Chapter 23: Steam Engines Done Right

Era span: 1698 Savery → 1900s compounds · Difficulty: high
Requires: Ch 22 iron, Ch 15 boring precision, Ch 20 pressure gauges ·
Unlocks: Ch 24 locomotion, Ch 26 prime power, factory independence from rivers

Steam is the first engine that makes power WHERE YOU WANT IT, burning fuel anyone can ship. History took 150 years to get it right; this chapter compresses the learning by stating the physics up front and the design sequence after.

23.1 The Physics First (History Did It Backwards)

• Water → steam expands ~1,600× at atmospheric pressure; condensing it back creates vacuum.
• Heat engines convert only a FRACTION of heat to work: Carnot bound η = 1 − T_cold/T_hot. Every efficiency gain in this chapter is secretly "raise T_hot, lower T_cold, waste less in between."
• Latent heat: boiling/condensing absorbs/releases enormous energy at constant temperature — why condensers dominate engine economics.

Knowing these three facts, you can skip a century of blind tinkering.

23.2 The Design Ladder

Stage 1 — Savery pump (skip if possible): no piston; steam pressure pushes water up, vacuum sucks it higher; boiler explosions capped it at ~10 m lifts. Historical dead end once Newcomen exists.

Stage 2 — Newcomen atmospheric engine: open-top cylinder; piston hangs from beam pump rods. Fill cylinder with low-pressure steam, spray cold water inside to condense, ATMOSPHERE slams piston down, stroke repeats. Robust, crude, ~0.5 % efficient — it burns coal ONLY where coal is nearly free (mine mouths), which is fine: its job is mine drainage (Ch 13).

Stage 3 — Watt's corrections (~1765–1790): 1. Separate condenser: keep the cylinder hot, condense steam elsewhere — fuel drops ~75 %. 2. Double-acting: steam pushes both strokes. 3. Parallel motion linkage: piston force transmitted without chains. 4. Centrifugal governor + throttle: automatic speed regulation — the first industrial feedback controller (Ch 45 ancestor). 5. Pressure gauge and vacuum gauge: instrumented operation (Ch 20).

Efficiency ~3–4 %. Watt's patent ironically DELAYED high-pressure development for decades (Dead end avoided: don't let one firm's IP strategy define your roadmap — patent around or leapfrog).

Stage 4 — High pressure (Trevithick/Evans): stronger boilers let steam WORK expansively above atmospheric; engines shrink 10×, become mobile. This unlocks boats and locomotives.

Stage 5 — Compounding & superheating (19th c. mature form): expand steam through successive cylinders (high→low pressure), superheat before admission to keep steam dry. Marine triple-expansion engines reach 15–20 % efficiency — good enough to make steamships outperform sailing freight economically (Ch 24).

23.3 The Machine-Tool Dependency

Watt's first engines leaked like sieves until Wilkinson's boring machine (1774 — originally cannon bores!) machined true cylinders. Lesson institutionalized: engine progress = machine-tool progress. Fund lathes/borers/planers continuously (Ch 15); every micron of cylinder fit becomes fuel saved for decades.

23.4 Boilers: Where Danger Lives

Most steam fatalities were boiler explosions, not engine failures. Non-negotiables:

• Water level gauges + fusible plugs (melt and vent if dry-fire begins).
• Spring-loaded safety valves sized generously, tested weekly under steam.
• Riveted/welded shells with computed thickness margins; pressure testing ABOVE working pressure with water (incompressible — failures are leaks, not bombs).
• Written operating logs per shift (Ch 11) — pattern analysis catches drifting maintenance before it kills.

23.5 Deployment Priorities

1. Mine drainage (pays instantly where water floods workings).
2. Mill drive in riverless towns; line-shafting identical to water mills (Ch 16) but sited anywhere near coal.
3. Blast blowing for ironworks (Ch 22) — hot blast synergy.
4. Locomotion — next chapter.
5. Electric generation — Ch 26; steam's final century of dominance.

Key threshold: an engine hall delivering 50–200 kW on demand converts any town into an industrial site. Track fuel-per-kWh relentlessly; thermodynamics says there's always headroom, and competitors who ignore the number die quietly.

