Compound Corliss engine

The Corliss engine is a horizontal mill engine with an unusual way of letting steam in: four small rocking valves, and a catch that lets go of the steam valve at a moment the governor picks. Compounding passes the same steam through a second, larger cylinder before it leaves. Thurston, listing what it brought together, records "a remarkable economy of fuel" (Thurston Manual p. 102). In the game this is the planned top of the power ladder, part of the later steel mod: it is sized and ruled, but it is not drawn or built, so this page has no render of it. For the engines before it see the horizontal engine and the engine ratings.

The crank end of a Mesta-Corliss cross-compound at the Republic Iron and Steel works, Youngstown, built for a 40 in blooming mill and photographed derelict. The crank and connecting rod stand in front of a flywheel 24 ft across.
The crank end of a Mesta-Corliss cross-compound at the Republic Iron and Steel works, Youngstown, built for a 40 in blooming mill and photographed derelict. The crank and connecting rod stand in front of a flywheel 24 ft across.
The same engine from its high-pressure side. The block with the round bolted covers is a cylinder, with its pipework above it. A steel works is where the game's planned engine belongs.
The same engine from its high-pressure side. The block with the round bolted covers is a cylinder, with its pipework above it. A steel works is where the game's planned engine belongs.
A cross-compound Corliss as its maker sold it, the Filer-Stowell heavy duty engine (Shillitto p. 177, Fig. 46, turned upright). Two engines lie side by side with one flywheel between them on one crank shaft. Steam works in one cylinder and then in the other, larger one. Side-by-side compounds set their cranks at right angles (Thurston Manual p. 107), so one side is always pushing hard when the other is at its dead point.
A cross-compound Corliss as its maker sold it, the Filer-Stowell heavy duty engine (Shillitto p. 177, Fig. 46, turned upright). Two engines lie side by side with one flywheel between them on one crank shaft. Steam works in one cylinder and then in the other, larger one. Side-by-side compounds set their cranks at right angles (Thurston Manual p. 107), so one side is always pushing hard when the other is at its dead point.

What the valve gear does

An engine's governor has to do something when the load changes. On the plain horizontal engine it closes a throttle (Thurston History p. 311), which wastes pressure before the steam has done any work. Corliss made the governor change how long the steam valve stays open, so the full boiler pressure always reaches the piston and the load only alters how far it expands. Thurston dates this governor-controlled cut-off to Corliss in 1849 (Thurston Manual p. 89).

Corliss valve gear on an engine built by the Murray Iron Works in 1903, with the valves named over the photograph. A single eccentric rocks the wrist-plate (Thurston Manual p. 101). Links radiate from the plate to all four valves, the steam valves on top and the exhaust valves below. Each link swings almost into line with its centres at the end of its range, so a valve opens and closes quickly and is "held open and nearly still" between (Thurston Manual p. 101). The steam links end in catches that a cam set by the governor knocks loose (Thurston Manual p. 98).
Corliss valve gear on an engine built by the Murray Iron Works in 1903, with the valves named over the photograph. A single eccentric rocks the wrist-plate (Thurston Manual p. 101). Links radiate from the plate to all four valves, the steam valves on top and the exhaust valves below. Each link swings almost into line with its centres at the end of its range, so a valve opens and closes quickly and is "held open and nearly still" between (Thurston Manual p. 101). The steam links end in catches that a cam set by the governor knocks loose (Thurston Manual p. 98).

Thurston's list of what made the design new is the clearest statement of it (Thurston Manual pp. 100-102): four valves placed so as to reduce clearance to a minimum; a rotating valve, cheap to fit and worked from outside the steam spaces; one wrist-plate on one eccentric for all four; a detaching gear that drops the steam valve shut; the governor directly setting the expansion, with the least possible work thrown on the governor itself. The separate valves matter too: with a steam- and an exhaust-valve at each end of the cylinder, the steam passages are short and the clearance small, "a source of some economy" (Thurston Manual p. 97).

A model in the Science Museum, cut open. The passages run straight from the barrel to the round valve chambers at its corners, so very little steam sits in the ports at the end of a stroke. Each valve only rocks through part of a turn.
A model in the Science Museum, cut open. The passages run straight from the barrel to the round valve chambers at its corners, so very little steam sits in the ports at the end of a stroke. Each valve only rocks through part of a turn.
A Harris-Corliss valve in section along its length (Thurston Manual p. 100, Fig. 36). Valve A is a cylinder partly cut away, with little more bearing surface than it needs, turned in a lathe. A slit lets the blade of the rock-shaft fit it without jamming, so it can wear down in its seat without leaking (Thurston Manual pp. 98-99). The stem has no stuffing box: a shrunk-on collar at F fits a recess at a and wears itself tight (Thurston Manual p. 99).
A Harris-Corliss valve in section along its length (Thurston Manual p. 100, Fig. 36). Valve A is a cylinder partly cut away, with little more bearing surface than it needs, turned in a lathe. A slit lets the blade of the rock-shaft fit it without jamming, so it can wear down in its seat without leaking (Thurston Manual pp. 98-99). The stem has no stuffing box: a shrunk-on collar at F fits a recess at a and wears itself tight (Thurston Manual p. 99).

