CMA CGM Petra cargo fire puts East Africa shippers on general average and delay alert

The Analysis

The box list I care about is short. The ship is the CMA CGM Petra, 8,048 TEU, and a TEU is a twenty-foot equivalent unit, so that is roughly four thousand forty-foot containers on one hull. A cargo-hold fire broke out on 23 August while she was in the Strait of Malacca, on voyage 0K13NW1MA of the KILIMA service, bound for Lamu. Salvors are still working the hold. CMA CGM put out a customer notice on 25 August: no crew injuries, no pollution reported. Cause not disclosed. Which holds were affected, not disclosed. How many boxes were involved, not disclosed.

Here is the part that reads differently on a customs desk than on a news desk. A fire is a physical event. General average is a paperwork event, and paperwork is where every shipper on that ship actually lives. The dates that decide your next two weeks are not 23 August. They are 7 August, when boxes were loaded at Qingdao, and 13 August, when boxes were loaded at Nansha. By the time the fire started, those bills of lading were issued, the export declarations were filed, and the manifests were transmitted to the carriers and the destination agents. You cannot un-issue a bill of lading. What you can still control is what happens at the release gate.

The wires will tell you the boxes face heat, smoke and water damage, and that East Africa sailings will be badly delayed. Both true, both secondary. The thing that will actually hold your container on the quay at Lamu, or Mombasa, or Dar es Salaam is a release condition: brokers expect the owner to declare general average, and if it does, every cargo interest on that ship provides security proportional to cargo value before the goods move, damaged or not. That is not a shipping problem. That is a document problem, and it lands on somebody's desk this week.

Let me take this one as a document, because that is the only way it turns into something you can act on before the container moves. Every box on that ship sits somewhere on one line: classification, declaration elements, inspection, duty, release. For the cargo loaded at Qingdao on 7 August or at Nansha on 13 August, the first two are already behind you. You filed an export declaration when the box left, and the HS code, the quantity, the declared value and the consignee were captured and transmitted in the manifest the carrier lodged with the destination authorities. None of that can be edited now. The file is what it is.

Be concrete about what frozen means, because shippers talk themselves out of this step. The declared value on the export side is the anchor. The quantity is the anchor. The description is the anchor. If your commercial invoice says one thing and the packing list says another, that mismatch is already sitting in a manifest somewhere, and it will surface at the worst possible moment, which is the moment an agent is deciding whether to accept your security. I have seen a release held up for less than a line of description that did not match, and the fix at that point costs days you do not have.

Inspection is behind you as well, or the box would not be on the vessel. Duty at the destination is a separate file owned by the consignee, and it is not the file that is going to bite on this voyage. That leaves release. Release is where the whole thing is now decided, and release on a general average voyage has one extra document standing in front of it that no export shipper ever plans for. That document is the average bond, and behind the bond, in most cases, a guarantee or a deposit from the cargo insurer.

Differentiate the two, because people use the words interchangeably and then get stuck. The average bond is your undertaking, as the cargo interest, that you will pay your share of the general average once the adjuster has done the arithmetic. The guarantee or the deposit is the money behind that undertaking, and it usually has to come from the insurer rather than from you. The agent at destination wants both: the signed undertaking and something that looks like it will actually be honoured. A bond with no guarantee behind it is a letter of intent, and letters of intent do not move containers off a quay.

Start with why this particular fire landed on this particular group of exporters, because the rotation explains it and the rotation is written down. The Petra skipped Shanghai and Ningbo because of typhoon and port-operation adjustments, and instead called Qingdao on 7 August and Guangzhou's Nansha on 13 August. Read that as a document problem, not a schedule problem. When a vessel skips a call, the boxes booked for the skipped port either roll to another ship or get re-routed to the next available China call, and on this voyage that is exactly what concentrated a set of China-origin boxes onto one hull. Their destinations read like one market: Lamu, Mombasa, Dar es Salaam.

The consequence is that one fire, one possible general average declaration and one release condition now sits across a cluster of exporters who have nothing else in common. Different HS codes. Different consignees. In many cases different insurers and different forwarders. The exposure is identical in form and completely different in size, and that gap between form and size is the entire story. A fire on a 8,048 TEU ship is a headline; a few hundred release files that all now need the same missing document is the actual work.

