Parsons' steam turbine is a multistage axial-flow reaction turbine that extracts shaft power from steam by expanding it across alternating rings of fixed and moving blades mounted on a rotor drum. Charles Parsons' first 1884 unit ran at 18,000 RPM producing 7.5 kW, and the principle now scales to single shafts above 1,750 MW in modern nuclear stations. The design solves the high blade-tip-speed problem of single-stage turbines by spreading the pressure drop across many small stages, and it powered the famous demonstration of Turbinia at Spithead in 1897.
Parsons Steam Turbine Interactive Calculator
Vary steam jet velocity, velocity ratio, stage count, and safe tip speed to see how splitting expansion reduces blade speed.
Equation Used
The article comparison notes that a 1000 m/s steam jet in a single-stage impulse design would need about half that speed at the blade tip. This calculator generalizes that idea by splitting the same expansion across N stages, so each stage velocity falls as 1/sqrt(N), then blade speed is found from the selected velocity ratio phi = u/c.
- Total expansion energy is split equally across N stages.
- Steam velocity scales with the square root of enthalpy drop.
- Velocity ratio phi equals u/c for the selected turbine row.
- Losses, wetness, leakage, and clearance effects are ignored.
How the Parsons' Steam Turbine Works
A Parsons turbine works by reaction. Steam enters the high-pressure end of a long rotor drum, passes through a ring of fixed (stator) blades that turn the flow and start its expansion, then through a ring of moving (rotor) blades where it expands further and pushes the rotor round. That fixed-moving pair is one stage. A typical large machine carries 50 to 100 stages in series, with blade height growing along the drum as the steam expands and its specific volume rises — inlet blades 25 mm tall, exhaust blades 1 m or more on a 660 MW low-pressure cylinder.
Why split the expansion this way? Because in a single-stage impulse design, getting useful work from high-pressure steam needs a blade velocity around half the steam jet velocity — and a steam jet at 1,000 m/s would demand a tip speed of 500 m/s, which tears a wheel apart. Parsons spread the pressure drop across many stages so each stage handles a small enthalpy drop, each blade row sees a modest steam velocity, and the whole machine can run at a sane tip speed of 200-350 m/s. Blade velocity ratio (u/c) sits near 0.7 for an ideal reaction stage, versus 0.5 for impulse — that's the design signature.
What happens when tolerances drift? Tip clearance is the silent killer. The radial gap between blade tips and the casing must hold to roughly 0.001 × blade height — typically 0.4 to 1.5 mm — and if a thrust bearing wears or the rotor grows differentially on a hot start, blades rub the casing and shed tips. Other failure modes you see in service: solid-particle erosion on the leading edge of HP stage 1 blades from boiler-tube scale, last-stage LP blade erosion from wet steam droplets above 12% wetness, and axial thrust runaway if drum balance pistons leak. None of these are subtle once they start — vibration trips and efficiency loss show up on the dashboard within hours.
Key Components
- Rotor drum: The forged steel drum that carries all moving blade rings. Diameter steps up along its length (drum diameter ratio typically 1:2.5 from HP to LP end on a single-cylinder machine) to accept the volume increase as steam expands. Modern drums are monoblock forgings to avoid the disc-burst risk of bolted construction.
- Fixed blade rings (stator): Rings of aerofoil blades fixed to the casing, between every pair of moving blade rows. They redirect the flow at the correct angle into the next moving row and carry roughly half the stage pressure drop. Blade pitch and stagger angle hold to ±0.25° on finished assemblies.
- Moving blade rings (rotor): Aerofoil blades root-fixed to the drum — fir-tree or T-root for HP stages, pinned-root for long LP stages. They extract work as steam expands across them. Tip clearance must hold 0.4-1.5 mm depending on stage diameter; rub means tip loss and immediate efficiency drop.
- Dummy (balance) piston: A stepped drum section with labyrinth seals at the HP end that uses upstream steam pressure to push the rotor against the natural axial thrust of the reaction blading. Without it, axial thrust on a 60 MW machine reaches 50-100 tonnes — far beyond what a thrust bearing can hold.
- Labyrinth seals: Non-contact strip seals at every shaft penetration and across the dummy piston. Radial clearance 0.3-0.6 mm. They limit steam leakage from one pressure zone to the next; worn seals cost 1-3% on heat rate before any other symptom appears.
- Thrust and journal bearings: Tilting-pad thrust bearing (Michell type) holds residual axial load after the dummy piston, typically 5-15 tonnes. Journal bearings are pressure-fed white-metal pads, oil film 50-100 µm. Bearing metal temperature alarm at 90 °C, trip at 110 °C.
