Week 43: Savannah Berth Waits Hit 24 Hours, Charleston Import Dwell Reaches 5.7 Days

The Analysis

Put the map on the table first. Take the southeast corner of the United States, the stretch of coast between Norfolk and the Florida line, and lay this week's numbers on top of it. Savannah's Garden City Terminal, the largest single container terminal in North America, is asking vessels to wait up to twenty-four hours for a berth. Charleston is holding import boxes 5.7 days at Wando Welch. Norfolk International Terminal is waiting eighteen hours. And four hundred miles up the same coast, New York's APM and Maher terminals report no berth waiting whatsoever.

Ten years ago that southeast corner was a different place, and not because it was quieter. The story then was expansion: deepening channels, lengthening berths, building rail inside the gate so a box could move from ship to train without touching a public road. Savannah was betting thirty years of its future on being able to take the biggest ships the Panama Canal could pass. Charleston was finishing a harbour deepening project that had been argued about for most of a decade. The bet was that if you built the berth and dredged the channel, the network would come.

The network came. What Hapag-Lloyd's week 43 update shows, on a page last revised 23 October, is what the bet looks like once it pays off and then keeps paying: a coast where the friction has moved off the berth and into the yard behind it, and where each of these terminals is congested for a different reason. That last clause decides everything that follows. Read these five terminals as one phenomenon called East Coast congestion and you will hedge against the wrong thing entirely.

Take the numbers one at a time, because they do not measure the same kind of thing and mixing them is how people end up drawing the wrong map. Savannah Garden City: berth waiting up to twenty-four hours, import dwell 4.3 days, rail dwell 0.9 days. The waiting applies to Gemini vessels and non-Gemini vessels alike, and that detail should stop you first. Norfolk International Terminal: waiting up to eighteen hours, import dwell 4.5 days, with the stated cause a lack of inbound export boxes. Virginia International Terminal and VIG: one crane down at each. Charleston: Wando Welch at 5.7 days of import dwell, North Charleston at 3.4 days. New York: APM Terminals and Maher with no berth waiting.

When a terminal makes every vessel queue for twenty-four hours regardless of whose network it sails in, you are not looking at a schedule design problem. Gemini's proposition to the market is reliability through a hub-and-spoke structure, tightly timed shuttles and controlled buffers. Put twenty-four hours of unplanned waiting into one node of that structure and the shuttle does not absorb the shock, it transmits it. A vessel that was supposed to arrive, work and leave inside a narrow window arrives late, works late, departs late, and lands late at whichever spoke was waiting for it. Non-Gemini ships waiting exactly as long tells you the constraint sits at the berth interface itself, it is shared by everyone on the coast, and it is not something one alliance can optimise its way around.

Norfolk is the most interesting line in the whole update and it runs to three clauses. Eighteen hours of waiting, 4.5 days of import dwell, and the blame placed on too few export loads being brought into the terminal. Read that slowly, because it describes something structural rather than cyclical. A container terminal is not a warehouse; it is a machine for turning containers over, and half of that turning is the loaded export that arrives by truck or rail, sits briefly, sails away, and takes itself off the property. When those boxes do not arrive, the terminal loses its most useful move: the one that brings a container into the stack and takes it out again quickly, running alongside discharge work, occupying no real estate for days at a time.

What happens instead follows arithmetic rather than drama. Imports come off the ship, go into the stack, and stay. Nothing exported arrives to make room in the flow. Every extra day an import container sits without moving is a day of yard space that cannot be sold to anybody else, and sooner or later density reaches the point where every remaining move takes longer: cranes wait on trucks hunting for their boxes, berth productivity falls, and the next vessel waits longer still. It is a queue forming behind itself. None of this requires an import surge. It only requires the far end of the lane to stop booking outbound loads, and the southeast has lived with that asymmetry for most of the last decade.

That is precisely why I want ten years placed alongside these numbers. Growth at Savannah and Charleston over that decade was overwhelmingly import-led. Goods came into these terminals, moved inland by rail and truck, and did not come back as loaded boxes in anything like the same quantity. The capital programme of those years, deeper channels, longer berths, more cranes, more rail inside the gate, was a bet placed entirely on the inbound leg, and on the inbound leg it worked. What week 43 records is the invoice for the outbound leg nobody financed, and invoices of that kind tend to arrive in instalments.

