Calcium carbide grades sizes and gas yield: what 300 litres per kilogram actually means

Every acetylene plant buys the same raw material and most buy it badly. Calcium carbide is sold by size and by quality, and the gas it will give depends on both: the 1931 British standard set 300 litres of dry acetylene per kilogram for lump carbide of 15 mm and above, but only 248 litres for 1–2 mm grains, and the same drum can lose a third of its yield in two days if it is left open. This page sets out what the grades are, what the yield figures mean, where they come from, and which sizes suit which generator.

Technical calcium carbide is not a pure compound. The material balance of a typical carbide furnace, as compiled by S. A. Miller (1965), gives a product of about 820 g of CaC₂ per kg, 120 g of free lime (CaO), 10 g of free carbon, and the balance as silica, alumina, ferrosilicon, silicon carbide, calcium sulphide, calcium phosphide and nitrogen compounds. The C.P.I., in its 1968 safety manual, puts commercial carbide at 80–86 % CaC₂ with the remainder mainly quicklime.

The impurities matter for two reasons. The free lime slakes exothermically in the generator and adds heat without adding gas, so a leaner carbide is hotter per cubic metre produced. And the calcium phosphide and calcium sulphide are the source of the phosphine and hydrogen sulphide that the purifier has to remove — see phosphine in acetylene and Indian carbide.

The stoichiometry is simple: 64 g of CaC₂ reacts with 36 g of water to give 74 g of Ca(OH)₂ and 26 g of acetylene. The C.P.I. (1968) gives the theoretical yield of 1 kg of pure carbide as 346.89 litres of dry acetylene at 0 °C and 760 mm, which corrects to about 366 litres at 15 °C. The C.P.I. also prints 372.66 litres at 15 °C, a figure that appears to include water vapour.

Commercial carbide never reaches that. Three things take the yield down: the free lime and other inert material that occupy part of every kilogram; the hydration that has already taken place on the surface of the lumps between the furnace and the generator; and incomplete reaction in the generator itself. The result is the trade reference figure of 300 l/kg, which Miller's mass balance equates to a carbide of about 82 % CaC₂, inside a practical range the C.P.I. gives as 265–325 l/kg and EIGA Doc 237 gives as 265–300 l/kg.

A yield figure without a reference state is meaningless. The European standards of the 1930s–1960s were all converted to dry gas at 15 °C and 760 mm for comparison; the US Federal Specification O-C-101a (1959) was written in cubic feet of wet gas at 60 °F and 30 inches of mercury; and IS 1040:1987 measures dry gas at 27 °C and 760 mm. A 300 l/kg carbide at 15 °C reads about 312 l/kg at 27 °C from the same gas. When comparing a supplier certificate with a works test, convert first.

The historical national specifications, compiled by Miller from the standards themselves, show the same pattern everywhere: yield rises with lump size and flattens above about 15 mm.

All figures are litres of dry gas per kg at 15 °C and 760 mm. The US specification is the outlier: a single figure of 4.5 cu ft/lb (about 276 l/kg) for everything from 1.4 mm to 108 mm, with 4.3 cu ft/lb (264 l/kg) for the two miner's-lamp micro-grades.

The physical reason for the slope is surface hydration. Technical carbide decomposes slowly on heating in moist atmospheres above 300 °C, and more slowly at ambient temperature, and that reaction happens on the surface of every lump. A fine particle is nearly all surface, so by the time it reaches the generator part of it has already become lime. Crushing makes fines, fines make dust, and dust is not inert: it foams, carries over and overheats in the generator. This is why the German specification tightened its dust limit to 1 % for lump grades and the proposed British revision moved from 5 % to 2 %.

IS 1040, Calcium Carbide, Technical — Specification, is written by the Inorganic Chemicals committee of BIS, not the gas-cylinder committee that writes IS 308 for dissolved acetylene, which is why the two are revised on unrelated cycles. The 1987 Third Revision was read in full for this page; it has since been superseded by IS 1040:2024, whose text KIGL has not yet obtained, so every figure below is labelled as the 1987 edition.

