In October 2024, the FAA's William J. Hughes Technical Center published a 29-page report with a title only a regulator could love: "An Evaluation of Parameters Pertinent to Dry Ice Sublimation." The report number is DOT/FAA/TC-24/24, the authors are Lindsey Anaya and Dan Keslar, the sponsor was the FAA Office of Hazardous Materials Safety, and anyone can download it free from the FAA technical library at actlibrary.tc.faa.gov. Almost nobody in pharmaceutical logistics has read it.

I have, several times, because it is the most current public dataset on how dry ice behaves in transport conditions, and because its findings cut directly against the way most of this industry plans dry ice shipments. This post is a walkthrough: why the FAA ran the study, what the researchers measured, what they found, and the conclusion sitting underneath all the numbers that matters more than any single one of them.

Why the FAA cares how fast dry ice disappears

Dry ice is a Class 9 dangerous good under both DOT and IATA rules. Not because it is cold. Because it disappears. Solid CO₂ does not melt into a puddle; it sublimates straight to gas, and inside the closed volume of an aircraft, CO₂ gas is a crew and passenger safety problem. The regulatory exposure limit is 0.5 percent CO₂ by volume, which is 5,000 parts per million. That figure appears in FAA Advisory Circular 91-76A, and it happens to match the OSHA 8-hour permissible exposure limit for workplaces on the ground.

How much dry ice an aircraft is allowed to carry falls out of a formula: X equals the CO₂ concentration limit, times the aircraft's volume, times the number of complete air exchanges per hour, divided by the sublimation rate. The first three terms are properties of the airplane. The last one is an assumption about the cargo. And because it sits in the denominator, that assumption does enormous work: assume a lower sublimation rate and the permitted dry ice load does not grow a little, it grows exponentially.

The assumption the industry has leaned on traces back to Advisory Circular 91-76, published in 1963, which put sublimation at 1 percent per hour per 100 pounds. Container manufacturers today advertise rates below 1 percent per hour, and every fraction of a point under the legacy figure translates into more dry ice per aircraft. So the FAA's hazmat office wanted to know whether the modern claims hold up, because if they do not, the loading math built on them is wrong in the unsafe direction. That is the study. It sits alongside ICAO Doc 9284, the international technical instructions for dangerous goods by air, and it builds on thin prior work: a 2006 CAMI study by Caldwell, Lewis, Shaffstall, and Johnson on dry ice in common air cargo quantities, and a 2023 University of Florida paper by Hafner, Welt, Pelletier, and Boz on how binding agent, density, and age affect sublimation. For a substance that moves through the pharma cold chain by the ton, that is nearly the entire published literature.

What moved the needle: pellets, boxes, and wear

The method was simple and disciplined: load dry ice into containers, control the environment, and weigh everything at intervals. The candidate variables were temperature, humidity, pressure, pellet size, container design, and container wear from reuse. Three of the six turned out to matter a lot.

Pellet size first. Smaller pellets sublimate faster. The team tested diameters from 0.10 inches up to 0.68 inches, and the pattern is exactly what geometry predicts: smaller pellets carry more surface area per pound, and sublimation happens at the surface. Two shipments packed with identical weights of dry ice can burn through it at different rates purely because of what came out of the pelletizer that morning. Nobody records pellet size on an air waybill.

Container design second. The study ran three commercial containers, different materials and different capacities, under matched conditions. They did not perform alike.

The same pound of dry ice is not the same shipment. The Large Thermal container averaged 0.53 percent per hour (1.04 lb/hr on a ~200 lb load). Medium Type A averaged 0.60 percent per hour (0.26 lb/hr on ~40 lb). Medium Type B averaged 0.71 percent per hour (0.29 lb/hr on ~40 lb). Between the two medium containers, that is roughly a third more dry ice consumed every hour, in a controlled lab, before real-world handling touches anything.

Container reuse third, and this is my favorite finding, because everyone in the industry knows it and almost nobody prices it. Sublimation rates rose as containers were reused, and the report catalogs the failure modes with unusual candor: chunks of EPS foam falling out of the insulation, cardboard warping from absorbed moisture, closures losing seal integrity, the fabric of the internal dry ice pod tearing. The rate on a spec sheet describes a container on its first trip. The container on its ninth trip has no published rating at all, and it looks identical from the outside.

The altitude question stayed open

The variable most relevant to aviation is the one the study could not close. In theory, lower pressure should speed sublimation, and pressure is the one condition that changes in flight by design: FAR 25.841 allows cabin pressure altitude up to 8,000 feet, which is about 10.9 psia against 14.7 at sea level. The team compared the two pressures directly, and the initial numbers pointed one way, but a container had to be replaced partway through the test series, which confounded the comparison. The authors call the pressure results inconclusive, and they are right to.

