Vincent Thomas Bridge Shuts Nov 7 for 16 Months After LA-LB Record September

The Analysis

The notice says this. Caltrans will close the Vincent Thomas Bridge from 7 November for about 16 months to replace the deck, with reopening targeted for early March 2028. The bridge is the road link between San Pedro, Terminal Island and Interstate 110. It carries about 53,000 vehicles a day, 8.8 percent of them heavy trucks, roughly 4,700 port-related truck trips, and about 8 percent of the port complex's drayage. The closure lands right after a record September: Los Angeles handled 1,042,652 TEU with loaded imports up 19 percent, Long Beach 997,177 TEU with loaded imports up 31.3 percent, and rail container dwell has already climbed to 7.19 days.

Put the map on the table first, because a bridge closure is never about the bridge. The Vincent Thomas opened in 1963, when a container was a thing people were still arguing about. The Gerald Desmond, its sibling span, was replaced in 2020 by the Long Beach International Gateway precisely because the old structure could not carry the trucks the port had grown into. Now the other span goes out for sixteen months, and every box that used to cross it will find another way, and the other way is already carrying traffic. This is how a port complex rebuilds its road network: one span at a time, and each rebuild pushes the queue onto its neighbour.

I have spent my working life inside terminal gates watching trucks arrive, and what I know about a closure like this is that the headline number is never the number the gate feels. Eight percent sounds survivable. It is survivable. But a gate does not experience percentages. It experiences a queue that forms at 06:40 and clears, or does not clear, by the end of the second shift. So let me take the numbers apart and put them back together in the units a terminal actually runs on: turns, hours, and where the line stops moving.

Begin with the arithmetic, because there is less of it here than the coverage implies. Fifty-three thousand vehicles times 8.8 percent is 4,664. The notice says roughly 4,700 port-related truck trips. Divide 4,700 by 53,000 and you get 8.87 percent. So the 8.8 and the 4,700 are not two measurements that happen to agree. They are one measurement wearing two hats, and the real number underneath is 4,700 trips. That matters for how you use it, because a derived figure inherits every assumption in the derivation. Whoever produced 8.8 percent divided a truck count by a vehicle count that includes every commuter crossing to San Pedro for lunch. The denominator is not the port. It is a road.

Now the thing the 8 percent hides, which is that a trip is not a unit of capacity. A drayage truck does not sell trips. It sells turns, and turns are a function of cycle time. Assume what a drayage move looks like at this complex: a two-hour-forty-minute round trip from the terminal gate, through the street network, to a warehouse in Compton or a rail ramp in City of Industry, and back, including the in-gate and the out-gate. In a ten-hour driving window that is 3.75 turns per truck per day. Add thirty-five minutes for the detour, which is my assumption and it is on the conservative side for a peak-hour reroute across the International Gateway, and the cycle becomes three hours fifteen.

Turns fall to 3.08. That is a capacity loss of 17.9 percent, and 17.9 percent applied to 4,700 trips is about 845 trips a day that the same fleet can no longer deliver. At roughly 1.4 TEU per drayage trip, call it 1,180 TEU a day. Against a September daily average of about 68,000 TEU across the two ports, that is 1.7 percent.

Sit with that number, because it is smaller than the headline and it is the honest one. The bridge does not break this port. What it does is remove a margin, and the margin was already thin. One point seven percent of daily capacity does not matter in a quiet February. It matters enormously in the fourth week of a peak when the terminals are holding imports for rail and the rail ramp is full. Capacity loss is not additive with congestion. It multiplies. A system running at 85 percent utilisation absorbs a 2 percent capacity cut without a visible change in queue. A system running at 96 percent does not absorb it at all, because the queue response to lost capacity is not linear, it is a curve that goes vertical near the ceiling. Everything about this closure depends on where the port is on that curve on any given Tuesday, and September put it high.

Here is the part of the story I have not seen reported, and it is the part that decides which terminals suffer. Eight percent is an average across the whole complex. Averages across a port are the most misleading number in our business, because a port is not one queue. It is a set of terminals, each with its own gate, its own yard, its own rail connection and its own access road. The Vincent Thomas serves Terminal Island. For the terminals on that island, the bridge is not 8 percent of the access. It is most of the direct access, and the alternative is a detour across the International Gateway and down the Terminal Island Freeway, which is a different road with a different merge and a different set of traffic signals.

