A Harrison Boiler is a sectional cast-iron steam boiler built up from a grid of hollow hemispherical units bolted together with wrought-iron tie rods, generating low-pressure saturated steam from a coal or wood fire. Joseph Harrison Jr. of Philadelphia patented it in 1859 after the Pacific Mills boiler explosion shook public confidence in shell-type boilers. The hemispherical shape resists pressure with thin walls, and any single unit failing vents safely without fragmenting. That made it the first boiler marketed specifically as explosion-proof, and hundreds were sold for mills, hotels, and public buildings through the 1870s.
Harrison Boiler Interactive Calculator
Vary sphere size, wall thickness, pressure, and section count to see spherical shell stress, joint separating force, heating surface, and water volume.
Equation Used
The Harrison boiler split pressure into many small cast-iron spherical sections. For a sphere, internal pressure creates membrane stress sigma = P r_i / (2 t), far lower than bending stress in a flat wall. The calculator also estimates separating force at a circular joint, outside heating surface, and internal water volume.
- Thin spherical shell membrane stress approximation.
- Outside diameter and wall thickness estimate the inner radius.
- Heating surface is approximated as full outside sphere area.
- Educational estimate only; no code rating, corrosion allowance, or casting defects included.
The Harrison Boiler in Action
The Harrison Boiler does one thing differently from every shell boiler that came before it — it splits the pressure vessel into dozens of small hollow cast-iron spheres, each about 8 inches in diameter, joined into a grid by short threaded necks and clamped together end-to-end with wrought-iron tie rods running the full length of the assembly. Fire and hot gas pass between the spheres on the outside; water and steam fill the inside. Each sphere is its own little pressure vessel, and the geometry is the point. A hemisphere under internal pressure sees pure membrane stress, no bending, so a 1/2-inch cast-iron wall handles 50-75 psig saturated steam with a comfortable safety factor. Try the same pressure in a flat-walled cast-iron box and you will crack it on the first warm-up.
The sectional cast iron boiler design also gives you a graceful failure mode. If one sphere develops a hairline crack — usually from a thermal-shock event when cold feedwater hits a hot section — that single unit weeps or pops, the tie rods hold the rest of the grid together, and the boiler vents to atmosphere instead of launching itself through the roof. That is the entire reason Joseph Harrison built the thing. After the 1854 Pacific Mills explosion in Lawrence killed 14 workers, the public would not trust shell boilers near factory floors, and Harrison saw a market for an explosion-proof boiler.
Where the design punishes you is in tolerances and water chemistry. The threaded necks joining the spheres need to seat dead flat — out-of-square by more than about 0.5° and you will leak steam at the joint or crack the neck on first pressurisation. The internal volume is small relative to heating surface, so scale buildup from hard feedwater chokes circulation fast; you want feedwater under 5 grains/gallon hardness or the lower spheres bake dry and crack. Wrought iron tie rods stretch over time under repeated thermal cycling, and a slack tie rod lets joints work loose at pressure. Re-torquing every 18-24 months of service was standard practice in period boilerhouses.
Key Components
- Hollow cast-iron spheres: The pressure-bearing units, typically 8 inch outside diameter with a 1/2 inch wall thickness, cast in green sand from grey iron. Each sphere has two threaded necks on opposing axes — one for the tie rod path, one for cross-connection to its neighbour. Wall thickness must be uniform within ±1/16 inch or the thin spot becomes the failure point at temperature.
- Wrought-iron tie rods: Long threaded rods, usually 3/4 inch or 1 inch diameter, running the full length of the sphere stack and clamping the assembly between cast end-plates. Tension preload sits around 4000-6000 lbf per rod. Too loose and the joints leak, too tight and the cast iron necks split — wrought iron not steel, because wrought iron yields gracefully where high-carbon steel would snap.
- Threaded sphere necks and lead gaskets: The sealing surface between adjacent spheres. Lead foil gaskets compress under tie-rod load and accommodate the small geometric mismatches between hand-fitted castings. Replace every overhaul — work-hardened lead leaks at the next thermal cycle.
- Furnace and grate assembly: External cast-iron firebox under and around the lower courses of spheres, fired with anthracite or bituminous coal. Grate area sized at roughly 1 sq ft per 12-15 sq ft of heating surface, matched to fuel calorific value and draft.
- Steam dome and feed connections: A larger casting on top of the assembly collecting saturated steam from the upper spheres, with a stop valve, safety valve set at 75 psig, and water-glass gauge. Feedwater enters at the bottom course, displacing cooler water upward through the natural-circulation path between spheres.
Where the Harrison Boiler Is Used
Harrison Boilers found their market in places where a shell boiler explosion would be catastrophic — populated buildings, mills with dense workforces, and installations where insurance underwriters demanded an explosion-proof boiler. They never competed with high-pressure water-tube boilers for power generation, but for low-pressure heating and process steam in the 50-75 psig range they sold strongly from 1860 through the 1880s before locomotive-style fire-tube boilers with proper safety codes pushed them aside.
