Metrics In Detail

How the signal works.

Live port data becomes a concise operating signal, with source freshness and route context kept visible.

Headline Layer

Congestion Signal

Headline score composition
Anchorage pressure 30% + Arrival surge 30% + Flow imbalance 20% + Dwell pressure 20% = 0-100 signal

Fresh source feeds are required before the dashboard publishes the headline score.

Inputs

Anchorage pressure, 6-hour and 12-hour arrival surge, positive net accumulation, and vessel dwell pressure. Terminal occupancy and purpose-of-call remain visible context, not headline-score drivers.

Scoring

Each component is normalized against current calibration bands before the weighted score is published.

Calibration Status

Empirical bands are accepted only after enough valid samples, covered days, and no large metric-history gaps. Until then, the dashboard exposes provisional calibration provenance.

Implemented Metrics

Signals by operating question.

M1 Demand Pressure

Anchorage Pressure

Counts slow or stationary vessels inside Singapore anchorage polygons from the latest vessel snapshot.

Used by M10. Indicative of waiting pressure, not proof every vessel is waiting for berth.
M2 Demand Pressure

Arrival Surge

Counts inbound ETAs due within 6, 12, 24, and 48 hours.

The forecast feed has ETA and port codes, but no inbound coordinates.
M3 Vessel Flow

Vessel Dwell Time

Matches arrivals to departures over a rolling 7-day window and reports P50, P75, and P90 dwell hours.

Needs continuous arrival/departure history before it becomes stable.
M4-M6 Vessel Flow

Arrival, Departure, Net Flow

Computes rolling 6-hour arrival/departure rates and net accumulation to show whether the port is clearing or building load.

Positive net accumulation is the M10 flow-imbalance component.
M7 Volume Context

Arrival Volume Intensity

Compares today's arrivals with the 30-day daily arrival average.

Useful context for unusual volume; not a direct congestion proof on its own.
M8 Terminal Pressure

Berth-Adjacent Occupancy

Counts cargo-like vessels below 1 knot within 50 m of reviewed berth slots, or wharf-edge fallback geometry.

Terminal percentages are capped at 100%; berth denominators have mixed confidence.
M9 Operations Context

Purpose-Of-Call Mix

Uses minimised arrival-declaration purpose values from the last 24 hours.

Taxonomy review is pending, so M9 is excluded from M10.
M10 Composite

Congestion Signal

Summarizes selected pressure indicators into LOW, MEDIUM, HIGH, CRITICAL, or STALE.

Presales monitoring signal only; not a validated decision trigger.
Route Overlay

Transshipment Delay Exposure

Applies a static Singapore route weight to the current congestion signal for selectable demo lanes.

Categories support monitor, notify, or commercial-review actions.
Route Overlay

How transshipment context is represented.

Route exposure formula
Congestion signal x Singapore route weight = Delay exposure

The route layer is a demo overlay, not a cargo-value or payout calculation.

Dashboard Rows

The latest congestion row is duplicated across selectable route scenarios in the metrics service's latest layer. That makes route switching fast without changing the route-neutral time-series archive.

Exposure Formula

Route delay exposure equals the Singapore congestion signal multiplied by the selected static route weight. The result is a monitoring/commercial-review signal, not real cargo-value exposure.

Displayed Context

Route ID, origin, Singapore transshipment port, destination, cargo archetype, primary user, driver, caveat, category, and recommended action are persisted for each scenario.

Future Upgrade

A real route layer should use licensed route geometry, ETA, distance, display rights, and cache rules. The current route overlay is deliberately static until those decisions are explicit.