So the correct reading of 8 percent is this: it is the complex-wide share, and the terminal-level share is much higher for some gates and close to zero for others. If you are an importer whose boxes land at a terminal on the island, plan against the higher number. If your boxes land in the outer harbour, plan against something much closer to zero. One number cannot serve both, and the notice gives you only one.

Then the clock, because what a gate feels is a clock. The closure starts on Saturday 7 November. Saturday is the lightest drayage day of the week, which is not an accident and is the single best decision in the whole plan: it buys 48 hours before the first real test, which is the Monday 06:00 inbound window on 9 November. That window is where I would put my attention. Importers re-cut appointments over the weekend, drivers who have never run the detour take it for the first time in the dark, and the queue forms not on the bridge, which is closed, but at the junction where the detour traffic merges into the Terminal Island Freeway. A closure moves a bottleneck. It never removes one. The engineering term is traffic diversion. The operational term is that your 07:00 appointment now arrives at 07:50 and the terminal's appointment system does not care why.

A terminal appointment is a contract about a window, and windows are enforced by a system that does not read the news. Miss your slot by fifty minutes and the system does not record a late arrival. It records a no-show, and a no-show costs you the slot, sometimes a fee, and sometimes a place at the back of tomorrow's book. That is the mechanism by which a road delay turns into a capacity loss even when the truck eventually arrives, because the appointment book is the terminal's production plan and a missed slot is a hole punched in it. Terminals that handle this well will widen windows for the first fortnight and then tighten them. Terminals that handle it badly will enforce the book and let the queue build outside the gate, which is the worst of the available outcomes, because a truck idling on an access road burns the driver hours that would otherwise have produced a turn.

Which brings me to the metric that actually degrades first, and it is not truck count. It is the dual-transaction share. A drayage turn is profitable when a driver drops an import and picks an export on the same visit, because the empty leg, the one that earns nothing, gets absorbed into a paid move. Dual transactions are what make a two-hour-forty cycle economic. Now lengthen the link between two terminals. If two gates that were twelve minutes apart become forty minutes apart, the arithmetic of the dual changes, and a dispatcher who is paid to protect turns will start refusing duals because the second leg no longer fits inside the appointment window. Watch that ratio, not the truck count. When the dual share drops, the empty legs come back, and every returning empty is a paid mile that produces no revenue, which is the quiet way a drayage market tightens.

Chassis deserve their own paragraph, because they are the part of the system that cannot be hurried. The complex runs on a pool, and a pool is a fixed stock of equipment that turns over at the speed of the cycle. If the cycle goes from 160 minutes to 195, every chassis in the pool is held about 22 percent longer per move, and the effective pool shrinks by roughly the same proportion without a single unit leaving. Assume a pool of a thousand chassis serving 4,700 trips a day, which is 4.7 trips per chassis and tight to begin with.

At the longer cycle the same pool supports about 3.8 trips per chassis, and 4,700 trips now need closer to 1,230 chassis. You cannot buy your way out of that in a week. Chassis are ordered months ahead, the pool is shared with every other user of the terminal, and a shortage shows up as a driver sitting at a yard waiting for wheels. So the first hard constraint is not the road and not the gate. It is equipment standing still, correctly, waiting for a truck stuck in a merge.

And this is where the empties enter, which nobody in the coverage has mentioned. The 4,700 trips are not all loaded. A meaningful share of port drayage is empty repositioning: returning empties to the terminal for a carrier, moving empties between depots for a landlord, hauling an empty to a shipper because the booking says so. Empty moves carry no revenue for anyone except the trucker, and they are the first thing dropped when capacity gets tight, because a dispatcher triages by revenue per turn. So the sequence I expect, and I have watched it before at this complex, is this.

Cycle times rise, turns fall, duals get refused, and the first casualty is the empty return. Empties then sit at shippers' yards and at off-dock depots, charter and chassis pools get depleted at the terminals, and yard density climbs for a reason that has nothing to do with import volume. A road problem becomes a yard problem within about three weeks, and a yard problem becomes a rail problem within six. That chain is not in the notice because the notice is about a bridge.