- Textile mills: Pacific Mills in Lawrence, Massachusetts installed Harrison Boilers after the 1854 explosion of their shell boiler, specifically because the sectional design could not fragment a populated weaving floor.
- Public buildings and hotels: The U.S. Capitol building in Washington used Harrison sectional boilers for steam heating from the 1860s, chosen for the explosion-proof reputation in an occupied legislative chamber.
- Naval and marine auxiliary: Several U.S. Navy yard installations used Harrison Boilers for shore-side steam supply where a wrought-iron shell boiler near ammunition stores was unacceptable.
- Sugar refineries: Havemeyer & Elder sugar refinery in Brooklyn ran Harrison Boilers on the low-pressure process steam circuit feeding pan coils, where steady 60 psig delivery mattered more than peak output.
- Hospitals and asylums: Pennsylvania Hospital and several state asylums specified Harrison Boilers in the 1870s for steam heating and laundry service, on direct recommendation from boiler insurance inspectors.
- Heritage demonstration steaming: A small number of preserved Harrison Boilers run today at industrial museums in the Mid-Atlantic United States, primarily on reduced pressure for educational demonstration of pre-code boiler design.
The Formula Behind the Harrison Boiler
The most useful number for sizing or rating a Harrison Boiler is the equivalent evaporation rate — pounds of steam per hour the assembly delivers from and at 212°F. It depends on total heating surface, fuel firing rate, and the mean gas-to-water temperature differential across the sphere walls. At the low end of typical Harrison operation — say 40 psig and a slow anthracite fire — you are looking at modest output and very long sphere life. At the high end, hard firing toward 75 psig, you push evaporation up but accelerate scale deposition and thermal cycling fatigue on the lower-course castings. The sweet spot for these boilers, historically, sat around 60 psig with a measured stack temperature of 550-650°F.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| We | Equivalent evaporation rate (steam from and at 212°F) | kg/h | lb/h |
| Ah | Total heating surface — outer area of all spheres exposed to fire and gas | m² | ft² |
| U | Overall heat transfer coefficient through cast-iron wall plus film resistances | W/m²·K | Btu/hr·ft²·°F |
| ΔTm | Log mean temperature difference between gas path and saturated water | K | °F |
| hfg | Latent heat of vaporisation at 212°F | kJ/kg (2257) | Btu/lb (970.3) |
Worked Example: Harrison Boiler in a heritage textile museum Harrison Boiler
You are predicting the equivalent evaporation rate of a recommissioned 1868 Harrison sectional cast-iron boiler being returned to demonstration steaming at a heritage textile museum in Lowell, Massachusetts, where it will supply low-pressure saturated steam at 60 psig to a small horizontal mill engine driving a single carding-room demonstration line. The assembly contains 96 hollow spheres of 8 inch outside diameter with total measured heating surface of 132 sq ft, fired on anthracite at a rate that produces a measured mean gas-to-water ΔT of 580°F, and historical performance data suggests an overall U around 7 Btu/hr·ft²·°F for these clean-tube cast-iron sectional designs.
Given
- Ah = 132 ft²
- U = 7 Btu/hr·ft²·°F
- ΔTm = 580 °F
- hfg = 970.3 Btu/lb
- pop = 60 psig
Solution
Step 1 — at the nominal 60 psig firing condition, compute the total heat transfer rate from gas path into the water side:
Step 2 — divide by latent heat at 212°F to get equivalent evaporation:
That is enough steam to feed a small single-cylinder mill engine of about 15-20 indicated horsepower running with reasonable cutoff — in line with what the original Harrison catalogues quoted for a 96-sphere unit.
Step 3 — at the low end of the typical range, dropping to a slow anthracite fire that gives ΔTm ≈ 420°F (40 psig operation, lower stack temperature):
That is comfortable steaming — long sphere life, minimal thermal stress on the lower courses, but only enough output for the engine on light load. The boiler will hold pressure all day and the firebox barely cycles.
Step 4 — at the high end, hard firing toward 75 psig with ΔTm pushed to 720°F:
That is the rated maximum for an assembly this size. You can run it there, but expect noticeably faster scale buildup in the lower spheres and visible thermal cycling on the cast-iron necks every time the engine load drops sharply. Period operators avoided sustained running above roughly 90% of rated evaporation for exactly this reason.
Result
The nominal predicted evaporation rate is 552 lb/h at 60 psig, which is the design sweet spot for a 96-sphere Harrison and matches the original 1868 catalogue figure within about 3%. The low-end output of 400 lb/h feels lazy — plenty of margin, gentle firing, sphere life measured in decades. The high-end 685 lb/h is achievable but punishes the castings with thermal cycling and accelerates scale deposition in the bottom course. If your measured evaporation comes in 15-20% below the predicted 552 lb/h, the most likely causes are: (1) scale on the water side of the lower spheres dropping U from 7 down toward 4-5 Btu/hr·ft²·°F, (2) air infiltration through worn lead gaskets at the sphere necks dragging firebox temperature down, or (3) a stretched tie rod letting joints leak combustion gas around the assembly instead of through the proper convective path.