I am not going to hand you a legal conclusion on general average; the boundary conditions of that rule belong to a different desk. What I will give you is the operating fact, the one that gets buried every time under the word risk. General average is an old principle of the sea: where extraordinary sacrifice is made to save the ship and the cargo together, the property that was saved contributes to the cost. That is why the sentence the brokers are using matters so much. Security proportional to cargo value. Not proportional to damage. Proportional to value. A box that never saw a flame still has a value, and a value is a contribution base.

So the choke point is not the fire report. The choke point is a stack of paper that has to be internally consistent before any agent at the destination will issue a delivery order. Bill of lading, commercial invoice, packing list, certificate of origin where the destination asks for one, insurance certificate, and on top of that the average bond and the insurer guarantee. The release question is narrow and brutal. Has the line's agent received an acceptable guarantee, and does the value on it match the file. Until that answer is yes, the cargo stays where it is, and the quality of your relationship with the forwarder does not change it by one hour.

Here is the piece I have not seen in the coverage, and it is the one that actually costs money. Everyone is reporting the fire and the delay. Nobody is reporting that the number the adjuster will use to fix your contribution is the value you yourself declared. The contribution base is built off declared cargo values, the invoice value and the insured value on the certificate. If your invoice was written down, or if the insured value sits below CIF, then the insurer's guarantee only covers what the policy says it covers, and the gap between your real exposure and the covered amount comes out at the destination counter, in cash, before anything is released. The fire did not create that gap. It revealed it.

There is a second, quieter version of the same trap inside the policy wording. Most shippers have never read the general average clause in their own marine cover, and cannot say off the top of their head whether contribution is covered at all, or whether the policy is written on a full-value basis. On a normal voyage nobody ever needs to know. On this voyage, that single clause decides whether the security is a phone call to the insurer or a wire transfer out of your own account.

Who gets hit, and how hard. A shipper with twenty FEU on that voyage, where FEU is a forty-foot box, has a large number but also a treasury, a broker who has done this before, and a marine policy written by someone who has thought about general average. The shipper with three boxes is the one I watch. The security requirement is the same in form, the paperwork is the same weight, and a three-box exporter does not have a person whose job this is. Then there is the consignee at the far end, and the consignee is often the first to find out, because the consignee is the one who walks up to collect and gets told that nothing moves until the bond is lodged. And remember that this book concentrated on three East African destinations: small importers, in many cases a buyer on an FOB or CIF contract who now has to front security that was never in the budget.

One more group that people forget, and it is the group the word saved is aimed at: the shippers whose cargo was not damaged at all. They lost nothing to the fire, and they still contribute to the cost of saving the ship. That is not a loophole someone forgot to close. That is what saved property means. If your mental model is that general average is a damage claim, you have the model backwards, and that misread is going to cost you real money at a counter seven thousand kilometres from where the fire happened.

There is also the group nobody writes about at all, the boxes that came off the ship. On a casualty voyage some cargo gets discharged and sits in a terminal or a container freight station while the inspection and the salvage work continues. Those boxes are not in limbo; they are in a file. They need a release instruction from wherever they were landed, and that instruction is issued by a different agent in a different place than the one you have been emailing about the sailing. If your consignee is told to expect the box at Lamu and the box is actually sitting elsewhere, the delivery order is not late, it is aimed at the wrong port. That is the kind of detail that costs a week, and it is only visible if you ask the carrier the right question in writing.

On timing, think in the units of a documentary file, not a sailing schedule. The fire is 23 August. The customer notice is 25 August. Today is 29 August. The salvors are still on site, so the physical picture is not closed, which means the adjuster's picture is not closed either, which means the security number is not final. What you can bank on is the sequence, and the sequence stacks rather than overlaps: a general average declaration, an average adjuster appointed, a security demand issued to each cargo interest, the security lodged, the delivery order issued, the release, then the inland leg to the consignee. Every one of those steps is measured in days, and none of them runs in parallel with the others.