23.6 The Engine Papers

• Hero's aeolipile (1st c. CE) proves reaction jets existed as temple toys for seventeen centuries — physics known, economics absent: fuel was expensive, labor cheap. Steam waited for the wage/coal price crossover (Ch 22 supplies it), not for genius.
• Savery's 1698 "Miner's Friend" patent claimed such broad wording (drawing water by impelling fire) that it taxed later inventors for decades — documented patent-drafting abuse predating modern IP law.
• Newcomen's first working engine ran at Dudley Castle colliery (~1712). Watt recorded his separate-condenser insight during a Glasgow Green walk in May 1765; the 1769 patent, Boulton partnership (1775), and a Parliamentary extension to 1800 made the firm rich and — per Cornish engineers' contemporary complaints — slowed high-pressure development until expiry. Both readings are documented; the book records the tension rather than adjudicating it.
• Wilkinson's cannon-boring patent (1774) became Watt's cylinder salvation — arms machining feeding civil industry again (Ch 49's armory loop).
• Cornwall quantified competition: mine engineers measured engines in "duty" (foot-pounds of work per bushel of coal), publishing league tables; Woolf's compound designs pushed duty records through the 1810s–30s. Efficiency racing as public sport — an institution worth rebuilding deliberately.
• Trevithick's Penydarren locomotive hauled iron along plateway rails (February 1804); Rocket won Rainhill Trials (October 1829) with the multi-tubular boiler + blast pipe package Ch 24 describes.

23.7 First Build: A 10 kW Stationary Engine From Computed Drawings

One buildable reference design: a double-acting simple high-pressure horizontal engine, non-condensing for its first season. Every dimension below is derived, not quoted — recompute them whenever one input changes.

Phase 0 — Size the cylinder (worked arithmetic).

1. Target 10 kW brake output; assume mechanical efficiency ~85 % → ~11.8 kW indicated required.
2. Boiler pressure 7 bar gauge (8 bar absolute, saturated steam ~170 °C); cut-off ~35 % gives a realistic mean effective pressure p_m ≈ 240 kPa.
3. Speed n = 300 rpm = 5 rev/s; choose stroke L = 200 mm.
4. Double-acting work per revolution = 2·p_m·A·L, so required piston area A = P/(2·p_m·L·n) = 11,800/(2 × 240,000 × 0.20 × 5) = 0.0246 m² → bore d = √(4A/π) ≈ 177 mm.
5. Adopt bore Ø180 mm × stroke 200 mm: A = 0.02545 m² → work/rev = 2 × 240,000 × 0.02545 × 0.20 ≈ 2,443 J → indicated power at 300 rpm ≈ 12.2 kW, ~10 kW at the belt.
6. Sanity checks before cutting metal: mean piston speed = 2Ln = 2 × 0.20 × 5 = 2.0 m/s — comfortably under the ~4 m/s where ring and bore wear escalate at this tech level. Steam demand: full displacement 2 × A × L × n = 50.9 L/s of 8-bar saturated steam (~4.16 kg/m³) at 35 % cut-off ≈ 75 kg/h; a good fire-tube boiler evaporating ~7 kg steam per kg coal puts coal burn near 11 kg/h ≈ 1.1 kg per delivered kWh.

Verification checkpoint: an indicator diagram (or careful gauge-and-stroke estimate) reading ≥10 kW sustained against a Prony brake for four hours confirms sizing; half that means leaking piston or dead passages, not wrong arithmetic.

Per-step failure modes:

• Knocking on stroke reversal: water slug in the cylinder → stop immediately; drain, fix priming source before restart.
• Speed hunts under load: governor friction or flyball spring sag → clean linkage, reset springs until ±3 % regulation.

Phase 1 — Boiler plate thickness (worked hoop stress).

7. Shell: seamless-wound or riveted drum, internal diameter D = 400 mm.
8. Hoop stress σ_h = pD/2t; solve thickness t = pD/(2·σ_allow·η). With wrought-iron/mild-steel plate ultimate ~320 MPa, safety factor 5 → σ_allow = 64 MPa; riveted longitudinal joint efficiency η = 0.70: t_min = (700,000 Pa × 0.40 m)/(2 × 64,000,000 × 0.70) = 280,000/89,600,000 ≈ 3.1 mm.
9. Adopt 6 mm plate — roughly double the computed minimum, covering corrosion loss, rivet-lap weakening, and shop damage. Flat ends are forbidden at this pressure unstayed: use dished heads with gusset stays, firebox crown stayed ≤150 mm pitch.
10. Tube bank: 24 × 65 mm firetubes × 1.5 m gives ≈7.4 m² heating surface — evaporates >100 kg/h against this engine's ~75 kg/h demand.

Verification checkpoint: hydrostatic test at 1.5 × working pressure (10.5 bar) held 30 min, gauge motionless, seams weeping nothing worse than damp. Any stream = strip, re-caulk, retest. Water only — compressed gas stores lethal energy.

