Ask ten cold chain teams how much dry ice belongs in a shipment and most of them will answer before you finish the question. "Forty pounds." "Whatever the box is rated for." "What we always use." The answer is fast and confident, and it is almost never the product of arithmetic.

Pack-out sizing is one of the last decisions in the pharmaceutical cold chain that still runs on inherited habit. The lane changes. The season changes. The box making its tenth trip is not the box it was on its first. And the number in the SOP stays at forty pounds, because forty pounds has never obviously failed. This post is about what actually sets the right amount, why the folklore survives, and why even careful math has a ceiling you should know about before you rely on it.

Where "forty pounds" comes from

Usually, from someone who left the company years ago. A number gets chosen during an early qualification, it works, it gets written down, and it becomes the standard. The people who follow it are not being careless. They are trusting a number that has a track record. The problem is that the track record is survivorship: the box arrived cold enough often enough that nobody went back to check the margin.

The industry's own baselines are older than most of the people using them. FAA Advisory Circular 91-76, written in 1963, assumed dry ice sublimates at roughly one percent per hour for every hundred pounds loaded. Modern container manufacturers advertise better, typically under one percent per hour for a fresh, well-sealed shipper. Those are useful anchors. They are not your shipment. A rule of thumb from the Kennedy administration and a spec sheet for a box in showroom condition are both averages, and an average is a poor description of any single trip.

Dry ice does not care about tradition. It sublimates, going straight from solid to gas at about -109°F, at a rate set by physics and conditions, not by what your SOP expects. So the honest version of the question is not "how much do we always use," it is "how much will this box lose between the moment we seal it and the moment a trained person puts the product into a freezer." That number has real inputs.

The five variables that actually set the number

The FAA's 2024 sublimation study (DOT/FAA/TC-24/24, still the best public dataset on how dry ice behaves in transport conditions) is a good place to see those inputs behave. Across the container types it tested, average sublimation ran between 0.53 and 0.71 percent per hour. That range is not noise. It is the fingerprint of the variables below, and each one moves your answer.

1. Transit time, measured door to freezer. Not carrier transit. Door to freezer. The clock that matters starts when you seal the box and stops when the product reaches controlled storage, and it includes every hour the shipment sits still: the tender queue, the dock, the customs hold, the gap between a signature and someone actually moving the box to the cold room. Teams size for the carrier's transit time because that is the number on the label. Sublimation runs on the total, and the idle hours are often the larger share.

2. Container quality and reuse. The FAA study found that sublimation rates climb as containers are reused. Insulation chips, cardboard warps from repeated condensation, and seals lose their bite. A shipper spec is measured on a fresh unit. The box on its ninth trip is running hotter than that spec, and nobody on the dock can tell by looking. If your fleet is reusable, "how much dry ice" depends on which specific box you grabbed, not on the model number.

3. Pellet size. Sublimation is a surface phenomenon, so more surface area per pound means faster loss. The study tested pellets from 0.10 inches up to 0.68 inches, and the smaller pellets sublimated faster. A pack-out filled from a machine making fine pellets will not last as long as the same weight of large pellets or block, even in the identical box on the identical lane.

4. Season and heat exposure. Here the study offers a useful nuance. As long as any solid CO2 remains, internal temperatures stayed remarkably stable, holding between -100°F and -116°F whether the outside was 40°F or 80°F. So heat does not ruin the payload while the ice lasts. What heat does is spend the ice faster and shrink the window, especially in the first hours and on a sun-baked summer dock. Season does not change the destination. It changes how fast you get there.

5. Lane variability. The four variables above are not fixed per lane. They shift shipment to shipment. The same origin and destination can produce a clean 44-hour run one week and a 60-hour crawl with a weather hold the next. A qualification done once, against one representative trip, describes an average lane. It does not describe the trip your product is on today.

Five variables, one answer What actually determines how much dry ice a shipment needs. None of them is "what we always use." 01 · TIME Door to freezer Count every idle hour, not just carrier transit. Dock dwell burns ice at the same rate as air. The idle hours are often the bigger share. 02 · THE BOX Quality & reuse Reused shippers lose ice faster. Insulation chips, cardboard warps, seals degrade with each trip. The spec is a fresh box. Yours may not be. 03 · THE ICE Pellet size Smaller pellets have more surface area, so they sublimate faster. Tested from 0.10" to 0.68". Same weight, different form, different clock. 04 · CONDITIONS Season & heat Payload stays near -100°F while ice lasts, but heat spends it faster. A summer dock shrinks the window. Heat changes speed, not the destination. 05 · THE LANE Lane variability The same route runs clean one week and slow the next. One qualification describes an average trip. The average lane isn't the trip you're on.
Four of these five change from one shipment to the next. A single static number cannot track all four at once.