When to Use a Harrison Boiler and When Not To
The Harrison Boiler made sense in its specific historical window — between the era of unregulated shell boilers and the introduction of proper pressure-vessel codes — and against modern alternatives it loses on almost every metric except the one it was built for. Here is how it stacks up against the two boiler types it actually competed with.
| Property | Harrison Boiler | Lancashire Shell Boiler | Babcock & Wilcox Water Tube |
|---|---|---|---|
| Maximum working pressure | 75 psig | 150 psig | 300+ psig |
| Evaporation rate per sq ft heating surface | 3-5 lb/h·ft² | 2.5-3.5 lb/h·ft² | 5-8 lb/h·ft² |
| Failure mode | Single sphere weeps, vents safely | Catastrophic shell rupture | Single tube rupture, contained |
| Capital cost (1875 dollars/IHP) | ~$45/IHP | ~$30/IHP | ~$60/IHP |
| Maintenance interval | 18-24 months retorque | 5-7 years internal inspection | 12 months tube inspection |
| Tolerance to hard feedwater | Poor — small water volume scales fast | Good — large water volume | Poor — narrow tube bore |
| Best application fit | Low-pressure heating in occupied buildings | Mill engine power, robust workhorse | High-pressure power generation |
Frequently Asked Questions About Harrison Boiler
This is almost always a tie-rod tension problem, not a gasket problem. Wrought iron tie rods stretch under thermal cycling, and a rod that was correctly torqued at ambient loses 10-20% of its preload by the time the assembly hits operating temperature. The lead gaskets between spheres need continuous compression to seal — drop below about 3000 lbf preload per rod and the joints open under internal pressure.
Diagnostic check: measure tie-rod elongation cold versus hot. If you see more than about 0.040 inch of additional stretch per foot of rod length when going from cold to 60 psig, the rod is past its useful service life and needs replacement. Re-torquing a stretched rod buys you one season at most.
If the boiler is for genuine demonstration of period technology and your insurance underwriter will accept it under a derated permit, the Harrison is the only honest choice. If it is a working boiler that happens to look old, pick a modern fire-tube — you will get double the evaporation rate per square foot, far better tolerance for municipal feedwater, and a proper ASME stamp.
The decision usually comes down to insurance and inspection. Most jurisdictions will permit a Harrison only at significantly reduced pressure (often 25-30 psig versus the original 75 psig design), which gives you about 40% of the original output. If your demonstration engine needs more steam than that, the Harrison is not the answer.
High stack temperature with unchanged firing rate means heat is not transferring from gas path to water side — and on a Harrison that almost always means waterside scale, not gas-side fouling. The narrow internal volume of an 8-inch sphere fills with scale fast on hard feedwater, and even a 1/16-inch scale layer drops the overall U from 7 down to about 4 Btu/hr·ft²·°F.
Pull the inspection plug on a lower-course sphere and look. If you see white or grey crystalline buildup more than about 1 mm thick, you have lost roughly a third of your heat transfer. Acid descaling on a sectional boiler is fiddly because the spheres do not drain cleanly — many heritage operators find it cheaper to swap individual spheres than to descale the whole assembly.
Mechanically yes, but the heat-flux profile is wrong. Natural gas burners produce a much shorter, hotter flame than the long radiant fire from an anthracite grate, and that concentrates heat onto the lower courses of spheres. Cast iron does not like local hot spots — you will see thermal cracking on the bottom course within 50-100 firing cycles.
If you must convert, install a low-NOx burner with a long lazy flame profile and add refractory baffles to spread the heat across the full sphere stack. Even then, expect to inspect the lower spheres twice as often as you would on coal. Most successful heritage Harrison installations stick with anthracite for this reason.
Harrison's claim was specifically about the fragmentation energy, not about whether failures could happen. A single 8-inch sphere holds roughly 0.2 cubic feet of water at 60 psig — when it lets go, the stored energy is small enough that the sphere weeps or pops, not detonates. A Lancashire shell boiler at the same pressure holds 200+ cubic feet of pressurised water, and when that flashes to steam on rupture you get an event that levels buildings.
The wrought-iron tie rods are the other half of the story. Even if a sphere splits, the tie rods hold the surrounding spheres in position and prevent the pressure event from propagating. Joseph Harrison demonstrated this in front of insurance underwriters in 1860 by deliberately overpressuring a unit to failure — the failed sphere vented, the rest of the assembly held, and he won his contracts on the strength of the demonstration.
Period safety valves on Harrison Boilers are deadweight or simple lever-and-weight designs, and both drift downward over time. Lever pivots wear, weight hangers stretch, and seat surfaces erode — all of which lower the effective lift pressure. A valve set at 75 psig in 1870 is probably lifting somewhere between 50 and 60 psig today.
Do not solve this by adding weight. Pull the valve, lap the seat, replace the lever pin, and recalibrate against a known dead-weight tester. If the valve body itself is eroded around the seat, replace it with a modern coded spring-loaded safety valve sized to the boiler's permitted pressure — most jurisdictions now require this for heritage operation regardless of the original valve's condition.
References & Further Reading
- Wikipedia contributors. Boiler explosion. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