The valve, once tripped, is pulled shut by a spring or a weight, helped by a vacuum chamber, and would slam. The dash-pot takes the shock: a piston on the valve gear drops into a shallow cup and cushions on trapped air. Sickels had used water for the same purpose; Corliss used air (Thurston Manual pp. 99-100).

The trip has a limit that follows from the geometry. The catch can only let go while the eccentric is still opening the valve, which ends at about half stroke. Past that the valve closes at the eccentric's own slow pace, so a release was wanted before then (Henthorn pp. 19-20).

Why cutting the steam off early saves it

A boiler's worth of steam made to work by its own expansion, from a textbook for Corliss designers, with the steps named over the plate (Tribe p. 51, Fig. 2). Steam at 95 lb absolute for the first fifth of a 60 in stroke, cut off, then expanding to 19 lb at the end. Tribe counts the mean pressure over the last four fifths as 39 lb, 23 lb above the back pressure, against 79 lb in the first fifth. Summed over the whole stroke, on his figures, the same steam does a little over twice the work it would do if let out at the cut-off.
A boiler's worth of steam made to work by its own expansion, from a textbook for Corliss designers, with the steps named over the plate (Tribe p. 51, Fig. 2). Steam at 95 lb absolute for the first fifth of a 60 in stroke, cut off, then expanding to 19 lb at the end. Tribe counts the mean pressure over the last four fifths as 39 lb, 23 lb above the back pressure, against 79 lb in the first fifth. Summed over the whole stroke, on his figures, the same steam does a little over twice the work it would do if let out at the cut-off.

Tribe puts the result of the Corliss gear plainly: its expansion curve "approaches very closely to the Hyperbolic curve", the ideal one, and "at least twenty-five per cent. of the fuel can be saved as compared with the slide valve arrangement" (Tribe p. 50). Condensing takes it further: the same engine exhausting at 7 lb absolute rather than 16 gains about 40% on the non-expansive, non-condensing engine (Tribe p. 51). A jet condenser does that work on the back pressure.

Why a second cylinder

If expansion is so good, a long expansion in a single cylinder should be better still. What stops it is the iron. Steam falls in temperature as it expands, and Tribe works the example of 200 lb absolute steam exhausting at 3.5 lb: a fall from 381 to 146 degrees F. If all this were done in one cylinder, "the walls of the same became alternately heated and cooled to nearly this extent" and there "would be great loss of condensation" (Tribe p. 90). Each fresh charge of steam meets iron just cooled by the exhaust, and part of it turns to water before it does anything.

Dividing the range between two cylinders is, in Tribe's words, "one of the chief values of the multiple expansion principle" (Tribe p. 90). His example at 150 lb absolute gives 358 to 258 degrees in the high-pressure cylinder and 258 to 158 in the low: a hundred degrees each in place of two hundred together (Tribe pp. 90-91).

A compound engine's two cards drawn on one set of axes, with the stages named over the plate (Tribe p. 69, Fig. 15). Boiler steam at 150 lb absolute enters the small cylinder and expands into a receiver at 28.7 lb. The large cylinder takes it from there and expands it to 6.5 lb, over the condenser's vacuum. The shaded areas are the work of each. One expansion curve runs through both, so to the steam the pair is one long stroke; to the iron each is a much narrower swing of temperature.
A compound engine's two cards drawn on one set of axes, with the stages named over the plate (Tribe p. 69, Fig. 15). Boiler steam at 150 lb absolute enters the small cylinder and expands into a receiver at 28.7 lb. The large cylinder takes it from there and expands it to 6.5 lb, over the condenser's vacuum. The shaded areas are the work of each. One expansion curve runs through both, so to the steam the pair is one long stroke; to the iron each is a much narrower swing of temperature.

The saving is real and moderate. A Corliss compound at a steam mill burned 1.62 lb of good fuel per horse-power an hour, where Thurston reckoned the same engine with its high-pressure cylinder disconnected could not do better than about 2 lb (Thurston Manual p. 109). The best Corliss compound mill engines used about 16 lb of steam per horse-power an hour (Thurston Manual p. 587). The Bristol engine in the table below, tested in 1894, took 13.5 lb per indicated horse-power an hour and 1 1/3 lb of coal (Denton p. 2). A second cylinder earns its keep where the load is steady. Tribe warns against it on rolling-mill engines, with "their alternate excessive overload and underload" (Tribe p. 14).