Industries That Rely on the Parsons' Steam Turbine
Parsons' turbine took over from reciprocating engines wherever shaft power needed to be steady, high-speed, and high-density. The 1894 Turbinia trial proved it at sea, and within 15 years every navy and major shipping line had converted. On land it became the universal prime mover of central electricity generation — and still is, even on nuclear plant where the heat source changed but the turbine layout did not. You will find the same blade-row architecture in everything from a 5 MW district-heating set to a 1,755 MW EPR low-pressure module.
- Electric power generation: Large central station turbo-generators — the Arabelle 1,755 MW unit at Flamanville 3 (EPR) and the Siemens SST-9000 fleet on combined-cycle plant both use Parsons-pattern reaction blading on the LP end.
- Marine propulsion (historic): RMS Mauretania (1907) carried four direct-drive Parsons turbines totalling 51 MW and held the Blue Riband for 22 years. The Royal Navy adopted Parsons turbines fleet-wide after Turbinia's 1897 Spithead demonstration.
- Naval propulsion (modern): Rolls-Royce MT30 derivative-class plant on Type 26 frigates uses geared reaction turbines downstream of a free-power section, lineage straight from the Parsons drum design.
- Nuclear power: Sizewell B's two Alstom turbo-alternators, 660 MW each, run a single HP cylinder and three LP cylinders of full reaction blading on saturated steam at 67 bar.
- Industrial cogeneration: Sugar mills, pulp plants and refineries use multi-stage Parsons-pattern back-pressure sets — Siemens SST-300 and MAN MARC ranges from 5 to 50 MW — to take HP boiler steam down to process pressure while generating site electricity.
- Steam-driven boiler feed pumps: Large fossil and nuclear stations drive their main feed pumps with dedicated Parsons-pattern turbines tapped from main-engine extraction steam, typically 25-40 MW per pump set.
The Formula Behind the Parsons' Steam Turbine
The single most useful number when sizing a Parsons stage is the blade velocity ratio, u/c — rotor tangential speed divided by steam velocity leaving the fixed blades. It tells you whether a stage is doing useful work or wasting steam. At the low end of typical operating range, u/c near 0.5, the stage extracts maybe 60% of its potential work and the rest leaves as exit kinetic energy you can't recover. At the design sweet spot of u/c = 0.707 (1/√2 for a 50% reaction stage) the diagram efficiency hits its peak around 92-94%. Push beyond u/c = 0.85 chasing higher speed and efficiency drops again because the relative steam velocity onto the moving blade falls and the blades start churning rather than extracting. This is why a Parsons machine carries so many stages — each one is sized to sit on its own velocity-ratio peak.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| ηdiagram | Stage diagram (blading) efficiency — fraction of available kinetic energy converted to rotor work | dimensionless | dimensionless |
| u | Mean blade tangential velocity at the pitch line | m/s | ft/s |
| c | Absolute steam velocity leaving the fixed (nozzle) blade row | m/s | ft/s |
| α | Nozzle exit angle measured from the tangential direction | degrees | degrees |
| ψ | Moving-blade velocity coefficient (ratio of relative exit to relative inlet velocity, accounts for friction) | dimensionless | dimensionless |
Worked Example: Parsons' Steam Turbine in a 25 MW geothermal turbo-alternator
You are setting blade-row geometry for stage 12 of a recommissioned 25 MW Parsons-pattern geothermal turbo-alternator being returned to service at the Wairakei power station in New Zealand, where the unit takes saturated steam at 13 bar from the wellhead separators and runs at 3000 RPM synchronous. Stage 12 sits mid-cylinder with a mean blade pitch diameter of 1.10 m, nozzle exit angle 18°, and a moving-blade velocity coefficient ψ = 0.88. You want to verify the stage is running near its diagram-efficiency peak, and check what happens at the typical operating envelope of u/c = 0.55 (off-design low) and u/c = 0.85 (off-design high) versus the design-point u/c = 0.707.
Given
- Dm = 1.10 m
- N = 3000 RPM
- α = 18 degrees
- ψ = 0.88 —
- (u/c)design = 0.707 —
Solution
Step 1 — calculate mean blade tangential velocity at synchronous speed:
Step 2 — at the design point u/c = 0.707, back out the required nozzle exit velocity and compute diagram efficiency:
That is 92.8% diagram efficiency — the sweet spot. The blade row is converting nearly all of the available kinetic energy into shaft work, and exit swirl is low so the next stator can pick the flow up cleanly.
Step 3 — at the low end of the operating envelope, u/c = 0.55 (this happens at part-load or after a reservoir pressure drop):
You have lost roughly 7.6 percentage points of stage efficiency. On a 25 MW machine that is around 1.9 MW of fuel-equivalent heat going out the condenser instead of out the generator terminals — you would feel this as a measurable rise in heat rate within the first hour of off-design running.
