BunkerStrategyIntelligence terminalReading listings, research log, and source-health evidence.
The archive's headline target is a risk-adjusted expected 2026 replacement cost. It is a point estimate built from the best-supported quantities, central installed rates, named risks and owner-side project costs—not an average of a very wide low/high range.
A point estimate is only useful when the specification and exclusions do not move inside the uncertainty range.
Recreate the documented facility in Sweden with current construction methods and 2026 prices while preserving documented capacity, protection, autonomy and redundancy.
2026 SEK · ex VATThe model separates direct works, identified uncertainty and the cost of investigating, designing and managing the project.
Best quantity × central installed rate × site factor, plus separately priced equipment packages.
Probability-weighted impacts from named risks; ordinary scope is not counted again.
Direct works plus expected risk, with investigation, design, owner costs and project management.
SEK 205m direct works + SEK 39m expected risks, then 22% investigation, design and owner costs:(205 + 39) × 1.22 = SEK 297.7m. Displayed at confidence-appropriate precision as SEK 300m.
A published rock volume overrides an area-derived proxy. Evidence rungs are not blended merely to make a smoother number.
If facility evidence is incomplete, a conservative shrinkage estimator can combine it with the relevant morphology prior: ln Q* = w ln Qfacility + (1 − w) ln Qmorphology. The initial weights above must be backtested rather than treated as permanent truth.
Rates must distinguish the complete installed scope from a material or supply-only observation:
The target rate is the weighted median of genuinely comparable current Swedish observations, adjusted for date, scale, access, geology and complexity. Supplier budgets are needed for protection systems.
The target estimate includes the expected value of specific risks, not the full impact of every adverse event at once.
A rock-facility register should identify events such as poorer rock, exceptional groundwater inflow, contaminated spoil and extra protected penetrations. Each entry carries a probability, direct impact, rationale, owner and correlation group.
Normal drainage, routine support and other high-probability work belong in direct scope. They must not reappear in contingency. Correlated risks and shared quantities receive explicit double-count checks.
If the evidence cannot establish whether an installation is a rock cavern or buried concrete, model each construction system separately. Do not add both systems or disguise the ambiguity inside a 10× band.
The probability-weighted value is only appropriate when the evidence supports branch probabilities; otherwise the archive withholds the point estimate and shows the unresolved branches.
A higher specification is a separate project brief, not the upper end of the functional-equivalent estimate.
Preserves the documented function, capacity, protection, autonomy and redundancy with current methods.
Uses a new structural, fire, accessibility, protection and supplier-performance brief.
Adds IT load, cooling, grid capacity, redundancy, autonomy and hardware boundaries.
Confidence is reported beside the estimate. It changes how precisely the number is displayed, not whether the archive answers the cost question.
This numeric score is a target framework. Where its six components have not been audited, the current site uses the reviewed evidence grade only and leaves the numeric score blank.
| Score | Grade | Display precision |
|---|---|---|
| 85–100 | A | Nearest SEK 1m |
| 70–84 | B | Nearest SEK 5m |
| 55–69 | C | Nearest SEK 10m |
| 40–54 | D | Nearest SEK 25m |
| Below 40 | E | One significant figure |
These modules are the work programme for improving evidence quality; their existence is not implied for records that have not yet been calibrated.
Choose functional-equivalent, modern hardened or data-centre-ready before costs are entered.
Trace halls, tunnels, shafts, portals and inner buildings into solid volume and support quantities.
Keep the unit, source, extraction note and provenance beside every quantity.
Record date, geography, scale, inclusions, exclusions, quote status and evidence quality.
Compare only within relevant morphologies, from rock command posts to fuel caverns.
Price doors, filters, penetrations, generators, fuel, EMP spaces and commissioning explicitly.
Separate survey, grouting, drains, sumps, pumps, dehumidification and monitoring.
Use probability × impact with correlation and double-count warnings.
Prevent a site and its included sub-installations from being summed twice.
Backtest error and bias by morphology as measured projects and supplier budgets arrive.
The framework follows the distinction between a most-likely base, risk and uncertainty in the UK Infrastructure and Projects Authority's Cost Estimating Guidance ↗, and the emphasis on WBS, assumptions, sensitivity, risk and confidence in the US GAO's Cost Estimating and Assessment Guide ↗. These guides inform the structure; they do not validate any individual bunker estimate.
Research methodology · 2026 prices · ex VAT · screening estimates are not tenders, approved budgets or market values