Port Klang yard hits 90-95% as berth delays stretch to 2-3 days

The Analysis

Put the map down first. Not a screen, an actual map. Find the Strait of Malacca, run a finger along the channel between the Malay peninsula and Sumatra, and stop where the shore opens into Port Klang. It sits about as close to Singapore as a major port can sit while still being in a different country, and for half a century it has grown in Singapore's shadow and grown anyway. That geography is why the numbers in Kuehne+Nagel's 15 to 21 July update carry more weight than a single week of numbers normally does.

What the update says is plain enough. Port Klang yard density has risen to 90 to 95 percent. Heavy berth congestion is producing average vessel delays of two to three days, against a seven-day average waiting time of 1.63 days, and several carriers expect further vessel bunching this week and next. The planned maintenance dredging has been postponed to 3 to 12 August. Manila is worse: a seven-day average wait of 4.48 days, off-window vessels delayed beyond five days, and overall vessel delays reaching ten to twelve days after the bunching that followed Typhoon Bavi. Singapore, sitting between the two, has calmed to a 1.25-day average with terminal waiting down to half a day to a day, though its yard is still around 80 percent.

I have spent a career reading ports on a ten-year scale, so let me say what I think this is before I say what to do about it. A yard at 90 to 95 percent is not a congestion statistic. It is a buffer that has been spent. Ports work because the yard absorbs the mismatch between when ships arrive and when trucks and trains come for the boxes. At 60 percent, a yard absorbs almost anything. At 90 to 95 percent it absorbs almost nothing, and every small shock goes straight through to the berth queue and out into your transit time. Ten years ago this would have been a bad week. What makes it different now is the network it sits in.

Spread the map out and the shape of it becomes obvious. Southeast Asia's container traffic runs on a small number of nodes strung along one strait: Singapore, Port Klang, Tanjung Pelepas, Laem Chabang, Tanjung Priok, and behind them the mainland Chinese ports that feed them. The same update puts Shanghai and Ningbo still congested after Typhoon Bavi, Qingdao recovering from post-typhoon bunching with waiting times expected to improve, and Yantian constrained by quay crane upgrades and berth restrictions with congestion risk expected to rise as arrivals concentrate from north and central China. Yantian is the one I would underline. A crane upgrade is not weather. It is a scheduled, dated, known-in-advance reduction in capacity, and it is happening at the same time as the post-typhoon backlog. Two independent constraints landing in the same month is how a bad week becomes a bad quarter.

Ten years ago this part of the map was a different thing. Port Klang was a regional gateway with ambitions. The transshipment traffic that defines it today was thinner, the mainland ports fed it less, and a two-to-three day berth delay would have been absorbed by slack that no longer exists. What really changed is not the port. It is the network. Southeast Asia has grown into a manufacturing and transshipment system dense enough that a shock at one node now propagates within days to every lane touching it, and dense enough that there is no longer any slack sitting anywhere inside it to soak the shock up. A yard at 90 to 95 percent is simply what a network with no slack looks like when something goes wrong upstream.

This is why I would take yard density over waiting time as the number to watch. Waiting time tells you what already happened. Yard density tells you how much capacity is left to absorb what happens next. Semarang, in the same update, sits at 82 percent with average vessel delays around two days and a waiting time near one day. Singapore is at around 80 percent. Port Klang is at 90 to 95. Those three readings are a ranking of how much more punishment each port can take, and Port Klang is at the bottom of it. If another bunching event lands in the next two weeks, as the update says several carriers expect, the yard is where it will show up first and the berth queue is where you will pay for it.

Manila deserves its own paragraph, because 4.48 days of average waiting is not the same kind of problem as Port Klang's. Port Klang is a dense, efficient port running out of room. Manila is a port where the queue itself has become the operating condition: off-window vessels waiting beyond five days, overall vessel delays of ten to twelve days in the same update, and the report's own expectation that congestion will persist while carriers work through backlogs. If you have boxes routed through Manila, treat 4.48 days as your planning number rather than as a worst case. A plan built on the average will be wrong on every off-window sailing, and off-window is where your urgent cargo tends to be.

