Acetone carry-over: why an acetylene cylinder should never be run below 2 bar

Acetone leaves an acetylene cylinder with the gas, and it leaves fastest at the end of the charge and in hot weather. The mechanism is vapour pressure: whatever the acetone's vapour pressure is at the cylinder temperature, that is its share of the gas going out through the valve. At full pressure the share is small; at 1 atm it is nearly a quarter. The residual-pressure rule exists to keep the cylinder off the steep end of that curve, and the arithmetic behind it is worth seeing in full.

Hölemann and Hasselmann measured the total pressure over acetylene–acetone solutions against the acetylene partial pressure (Forschungsberichte des Landes Nordrhein-Westfalen, 1952–1959). As compiled by S. A. Miller (1965, Table II.33), the gap between the two is the acetone's own vapour pressure, so the acetone fraction of the withdrawn gas is simply one minus the ratio of partial to total pressure. Computed from that data:

Three things follow directly.

The same result for a single discharge was compiled by S. A. Miller (1965): if a cylinder were discharged isothermally at 15 °C down to 1 atm, the acetone vapour content of the gas would rise from 0.9 % molar at the start to 13.2 % at the end. In practice a discharge is rarely isothermal, because desorption cools the cylinder, and rarely goes to 1 atm, so the higher figures are not reached. That observation contains a counter-intuitive point: a slow discharge, which stays closer to isothermal, reaches higher acetone fractions than a fast self-cooling one. Slow draw does not reduce vapour carry-over. What slow draw prevents is liquid carry-over by frothing, which is the subject of the withdrawal-rate page.

Integrating the table over a typical discharge shows where the loss sits. Draw an illustrative 7 m³ cylinder at 20 °C from 15 atm down to 1.5 atm, weighting each pressure interval by the gas released in it: the acetone loss comes to roughly 0.09–0.12 kg per fill, about 1 % of a 13 kg charge, and almost all of it in the last 2 atm. Run the same cylinder down to 0.3 atm and the loss roughly triples; run it at 40 °C and it triples again.

Two independent trade figures confirm the practice. As compiled by S. A. Miller (1965), industrial users returned DA cylinders at about 3 atm absolute, and high solubility at low pressure was therefore a disadvantage in a solvent, since that gas is never delivered. Soviet cylinder practice specified a return pressure of 2 atm gauge (Strizhevskii and Gusev, Avtogennoe Delo, 1949). The modern rule of returning cylinders at not less than about 2 bar gauge (about 3 atm absolute) matches both. The gas left behind is not trivial: at 20 °C acetone still holds 94 g/kg at 3 atm against 472 g/kg at 15 atm, about 20 % of the charge, but that fraction cannot be recovered without the acetone going with it, as the pressure-and-contents page explains.

AIGA 037/16, which adopts EIGA's permissible-charge document and cites EN 12755 for the figures, gives the average solvent loss per fill cycle as 0.0075 kg per litre of water capacity for acetone and 0.00025 kg/l for DMF, a ratio of 30 to 1. For a 40-litre cylinder that is about 0.3 kg of acetone per cycle in normal service. The number of consecutive fillings before replenishment follows from the permitted solvent range divided by the loss rate, and AIGA 022/26 states the operational result: acetone bundles are typically replenished after 6 fillings and DMF bundles after 100. AIGA also records that solvent carry-over, especially acetone, occurs during rapid blow-down at the filling plant, and that a large full cylinder is blown down over 8 to 10 hours for that reason.

In India replenishment is required at every fill. IS 16271:2014 cl 3.3.1 requires the solvent content to be established from the cylinder weight after deduction of residual gas, with any loss below the stamped tare made up with the same solvent before filling; the filling inspection page gives the procedure. Rule 41 of the Gas Cylinders Rules 2016 requires the dates on which solvent was added to be recorded against each cylinder.

The Explanation to Rule 37 of the Gas Cylinders Rules 2016 states both halves of the solvent problem: excess solvent can make a cylinder hydraulically full and develop extreme pressures on heating, while a shortage of solvent makes the cylinder less resistant to decomposition caused by flashback. Carry-over moves a cylinder steadily towards the second condition. The acetylene-to-solvent ratio fixed at type approval is the backfire line; every fill cycle without replenishment moves the cylinder closer to it. Two published investigations of acetylene cylinder fires at filling plants, in Taiwan in 2005 and Great Britain in 2010, both named a solvent shortfall among the causes (the British finding was too little acetone and/or too much acetylene); the British cylinder was a 75-year-old charcoal-mass design, so its read-across to a modern monolithic fleet is limited.

Acetone vapour at the torch is also the reason a run-down or hot cylinder gives a flame that behaves differently from a fresh one. Acetone burns, and its carbon and oxygen alter the flame chemistry; liquid acetone reaching the torch disturbs the flame visibly and can make it spit. Keeping the cylinder above residual pressure, upright and within its withdrawal rate keeps the delivered gas as the regulator setting assumes.

Acetone boils at 56 °C and has a vapour pressure of about 185 mm Hg (24 kPa) at 20 °C. N,N-dimethylformamide boils at 152 °C and has a vapour pressure of about 2.7 mm Hg at 20 °C; expressed at 25 °C the comparison is about 0.5 kPa against 30 kPa, roughly sixty times lower. Applying the same arithmetic as the table, the DMF fraction of withdrawn gas at 20 °C and 1 atm is of the order of 0.4 %, against acetone's 23 %. That, and not solubility, is the engineering case for DMF: on a per-kilogram basis at cylinder pressure acetone actually dissolves marginally more acetylene (472 g/kg against DMF's 452 g/kg at 20 °C and 15 atm, as compiled by S. A. Miller, 1965), and DMF's advantage in capacity comes only from its higher density. KIGL uses DMF for its ultra-high-purity acetylene, where solvent in the delivered gas would defeat the purpose; the full comparison is on acetone versus DMF. Even with DMF, the 1965 compilation cautions that solvent loss is not always vapour loss, since liquid entrainment can occur with some porous masses, and that replenishment is reduced, not eliminated.