Chapter 22: Coke Smelting and Cheap Iron

Era span: 1709 Darby → 1850s · Difficulty: mid–high
Requires: Ch 13, Ch 14, Ch 16
Unlocks: Ch 23, Ch 24, Ch 26, Ch 27

Charcoal iron hit a ceiling: each ton of bar iron consumed the annual growth of ~10+ acres of forest. England's answer — smelting with coked coal — is one of the most consequential substitutions in industrial history. This chapter is deliberately short on chemistry (already covered in Ch 14) and long on SCALE engineering.

22.1 Why Raw Coal Fails and Coke Works

Coal straight from the ground carries sulfur (embrittles hot metal), volatiles (tar/smoke wreck furnace chemistry), and crumbles under load (chokes the air column). Coke fixes all three: heat coal in oxygen-starved ovens to drive off volatiles (~20–30 % mass loss), leaving porous carbon skeleton that burns hot, stays strong in a loaded shaft, and (with careful coal selection or washing) leaves sulfur behind.

• Beehive ovens: dome chambers batch-coked; simple, wasteful of byproducts.
• Byproduct retorts (later): capture tar, ammonia liquor, gas — the same family of chemicals as Ch 21's industry.

22.2 The Shaft Furnace Grows

Safety warning: shaft furnaces scale up blast, heat, and gas together — top-gas CO kills, slag/water contact explodes, and taller stacks drop scaffolds and tuyère blowbacks onto crews. Run with gas-tested tops, dry runners and ladles, blast-failure drills, and exclusion zones at tapping and casting; tap slag and iron only on rehearsed signals.

Coke's crush strength permits TALLER furnaces than charcoal ever allowed — taller shaft = better heat exchange = less fuel per ton. Additions that compound:

1. Water- or steam-driven blowing engines: continuous high-pressure blast replaces bellows; furnace temperatures stabilize.
2. Hot blast (Neilson, 1828): preheat incoming air through stoves heated by furnace's own exhaust gases — cuts fuel ~⅓ immediately. Recycling waste heat becomes standard design doctrine from here forward.
3. Slag control: limestone flux charges calculated per ore analysis (Ch 20 assays); tap slag continuously, cast pig iron into ingots ("pigs") for transport to refineries.

22.3 Refining Pig to Wrought

Pig iron (~3–4 % carbon) is brittle. Two historical routes:

• Finery/forge charcoal routes — obsolete once volumes grew.
• Puddling (Cort, 1784): stir molten pig in reverberatory furnace (flame passes OVER the metal, fuel never touches it) while carbon burns out; the pasty ball is then squeezed/rolled to expel slag. Brutal manual labor, but output jumped from hundreds of tons to tens of THOUSANDS annually per works. Mechanized puddling failed historically until Bessemer made it moot — don't over-invest there (Jump below).

22.4 Rolling Mills

Rolling (squeezing hot metal between powered rolls) replaces hammering for stock production: bars, plates, rails, structural shapes. Rolling rail directly (rather than hammering strap rail) multiplies railway build-out feasibility in Ch 24. Universal rolling mills (adjustable roll gap via screws) make one mill produce many sections — flexibility as capital efficiency.

22.5 Numbers That Tell the Story

Britain pig-iron output: ~17,000 t (1740) → ~125,000 t (1796, coke dominant) → >2,000,000 t (1850s). Iron price fell ~an order of magnitude across the century. Every downstream chapter's economics reprice accordingly: machinery cheap enough to build MORE machines — the self-replication threshold crossed at last.

Key threshold: iron so cheap that replacing wooden machine parts, bridges, ship hulls, and rails with iron pays back within years. When you see that arithmetic locally, scale the works aggressively.

22.6 The Jump to Steel

Jump: with hot blast + assay discipline, push experiments toward direct steelmaking rather than perfecting wrought iron: Henry Bessemer's converter (1856 — air blown THROUGH molten pig burns carbon in about twenty minutes, against a couple of hours per puddling heat) needs only phosphorus-free ores OR basic linings (Gilchrist-Thomas, 1879) to conquer everything. Full treatment lands in Ch 27; flag here that puddling-era investments should be sized as BRIDGES, not destinations.

Dead end avoided: deforestation-by-smelting. Charcoal iron at scale strips whole regions of forest (the historical Mediterranean, the Weald, China's loess margins). Coal geology, not forest stewardship, is what lets iron production compound without ecological bankruptcy.

22.7 The Coalbrookdale Papers

• Abraham Darby I leased Coalbrookdale's furnace in 1709 and coked Shropshire clod coal for smelting thin-walled castings (cookware first — the cheap-castings business funded the metallurgy). His son Darby II solved coke-furnace blowing problems and cast Newcomen cylinders (~1722) — the steam age's cylinders came FROM the coke-iron complex, closing the loop between chapters.
• Darby III built the Iron Bridge (1779) — 30 m span, first major cast-iron bridge, assembled from standardized castings: marketing monument for the family furnace.
• Henry Cort's ruin is documented incentive-system evidence: his puddling/rolling patents (1783–84) made him wealthy briefly; his financial agent embezzled government funds into Cort's works; when the agent died insolvent, the Crown held liens — Cort died poor (1800) while his patents lapsed into free use, accelerating national output. Britain's puddled-iron boom was partly built on a bankrupted inventor's orphaned IP: record it, plan around such outcomes (Ch 47).
• Neilson's hot blast (1828) emerged from Glasgow Gas Works troubleshooting (heated air improving iron-smelting furnace combustion), cut fuel consumption roughly a third, and enabled raw-coal smelting in Scotland — waste-heat recovery paying for itself within months wherever adopted.

