How to see port congestion coming three weeks early
Berth waiting time tells you a port is already congested. Three leading indicators — schedule reliability drift, yard density and inland dwell — tell you it is about to be.
Key takeaways
- Berth waiting time confirms port congestion that already exists — by the time vessels queue, your booking is inside the queue.
- Three indicators lead it by two to four weeks: schedule reliability drift on the services calling that port, yard density and container dwell, and inland evacuation capacity.
- Terminal productivity does not degrade smoothly — past roughly 80–85% yard utilisation each extra box costs disproportionately more rehandling.
- Most ports congest from the land side, so watch dwell and empty-depot behaviour before the anchorage.
Most published congestion data — anchorage counts, waiting times, TEU backlogs — describes something that has already happened. The question a forwarder actually has is different: not is this port congested, but will it be in three weeks, when my boxes get there. That has a usable answer, because congestion builds in a physical sequence whose early stages are measurable well before a vessel queues.
Why berth waiting time is a lagging measure of port congestion
A vessel at anchor is the last event in a chain, not the first. The waiting time reported this week is the product of cargo loaded five weeks ago, a rotation that slipped three ports back, and a yard that began filling a fortnight before anyone noticed.
The metric is also asymmetric. Queues form quickly and clear slowly: arrivals can double in a day, but a terminal cannot double its moves per hour, and a full yard is a slow yard. A backlog built in one week routinely takes two or three to clear. By then the commercial consequences are in motion — calls omitted, congestion surcharges filed, alternates taking bookings. Port Watch, which tracks berth waiting times and yard utilisation, is a confirmation layer. The decisions belong earlier.
The three leading indicators
Schedule reliability drift on the services calling that port
The useful measurement is not average delay. It is bunching. Three vessels arriving within 24 hours and three spread over five days produce identical average lateness, and only one of them overwhelms a berth window.
Track the spread of arrivals against the proforma schedule, on the services you book rather than the port as a whole. Then look one step upstream: a ship two days behind at its previous call will not recover on a short leg, so reliability at the port before yours is the cheapest forward view available. Watch for lateness correlated across services — a shared upstream port, a weather event, a routing change — because that is what produces a bunched wave.
Yard density and container dwell time
Yard utilisation is occupied ground slots against available ones; dwell is how long the average box sits before leaving. Together they say whether the terminal is a buffer or a bottleneck.
The relationship is not linear. Below about 75% a yard absorbs shocks. Somewhere around 80–85%, depending on stacking equipment and block layout, the crane spends a rising share of its time on unproductive moves — digging out the box under three others. Moves per hour fall, discharge takes longer, berth occupancy stretches, the next vessel waits. A terminal at 88% is not marginally worse than one at 78%; it is on a different part of the curve.
Dwell is the more honest of the two, being the ratio between what arrives and what leaves. Rising dwell on flat volume means evacuation is failing.
Inland evacuation capacity
A port almost always congests from the land side. Sea-side arrivals are lumpy but bounded by a published schedule; the land side is where boxes actually leave, and if they stop leaving, everything backs into the yard.
The signals worth having: rail wagon and slot availability; barge waiting time at the quay, which in the Rhine ports moves before anything else; how many days out your haulier’s first collection slot is; and empty depot behaviour. Empty returns are the tell. When depots refuse returns, demand pre-notification, or divert to a facility 40 kilometres away, the system is out of space — usually a week or two before the anchorage shows it.
How the three feed each other
These are not independent gauges but one loop. Bunched arrivals push more boxes into the yard than the plan assumed; higher density cuts crane productivity; lower productivity extends berth occupancy, delaying the next vessel, which then sails late and adds to reliability drift at its next port. Boxes that cannot leave inland raise dwell, which raises density again. Carriers respond by omitting calls or blanking sailings to restore rotations, moving the congestion sideways to a neighbour.
Why the loop matters for timing
Every stage has to unwind in sequence for a port to recover, which is why clearance takes two to three times as long as build-up — and why one indicator improving means little. Yard utilisation can fall simply because the carrier stopped calling.
Secondary signals
- Labour negotiation timetables. Expiry dates are known months ahead, and disruption often arrives before the deadline: shippers front-load cargo ahead of a possible stoppage, and the front-loading is itself the congestion.
- Weather seasonality. A 48-hour typhoon closure in south China produces a bunched arrival wave in North Europe two to three weeks later, and low water on the Rhine in late summer removes barge capacity from Rotterdam and Antwerp without touching a vessel.
- A neighbouring port’s problems. Congestion is conserved. When one gateway falters, the alternates absorb the volume into inland networks not sized for it.
- Omitted calls and surcharge filings. An omission notice is a congestion report written by someone with better data than you have. A newly filed port congestion surcharge carries a notice period, so it appears before the peak — see the guide to container shipping surcharges and their triggers.