That sequence is why waiting is the expensive choice. A shipper who decides to wait until the adjuster's letter arrives has already burned the window in which the insurer letter could have been drafted, the value reconciliation could have been done, and the consignee could have been warned. The physical voyage takes weeks to recover; the document chain has to run ahead of the ship, not behind it. On the schedule itself, when the vessel actually turns up at Lamu, that is Vivian Zhao's column and her numbers; my interest is narrower and more boring, which is simply which document the line's agent will ask for, and on what date.

Let me put an illustrative number on it, with the assumptions written out so nobody mistakes this for a quotation. Assume you had three forty-foot boxes on that voyage. Assume each is invoiced and insured at eighty thousand dollars CIF, so your declared value is two hundred and forty thousand dollars. That declared value, not the freight you paid, is the number the adjuster looks at. Suppose, purely for illustration, that the security is set at two percent of cargo value; the real figure comes from the adjuster, not from me. Two percent of two hundred and forty thousand is four thousand eight hundred dollars, and that is money or a guarantee you place before anyone hands you a delivery order, before you even know whether your own boxes were touched.

Now stretch the shape of it. The shipper next to you on the same voyage with ten boxes at the same declared value is looking at sixteen thousand dollars. Same ship, same fire, same notice, and a spread of more than threefold at the counter, purely because of box count and declared value. That is the arithmetic of general average, and it is exactly why I keep saying the document, not the damage, sets your cost. The freight you paid is a rounding error next to the cash you may have to place.

Where does this flip? Two places, and both are worth tracking. One is if the owner never declares general average, if the casualty is handled as a hull claim and the cargo moves on an ordinary delivery order. In that case everything above collapses back into a delay file and the only currency is time. The brokers expect a declaration; an expectation is not a declaration, and the gap between the two is precisely the window you should be using right now. The other is if cargo from the affected part of the ship is landed or transshipped en route.

If that happens, the delivery order is not issued at Lamu at all, it is issued wherever the cargo comes to rest, with a different agent and a fresh value check attached. Which of those two you are in depends on facts that, as of today, have not been disclosed. Affected holds and box count are both still open. That is not uncertainty for its own sake. It is a specific list of questions, and those questions go to your carrier in writing.

Alternatives, and the traps sitting next to each of them. You cannot move this cargo to another ship, so the levers are documentary and commercial. On the documentary side, get the guarantee structure right before the ask arrives, and reconcile invoice value, insured value and CIF value onto one page. If those three numbers do not agree, the adjuster's number and the insurer's number will not agree either, and the difference is yours to fund. On the commercial side, talk to the consignee now, not when the agent calls. An FOB buyer may believe the risk is the seller's; a CIF buyer may believe the opposite; either way the demand lands on the cargo interest at destination and somebody has to have the money ready. Read your delivery terms before you read your email.

The traps, in the order I see them most often. Assuming no damage means no exposure, which is the single most common misread of general average. Assuming the forwarder can post the bond for you, when the bond has to be backed by the cargo insurer or the cargo interest. Assuming the destination broker can stand in for the adjuster, which he cannot, because the adjuster is appointed by the process and not by you. Assuming the delay is counted from the fire date, when the part you actually control is counted from the hour the security demand lands. And the quiet one: making sure the certificate of origin and the invoice match the release file, because a mismatch there turns a normal release into an inspection, and an inspection adds its own days and its own cost on top of everything else. This is the part where the paperwork looks beautiful in the folder and the floor does not care.

Now the practical part, in the order the clock demands it. Today, 29 August, before close of business wherever the cargo owner sits: written confirmation from the carrier or its destination agent on whether general average has been declared on voyage 0K13NW1MA. Same day: get invoice value, insured value and CIF value onto one page and find the gaps. Within forty-eight hours: written confirmation from the marine insurer on whether contribution is covered and whether they will issue the guarantee, because the insurer letter is the slowest item in the whole file. Within seventy-two hours: brief the consignee at Lamu, Mombasa or Dar es Salaam with the number they may have to front, so the cash is ready the moment the demand arrives. And keep one named person on the adjuster file from the day it opens, not the day the container is discharged. This cargo can be saved. It just will not save itself, and the paperwork has to move before the ship does.

↑ Back to the key points

— By Derek Xu