22.8 Hot Blast: Teaching Waste Heat to Work

§22.2 names hot blast; this section teaches it, because it is the cheapest single fuel win available to any shaft-furnace economy and the founding doctrine of industrial waste-heat recovery.

Mechanism. Cold blast steals heat from the hearth: every cubic meter of air must be warmed from ambient to flame temperature by burning fuel inside the furnace. Preheating that air outside the burden means less fuel burned for the same flame temperature — or a hotter flame from the same fuel, which smelts faster and tolerates poorer fuels. The exhaust gas leaving a furnace still carries 400–800 °C of usable heat; routing it backward into the incoming blast converts a smoke nuisance into fuel.

Documented numbers. James Beaumont Neilson's patent (1828) began with blast heated only ~80–150 °C through cast-iron pipes; Clyde Iron Works' coal consumption per ton of pig fell by roughly a third almost immediately, and Scottish works adopted the practice within a few years. Higher blast temperatures followed better stoves: 300 °C common by the mid-1830s, 400–600 °C with improved pipe and stove materials. The second documented payoff: hot enough blast let Scottish furnaces run raw bituminous coal instead of coke where local chemistry allowed (self-fluxing blackband ores), deleting the entire coking step's cost. A century later Cowper regenerative stoves pushed blast beyond 1,000 °C — the same doctrine scaled.

Worked stove sizing (the arithmetic that tells you the stove pays before you build it). Assume a works furnace drawing blast at 100 m³/min and a target of heating it from ambient to 300 °C:

1. Air mass flow: 100 m³/min × ~1.2 kg/m³ ≈ 120 kg/min.
2. Heat duty: 120 kg/min × 1.0 kJ/kg·K × (300 − 20) K ≈ 34 MJ/min ≈ 560 kW continuous recovered duty.
3. Pipe area: gas-to-gas heat transfer through cast-iron pipe walls runs conservatively near U ≈ 15 W/m²·K; counterflow ΔT ≈ 315 K → area = Q/(U·ΔT) ≈ 560,000/(15 × 315) ≈ 120 m² — e.g., sixty 150 mm-diameter pipes of ~4 m length in the flue chamber.
4. Compare against fuel: 560 kW recovered around the clock is ~48 GJ/day — roughly 1.5 t of coal per day (at ~30 GJ/t) not burned for the same smelting work, which is why adoption was measured in months, not years.

Scale linearly for bigger blast; Clyde-class furnaces needed stove buildings, not pipe arrays, and got them. Two honesty notes: measured savings ran far ahead of this arithmetic. The sensible heat recovered is only a few percent of what a cold-blast furnace burned, but it arrives in the hearth as high-temperature heat, where each kilojoule displaces several kilojoules of fuel that would otherwise burn just to reach smelting temperature — most of the historical one-third saving comes from that leverage, not from the heat balance alone; and the raw-coal route is conditional — high-volatile caking coals low in sulfur on self-fluxing ores — otherwise stay with coke (§22.1).

Deployment steps.

1. Build a brick recuperator chamber in the exhaust flue; run blast air through an array of cast-iron pipes heated from outside (counterflow: hottest exhaust meets outgoing hottest air).
2. Start conservative — target blast 150–200 °C; iron pipes scale and burn above their capacity, and a burst pipe dumping cold air into a hot furnace costs more than it saves.
3. Log fuel per ton of pig weekly (§22.5 numbers) and plot on an individuals (I-MR) chart per Ch 47 §47.13 alongside the per-charge logs — the fuel drop is large enough to see immediately, which makes hot blast the perfect first exercise in measured process control.
4. Upgrade stove materials toward 400–600 °C blast once weeks of logs show stable operation.
5. When furnace output justifies gas-tight top closure, switch to true regenerators: Cowper stoves (1857) burn cleaned furnace gas on checker-brick chambers in alternating pairs, 20–60 minute swings, delivering 600–1,000 °C blast continuously.

Instrumentation and routine (the stove is a process, not a monument):

• Blast thermometer at the tuyère inlet — the stove's single most important gauge; every other number on this section keys to it.
• Furnace-pressure taps either side of the stove circuit: rising pressure drop across the pipes means fouling/scale buildup inside them.
• Swing procedure for regenerator pairs: heat-on chamber burns gas until its checker top nears material limit; swing valves; the fresh chamber gives its stored heat to blast while the spent one reheats. Miss a swing and the on-line stove chills mid-blast — the furnace feels it within minutes as falling tap temperature.
• Monthly: inspect pipe sections and checker crowns; log wall-thickness checks on the hottest pipes; retire thin sections before they burst into the flue.