One more thing that a single week of numbers hides. The update mentions Typhoon Bavi twice, once for the mainland Chinese ports and once for the bunching it caused at Manila, and it mentions flooding risk at Laem Chabang in the same breath as rising climate exposure for export infrastructure. July to September on this map is typhoon season. That means the congestion you are reading about is not a one-off to be waited out. It is the shape of the next eight weeks, and the maintenance windows a port would normally use to recover are competing with the weather for the same calendar. Plan the quarter, not the week. A routing decision made for July and revisited in October is a different decision from one made for the season.

Let me put numbers on it, with the assumptions stated. Assume you move 30 boxes a month through this corner of the map, 20 through Port Klang and 10 through Manila. Take the middle of the Port Klang range, two and a half days, and apply it to the 20 boxes: 50 container-days. Take the Manila seven-day average, 4.48 days, and apply it to the 10 boxes: 44.8 container-days. Together that is 94.8 container-days a month in which your cargo is not where your plan says it should be. Assume you hold 21 days of cover. That average delay of 3.16 days per box has just consumed 15 percent of your buffer in a single month, without you changing a single order.

Run it for two months and roughly a third of your safety stock is gone, and the replenishment order that would restore it has to sail through the same congested water. If you need a figure for the internal memo, multiply those 94.8 container-days by your own cost of a container-day. I am not going to quote you a freight rate here. That is not what this column is for, and a rate you read in a column is not a rate you can negotiate against.

The dredging date is the item in the update that most people will skip, and it is the one with a calendar attached. The update gives a reason beside the date: to help ease congestion, the planned maintenance dredging has been moved to 3 to 12 August. I read that the way I think it is meant, that dredging gear working in and around the berths is itself a constraint, and moving it out of the way buys the port some room now. But do not miss the second half of the sentence. The work now falls inside the first third of August, it will not be moved twice, and August is when your peak-season volume starts to build. Put 3 August and 12 August on your own calendar as dated events, not as somebody else's operational footnote.

Underneath all of this there is a longer pattern I think is worth saying out loud. Hubs in this region have always competed on one simple axis: who can turn a ship fastest and hold a box cheapest. For most of the last thirty years Singapore won that contest so consistently that its neighbours competed for the overflow. What has changed is not that Singapore got worse. It is that the overflow got big enough to matter. When the spill is large, the spill port becomes a primary node whether or not it was planned as one, and it inherits primary-node congestion without necessarily having primary-node buffer. That is the structural read on Port Klang at 90 to 95 percent, and it is why I treat this as a network story rather than a bad week at one terminal.

Here is what I would do, and I would start this week. By 24 July, have your bookings team pull forward every August shipment that can be pulled forward, targeting 80 percent of August volume tendered before 31 July, because the dredging window opens on 3 August and anything on the water inside that window has no slack to fall back on. By 31 July, re-baseline your transit promises: add 2.5 days to anything routed through Port Klang and 4.5 days to anything through Manila, and tell your customers the new number before they discover it themselves. The trigger to go further is yard density. If the next update still shows Port Klang at or above 90 percent after the dredging window closes on 12 August, stop treating this as a July event and start treating it as a fourth-quarter routing question.

The alternatives are few and each has a price. Shifting transshipment from Port Klang to Singapore buys you a calmer terminal waiting time of half a day to a day against Port Klang's 1.63-day average, but Singapore's yard is already around 80 percent and it is not absorbing the region's overflow for nothing. Direct calls instead of transshipment remove a node from your route entirely, which is the only alternative that actually shortens the chain rather than moving you along it, and it usually requires volume you may not have on that lane. Holding more inventory converts a transit problem into a working-capital problem, which is a legitimate trade when your buffer is 21 days and you have just spent 15 percent of it in a month. What you should not do is keep booking the same routings on reflex and absorb the delay quietly inside your own service levels.

The pitfalls are mostly calendar problems rather than geography problems. Assuming the delay is already built into your transit promise when it is not. Booking on the assumption that the dredging postponement relieves congestion for good, when it moves the work into a dated window in August. Treating Manila's 4.48 days as an average to plan against rather than a floor to plan from. Forgetting that Yantian's crane work is a scheduled constraint and will still be there next month. And the oldest one on this map: routing by habit because Port Klang has always worked, when the reason it always worked was slack, and the slack is gone.

It helps to read the whole update as one map rather than as a list of countries, because the pattern only shows up when the ports are put next to each other. Kobe and Yokohama both sit at a 1.2-day seven-day average, with Yokohama still reporting waiting around 1.5 days on cargo overflow and continued bunching. Semarang is near one day of waiting, with berth congestion and average vessel delays around two days, and its yard has climbed to 82 percent. Laem Chabang is around one day. Singapore is 1.25 days, with terminal waiting down to half a day to a day.

