Planning, controls, risk and delivery lessons from twelve landmark programmes

Scope was largely resolved before contract award, the supply chain was locked early, and owner-led procurement discipline kept package prices stable even through the 2008 financial shock.

Report against the physical workface where the critical path actually lives — in this case the top of the core — rather than aggregated S-curves that hide the only signal that matters.

Identify the one location (a riser, a cutover window, a regulatory submission) where the critical path physically sits, and require daily variance reporting on that workface only.

Civils were reported separately from systems for years. By the time integration risk crystallised, civils contingency had been consumed and the December 2018 date was already politically embedded.

An independent reset combined deeper assurance authority with a single integrated dashboard for physical progress and systems readiness — removing the single-discipline blind spot.

Any programme with a long physical or build phase plus a software integration phase needs unified reporting from day one. Software risk reported in a separate forum is software risk that will be missed.

Awarding to the lowest fixed price on incompletely characterised geotechnical scope pushed real cost growth into the claims process — exactly where contracts that under-price risk usually end up.

Pre-agreed dispute boards and performance bonds were written into procurement, not added later. Several major claims were resolved at board level without escalating to international arbitration.

Treat procurement design as project controls. Dispute boards, bonds, indexation clauses and assurance triggers decided before contract award shape every variance and reserve drawdown for the next decade.

Construction sat under the integrated T5 Agreement; operational readiness sat in a separate governance silo, and the baggage system was never tested at full operational scale before go-live.

Extend the project baseline and risk register through commissioning, training and ramp-up. Apply the same EVM and assurance rigour to operational readiness as to structural completion.

No. The collaborative T5 Agreement delivered the build on time and to budget — a strong result for a programme of its size. The failure was that operational readiness lived outside that agreement.

A fixed political opening date (09/09/09) forced schedule compression and parallel commissioning during a boom-time labour and materials market, with supply agreements that were not locked in early enough.

That is a sponsor decision — the controls function's job is to make the trade-off visible. Dubai Metro hit its date because a single empowered owner accepted the cost of acceleration with eyes open.

A short, clear decision chain accelerates delivery. Multi-headed sponsors slow every variance, every change and every risk acceptance — and the cost of that drag rarely appears in any report.

Stable inter-governmental agreements bridged the 1997 sovereignty handover, so the programme kept a consistent delivery structure through a political change that would have derailed most mega-projects.

Cargo and baggage handling cutover were under-rehearsed relative to the scale of change. Construction was effectively on time; the weakness was operational readiness, treated as a hand-off rather than a project.

Daily program-level rollup of cost, schedule and risk — not monthly. At ten or more sub-projects, monthly reporting is too slow for the integration office to act before signals decay.

The ODA built a master schedule with one-year-early completion targets and defended that buffer. When a late-phase security contractor shortfall hit, the buffer absorbed the shock.

The clean separation between sponsor (government), delivery (ODA) and operator (LOCOG), defined before the bid was won. Role ambiguity at this layer is its own cost driver.

Set internal completion at least one full review cycle before the external date, baseline against the internal date, and treat any erosion of that buffer as a serious leading indicator.

Construction controls and the BOT finance model were managed apart. Scope additions, particularly safety redesigns, were absorbed into cost without a matching finance-side response, leading to repeated refinancings.

Construction EVM and the project finance model must be reconciled in the same forum. A governance gap between the two is where late-stage scope growth quietly turns into debt.

Engineering execution — the tunnels broke through to plan and the service has operated reliably since 1994. The lesson is financial-controls integration, not engineering.

Construction started before the shell geometry was buildable. Without a resolved design there was no baseline to control against, and the politically driven start date locked in years of late redesign.

Probably not on their own. The decisive failure was upstream: starting construction without a design freeze. No EVM regime can stabilise scope that has not yet been invented.

Starting a transformation programme on an unsigned target operating model, or breaking ground on a hospital before clinical adjacencies are agreed. Both compress invention into the construction phase.

A long programme with weak owner-side technical oversight, inflation not modelled until later, and assurance gaps that only tightened after the 2006 ceiling collapse. Independent assurance came late.

Independent assurance was strengthened, portfolio-level EVM was put in place and the quality regime was overhauled. The lesson is to start with that posture, not arrive at it after a public incident.

Independent assurance, inflation indexation and quality oversight have to be set up at sanction, not retrofitted after a failure. Year-of-expenditure forecasting on long programmes is non-negotiable.

Several assets have been re-baselined and phased rather than abandoned. Until publicly disclosed scope and dates stabilise, traditional overrun terminology does not yet apply cleanly.

Estimate in ranges, not points. Maintain a decision log of every scope change, and re-baseline assets when scope shifts rather than burying the change inside contingency drawdown.

Treat novel scope as research until it is buildable, then baseline. Probabilistic estimating and per-asset accountability are how a portfolio of first-of-a-kind assets stays controllable.

The original 2012 envelope was published as a point estimate without consistent inflation indexation, against a Y-network scope that later changed. The visible movement reflects both inflation and scope decisions.

Publishing P50 and P80 ranges from day one rather than political point estimates, and indexing to a transparent base year. Point estimates on long programmes are a communication choice that always ends badly.

Probabilistic reporting, indexed base year and a published decision log of scope changes. These three habits separate a re-baselining from a credibility crisis.