One more cut at the September figures, because the total hides the mix. 1,042,652 plus 997,177 is 2,039,829 TEU across the two ports. Loaded imports rose 19 percent at Los Angeles and 31.3 percent at Long Beach, and those two growth rates ran ahead of the totals, which tells you the import leg grew faster than the rest. Imports and exports do not generate the same drayage. An import container almost always leaves by truck or by rail, and the truck share is what fills the island gates. A trans-Pacific trade that is heavy on empties back out generates drayage moves with no cargo revenue attached. So a month in which loaded imports outperform the total is a month in which the truck-heavy share of the flow rose faster than the headline, and drayage demand per TEU went up. That is the shape of the demand that entered this closure. A record September is not incidental to the story. It is the setup.

Now the rail number, because 7.19 days is the number I would underline if I could underline only one. The port complex entered this closure with rail container dwell already at 7.19 days. That is the dwell on boxes waiting for a train, and it is the clearest evidence that the binding constraint at this gateway is not the road. The road moves the box from the terminal to the ramp. The ramp is where the box then waits a week. Remove 8 percent of the road capacity and you have not created the congestion.

You have removed the relief valve from a system whose bottleneck was already downstream. That distinction changes what you should do about it, and it changes who should pay. If rail dwell is the constraint, then a drayage surcharge is a tax on the wrong party. The box is not late because the truck was slow. The box is late because the train was full, and the truck is now being asked to absorb a cost that belongs to the ramp.

To the counterfactual, because a closure this long deserves one. Suppose the bridge had stayed open. September would still have produced a 7.19-day rail dwell, because a record import month with loaded imports up 19 percent at Los Angeles and 31.3 percent at Long Beach does that on its own. So the bridge is not the cause of the congestion. It is an amplifier, and amplifiers matter at the top of the range and nowhere else. Suppose instead the closure had begun in February rather than November. You would still lose 16 months, but you would not put both Thanksgiving peaks and both pre-Lunar New Year surges inside the window. The start date is the variable the port could have argued about, and it landed on the worst quarter for retail. That is a scheduling decision, not an engineering one, and it is the one that will be remembered.

The duration deserves its own note, because the arithmetic is unusually honest. Sixteen months from 7 November 2026 is 7 March 2028, and the target says early March 2028. Infrastructure dates usually do not survive their own arithmetic, and this one does, which I take as a sign that the deck replacement is a known quantity rather than a rebuild with unknowns in it. But sixteen months means two full peak seasons inside the window, two trans-Pacific contract seasons in May, and one complete annual cycle in which nobody at any importer's office will remember why a particular lane got slower. The risk in month fourteen is not the detour. It is institutional forgetting. People who joined after November will route trucks the way the map tells them to and will not know there was ever a bridge.

So what should you actually watch, and what number should pull a trigger. I would run four indicators at the terminal level, weekly, not the port level. Gate-to-gate turn time per terminal, which for a Terminal Island gate starts at that 160-minute baseline and should not be allowed to sit above 200 for two consecutive days. Dual-transaction share per carrier, which should not fall by more than a fifth from its pre-closure level. Empty dwell at your off-dock depot, which is the earliest warning that capacity is being triaged. And rail dwell at the ramp you actually use, because at 7.19 days the ramp was already the constraint and a road closure will not improve it. Those four, in that order, tell you more than any port-wide percentage will.

There is one mitigation that works and it is not on the road, and it deserves to be said plainly. Shift the turns into the night. The detour is a peak-hour problem; at 02:00 the International Gateway is an empty bridge and the cycle can come back under 160 minutes. So the terminal that keeps a night gate open through the winter does more for this closure than any road improvement Caltrans can deliver, and the quantity that then binds is not lane capacity but gate labour hours and the number of chassis in the pool at 02:00. Importers who can move their appointments should ask for night windows in writing before the first fortnight ends, because night capacity is allocated on a first-come basis and it will be gone by December.

I will close with the map, since that is where I started. The Vincent Thomas has been carrying this harbour since 1963. The Gerald Desmond was replaced in 2020 because it could not take the trucks. Now the second span goes out for sixteen months, and for a year and a half the harbour will run on one road where it used to have two. That is not a crisis. Ports are resilient things, and this one has absorbed far worse. But it is a reminder that the road network under a port is not infrastructure in the abstract. It is the thing that decides whether a box that arrived at 04:00 clears the gate by noon, and there is no model, no platform and no dashboard that fixes a bridge that is not there.

↑ Back to the key points

— By Tomás Ferreira