Durban port calls blow past 12 days as SAFLA and RFA demand Gateway Terminal recovery plan

The Analysis

Twelve days. That is where independent monitoring put the average port call at Durban by late August, against under five days at the end of June, and I have been sitting with that number for a week because I am the one who pays for every day in it. Not in the abstract. In demurrage, in detention, in units that are not on the shelf while the season they were bought for is being sold.

Let me say which chair I am in before anything else, because this story is being told from a lot of chairs at the moment and they do not reach the same place. I buy. I do not operate the terminal, I do not run the system that was cut over in mid-August, and I am not going to spend a paragraph deciding who is at fault. What I do is commit to units six to ten weeks before a customer walks in for them, and then answer for the margin on those units when they land. Twelve days of port call does not appear in my world as a port statistic. It appears as seven extra days of inventory I own, as a bill that accrues while the box is not moving, and as a shelf date that either holds or does not.

The shape matters more than the level, so let me lay the shape out. In July vessels averaged eighty hours at anchorage and one hundred and six at berth, which is one hundred and eighty-six hours, or seven point seven five days. By early August it was one hundred and sixty-six hours at anchorage and about one hundred and sixteen at berth, two hundred and eighty-two hours, or eleven point seven five days. That is the twelve days. Alongside it, weekly throughput fell twenty-six percent after the mid-August Navis N4 cutover, and monthly berth calls dropped from thirty-four in May to nineteen in August. Three numbers, one message: the terminal is producing less, not my goods arriving in more volume. And the terminal in question carries more than forty percent of South Africa's containers, which is the part that removes my easiest exit.

Put it against a real book, because a port statistic only becomes a decision when it is next to your own units. Assume a consumer goods importer bringing sixty containers a month in through Durban, seven hundred and twenty a year. Assume nine hundred units per container, so fifty-four thousand units a month and six hundred and forty-eight thousand a year. Assume an FOB value of fifty-four thousand dollars a container, which puts unit FOB at sixty dollars exactly. Assume landed cost per unit before any delay charges of eighty-two dollars forty, which is my assumption covering freight, duty and inland together.

Assume a retail price of one hundred and twenty-nine dollars ninety-nine. Contribution per unit is therefore forty-seven dollars fifty-nine, and gross margin is thirty-six point six percent. Annual retail value on that book is eighty-four point two three million dollars, annual gross profit is thirty point eight four million, and monthly landed flow is four million four hundred and fifty-two thousand dollars. Every figure from here is arithmetic on those assumptions plus the published port numbers. None of it is a fact about anybody else.

The first cost is the one finance sees, and it is the smallest. Seven extra days of port call is seven extra days of pipeline. On a monthly landed flow of four million four hundred and fifty-two thousand, seven days is seven over thirty of that, which is one million thirty-eight thousand eight hundred dollars of inventory that did not exist in June. At a carrying cost of eleven percent a year, my assumption for this market, that is one hundred and fourteen thousand dollars a year, or about nine thousand five hundred a month. It is real and it is not the number that will hurt me.

The second cost is the one planners forget, and it is bigger than people expect. Safety stock is calibrated to lead time, and lead time just changed. Assume my baseline order-to-shelf was the five day port call plus three days inland, my assumption, plus the average gap between berth calls. In May there were thirty-four calls in the month, so thirty over thirty-four is nought point eight eight days, and the baseline is eight point eight eight days. In August there were nineteen, so thirty over nineteen is one point five eight days, and the current figure is twelve plus three plus one point five eight, which is sixteen point five eight days.

If safety stock scales with the square root of lead time, which is the usual approximation and my assumption here, then a baseline of fourteen days becomes fourteen times the square root of sixteen point five eight over eight point eight eight, which is fourteen times one point three six six, or nineteen point one days. Five point one days of extra safety stock. On the same monthly flow that is seven hundred and sixty-one thousand dollars of additional inventory, and at eleven percent it is eighty-three thousand seven hundred dollars a year. Nobody signed for that. It just happened when the port call doubled.

The third cost is the one with an invoice attached. Operators are reporting mounting demurrage and detention, and the release says some shippers have been pushed to air freight. I do not have a published rate, so here is my assumption and I will label it as one: assume free time runs to four days, assume combined demurrage and detention accrues at one hundred and ten dollars per container per day once free time is exhausted, and assume the seven extra days all land inside the billable window.

Seven days at one hundred and ten is seven hundred and seventy dollars a container. On sixty containers a month that is forty-six thousand two hundred dollars a month, and on seven hundred and twenty a year it is five hundred and fifty-four thousand four hundred dollars. Spread over nine hundred units, that is eighty-six cents a unit, or one point zero four percent of landed cost. Against annual gross profit of thirty point eight four million, it is one point eight percent.

