The Certification Papers a Lithium Battery Shipment Cannot Leave Without

A lithium battery shipment leaves the dock on two documents, and everything else supports them: the UN 38.3 test summary that covers the exact battery design on the pallet, and the dangerous goods declaration whose classification fields point back to that same summary. If the two don't describe the same battery, the carrier's dangerous goods acceptance desk stops the shipment, and no label, placard, or phone call gets it back onto the aircraft or the trailer.

The email usually lands around 4:30 in the afternoon. The packs are already wrapped, the pickup is booked for the morning, and the forwarder's note is one line: "Please send the UN 38.3 test summary for this shipment." No explanation of what they intend to do with it. After enough of those, I stopped treating it as a document request and started treating it as a cross-check. Someone is about to set that summary next to the declaration I already sent and read the two as one document.

So I don't just forward a PDF. I open both files side by side and look for the gap before they do.

The declaration is a set of pointers, not a set of facts

A dangerous goods declaration for lithium batteries is short. UN number, proper shipping name, hazard class, packing group where applicable, quantity, emergency contact, signature. Six lines. What took me a while to internalize is that none of them is an isolated fact about the box on the pallet. Each one is a pointer to something else in the file.

UN 3480 is the number when cells or batteries travel on their own, a pallet of bare packs with no host equipment. UN 3481 is when they're inside equipment or packed with it. Same pack, same cells, different number, because the classification follows the shipment configuration and not the part number. I watched a shipment sit for a day once because a pack destined to be installed into a customer's unit got declared 3480 out of habit. It was the correct number for a previous shipment of the same battery. It was the wrong number for that one.

Packing group is the tell. Lithium-ion cells and batteries in Class 9 are not assigned one, so that column stays empty, and an empty column is the correct entry rather than an oversight. When I see II or III sitting in that slot on a draft, I know it was lifted from a template built for something else, and I go back and read every other line on the form with more suspicion than I brought to the first one.

Quantity is where the rounding happens, and it's the field with the least tolerance for it. Net quantity per package is what determines which section of the packing instruction the shipment falls under, and those sections carry different per-package limits, different labels, and in some cases different aircraft rules. If the box on the scale and the number on the form disagree, the form is the one that's wrong.

The emergency contact has to be a number that gets answered while the material is in transit, not an office line that rolls to voicemail at five. Most US shippers hold an account with a registered emergency response service precisely so the number on the declaration doesn't change when the person who opened the account leaves the company. That stability is the point. A declaration carrying a stale contact number is a declaration that gets bounced.

The summary describes a design, not a part number

The UN 38.3 test summary is one or two pages. It is not the test report, and it is the single most requested document in this whole process, because carriers ask for it as a condition of accepting cells and packs, and forwarders ask for it before the carrier ever does.

What's on it is the part that matters. Identification of the cell and the battery, the laboratory that ran the work, a list of the tests in the T1 through T8 series, the results, the issue date, and a description of the battery configuration: which cell is inside, how many of them, how they're assembled, and what protection circuitry or management electronics are part of the design.

That last piece has cost me the most time. The summary doesn't certify your part number. It certifies a design, meaning a specific cell, a specific arrangement, a specific set of electronics. The declaration has no field for the battery model, so nothing on the shipping papers tethers the freight to a tested design except this summary sitting in the folder. That is exactly why the acceptance desk wants it.

The temptation is to file the summary once and treat it as permanent. It isn't. It belongs to a revision.

The swap that breaks it

Here's where it usually goes wrong, and it isn't a paperwork failure at the root. It's a sourcing decision that surfaces later as a paperwork failure.

We qualify a second cell source. Same form factor, same nominal voltage and capacity, an approved alternate on the BOM. Engineering signs it off, procurement buys it, production builds it, and the packs go onto a pallet under the same part number they've always had. Then the pallet ships against the UN 38.3 summary for the original cell, and the document in the folder now describes a battery that isn't on the pallet. The summary doesn't say "configuration may vary." It names the cell. Read the two documents together and the mismatch is not subtle.

I don't decide whether a change is small enough to be treated as the same battery type under the rules. That call belongs to the lab that ran the tests and to whoever signs the summary, and it should be documented rather than assumed. My job is to make sure the question gets asked at the BOM change instead of at the dock, because at the dock the only remaining answers are "hold the shipment" or "send it and hope."

What rides with the shipment, and what stays in the file

Some of this paperwork travels. The declaration goes with the freight. Labels and marks go on the package, the Class 9 lithium battery label, the lithium battery mark where the section calls for it, and for air, the handling documentation and the state-of-charge statement. The carrier's acceptance checklist, where they have one, gets signed at handover.

The test summary usually doesn't ride on the truck. What the rules require is that the shipper be able to make it available, and what a carrier requires in practice is that they see it before they accept the freight. Those are two different obligations, and a lot of confusion comes from mixing them up. Ground shipments get held over a missing summary all the time, even where nothing says the summary has to be inside the trailer.

If it flies, add a layer. Lithium-ion cells and batteries shipped alone by air are subject to a state-of-charge ceiling, and 30 percent is the limit most shippers work to. Aircraft restrictions on UN 3480 have also tightened more than once in the past decade. Older versions of the packing instruction that people still have saved on their desktops describe options that no longer exist. I don't re-derive any of that from memory. I confirm the current packing instruction edition and the current state-of-charge language with our dangerous goods provider, in writing, and that confirmation goes into the shipment file. The person at the acceptance desk is working from the current edition whether or not anyone in my building is.

Building the file so a question takes two minutes

What I do now is assemble a per-shipment packet and name it with the shipment reference rather than the customer name, because shipment references are what people search for when they call.

The packet is not thick:

• The signed declaration, matching the version that goes with the freight.
• The UN 38.3 summary for the BOM revision actually on the pallet, with that revision and the summary issue date written at the top of the file so the pairing is visible at a glance.
• The state-of-charge statement, for air.
• The emergency response service account confirmation that sits behind the phone number on the declaration.
• The packing list showing net quantity per package, so the quantity field can be checked against something real.

Then one page in front of all of it: an index mapping each declaration field to the attachment that backs it up. UN number, quantity, emergency contact, test summary. Four lines pointing at four places.

Nobody requires that index. It's the thing that turns a carrier question from a two-hour email chain into a two-minute answer, because the person asking usually doesn't know what they're looking for either. They know they have a rule to satisfy and a document that has to satisfy it. Handing them the map is faster than watching them build one.

What I'd tell myself five years ago

The held shipments I've dealt with were almost never missing a document. They had all of them. What they had was a set of documents that didn't agree with each other: a summary for a design that had changed, a quantity that had been rounded, an emergency number nobody updated after the last acquisition. Every one of those is a real failure, and not one of them is a filing failure.

So the question I ask before the pallet gets wrapped isn't whether we have the papers. It's whether these papers describe the thing on this pallet. Two documents, same battery. If I can't say yes to that, the rest of the file doesn't matter.