Why an acetylene cylinder delivers less gas in cold weather and what to do about it without a flame

Every winter, a cutting station that worked perfectly in October starts complaining in December that the acetylene cylinders are short, that the flame goes thin and oxidising halfway through a job, and that the regulator gauge drops faster than it should. The cylinders are not short. The acetone inside them is cold, and cold acetone holds acetylene more tightly than warm acetone does. This page explains the mechanism with the published figures, shows why drawing harder makes it worse, and sets out the only remedies that are safe.

A dissolved acetylene cylinder holds its gas in solution in acetone soaked into a porous mass. How much acetylene a litre of acetone holds at a given pressure depends on temperature, and it moves the opposite way to what most people expect: colder solvent, more gas held, less gas released.

The C.P.I. 1968 safety code gives the Bunsen solubility coefficient of acetylene in acetone near atmospheric pressure as a temperature series, and it is the cleanest statement of the effect in the literature:

Read upward, the table says what a winter morning does. A cylinder filled and settled at 15 °C, then left in a yard at 5 °C overnight, has acetone that grips the gas harder still. The pressure it shows is lower, the rate at which it can push gas through the mass and out of the valve is lower, and the torch sees a weak supply. The C.P.I. code attributed the slower winter output to exactly this rise in solubility. Nothing has been lost. Weigh the cylinder and the gas is all there, which is why contents are checked by weight and not by gauge; the arithmetic is on the page on why cylinder pressure does not show contents.

The same physics is what makes acetylene storage work at all. Acetone holds roughly 25 volumes of acetylene per volume per atmosphere at 15 °C, which is the whole basis of the dissolved acetylene cylinder, explained on why acetylene is dissolved in acetone. Cold weather is the solvent doing its job slightly too well.

Taking acetylene out of solution costs heat. The C.P.I. 1968 code gives the heat liberated on dissolving one gram of acetylene in acetone under cylinder-filling conditions as about 160 calories; the same quantity is absorbed when the gram comes back out. Draw a kilogram of gas in an hour and the cylinder must find 160 kilocalories from somewhere. In summer the steel shell takes it from the room. In winter the room has less to give, the acetone cools, solubility climbs, and the delivery falls further. The cylinder refrigerates itself.

The visible symptom is frost or condensation on the lower part of the shell during a heavy draw. The invisible one is acetone blowing: when gas is demanded faster than the mass can release it cleanly, liquid acetone is entrained as droplets and carried into the regulator and hose. Acetone in the torch gives a sooty, unstable flame, attacks nitrile seals and spoils a weld. EIGA's 2025 filling document records that acetone losses in customer use, about 60 g per kg of acetylene withdrawn in temperate climates, rise to 100 g per kg in warm climates and higher still with spitting; over-drawing a cold cylinder produces the same spitting in winter. The effect on weld quality is covered on the acetone carry-over page.

This is why there is a withdrawal limit at all. CGA G-1 sets it at one-seventh of the cylinder's contents per hour for welding and cutting, one-tenth for intermittent solvent-sensitive work and one-fifteenth for continuous use. KIGL specifies one-seventh. The 1965 engineering literature, as compiled by S. A. Miller, recorded that a monolithic-mass cylinder would sustain 20 % of its contents per hour only until 65 to 80 % of the charge was gone, that makers recommended a maximum of 12.5 or 14 % per hour, and that cylinders should be coupled above that rate. In cold weather the sustainable rate is lower than any of these figures, and the one-seventh rule, explained on its own page, becomes a ceiling rather than a target.

A cutting nozzle is set for a fixed acetylene flow at a fixed pressure. Starve it and the flame does not simply shrink; it goes oxidising, because the oxygen side is unaffected by the cold. An oxidising flame has a short inner cone, a purplish envelope and a hiss, and it leaves slag inclusions and porosity in a weld and a ragged, over-oxidised cut. The neutral flame that reaches about 3,100 °C, the equal-volume mixture of acetylene and oxygen in the C.P.I. 1968 tables, is lost before the operator notices. Flame settings are on the flame types page. The point here is that a cold cylinder degrades the work before it stops the work, which is why the complaint arrives as quality trouble before it arrives as supply trouble.