Add the three. One hundred and fourteen thousand of working capital carry, eighty-three thousand seven hundred of safety stock carry, five hundred and fifty-four thousand four hundred of delay charges. Seven hundred and fifty-two thousand dollars a year while this condition holds, which is two point four four percent of gross profit on the book above, one point four percent of landed cost, or a quarter of about one hundred and eighty-eight thousand dollars. That is what twelve days costs me. Not the port. Me.

Can I price it back? This is where I want to be careful, because the easy version of this arithmetic is wrong. Seven hundred and fifty-two thousand over six hundred and forty-eight thousand units is one dollar sixteen per unit. To recover that through price, the ticket goes from one hundred and twenty-nine ninety-nine to one hundred and thirty-one fifteen, a rise of nought point eight nine percent. That is the correct figure, and it is much smaller than the two point four four percent of gross profit, because raising a price on a fixed unit cost adds the whole increase to contribution rather than a margin share of it.

Now the catch. At one hundred and thirty-one fifteen my contribution per unit is forty-eight seventy-five instead of forty-seven fifty-nine, so I can afford to lose one minus forty-seven fifty-nine over forty-eight seventy-five, which is two point three eight percent of my units, before the increase costs me more than it earns. Two point four percent of volume. That is the entire elasticity budget on this book. If the goods are the kind a customer compares against three substitutes on a shelf, I will lose more than two point four percent for nothing and the increase is a mistake. If they are the kind a customer comes in specifically for, I will probably hold the volume. So the decision is not whether to raise prices. It is which SKUs are in the second group, and that is a list I can actually produce.

Now the two operating numbers, and I want to use them for planning rather than for blame. Weekly throughput down twenty-six percent after a system cutover and berth calls down from thirty-four to nineteen are not the same fact twice. Fewer calls means fewer discharge windows in the month, and the gap between them went from nought point eight eight days to one point five eight. That sounds trivial next to twelve days, and it is not, because it compounds with the twelve: my exposure is not the port call plus a day, it is the port call plus the wait for the next window, and the second number has nearly doubled while the first has more than doubled.

The twenty-six percent tells me something different. A terminal that is moving a quarter less cargo per week than it was before the cutover is a terminal where my box is in a slower queue, regardless of how my own volume looks. I am not going to assign responsibility for that. I am going to plan against it, and the plan is a number of days, not a feeling.

Which brings me to air freight, because some shippers have already gone there and I want the arithmetic on the table before anybody else does. Assume a unit with two point eight kilograms of chargeable weight, my assumption. Assume the ocean freight line inside that unit is four dollars ten. Assume air at three dollars ten a kilo, so two point eight times three ten is eight dollars sixty-eight, and the uplift over ocean is four dollars fifty-eight a unit. Airing one unit costs four fifty-eight. Not airing a unit that misses its selling window costs me the whole contribution of forty-seven fifty-nine. So air is justified whenever the unit would otherwise be genuinely lost and would otherwise have sold, and the break-even is four fifty-eight over forty-seven fifty-nine, which is nine point six percent. If I think there is better than a one in ten chance the unit is lost without air, air pays.

The trap is that people air containers rather than units. A container holds nine hundred units, so airing the whole box costs nine hundred times four fifty-eight, which is four thousand one hundred and twenty-two dollars. Assume twenty-two percent of the units in an affected box are the ones actually at risk of missing the window, my assumption. That is one hundred and ninety-eight units, and one hundred and ninety-eight times four fifty-eight is nine hundred and seven dollars. The other three thousand two hundred and fifteen dollars is money spent flying units that would have arrived perfectly well. Assume forty percent of my sixty monthly boxes carry affected SKUs, so twenty-four boxes.

Airing all of them at four thousand one hundred and twenty-two is two hundred and forty-seven thousand three hundred and twenty dollars a month. Airing only the at-risk twenty-two percent in those same twenty-four boxes is twenty-one thousand seven hundred and sixty-eight dollars a month. The difference is two hundred and twenty-five thousand five hundred and fifty-two dollars a month, which is more than four times the entire delay charge exposure I am trying to avoid. Air is a scalpel here, not a strategy, and the unit of account is the unit, not the box.

The alternative I would rather run is inventory rather than mode. Pulling thirty days of additional cover forward costs me the carrying cost on thirty days of landed flow: four million four hundred and fifty-two thousand at eleven percent for a quarter is one hundred and twenty-two thousand four hundred dollars. Compare that with the one hundred and eighty-eight thousand a quarter the twelve-day condition is costing me, and it looks cheap. But be honest about what it buys. Pre-building solves whether the shelf is empty.