Failure modes: pipe burnout between inspection intervals → stagger spare pipe sections and inspect monthly; check valves fouling with tar → fit cleanouts; blast too hot for tuyère refractory → water-cooled tuyère jackets before pushing temperatures further; missed swing cycle → manual override drill rehearsed quarterly.

Scale boundary — when hot blast is the wrong investment. The stove's recovered heat must exceed what it loses through its own walls and the pumping penalty of dragging blast through long pipe runs — runs that bellows-driven blast, with its stroke-to-stroke pressure pulsing, handles worst of all. Below roughly the Ch 14 bloomery class (single small shafts, bellows-driven), stove losses and added resistance swallow the gain: keep cold blast until a furnace consumes fuel fast enough to feed a recuperator continuously — as an ordering rule, hot blast enters the plan when the works burns multiple tons of fuel per day, not per week. Above that line the economics only improve with size, which is why every large ironworks on record converged on it within a decade of availability — and why small charcoal furnaces historically kept cold blast for decades after 1828 without being wrong to do so.

Synergy with blowing power. Hot blast does not reduce the volume of air a furnace wants, but it raises what each cubic meter does (Ch 16 water-driven blowers suffice at small scale). At coke-furnace scale the steady high-pressure blast comes from steam blowing engines (§23.5): the engine burns fuel to move air, the recuperator recycles heat into that air, and both fuel streams land in the same per-ton ledger — measure them together or optimize neither.

Key threshold: blast ≥200 °C cuts fuel per ton by about a third; below ~100 °C the gains drown in stove losses. If your logs cannot show the difference, your instrumentation is the first repair.

Worked fuel ledger (per ton of pig, rounded from published Clyde-era returns): cold blast consumed roughly 8 t of coal per ton of pig; hot blast at ~300 °C cut this toward 5 t — three tons saved per ton of iron, call it ~90 GJ of fuel per ton of product at coal's ~30 GJ/t. A works casting 50 t of pig per week therefore saves ~150 t of coal a week, before counting the coking step deleted wherever raw coal proves workable. Run this ledger monthly from your own logs (§22.5); the stove's payback time falls out of two columns of arithmetic.

Institutional footnote worth carrying into Ch 47: Neilson defended his patent through litigation (Neilson v Harford, finally decided in 1841, upheld it) and died wealthy from royalties — the incentive system working exactly as designed, a pointed documented contrast with Cort's ruin one section above. Both cases are on the historical record; rebuilders should plan for both outcomes.

22.9 The Foundry: Patterns, Sand Moulds, and the Cupola

Pig iron is only half useful until it can be remelted and poured into shapes: pump barrels, engine cylinders, pipes, gears, machine beds, stove plates, and the castings of the Iron Bridge (§22.7). Lost-wax and stone moulds (Ch 10 §10.3) do not scale to tonnes. The foundry's answer is a reusable pattern and an expendable sand mould.

Safety warning: molten iron pours at roughly 1,300–1,450 °C, and any moisture it meets — damp sand, wet tools, wet or rusty scrap charged into the melt, a puddle on the casting floor — flashes to steam and throws metal across the shop. Cupola tapping also exposes crews to carbon monoxide and radiant burns. Dry and preheat every ladle, tool, and mould surface; charge only dry metal; keep the floor dry and sanded; wear foot, leg, face, and hand protection; take rehearsed positions for every pour; and keep onlookers out of the casting floor.

• Patterns: a wooden (later metal) replica of the part, made oversize to allow for the metal's shrinkage on cooling — patternmakers' contraction rules allow about 1 % for grey cast iron and roughly 1.5–2 % for brass and steel — with draft (a slight taper) on vertical faces so it lifts cleanly out of the sand, and extra metal wherever a surface will be machined.
• Green-sand moulding: silica sand bonded with several percent of clay and dampened just enough to hold a sharp edge is rammed around the pattern in a two-part box (the cope above, the drag below). The pattern is lifted out; a sprue, runners, and gates are cut so metal can enter; risers hold a reservoir that feeds the casting as it shrinks; vents let gas escape.
• Cores: hollow parts — pipes, cylinders, water jackets — need cores of sand bonded with linseed oil or a similar binder, shaped in a core box and baked hard, then set in the mould before it is closed.
• The cupola: a vertical shaft lined with firebrick (Ch 5 §5.5), charged in layers of coke, pig and scrap iron, and limestone flux, with air blown in near the bottom. Molten iron gathers in the well and is tapped into ladles. John Wilkinson patented an early form (1794), and the cupola remained the standard remelting furnace for cast iron for more than a century and a half.
• Defects and their causes: blowholes (wet sand, poor venting), cold shuts (metal too cool or poured too slowly), shrinkage cavities (risers too small), sand inclusions (mould surfaces washed away by the stream), and hot tears (sections restrained as they cool). Pour test bars from every melt, break and inspect them, and log melt, sand moisture, and pour temperature with each batch on an I-MR chart (Ch 47 §47.13).

Key threshold: a foundry that can pour a pump barrel or engine cylinder sound enough to bore (Ch 15 §15.4) without blowholes in the bore is the hinge between cheap iron and steam power (Ch 23).