A person needs about 2000 kcal of food energy per day, which, according to conventional nutritional thinking, should ideally be broken down as about 40% carbohydrate, 30% protein, and 30% fat: 800 kcal of carbohydrate (weighing 160 g), 600 kcal of protein (weighing 120 g), and 600 kcal of fat (weighing 67 g), for a total weight of 350 g of macronutrients.
But foods do not consist entirely of macronutrients, even in storage. To take an example food, soybeans are normally required to have ≤11% moisture content for food use; 172 g of cooked soybeans contains (according to one source which cites USDA SR-21) 17 g carbohydrate, 15 g fat, 29 g protein, and 1.6 g of micronutrients, totaling 63 g, but 10 g of the carbohydrate is indigestible fiber, about 16% of the total macronutrient mass. So only about 75% of dry soybeans is made of digestible macronutrients, a number which I will take as typical for other dried foods.
This means you need about 470 g of stored dried foods per day to get to 2000 kcal. This is an amount that is reasonably tractable to store; 64 years’ worth is just 11 tonnes. (If we take soybeans as a typical bulk food, they cost about US$10 per bushel, which is 60 pounds; 11 tonnes is 404 bushels or US$4040, US$370 per tonne.) But without further work, stored foods will deteriorate in nutritional value during 64 years, and may even rancidify, producing free radicals and other poisons.
The only way to prevent this nutritional deterioration is to keep the food cold in a freezer, which generally requires three components: insulation, refrigeration, and passive thermal storage (either thermal mass or phase-change materials).
In habitable regions of Earth, the ambient temperature is too high for long-term food storage, so insulation is needed to prevent heat exchange between the food and the environment. You can measure the thermal resistance around a thing in K/W, kelvins per watt — generally the heat flow is proportional to the temperature difference. This thermal resistance is proportional to the thickness of the insulation, inversely proportional to the surface area through which the heat is flowing, and proportional to the insulance of the material. Insulances range from 1000 W/m/K for diamond down to 20 mW/m/K for silica aerogel. Water’s insulance is 580 mW/m/K, though it rises to 2200 mW/m/K when frozen, and commonly-used low-temperature insulating materials range from straw at 90 mW/m/K to styrofoam at 33 mW/m/K. (Refractory insulators are trickier; firebrick is normally around 500 mW/m/K). Minerals are less insulating; quartz is 3 W/m/K, sandstone is 1.7, graphite is a metal-like outlier at 168, copper is 400. Water-saturated dirt is about 2–4 W/m/K, while dry sand is about 150–250 mW/m/K, an order of magnitude better.
Because the heat leakage is proportional to the surface area, larger freezers are more efficient — they have less surface area per unit volume, so they lose less heat per unit of food stored, if we hold constant the insulation thickness and material. A cubic-meter sphere has about 4.8 m² of surface area; a thousand-cubic-meter sphere has 480 m² of surface area, only 0.48 m² per m³. So it will warm up ten times more slowly, and it will need only a hundred times as much refrigeration — one-tenth as much per unit of food stored.
Refrigeration is needed both to initially cool the food down and to compensate for heat lost through the insulation. In the steady state, the refrigeration needed is proportional to the temperature difference (to ambient) and inversely proportional to the thermal resistance (to ambient).
Passive thermal storage is needed to keep the food cold when the refrigeration is turned off, either because the thermostat has turned it off or because energy is not available to power the refrigeration. (A solar-powered refrigeration system might only work on sunny days, for example, and a grid-tied system won’t work during blackouts.) As with refrigeration, the passive thermal storage needed is inversely proportional to the thermal resistance of the insulation. The food itself has a certain thermal mass, typically about 2kJ/kg/K, and it can be supplemented with a phase-change material such as water ice, which has an enthalpy of fusion of some 79 kJ/kg. XXX XXX I think this is wrong and the correct figure is 333 kJ/kg. Adding phase-change materials increases the volume that must be insulated, which increases the surface area, requiring a greater depth of insulation over a larger area. Consequently, there is a tradeoff between resiliency to power outages and energy efficiency.
Water-saturated dirt is not a very good insulator, with insulances of around 3 W/m/K, about 30 times worse than straw and 90 times worse than styrofoam. But you can get 30 meters of water-saturated dirt insulation just by drilling a 30-meter well, while the equivalent meter of straw insulation requires you to buy and then move around cubic meters of straw. Better still, dry sand is only a factor of 3 worse than straw, so it’s easy to get an enormous thermal resistance in dry areas. Straw is by far the cheapest decent insulation material; it costs at a minimum about US$25 per tonne just to replace its fertilization value and bale it; in places the cost can be two or three times that, depending on competing demand for livestock feed.
If you were to try to build a mound of dry sand aboveground in order to get dry-sand thermal insulances in an otherwise swampy area, you’d need the sand to cost less than about US$10 per tonne to make it cheaper than straw. Construction sand for concrete costs about US$20–40 per tonne delivered here in Buenos Aires, but if you can instead just move earth around locally (producing a pit and a mound), you could conceivably improve on that. However, excavation costs are typically close to that same US$20–40 per tonne using construction-scale hydraulic earthmoving equipment.
So probably straw insulation is the