Per-step failure modes:

• Seam weep under test: mark, depressurize, re-caulk seam, retest from zero.
• Gauge creep during hold: thermal expansion still equalizing; restart the clock only after full temperature soak.

Phase 2 — Safety hardware before first fire.

11. Spring safety valve set 7.5 bar gauge, sized to hold pressure ≤110 % with fires at maximum; hand-ease it weekly under steam.
12. Fusible plug (lead–tin alloy, melts ~230 °C) in the firebox crown — dry-firing melts it and vents the box instead of collapsing the crown.
13. Two independent water gauges plus try-cocks; feed pump AND injector-class backup.

Verification checkpoint: pop test — valve lifts within ±2 % of set pressure twice consecutively; fusible plug sits flush in crown, unobstructed.

Phase 3 — Valve gear: lap and lead (where most homemade engines go wrong).

14. Mark steam ports with the piston at each dead center (rock the flywheel over center by feel — the "rock-over" method — and scribe crossheads/crank marks at both).
15. Set the slide valve's outside lap symmetrically on both steam ends: lap is what creates cutoff; more lap = shorter admission = earlier expansion. For ~35 % cut-off at this geometry, outside lap near one-third of port width is the starting point; verify by measuring port opening at dead center.
16. Set lead — slight port opening (~1–3 mm) at dead center so steam is already entering as the stroke begins; too much lead wastes throttle, none makes the engine knock on reversal.
17. Exhaust lap ≈ zero for a first engine (inside edges of valve flush with exhaust ports).

Verification checkpoint: engine turns over slowly on hand-barrow air/steam, runs in BOTH directions from mid-position, and reverses cleanly — a mis-lapped valve hunts or refuses one direction.

Per-step failure modes:

• Runs one way only / dies at half-revolution: unequal lap or eccentric set wrong → re-scribe dead centers, reset eccentric to crank at 90° plus lap angle.
• Knocks at each dead center: clearance lost in crosshead or bearings, not the valve → check brasses first.

Phase 4 — First-fire procedure.

18. Cure refractory/masonry gently: wood coals only, day one; never raise metal temperature faster than ~50 °C/h thereafter.
19. Fill with the softest water available (scale is silent murder), glass at mid-level, vent open until steady steam shows.
20. Raise steam over 2–3 h; blow down 10 % twice in the first day to skim oil and salts.
21. Admit steam to cylinder with engine on light load; run 30 min, shut, retighten gland bolts warm, then couple the load.

Verification checkpoint: gauge holds 7 bar under full load with safety valve quiet; exhaust clear (no white priming carryover); bearings cool enough to rest a hand.

Per-step failure modes:

• Priming/carryover: erratic gauge, water hammer, wet exhaust → drop load, check water level and quality, reduce firing rate.
• Dry firing: fusible plug vents with hiss while gauge falls → kill fires, cool completely, replace plug (spares stocked), find feed failure first.
• Gland leak growing: steam feathering at rod → tighten evenly warm; never crank cold.

Phase 5 — Condenser-trip procedure (when the jet/surface condenser is added).

22. Normal running: condenser vacuum ≥70 kPa below atmospheric; log it hourly (§23.4 log discipline).
23. Trip drill monthly: throttle cooling water deliberately → when vacuum falls through 50 kPa, the atmospheric break-valve must open within seconds, dumping exhaust to roof; engine keeps turning non-condensing at reduced economy.
24. Pass criterion: no stall, no overspeed, vacuum recovered within 5 min of cooling-water restoration.

Verification checkpoint: three consecutive clean trips logged; then the drill joins the weekly safety-valve round.

Key threshold: fuel burn near 2 kg coal per kWh delivered marks an honest first build; Cornish-style discipline (§23.6 duty tables) exists precisely to halve that number. Log fuel and feed water every shift from day one — the engine that isn't metered regresses silently.

Duty, converted into your terms. §23.6's league-table metric is work per bushel of coal (84 lb ≈ 38 kg): duty (ft·lb) = output work ÷ bushels burned, and since 1 kWh = 2.66 million ft·lb, any metered engine converts straight across. The documented trajectory: Newcomen-era engines ran roughly 4–6 M ft·lb/bushel (≈20 kg coal/kWh); Watt's separate condenser lifted practice to roughly 20–30 M (≈3–4 kg/kWh, matching its 3–4 % thermal efficiency above); the best Cornish compounds peaked around 90–100 M (≈1 kg/kWh). This section's own 2 kg-coal/kWh acceptance bar corresponds to ~50 M duty — a Watt-class result, with Cornwall's numbers marking what disciplined iteration buys. Compute yours weekly from the shift logs; publish the table internally exactly as Cornwall did (§23.6).