Here is the figure almost nobody pulled out of this update, and it is why I think most commentary got the story backwards. Savannah's rail dwell is 0.9 days. Under a day, in the same week the general import dwell is 4.3 days and ships are queueing for a berth for twenty-four hours. Read those two together. The intermodal ramp is clear. The trains are not the problem. If rail were congested, boxes would pile at the railhead and it would be rail dwell leading every headline, and it is not. So whatever is eating four days out of an importer's supply chain at Garden City happens between discharge and the moment somebody puts that box on a train or through the gate: in the yard and at the truck gate, not on the water.

That changes what a remedy looks like and who ends up paying for it. If the constraint were berth capacity, the answer is more berths and more cranes and you wait three years for it. If the constraint is yard density, gate throughput and appointment availability inside the hours a terminal actually operates, the answer is operational and commercial: longer gate hours, night and weekend gates, appointment systems that clear rather than ration, storage pricing that discourages using the terminal as cheap warehousing. None of that pours concrete and all of it is cheaper than a berth.

A second silence in this data is worth naming. Every figure in the update is a boundary, not an average: up to twenty-four hours at Savannah, up to eighteen at Norfolk. Nobody publishes the distribution behind those words. Yet anyone who has ever built a delivery promise knows you plan against the tail of a distribution rather than its middle, and that the cost of planning to the average is paid entirely on the bad days. A vessel waiting six hours and a vessel waiting twenty-four make completely different demands on your chassis pool, your driver hours and whether your appointment is worth keeping at all, and both of them are folded into the phrase up to twenty-four hours.

Then there is New York, sitting in the same update with nothing to report. No berth waiting at APM, none at Maher. If the East Coast were genuinely full, the largest consumer market on that coast would fill first, because that is where every network wants its metal to go. The absence of waiting in New York alongside eighteen hours of it two berths down at Norfolk is not a curiosity, it is the argument in full. This is not a coast-wide event. It is a set of local failures inside one region's terminals, and local failures have local remedies, which is good news and bad news depending on where your boxes discharge.

Note also the two cranes, one down at Virginia International Terminal and one at VIG. Small number, disproportionate consequence. Berth productivity is not a smooth function of how many machines are standing; it behaves like a step function. Removing one crane from a smaller secondary terminal takes out a large fraction of that berth's productive capacity at the exact moment everything else is under strain, and the queue that forms does not drain evenly once the crane comes back. Any terminal running below its planned crane count is running below its published capacity in reality, no matter how long the quay is.

When does any of this reach somebody's money? Not immediately, and the travel time is worth planning for explicitly. Assume your boxes discharge at Garden City this week. Add the published 4.3 days of import dwell, then whatever your drayage and appointment situation adds, then inland transit, say three to five days to a Midwest distribution centre given that the rail ramp itself is turning boxes in 0.9 days. Goods planned to be on the shelf early in November now arrive mid-November, and the decision those goods were needed for, a promotion, a production start, a replenishment, has already been made without them. That ten-day gap between a queue forming at the berth and a consequence arriving at your warehouse is the planning window you actually have, and most importers discover it from the wrong end.

Exposure is not shared evenly. The importer with a diversified footprint able to land at New York or through a Gulf port and haul inland either way holds optionality, and optionality is the only real hedge against a queue you cannot influence. The importer whose entire Southeast volume funnels through Wando Welch at 5.7 days carries concentration risk in the one terminal currently slowest on that coast, and concentration risk at a single gate is worth more attention than any rate negotiation this quarter. The exporter is exposed too, though less visibly: it is a thin export book that stops the yard turning at Norfolk, which means importers and exporters who have never spoken to each other are linked through the same stack.

Where might I be reading too much structure into an ordinary bad month? Two places, chiefly. If week 43 is a peak-season pulse, vessels bunching into a narrow window, a few weather days, a holiday somewhere upstream, then excess waiting fades within weeks and the right response is to absorb rather than reroute. Telling the two apart takes one more month of data and one simple test: does berth waiting at Garden City fall back while import dwell holds at 4.3 days, or does dwell itself come down? Waiting that clears while dwell holds is a pulse. Dwell that sits for another six weeks is the structure talking, and it is saying something you can plan around.

And here is the condition that would genuinely change the conclusion, which is why the stated cause at Norfolk matters so much. If outbound export bookings from the Southeast recover, resins, pulp, agricultural loads, whatever puts weight in a box leaving that coast, the missing half of the terminal's throughput returns, the yard starts turning again, and Norfolk's dwell improves with no capital spent and no network decision made by anybody shipping through it. One parameter outside the fence moves the whole system. Watch the export book rather than the berth queue: the queue tells you what has already happened, the export book tells you what happens next.