A worked example, on a clean 48-hour lane

Let's run the folklore against the math. Take a mid-size insulated shipper packed with 40 pounds of dry ice on a lane that runs 48 hours door to freezer. Use the FAA study's rate for a mid-size Type A container, about 0.60 percent per hour, which on a 40-pound load works out to roughly 0.26 pounds lost every hour, around the clock.

Forty-eight hours at 0.26 pounds per hour is about 12.5 pounds consumed. That leaves roughly 27 pounds at the freezer door. On paper, forty pounds looks generous, and this is exactly why the folklore survives: on a clean lane, in a good box, the standard number wins with room to spare. Nobody who ran that trip has any reason to question the SOP.

Now change the shipment, not the lane. Swap the fresh box for one on its ninth trip, running closer to the 0.71 percent per hour the study saw for a more degraded container. Fill it from a machine making small pellets. Ship it in July, off a dock that sat at 85°F. Then add one unplanned twelve-hour dwell that nobody counted, because the receiving team was short-staffed. Suddenly the same "48-hour lane" is a 60-hour trip in a hotter box burning ice faster, and consumption is not 12.5 pounds. It is north of 20, and the 27 pounds of paper margin is a good deal thinner in the box.

The clean math: 40 lb pack-out, 48-hour door-to-freezer, fresh mid-size shipper at ~0.26 lb/hr = about 12.5 lb consumed, ~27 lb to spare. The real math: a reused box near 0.71%/hr, small pellets, a hot summer dock, and one unplanned 12-hour dwell can push consumption past 20 lb on the identical lane. Same SOP. Same forty pounds. Very different box on arrival.

Notice what did the damage. Not one big failure, but four ordinary variables each nudging the same direction at the same time. None of them alone would have raised an eyebrow. A slightly older box, a slightly smaller pellet, a warm afternoon, a short delay: any one is a shrug. Stacked, they quietly ate half the buffer. That is the trap in static sizing. You can pick a number that survives the average trip and still get caught on the trip where the variables line up against you, and you will not know which trip that was until you open the box.

Good static math answers a question the shipment isn't asking. It tells you what an average box loses on an average lane. It cannot tell you what this box, on this lane, this week, has left.

Why even good math still isn't enough

Suppose you do the arithmetic properly. You size for door-to-freezer time, you add headroom for reuse, you account for pellet size and season. You are now far ahead of the "forty pounds" crowd. And you have still made an estimate, because the inputs are estimates. You do not know, in advance, that this box would sit twelve extra hours. You do not know that this particular reused shipper lost a chunk of insulation last month. You do not know the pellets that came off the machine this morning ran small.

The math also fights itself in ways a flat rate hides. The FAA study found sublimation is not steady. It runs hot early, around 1.15 percent per hour in the first couple of hours while the container equilibrates, then settles toward 0.50 percent per hour after about six hours. Flight phases move it too: ascent averaged 1.77 percent per hour, cruise 0.49, descent 0.41. So a shipment with more short legs and more takeoffs burns ice differently than one with a single long haul, even at the same total transit time. A single average rate smooths over all of it.

This is the honest limit of sizing. Better inputs give you a better estimate, and a better estimate is worth having. But an estimate made at pack-out is a prediction about a future the shipment has not lived yet. The variables interact, they vary trip to trip, and the one that gets you is usually the one you could not have known when you sealed the box.

Stop estimating the ice. Measure it.

The shippers furthest along have made peace with this. They still do the sizing math, because you have to pack something. But they have stopped treating the estimate as the answer. Instead of trusting a number chosen at the dock, they watch the actual quantity of dry ice in the actual box, in real time, against the actual clock.

That change reframes the whole question. "How much dry ice does this shipment need" is a planning question you answer once, imperfectly, with folklore or with math. "How much dry ice does this shipment have left right now" is an operational question you can answer continuously, and it is the one that actually protects the product. When you can see remaining mass, the twelve-hour dwell stops being a hidden risk. It becomes a line on a chart that either clears the threshold or does not, while re-icing or re-routing is still a phone call instead of a claim.

Folklore sizing was never really about dry ice. It was a workaround for not being able to see inside the box. When the box is a sealed guess, a round number you can defend in an audit is the best tool you have. Once you can see the mass, the argument over whether it should be forty pounds or forty-five mostly goes away. You pack a sensible amount, and then you let the box tell you the truth about the trip it is actually on. That is the instrument we build, and it is the difference between hoping the estimate held and knowing it did.