IS 1040:1987 defined three qualities, A, B and C, with Quality C in two grades, and twelve graded sizes: 0.5–1.5, 1–2, 2–4, 4–7, 4–15, 4–80, 7–15, 15–25, 25–50, 25–80, 50–80 and 80–120 mm. Grading uses round-hole sieves; the gas-yield and impurity tests use a 4 mm square-hole sieve. Dust is defined as material passing a 1.0 mm round-hole screen for Qualities A and B, but a 0.5 mm screen for Quality C — a detail that catches out works laboratories applying one procedure to every delivery.

The gas yield in the 1987 edition is measured at 27 °C and 760 mm, and an individual sample may be 5 % below the tabulated value provided the lot average conforms. The yield table in the scanned copy cannot be read reliably, so no IS 1040 yield figure is quoted here; it must be taken from the print copy. The standard also sets limits on the acetylene the carbide evolves: at least 99.0 % C₂H₂, not more than 0.15 % H₂S, 0.06 % PH₃ for Qualities A and C and 0.08 % for B, 0.001 % AsH₃ and 0.10 % NH₃, all by volume. Note that these are limits on carbide gas; the dissolved acetylene in a cylinder must meet the far tighter IS 308:2024 limits of 0.01 % PH₃ and 0.02 % H₂S.

Packing under IS 1040:1987 is in moisture-proof steel containers of 50 or 100 kg unless otherwise agreed, one graded size per container, marked with quality, grade, size, net and gross mass, month and year of packing, lot number and the words CALCIUM CARBIDE — DANGEROUS, IF NOT KEPT DRY with the IS 1260 danger symbol.

KIGL quotes 300 l/kg for 25–50 mm carbide and 260 l/kg for 4–7 mm, with 280 l/kg for 15–25 mm; the same schedule lists 300 l/kg for 50–80 mm and 260 l/kg for 7–15 mm. Both end figures come from legitimate published standards. The 300 l/kg top figure is BS 642:1931 exactly; the 260 l/kg figure for 4–7 mm is the proposed international figure of the 1960s exactly. The schedule is therefore a hybrid: the 1931 British figure at the coarse end and the 1960s international figure at the fine end. BS 642:1931 would have allowed 277 for 4–7 mm, and the 1960s international schedule would have allowed only 295 for 25–50 mm. The 13 % spread between the two ends of KIGL's schedule is real, is supported by every schedule in the table above, and is worth more than every other loss path in a plant combined.

Buyers often hear that carbide may deviate 3 % from its nominal yield and assume it describes laboratory scatter. It does not. The footnote to Miller's compilation of the national schedules states that the claim limits allow for a 5 % tolerance below standard in the British specification and 3 % in the other specifications, with no tolerance in the Japanese. The 3 % is the delivery tolerance below which a buyer may claim. Laboratory repeatability is a different number: the Knapsack-Griesheim microgasification method (Mewes, Chem.-Ing.-Tech. 1951) requires three determinations within 1 % of one another, and the US gas-meter method requires three 50 g determinations spread by less than 0.2 cu ft/lb.

In practice, a 300 l/kg lot bought on a 3 % tolerance has an effective floor of 291 l/kg. The German and international claim limit for 25–50 mm sat at 286 l/kg, and outright rejection, where it was provided at all, bit at 271 l/kg in the British schedule and 257 in the German. See carbide quality tests and acceptance for the test methods.

The C.P.I. (1968) records wide-range generators accepting 15–80 mm or even unsorted 1–80 mm, narrower designs taking 50–80 or 25–50 mm, medium-grain machines taking 4–7 or 7–15 mm, and 0–2 mm fines needing special precautions. By the 1960s, as compiled by S. A. Miller (1965), the tendency, at least for wet generators, was towards composite carbide of 4–80 mm or even 0–80 mm, which is exactly why the proposed British revision added 4–80, 1–80 and 0–80 mm grades with their own yields of 290, 290 and 255 l/kg.

KIGL's DA-30 carbide-to-water generator is specified for 4–25 mm recommended and 4–50 mm allowed, with about 4,100 litres of generator water. The recommended band spans the 4–7, 7–15 and 15–25 mm grades; the allowed band adds 25–50 mm. IS 1040's 4–80 mm size is a legitimate carbide grade, but it is not a legitimate feed for these machines: the limit is set by the generator, not by the carbide standard, and a buyer reading IS 1040 alone would get this wrong. See the KIGL acetylene gas plant specifications and the guide to sizing an acetylene plant.