A second consequence hides in the empty half of that trade, and it belongs to the carriers rather than to the terminals. A lane that imports far more than it exports does not merely leave the terminal short of useful moves; it leaves the equipment itself on the wrong side of the map. Boxes accumulate inland where they were consumed, empties have to be repositioned back to wherever the next booking actually is, and every repositioned empty is a move that earns nothing while occupying yard space, rail slot and chassis exactly like a loaded box. Fleets are finite. So a thin export book does not politely confine its damage to dwell statistics: it quietly consumes equipment availability and a little slot capacity that imported cargo also needs. Shippers usually meet this much later and under a different name, either as no equipment available at the pickup end or as a surcharge nobody ever connects back to an export booking that never happened.

One more comparison belongs in this picture, because it is the one everybody reaches for and it misleads. The congestion of 2021 and 2022 only looked like this from the outside. Back then vessels queued because demand exceeded every step of coast-wide capacity at once, dwell rose because nothing could move inland, and New York was full along with everybody else. The distinguishing test is not severity but correlation. When the largest terminal on a coast has no queue while its neighbour two berths away waits eighteen hours, the cause is local by definition. Local causes make routing decisions worth something. Coast-wide causes do not, because there is nowhere left to route to. Week 43 hands us a week in which routing decisions have a little value again.

Pulse or turning point, then, and I will answer it plainly. This is a pulse in the queue and a turn in the structure. The twenty-four hour wait goes away, because queues always do. The reason it formed does not, because it was never an operations problem: a decade of import-led growth built berths faster than it built any reason for containers to leave that coast loaded. The map tells you where the network grows next, and the answer is written in Norfolk's 4.5 days rather than in any press release. Capacity in this region will increasingly belong to whoever solves the outbound leg, the empty flows, the inland delivery density. The redrawing is happening in the part of the map nobody photographs.

Put money on it with the assumptions visible, since assumptions are what let you argue with me. Assume one importer moving 200 containers a month through Savannah. Assume each container holds 60,000 dollars of goods and that working capital costs 8 percent a year, so roughly 13 dollars a day per box simply for money sitting still. Assume published dwell slips by one day, well inside the range above. That is 200 boxes and 2,600 dollars a month in carrying cost alone, before demurrage enters the picture. Now assume diverting those boxes to a New York landing costs an extra 350 dollars each in inland haulage and adds two days of inland transit. Spending 70,000 dollars a month to avoid 2,600 dollars of carrying cost is not a trade, it is a donation.

Change one assumption and the trade turns over completely. Assume dwell at Wando Welch runs to 8 days instead of the published 5.7, assume the contract gives three free days, and assume demurrage and detention together cost 150 dollars per box per day once free time expires. Five chargeable days across 200 boxes is 150,000 dollars in a month, and now the 70,000 dollar diversion is the cheapest decision on the page. Nothing about the port changed between those two sets of numbers. One contractual variable moved. That is not hedging, it is the point: the two things worth tracking every single week are dwell days measured against your free time, and whether the contract in your drawer actually grants the free time you believe it does.

What I would do with this map, keeping it practical, starts with getting your own figures broken out per terminal rather than per carrier. It is the terminal that holds your boxes, not the vessel that brought them, and most reporting from carriers is structured around their own network rather than around your exposure. Then price that exposure against your own paper: how many free days, what each day costs after that, and which terminal every routing you buy actually lands in. Half a day's work, and most importers come out of it discovering that a chunk of what they thought was a rate problem is really a free-time problem.

Alternatives, priced honestly. Routing around the problem by landing in New York and hauling inland costs real money and buys back predictability only for as long as New York stays open, which it is today and need not be in two quarters. Swapping Norfolk for Savannah trades one constraint for a different one, and in Norfolk's case the constraint is the export book rather than the crane count. Adding pipeline inventory is the dullest answer and usually the cheapest: one extra week of stock on your highest-volume Southeast SKUs costs carrying charges and nothing else. The pitfall shared by all three is checking once. These figures get revised weekly for a reason, so put the review into somebody's Monday morning, let it run for a full quarter, and refuse to believe any single week, including this one.

↑ Back to the key points

— By Tomás Ferreira