Port Klang is 1.63 days. Manila is 4.48 days. Line those up and you have a gradient running from the calm end of the map to the strained end of it, and the two ports at the strained end are the two that carry the same update's warning about bunching. That is not coincidence. Bunching is what happens when a schedule breaks upstream and every delayed ship arrives at once, and the ports with the least yard are the ones that turn a bunch into a queue.

There is a second-order effect of a full yard that I rarely see mentioned, and it bites exporters before it bites importers. Yard space is not carved up separately for boxes leaving and boxes arriving. When a yard is at 90 to 95 percent, the terminal has to decide what it will accept through the gate at all, and the boxes it defers first tend to be outbound ones, because a box waiting to be loaded can wait somewhere else. If you ship out of Port Klang, the delay to worry about is not only the two-to-three day vessel delay the update reports. It is the gate appointment you could not get, the cutoff you missed by six hours, and the sailing your box was rolled to. Ask your forwarder this week what the earliest gate-in window they can actually secure is, and compare it with what they could secure in June. If the window has shortened, your export cutoffs have moved even though nobody announced it.

Practical detail, because the distance between reading this and acting on it is one phone call. Ask three questions of whoever books your cargo, and ask them this week rather than at the next review. Which of my sailings in the next six weeks are transshipped through Port Klang, and how many boxes sit behind each one. What is the alternative routing on each of those sailings, measured in transit days rather than in money. And what is the latest gate-in I can be given for my export boxes out of Port Klang in the week of 3 August, given that dredging is scheduled then. Those three answers will tell you more than another week of updates will. If the answers are slow in coming, that is itself an answer, and it usually means nobody has mapped your exposure to this port at all.

The strait is worth a sentence of its own, because it is the reason none of this stays local. Everything moving between the Indian Ocean and the manufacturing coast of Asia passes through a channel with no spare width, and the ports along it are the valves on that channel. A valve running at 90 to 95 percent of its storage capacity is a valve with no surge tank behind it. Over a decade, that is the kind of detail that decides which hubs win share and which lose it, because shippers and carriers remember which port held their cargo for three days in a bad month, and they quietly re-plan around it for years afterwards. Reputations on this map are built slowly and lost inside a single season, and the season is running now.

On the inventory side, be careful how you rebuild the buffer, because the obvious move is the wrong one. The instinct after a month like this is to place one large replenishment order to refill what was spent. That order arrives as a single lump, lands in a port whose yard is already at 90 to 95 percent, and turns a shortage problem into a storage and dwell problem at precisely the wrong address. Spread it instead. Four smaller orders across four weeks will clear a congested terminal faster than one large one, and each gives you another reading on how the port is actually performing. The target I would set is plain: no single replenishment order should represent more than 25 percent of the buffer you are restoring, and every order should be tracked from gate-in to delivery, so that by the end of the quarter you are working from your own measure of this port rather than from somebody else's.

One caution, because it would be easy to read this as advice to abandon Port Klang, and that is not what I am saying. Port Klang earned its place on this map over fifty years, and one congested fortnight, even one at 90 to 95 percent yard density, does not undo that. What I am arguing for is narrower and more useful: stop treating any single port on this strait as a fixed input to your plan. It is a variable, and the update gives you a fresh reading on that variable most weeks. Build the plan so that a two-to-three day delay at one node is an adjustment rather than an emergency, and it will matter much less which port is having the bad month. The shippers who suffer through periods like this are almost never the ones who picked the wrong port. They are the ones who picked a port and then stopped looking at it.

Step back from the week and look at the decade. Ports do not fail, they change function. The Port Klang of twenty years ago and the Port Klang of this update are the same stretch of coast doing a different job inside a denser network, and the delays it is showing are the sound of a hub carrying more than it was built to absorb, during a maintenance window it could not keep. Somewhere in the next few years this corner of the map gets more berth, more yard, more rail, or it loses share to whichever neighbour built first. I do not know which, and anyone who tells you they do is selling something. What I know is that the network got tighter, the buffer got thinner, and a two-to-three day delay in July is the shape of that. Plan for the tighter network rather than for the July week, and neither one will surprise you.

↑ Back to the key points

— By Tomás Ferreira