The flight profile tests are where careless readers will get into trouble. Averaged across containers, sublimation ran 1.77 percent per hour during ascent, 0.49 during cruise, and 0.41 during descent. It is tempting to conclude that climb-out is the danger zone. But the report notes something quieter and more interesting: every test, in the air or on the ground, starts hot. Rates ran around 1.15 percent per hour over the first two hours of any test, settling near 0.50 percent per hour after about six, as the container interior acclimated. Ascent happens at the start of a flight. The spike may be acclimation wearing a pressure costume, and to the authors' credit, they say so instead of claiming a cleaner result than the data supports.

And the fifth finding, the one that should reframe how this industry thinks about monitoring: temperature and humidity, the two things everyone actually measures, mattered least. For well-insulated containers, external temperature swings between 40°F and 80°F barely registered; internal temperatures held between -100°F and -116°F throughout. That is insulation doing its job. It is also the proof, from the government's own instruments, of something we write about constantly: the internal temperature of a dry ice shipper is flat and reassuring right up until the moment the dry ice is gone. The mass drains for the entire trip. The thermometer mentions none of it.

Five findings from DOT/FAA/TC-24/24 What the FAA's lab learned about dry ice sublimation, October 2024 01 · PELLET SIZE Smaller burns faster Diameters from 0.10" to 0.68" tested. More surface area per pound means faster loss. Nobody records pellet size on a waybill. 02 · CONTAINER 0.53 to 0.71 %/hr Three containers, matched conditions, significantly different average rates. The box picks the rate, not the dry ice. 03 · REUSE Wear raises the rate EPS chips out, cardboard warps, seals loosen, the dry ice pod fabric tears. Trip nine has no spec sheet. 04 · PRESSURE Inconclusive, stated plainly Altitude test confounded by a swapped container. Ascent spike (1.77%/hr) may just be early acclimation. The aviation question is still open. 05 · TEMP & HUMIDITY Least impact of all Internals held -100°F to -116°F while outside air swung from 40°F to 80°F. The logger sees a flat line the whole way down.
Five findings, one theme: the variables that drive sublimation are the ones nobody can see from outside the box.

The finding underneath the findings

Now zoom out and look at what the data adds up to. Pellet size matters, and varies batch to batch. Container design matters, and the differences are large. Wear matters, and accumulates invisibly. Pressure might matter, but it is tangled up with acclimation. And time inside the container matters on its own: the first two hours of any test ran at more than double the settled rate. These are not independent knobs. A fresh Large Thermal packed with large pellets on its first trip and a tired Medium Type B packed with fines on its ninth are different physical systems wearing the same label: "dry ice shipper, sublimation under 1 percent per hour."

Which means any pre-shipment estimate of how long a pack-out will last is an average taken over conditions you cannot verify at the dock. The study could produce trustworthy averages because it held everything constant: same containers, characterized pellets, controlled chamber, calibrated scale, and a team whose entire job was watching the weight. No shipper, 3PL, or freight forwarder has any of that at 6 AM on a Tuesday. They have a box, a rate from a spec sheet, and hope.

None of this is a criticism of the study. It is careful work, and where the data was confounded, the authors said so instead of rounding to a convenient answer. The criticism is reserved for anyone who reads 0.53 percent per hour off a lab report or a spec sheet and treats it as an operational guarantee for a specific box on a specific lane in a specific week.

The lab had a scale. Your shipment doesn't.

Notice how every finding in this report was produced: the researchers weighed the containers. Weight loss over time is the ground truth of sublimation. Everything else, temperature, pressure, humidity, is context around that one measurement. And a calibrated scale under the container is precisely the instrument no shipment gets to carry. Once the box leaves the dock, into a truck, a belly hold, a transfer facility, the ground truth goes dark until someone opens the lid at the destination and finds out how the story ended.

So the industry runs on the two things it can do. It monitors temperature, which this study demonstrates stays flat and comforting until the dry ice is already gone. And it predicts, using rates that this study demonstrates depend on pellet size, container model, container age, and elapsed time, none of which are visible from outside the box. The gap between what a lab can measure and what a shipment in motion can measure is where cold chain product quietly dies, and this report, without ever intending to, maps that gap more precisely than anything else in the public record.

That gap is the problem we started CryoTrak to close. Our patent-pending approach is built to answer the question the FAA answered with a scale, but for a box in motion: how much dry ice is actually left, right now, in this specific shipment. Not a fleet average, not a spec sheet rate, not a lane qualification from last winter. The FAA's own findings make the case better than any pitch could: containers differ by a third, wear degrades them invisibly, the early hours run at double rate, and the temperature trace tells you nothing until it is too late. Every one of those findings is an argument that the only sublimation rate that matters is the one being measured in your box while it moves.

Read the study. It is 29 pages, it is free, and it is the best education in dry ice behavior available anywhere at that price. Then pull up your own dry ice SOP and ask one question: the sublimation rate we plan around, which container was it measured on, with which pellets, on which trip of its life? If nobody can answer, you are not planning with a rate. You are planning with an average of other people's shipments.