It does nothing about the box lying at the terminal, because those containers are already there and already accruing. So the test is not against the delay charges, it is against the stockouts. Assume a stocked-out unit costs me its full contribution of forty-seven fifty-nine, my assumption that the sale simply does not happen this season. One hundred and twenty-two thousand four hundred over three months is forty thousand eight hundred a month, and at forty-seven fifty-nine a unit that is eight hundred and fifty-eight units a month, or one point six percent of fifty-four thousand.

If I expect to lose more than one point six percent of monthly units to stockouts, the pre-build pays. If I expect to lose less, it is a hundred and twenty-two thousand dollars spent to move the same problem three months down the road, plus a warehouse full of goods I would otherwise have bought later.

Changing gateway is the alternative everybody mentions and almost nobody costs properly. This terminal handles more than forty percent of South Africa's containers, which means the phrase use another port is not as available here as it would be somewhere else. And I am going to measure it the way I always do, which is not in freight cost. A gateway change is a change in total days to my distribution centre, and days are what convert into safety stock and shelf dates. A routing that costs less and takes four more days is not cheaper, it is four days of inventory I now finance and four more days of exposure to whatever breaks next. Measure the days, multiply by your monthly landed flow over thirty, multiply by your carrying cost, and put that number next to the saving. If it is bigger, you have not found an alternative, you have moved the cost to a line nobody owns.

The traps in this particular situation are worth naming because they are specific. The system cutover is the first one. Assume that container status in the terminal system is not reliable while a new operating system settles, and do not schedule collection or promise a shelf date off a screen that says discharged. Confirm against your own container events, not against someone else's dashboard. The second trap is the basis on which free time is counted, because a demurrage clock that runs on calendar days and a detention clock that starts on a different event produce very different bills for the same seven days, and the difference between my assumed seven billable days and a real ten is three hundred and thirty dollars a container, or nineteen thousand eight hundred a month on sixty boxes.

Get the counting basis in writing before you argue about the rate. The third trap is allocation: demurrage and detention sit wherever the contract puts them, and if your agreement is silent on delays attributable to the terminal, they sit on you. That is not a port problem, it is a drafting problem, and it is the one on this list I can fix myself.

There is an industry ask on the table as well. SAFLA and the Road Freight Association have called for a recovery compact at Durban Gateway Terminal, and the terminal has been run by ICTSI since the first of January, which puts roughly seven and a half months between the operator taking over and the system cutover in mid-August. I am not going to use that to assign blame; new operators and new systems take time to settle and I have no standing to judge it. What I will say is that a recovery compact is only useful to a buyer if it comes with a number and a date.

Average port call in days, measured the same way the monitors measured it, published on a cadence, with a threshold that triggers something. Vague commitments to improve are not something I can plan against. Give me twelve going to eight by a stated date and I can size inventory against it. Tell me it is being addressed and I have to assume twelve forever, and assuming twelve forever is the most expensive assumption on this page.

One thing I am leaving out on purpose: what happens in the day or two after a box is released, when it is the inland leg, the appointment and the yard that decide whether the shelf date holds. Nadia Pryce works that end of the week and reads it better than I do. My job stops at deciding what to buy, how much cover to hold, and how much of this delay I have already paid for in the price.

If I were signing this as a buyer, here is what I would sign. A demurrage and detention clause that names the cause and allocates the cost by cause rather than by default, with the terminal-attributable share capped or recoverable. A service term stated in days of port call rather than in intentions, with a threshold at eight days and a review at twelve. A standing rule that air freight is approved per unit against a documented at-risk list and never per container, with the at-risk list owned by planning and refreshed weekly. And a pre-build trigger written as a number rather than a mood: thirty days of additional cover goes on when forecast stockouts exceed one point six percent of monthly units, and comes off when they do not. None of those four costs me anything to ask for. All four move money that is currently leaking out of the book without anybody deciding it should.

That is what twelve days means from where I sit. Not a terminal story and not a system story. It is seven days of inventory I now finance, a safety stock calibration that quietly reset itself, and a bill that accrues while I wait. Seven hundred and fifty-two thousand dollars a year on the book above, two point four percent of my gross profit, decided by somebody else's operating schedule. I can price for it, and the arithmetic says I have about two point four percent of volume to play with before pricing makes it worse. Or I can plan for it, which is cheaper, slower to show up, and entirely within my own control. The terminal will publish its recovery when it publishes it. My shelf dates do not wait that long.

↑ Back to the key points

— By Grace Yang