At less than full volume, the lift force of the VBP platform is directly proportional to the number of moles of lift gas in the lift cells. For a sealed zero pressure balloon in thermal equilibrium, lifting some weight W will always require n.w. moles of lift gas. The number of moles trapped is independent of cell pressure, volume or temperature, and the amount of air displaced ultimately varies only with temperature. A balloon can get by with less than n.w moles of lift gas only if it can heat its lift gas above the temperature of the surrounding air. If the lift cells trap less than n.w moles, a platform in thermal equilibrium will fall.

Changing a lift cell requires deflating it. If the lift gas inside is vented to the atmosphere and not replaced, the platform has less than n.w moles and drops. The lift gas remaining in the other cells won't stop this drop down until the temperature starts to rise again below the thermocline, at which point it is well into the layer of high force winds. Thus, the lift gas must be replaced before the thermocline.

As the platform descends, the other lift cells will be compressed. However, this won't accelerate the platform's fall because their individual lift force will not decrease with compression. As long as the pressure differential is zero and the balloon is roughly in thermal equilibrium, the lift force exerted by a compressed lift cell remains the same. This isn't enough to keep the platform at altitude, of course, but because the volume of gas in them is reduced, there is extra room in the lift cells for more gas. A partially compressed lift cell can be reinflated to full volume at a lower altitude, adding extra moles of lift gas to the configuration.

If it still retains both its integrity and concentration of lift gas, gas from the cell to be deflated can be slowly pumped into another cell. In this way, the total number of moles in the platform's array won't change, and it will not lose significant bouyancy. The free space under the shroud at full inflation is just large enough to raise a nearly empty new lift cell (carrying just enough gas to raise itself and little else) between the old ones, then pump out the old one into the new one. Doing this without raising a new cell to hold the gas will require compressing the gas in other cells beyond the zero-pressure state, which will cause the platform to drop for a time. However, eventually the cells will be compressed back into a zero pressure state at a lower altitude, at which point the balloon will stabilize again. This can be done slowly enough that the pressure differential never grows too high for the balloons to handle, and the gas can be vented back into the new cell. A slow, controlled descent over short distances can be done in this fashion.

Compressing a whole lift cell worth of gas is a job of work at just 50mb. Fortunately, the pressure inside a cell need only be raised a tiny amount to cause a change in altitude. Less than a millibar will serve. This can be controlled rapidly with a relatively small compressed air bladder, or accomplished over days using a small compressor.

Trouble comes with poisoned and ruptured lift cells, ones whose gas can't simply be shifted about but must be completely replaced. Lift cells are so huge compared to their typical leaks and losses that replacing them is almost wholly different from simply maintaining their mixture day to day. Poisoned or dilluted lift gas would be the problem encountered most often. If an intact lift cell receives a regular trickle of new lift gas while used gas is slowly vented from the bottom, then dillution should only be a problem if there is a hole in the lift cell fabric for air to diffuse in. If its low enough to the keel, we could even consider patching the rent fabric. A cell with a hole in it needn't deflate unless there is a path for the lift gas to rise up through the hole, but if that happens the cell will eventually deflate and must be considered ruptured. Loss of lift through poisoning of the lift gas is a slow process that may happen over days or weeks, but a ruptured lift cell may lose lift catastrophically. A poisoned cell will usually be able to wait for a crew to go out and replace it. A ruptured cell may require rapid action that must be taken from inside the hab because there simply isn't time to send a repair crew out before the platform falls from the sky.

For the kind of emergency replacement we'd need upon completely losing a lift cell, there are two options: a regular schedule of complete replacement of a liquid hydrogen supply, ultimately carrying up as much liquid hydrogen as any other supplies; or the inclusion of sufficient ionic hydrides and water for emergency inflation. Both have their problems.

Liquid hydrogen stores well if it can form a slush with methane or some other gas. The large amount required for a cell (>100kg) should retain its chill well over time. However, there will still be significant losses over the course of a few months. While a few dewars full of liquid hydrogen may be just the ticket for a planned replacement of a lift cell and even for making up lost gas on short notice, a single load can't stay in storage for months just in case we have an emergency.

Solid ionic hydrides, particularly calcium hydride (CaH2) and magnesium hydride (MgH2), are typically used to provide hydrogen sources for weather balloons and other uses at remote locations. Upon adding water, they form a hydroxide and release hydrogen gas & water vapor. The total amount of hydride and water required is several tons for a single cell, with the magnesium hydride being lighter by half but still at least 1.5 tons. Their weight is discouraging, but we would end up lifting more liquid hydrogen than hydride over time. The primary advantage of a store of ionic hydride is that it would stay stored for years and all that we would have to do is punch a button to use it. The primary disadvantage is that the weight wouldn't end when we used it. Both hydrides leave behind useless byproducts that weigh almost as much as they do.

The primary advantage of liquid hydrogen is that our entire replacement store can be carried up on the elevator, and we'll never have to carry any down. The primary disadvantage is that we would have to carry up our entire replacement store so often because there's never any to carry down.

All in all, I believe we should choose the lesser of two evils and go with replacing lift cell gas from a reserve of liquid hydrogen. It's normal evaporative losses can be tapped for hydrogen replacement to the cells, so we can at least get some use from the vast amounts we'd otherwise vent. Keeping a large stash of it in preparation for emergency replacements could thus reduce or eliminate the need for deriving hydrogen from life support processes to maintain the gas mixture in the cells.