Why your acetylene yield per kg of carbide is falling and where the gas is going

A falling yield per kilogram of carbide is the earliest warning an acetylene plant gives. It moves weeks before any alarm trips, which is why the KIGL shift log computes it every shift. When it moves, operators usually look at the generator first. The numbers say they should look at the carbide first: the difference between coarse and fine grades on KIGL's schedule is about 13 %, two days in an open drum can cost 20–40 %, and those two together outweigh every loss inside a well-run plant. This page ranks the causes by size and gives the check for each.

Set side by side, the figures show that storage and grading dominate: inside a well-run plant, the 13 % grading spread is worth more than every in-plant loss path combined. IS 8471:2003 cl 5.2 sets the generator's own target, that recovery of acetylene shall be not less than 90 % of the yield rate of the carbide used. A plant achieving 90 % on good carbide and 90 % on poor carbide is a good plant both times; the yield per kg will still differ by as much as the carbide does.

Yield falls with lump size because fines have already partly hydrated and are easily coated with deposits that interfere with the first free access of water. Measurements on one Russian lot (Strizhevskii, Zaitseva and Shelechnik, 1951, as compiled by S. A. Miller, 1965) gave 233, 247, 251, 258 and 268 l/kg across 2–8, 8–15, 15–25, 25–50 and 50–80 mm, a 15 % fine-to-coarse spread measured against the finest grade. KIGL's 260 to 300 l/kg schedule gives 15.4 % on the same basis, which is the same 13 % used above when expressed against the 300 l/kg figure. The standard schedules are set out in carbide grades, sizes and gas yield.

Check: compare the graded size on the drums with the previous lot. A delivery labelled 4–25 mm that is mostly at the bottom of the range will yield like 4–7 mm. Sieve a sample to IS 1040 if the fall is persistent; the method is under carbide quality tests. Ask also whether the dust content changed: dust has the lowest yield of all and foams in the generator.

Carbide reacts with atmospheric humidity. Two days in an open container cost 7–25 mm carbide 20–40 % of its gas in the measurements compiled by S. A. Miller (1965), ten times the loss across the producer's whole process. The headspace of a closed drum without a CO₂ blanket can reach 26 % acetylene (Van Ness, USP 2,899,052, 1959), so a drum that has hissed on opening has already given some of its gas to the store. See carbide storage.

Check: how many drums were opened at the start of the shift, how long they stood, whether any part-drums were carried over, whether the store has a leaking roof or a water pipe through it, and whether the monsoon has started. A fall that tracks the weather is a storage problem. Look also at the carbide as it is charged: a white powdery coating on the lumps is lime, and every gram of it is gas already lost.

Acetylene dissolves in the lime slurry essentially as in water, 0.685 g per litre at 60 °C and 0.438 g per litre at 80 °C, and leaves with every litre of slurry discharged. At 60 °C this is about 1 % of the plant's make; at 75 °C it is 0.5–0.6 %. A generator run cool, or run with excess water so that it never warms up, loses more. EIGA Doc 237 gives the reaction-water temperature for maximum yield as 70–80 °C and a water-to-carbide ratio of about 8:1 within a range of 7:1 to 10:1; KIGL's own band is 6–8 litres per kg, discussed under water-to-carbide ratio.

Aftergeneration is the related loss. Lumps coated in lime before they finish reacting are carried out with the sludge and gasify in the pit, where the gas is both lost and a hazard. Large lumps at 20 % slurry solids coat readily; the data compiled by S. A. Miller (1965) show 50–80 mm carbide coating at every temperature up to 80 °C while 8–25 mm essentially never coats. Oversize carbide, poor agitation and thick sludge all push this up.

Check: generator temperature against its set point, water meter total against carbide weight for the week, agitator running its cycle, and the pit for continued bubbling after discharge.

A leak takes gas between the generator and the cylinder and shows as a gap between generation and filling. Gasholder seals, hopper fill caps, screw-feed shaft seals, purifier lids and drier drain valves are the usual places; the troubleshooting register lists them by component.

Check: soap solution on every joint, never a flame. Repair only with the plant stopped and purged; a leak is never fixed while the plant is running.

A purifier that is not drained of condensate from the bottom every shift floods, channels and restricts flow. Pressure builds upstream, relief valves lift, joints weep, and the compressor runs at low suction. The same flooding causes the bed to fail the silver nitrate test early. Falling yield and a failing purifier at the same time point here.

Check: purifier drain records, pressure drop across the bed, and the silver nitrate result at the outlet.

The generator itself loses gas when it vents: pressure above the working maximum after water was not added following a cold shutdown, a safety-valve vent lever left open, or a high-pressure switch set too low. It produces less than its carbide could give when it overheats and polymerises part of the acetylene, which shows as a tar smell and brownish-yellow residue and is covered under generator overheating. And it under-gasifies when the slurry is too thick: as compiled by S. A. Miller (1965), on poorer carbide of 250 l/kg the slurry reaches 21 % solids at a 6:1 ratio, past the 18 % fluidity limit, so with poor carbide the water ratio must rise.

Check: venting records, the temperature pattern on the tank, slurry consistency at the drain, and the hopper weights against the feed timer.

Yield per kg is only as good as the two numbers behind it. Carbide is weighed per hopper charge; gas is measured at the generator meter or inferred from cylinders filled by weight. A scale reading 2 % heavy hides a 2 % loss. Calibrate the carbide scale and the filling scales on a schedule, and record which carbide lot each charge came from so a fall can be traced to a delivery. IS 1040's gas-yield test on a sample of the lot, described under carbide quality tests, tells you what the carbide should have given; the shift log tells you what it did give. The difference is the plant's loss, and IS 8471 expects it to be no more than 10 %.