Turning a warning into a decision
A signal you do not act on is expensive to have collected. The realistic responses:
| Signal | Typical notice | Decision it supports |
|---|---|---|
| Arrival bunching on the service | 2–4 weeks | Buy transit buffer; prefer a direct call with fewer upstream ports |
| Yard utilisation moving into the 80s | 1–3 weeks | Consider an alternate discharge port; pre-alert the customer |
| Rising import dwell, empty depot restrictions | 1–2 weeks | Fix free time in writing; book collection and empty returns early |
| Omitted calls, congestion surcharge filed | Days | Damage control: communication and inland recovery |
The most valuable response is usually the least dramatic. Detention and demurrage free time is cheap to negotiate in a calm market and impossible in a squeeze, so it belongs in the annual round alongside the rate — one of several points where the clauses matter more than the number.
The crowded alternate
If the congestion is visible to you, it is visible to everyone booking the corridor. Rerouting to the obvious alternate often means arriving into a yard that has just absorbed several thousand diverted boxes, behind an inland network sized for normal volume — and a longer, dearer road leg at the end.
Common misreadings of the data
Comparing waiting times between ports. The definitions differ. Some measures start at the pilot station, some at anchorage entry, some at notice of readiness; some count a vessel idling because it arrived early for its own window, which is not congestion. Check that two numbers measure the same thing before comparing them — the discipline set out in the notes on how the benchmarks are built.
Treating one week as a trend. A port with a dozen deep-sea calls a week is a small sample, and one crane breakdown moves the average. Use a four-week rolling view, and compare a port against its own history rather than a peer’s.
Reading average dwell without the distribution. A tail of customs-held or abandoned boxes inflates the mean while saying nothing about flow; the share of boxes over five days is more useful. And because vessels slow-steam to arrive into their window, a thin anchorage count can hide a queue.
A weekly monitoring routine
Half an hour, the same morning each week, on your main services and terminals:
- Pull the four-week rolling arrival spread against proforma — the spread, not the average delay.
- Check performance at the port immediately upstream on each rotation.
- Note yard utilisation and import dwell at your discharge terminals, recording direction rather than level.
- Ask your haulier how far out the first available collection slot is, and whether it has moved since last week.
- Check whether empty depots have added restrictions or changed return locations.
- Scan carrier advisories for omitted calls, rotation changes and surcharge filings.
- Write one word per port — better, flat or worse — and keep the log; the four-week trend is the signal.
When the log turns worse two weeks running, put three questions to the carrier: which vessels are outside their berthing window; if the call is omitted, who pays the inland leg; and what free time will you commit to in writing for the next sixty days.
Frequently asked questions
How far in advance can you predict port congestion?
Two to four weeks of warning is realistic from the leading indicators, and rarely more than that. Schedule drift on the services calling a port is visible while the ships are still at sea, and yard utilisation and container dwell start climbing before any vessel queues at the anchorage.
Longer horizons are possible only for scheduled causes — a labour contract expiry, a typhoon season, a known rail engineering programme — where the calendar is public months ahead.
What is a healthy yard utilisation for a container terminal?
Below roughly 75% a container yard behaves as designed, and above roughly 80–85% productivity starts to fall away sharply. The reason is rehandling: the fuller the stack, the more unproductive moves a crane makes to dig out the box it actually needs.
The exact threshold depends on stacking equipment, block layout and how much of the yard is committed to empties, so track the direction of travel at your own terminal rather than a universal number.
Why does berth waiting time keep rising after arrivals slow down?
Because a port queue clears more slowly than it forms. Arrivals can spike in a day, but the terminal can only discharge at whatever rate its yard allows — and a congested yard is a slower yard, so throughput falls exactly when it is needed most.
That asymmetry is why waiting time is a poor early warning. It confirms a problem that formed weeks earlier and will take longer than that to unwind.
Is port congestion usually caused by too many ships?
Usually not. Most ports congest from the land side: containers arrive faster than rail, barge and road can evacuate them, the yard fills, terminal productivity drops, and only then do vessels start waiting for a berth.
Sea-side arrivals are lumpy but bounded by the published schedule. Inland capacity — wagons, barge slots, drivers, chassis, empty depot space — is the variable that actually decides whether a terminal keeps flowing.
Should I rebook to an alternate port when congestion starts?
Only after pricing the inland leg, because the alternate port is normally further from the consignee and the extra haulage often eats the whole saving. Rebooking also tends to be a crowd decision: if the congestion is visible to you, the nearby alternates are already absorbing diverted volume.
Buying transit-time buffer, or securing free time and collection slots early, is frequently the cheaper response.