How the game plans to follow it

The planned engine is the steel works' single mill engine, carrying its blowing, hot blast and rolling on one crank. It is taken from the one compound Corliss whose every term was measured: the Bristol cotton-mill engine of 1894, a cross-compound with cylinders of 30 and 56 in by a 72 in stroke (Denton pp. 1-2). The game rates it by the same arithmetic the horizontal engine uses, with the test's own mean pressure, so the cylinders' real losses are inside the number and not modelled one at a time. Its steam comes from a battery of Lancashire boilers.

Its 161 kN*m at the crank is far more than a cast main can carry, so the flywheel is toothed and drives three mains through a 4:1 step-up, as the 30 ft geared flywheel of the Centennial engine drove its main shaft in 1876 (Sandhurst pp. 362-368).

Period basis, and what is the game's own

  • Period: four rocking valves a cylinder, steam above and exhaust below, one wrist-plate and eccentric, a governor setting the trip (Thurston Manual pp. 97-102). The plan is the same gear on each side, the steam valves worked from one wrist-plate.
  • Period: tandem was the usual way to compound a Corliss (Thurston Manual pp. 107-108), and two engines side by side were "oftener practised in Europe than in the United States", especially for large powers (Thurston Manual p. 109). The plan is a cross-compound because the Bristol engine, the one fully measured, was one. The horizontal engine's tandem refit is a separate, smaller thing.
  • Period: Tribe doubts compounding on a rolling-mill engine (Tribe p. 14); the books give no horse-power for a rolling-mill Corliss. The game's own: one engine carries the whole works, rolling included.
  • Period: the Centennial engine's geared flywheel, 30 ft across with 216 teeth (Sandhurst pp. 362-368). The game's own: the 4:1 ratio and the three mains are set by what a cast main can carry.
  • The game's own: nothing here is built. No shape is drawn, and every number below is planned. See vanilla against reality: the base game has no steam engine.

The numbers

Engines keep period figures (see the scaling page), drawn at true size (scale and units). Every value below is planned. The context among the other engines is on the ratings page.

In the gamePeriod figureHow it was scaledGame value
Cylinders30.025 and 55.915 in bores (Denton pp. 1-2)True sizePlanned 0.76 m and 1.42 m
Stroke72 in (Denton pp. 1-2)True sizePlanned 1.83 m
Speed65.2 rpm, 783 ft a minute of piston (Denton pp. 1-2)True valuePlanned 6.83 rad/s at the crank
Steam125 lb on the gauge (Denton p. 1)True valuePlanned 125 lb, 8.5 atm
Cut-off0.30 in the high-pressure cylinder (Denton pp. 1-2); release before half stroke (Henthorn pp. 19-20)The test's own, the most takenPlanned 0.30 rated
Mean pressure27.35 psi referred to the low-pressure piston (Denton pp. 1-2), hyperbolic expansion (Tribe p. 50)The test's own, 0.726 of the hyperbolic figurePlanned 27.35 psi
FrictionNone in the testThe game's ownPlanned 2 psi on the low-pressure piston
Power, cut-off 0.30, condensing1592 ihp (Denton pp. 1-2)Mean pressure x piston x stroke x strokesPlanned 1187 kW indicated, 1100 kW useful, 1476 hp
Power, cut-off 1/5, condensingTribe's worked cut-off (Tribe p. 51)The same rulePlanned 852 kW indicated, 766 kW useful
Power, ventedBack pressure 18 psi absolute (Rankine p. 409)The same rule, cut-off 0.30Planned 746 kW indicated, 659 kW useful
Steam, rated13.5 lb per ihp-hour (Denton p. 2), 2.71 kg/s at the testCylinder volume x density at 139.7 psiPlanned 2.51 kg/s, 8.2 kg per useful kWh
Coal1 1/3 lb per ihp-hour (Denton p. 2)Period ratePlanned 963 kg an hour
BoilersNo boiler count in the testRated steam over a Lancashire's firingPlanned a battery of 5 to 7
FlywheelAbout 4 strokes across, rim at most 4500 ft a minute (Henthorn pp. 93-94)Rim speed sets itPlanned 22 ft, 6.70 m across, rim 22.9 m/s
Torque at the crankPower over speedArithmeticPlanned 161 kN*m, 9.5 times a cast main's limit
Gearing to the mainsA geared 30 ft flywheel (Sandhurst pp. 362-368)Set by the mains' limit, the game's ownPlanned 4:1 to three mains, 13.4 kN*m each

Open, and what you can tell us

The books disagree on how fast such a piston should go. Henthorn says the average piston speed "should not exceed 500 ft. per minute" (Henthorn p. 93); Tribe allows up to 800 for a long stroke with Corliss gear (Tribe p. 15); and the Bristol engine ran at 783 in a measured test (Denton pp. 1-2). The game uses the Bristol's. Thurston also says the compound Corliss "was first introduced by other builders" (Thurston Manual p. 107) and does not say who. If you know a source for either, we would like to read it.

Picture credits