Super Typhoon Champi tracked toward Taiwan and eastern China in the week of 20 October, and every shipper with export volume out of the Yangtze delta asked the same question: what happens to the boxes we already pushed out the door. The timing was awkward. Drewry had just shown average vessel waiting time at Shanghai falling to 38 hours from 58 the week before, and Ningbo to 34 from 40, so the backlog had finally started to shrink. Forwarders were telling customers to expect pilotage halts and temporary gate suspensions at Ningbo, Shanghai, Xiamen and Fuzhou from Friday night through Monday rather than a full terminal shutdown. Carriers were already skipping calls to protect schedule integrity.
A port closure at origin is a strange kind of disruption, because nothing actually breaks. Production keeps running, the cargo is fine, the carrier still exists, and the port will reopen. What stops is the exit. For the two to ten days it stays stopped, every decision you make about cargo gets made with worse information than usual and a shorter clock than you would like. What separates the teams that come out of it well is not foresight. It is having a sorting rule, and applying it before the gates shut.
Sort cargo by where it stands, not by when it is due
The instinct is to pull the list of shipments due this month and work it from the top. That is the wrong axis. During a closure, the thing that decides what you can still do about a container is where it is physically standing when the gate shuts.
Not yet released at the supplier. Everything is still open. You can hold, re-route to an alternative load port, split the order, or pull a partial by air.
On the road to the port. You can turn the truck around at the cost of a wasted leg and a storage day. That is cheap.
Gated in at the terminal. Options close here. The box sits inside a facility that is not operating, you cannot get it back, and it moves when the terminal says it moves.
Loaded but not sailed. Slightly better, in that it is on the right vessel. Considerably worse, in that the vessel may now skip ports downstream to recover its schedule.
Already on the water. Nothing left to decide at origin. The work moves to the destination end.
Sorting takes twenty minutes if the data sits in one place and most of a day if it does not, which is the real reason teams start this too late. Supplier release lives in the ERP, the drayage leg lives with the forwarder, the gate-in lives with the terminal, and the load confirmation lives with the carrier. A command center like Orkestra exists to collapse those four sources into one record per container, so "where is this standing right now" is a filter rather than four phone calls and an inbox search. In a closure that difference decides everything, because the window in which the answer still changes an outcome is measured in hours.
The useful decisions are the ones that expire
Most of the value in a closure is captured before it starts, during the watch period when the forecast is public but the gates are still open.
The asymmetry is worth stating plainly. A truck held at the supplier for two days costs a storage day and a rebooking fee. A container gated into a closed terminal costs you every option you had, and it starts the destination clock in a yard that will be congested when the backlog lands. The two outcomes are not close, which means the default during a watch period should be to slow the flow into the port and keep cargo where you still control it.
That default needs limits, or it becomes its own problem. Apply it to the lanes the forecast actually touches, name an owner, and set a time when the hold is reviewed rather than letting it run. The same discipline applies when the stoppage is on the production side instead: see planning around a factory shutdown at origin for how to decide what to pull forward and what to let slip.
Expect an omission, not a delay
The most common mistake is to model a closure as lateness. Carriers do not generally run a week behind and then catch up. They protect schedule integrity by dropping calls, which is exactly what they were doing through the Champi watch. Your container does not sail late. It does not sail.
The practical consequence is that the number to watch is the port rotation, not the vessel ETA. A vessel can be running close to schedule and still have removed your load port from its string, and the ETA field will look healthy right up until someone reads the revised rotation. Catching that change on the day it is published rather than a week later is the whole skill, and it is the same detection problem covered in blank sailings and port omissions. A dropped call usually costs seven to ten days to the next string, and longer when the next vessel is already full, which is normal in a backlog.
This is also where the detection gap shows up most expensively. Around 62% of supply chain exceptions are still found after the customer has felt them, and a port omission is a textbook case: the information exists in a carrier advisory days before anyone internally reads it.
The reopening date is not your date
Gates reopening does not mean your box moves. The restart is governed by yard occupancy, appointment availability, the pilotage queue and the vessel bunching that follows any pause. A terminal that reopens on Tuesday and clears 60% of a normal day's volume is still adding to its own backlog.
Two numbers give you a realistic restart curve: boxes cleared per day against boxes waiting. Ask the terminal or the forwarder for both, and re-ask every day, because the second number falls far more slowly than anyone expects. That curve is what you re-date customer commitments against, not the reopening announcement.
Then plan for the other end. Everything that queued leaves in a compressed window and arrives the same way, so the destination sees a month of volume in ten days. That is a separate piece of work, and it starts before the ships arrive: see managing an inbound container surge.
Re-date the orders before anyone asks
The operational failure in a closure is rarely the delay itself. It is the silence while the team works out what the delay means, which can run for a week and is where customer trust goes.
Re-dating needs the container-to-order link to be live, so a rolled sailing turns straight into a named list of orders and customers rather than a reconciliation project. Once that link exists, triage becomes a ranking exercise instead of a scramble, which is the same discipline as prioritizing exceptions when everything is late at once. It is also what moves resolution time: OIA Global cut exception resolution time by 60% in a quarter by working from one connected view rather than assembling it per incident.
Write it down while it is happening
Closures are frequent enough to learn from and far enough apart that nobody remembers the detail. Keep a running log during the event: which suppliers released late, which forwarder answered within the hour, how long from reopening to your first gate slot, how many boxes rolled and by how many strings. Those numbers are the only honest input into the next contingency plan, and they feed directly into a lane-level risk score that ranks where this will hurt next time.
What to do Monday
Take your top three origin ports and write the five-bucket sort for each one, today, while nothing is closed. Name who pulls the list, where each field comes from, and what the hold rule is during a watch period. Then run it as a drill against last week's live cargo and time it. If the sort takes longer than an hour, the problem is not your contingency plan, it is that the record does not exist in one place yet.
That is the part Orkestra's exception management layer is built to remove: a single record per shipment across carriers, forwarders, terminals and your own systems, with the order link attached, so a closure becomes a filter and a ranked list on the first morning instead of the third.
Book a demo to see a closure worked against your own lanes.
