-only contract with the District Health Board as principal and a separate clinical fit-out package held by a specialist medical equipment supplier. The programme at award was 36 months with a fixed clinical commissioning and opening date driven by a winter respiratory season political commitment. Liquidated damages were modest at NZD 12,000 per day, capped at 5% of contract value, which became important when the acceleration arithmetic was finally done.
The fixed opening date that shaped every decision
The opening date had been announced publicly 18 months before construction started and reaffirmed twice in cabinet. From day one the project carried a political risk that no commercial recovery option could touch. Every monthly steering review assumed the opening date was immovable, and every recovery option was sized against that assumption. The problem with a fixed end date is that it makes every delay look acceptable in isolation — a two-week slip is always recoverable when the end is still 18 months away. The cumulative effect is invisible until the recovery cost passes the liquidated damages exposure, at which point the rational commercial choice is to accept the LDs. On this project, that crossover happened around month 22 but was not surfaced until month 28.
Why RAG status stayed green for six cycles
The monthly steering report used a red-amber-green rating against cost, schedule, quality, safety and risk. The criteria for each colour were not defined numerically — the rating was set by the project director based on judgement. From month 12 to month 18, the schedule rating was green or amber-green every month even though the SPI had dropped from 0.98 to 0.91 over the same period. The reason was that the project team genuinely believed the slip was recoverable and the rating reflected confidence rather than data. The lesson is that any RAG system without numerical thresholds will reflect the optimism of the person setting the rating. The fix is mechanical: define the thresholds, publish them, and let the data set the colour. On this project, a CPI below 0.95 should have been amber and below 0.92 should have been red — both crossed in month 16 without a change of colour.
The 16-week acceleration and the arithmetic that failed
By month 24 the project was 12 weeks behind. A 16-week acceleration plan was approved by the steering committee, costed at NZD 8.4 million, comprising weekend and night shifts, an additional tower crane, and premium time on the mechanical, electrical and partitioning trades. The liquidated damages exposure at the time was approximately NZD 6.7 million if the project opened 16 weeks late. The acceleration was therefore NZD 1.7 million more expensive than the LDs, before counting the disruption cost and the quality risk of compressed commissioning. Nobody did this calculation explicitly at the steering meeting. The acceleration was approved on the basis of the political commitment, which is a legitimate reason — but it should have been recorded as a political-cost decision, not a commercial one. The project opened nine weeks late despite the acceleration, with LDs of NZD 3.8 million and acceleration costs of NZD 7.9 million actually spent.
TCPI above 1.10 as a recovery signal
The to-complete performance index sat above 1.15 from month 18 onwards. TCPI above 1.10 means the project needs to perform meaningfully better than baseline for the entire remaining duration to hit the budget. Sustained TCPI above 1.15 is mathematically a recovery signal — it is telling the project that the budget is no longer achievable through normal performance. On this project, TCPI was reported every month but framed as a forecasting metric rather than a decision metric. The lesson is that TCPI is the most under-used metric on construction projects. CPI tells you what happened. TCPI tells you what has to happen next. A project that ignores a sustained TCPI above 1.10 for six months has effectively decided to over-spend without saying so.
Risk register decay and contingency burn
The risk register at award had 96 items with a quantified contingency of NZD 11.5 million. By month 18 the register had not been refreshed against actual project conditions. New risks that had emerged — the central energy centre commissioning sequence, the live emergency department interface, the medical gas certification process — were not in the register. Contingency had been drawn down to NZD 4.2 million against a re-assessed exposure that, when finally calculated at month 22, was closer to NZD 9 million. The lesson is that the risk register is a living document or it is wallpaper. A quarterly re-baselining session that costs the team two days saves more contingency than any commercial recovery action.
What this project says about fixed-date healthcare builds
Hospital projects with politically fixed opening dates are a known failure mode. The recovery options shrink to two: accept the LDs and protect the budget, or accept the cost and protect the date. Pretending both can be protected is the most expensive option on the panel. The right governance for these projects is a steering committee that is explicitly asked, at the monthly meeting, which of the two protections is being chosen this month — and to record that choice in the minutes. On this project, the decision was made by default rather than deliberately, and the cost difference between a deliberate decision in month 16 and a default decision in month 24 was approximately NZD 4 million.