Step 4 — at the high end u/c = 0.85 (overspeed transient or a partial-arc admission excursion):
Slightly higher diagram efficiency on paper, but in practice you will not see this number — at u/c above 0.8 the stage starts to develop negative reaction at the root, the moving blades go into churning, and rotor windage loss eats the gain. Useful operating band is 0.65 to 0.78.
Result
At the design point u/c = 0. 707, stage 12 delivers 92.8% diagram efficiency with a nozzle exit velocity of 244 m/s and blade tangential speed of 173 m/s. That is the sweet spot Parsons sized for — small exit swirl, clean handover to the next fixed row, and full reaction split across the blading. At u/c = 0.55 efficiency falls to 85.2% (a 1.9 MW heat-rate penalty on a 25 MW set), and at u/c = 0.85 the textbook number rises slightly to 93.7% but real-world windage and root negative-reaction kill the gain. If you measure stage efficiency 3-4 points below predicted on test, suspect three things in this order: nozzle-block erosion opening α beyond the blueprint 18° (steam now lands at the wrong angle on the moving blade leading edge), moving-blade tip rub from a thrust-bearing wear-back (radial clearance opens past 1.5 mm and tip leakage doubles), or fouled labyrinth seals across the diaphragm letting stage steam bypass the blading entirely.
Parsons' Steam Turbine vs Alternatives
The Parsons reaction turbine is not the only way to extract shaft power from steam. Three architectures dominate the design space: pure reaction (Parsons), pure impulse with velocity-compounding (Curtis), and impulse with pressure-compounding (Rateau / De Laval staged). Each wins on a different axis, and large modern machines often combine them — Curtis first stage for governing, Parsons reaction for the bulk of the expansion.
| Property | Parsons reaction turbine | Curtis (velocity-compounded impulse) | Rateau (pressure-compounded impulse) |
|---|---|---|---|
| Optimum blade velocity ratio (u/c) | 0.70-0.78 per stage | 0.20-0.25 (2-row Curtis) | 0.45-0.50 per stage |
| Stages required for 600 MW expansion | 50-90 | Not viable alone — 4-6 Curtis wheels max | 20-30 |
| Typical peak diagram efficiency | 92-94% | 65-75% | 82-86% |
| Relative manufacturing cost (large utility scale) | High — many small stages, tight tip clearance | Low — few stages, loose clearances | Medium |
| Tip clearance sensitivity | High — 1% extra clearance costs 1.5% efficiency | Low — impulse blading less sensitive | Medium |
| Best application fit | Large utility, marine, nuclear LP | Governing stage, small auxiliaries, mechanical drives | Mid-size industrial, back-pressure cogen |
| Axial thrust | Large — needs dummy piston | Small — pressure drop in nozzle, not blading | Moderate — diaphragm seals balance most of it |
| Tolerance to wet steam (last stages) | Good with hardened LE shields | Fair | Fair to poor |
Frequently Asked Questions About Parsons' Steam Turbine
The Curtis stage is there to drop a large pressure ratio in one wheel under partial-arc admission, so the governor valves only feed steam to a fraction of the nozzle ring at part load. You cannot do partial-arc admission on reaction blading because reaction needs full circumferential pressure drop across the moving row — admit steam to half the arc and the unloaded half acts as a brake.
So you pay 10-15 efficiency points on stage 1 to get clean part-load control, and the downstream Parsons stages then run at full arc near their efficiency peak. On a typical 600 MW machine the Curtis stage handles 80-150 °C of the total 350 °C drop.
Mass continuity, applied stage by stage. The volumetric flow ṁ × v rises as steam expands (specific volume v can grow 200× from HP inlet to LP exhaust on a condensing machine), and you size each stage's annulus area A = ṁv/cax at a chosen axial velocity, typically 80-120 m/s.
That gives you blade height h = A/(π × Dm). Hold axial velocity roughly constant down the cylinder and the heights fall out automatically — 25 mm at HP inlet, 200 mm mid-cylinder, 1+ m at LP exhaust on a large machine. CFD is for tuning aerofoil profile loss, not for setting the macro geometry.
Check the surface finish of the HP fixed blades. Solid-particle erosion from boiler-tube exfoliation roughens the trailing edge progressively — Ra creeps from a polished 0.4 µm to 3-5 µm over a decade, and aerofoil profile loss roughly doubles. Borescope cameras don't pick this up because the blades still look the right shape; you need a finger-and-fingernail check or a replica tape.
The other slow killer is dummy-piston labyrinth wear. Each strip wears 0.05-0.1 mm per 10,000 hours, and after 80,000 hours leakage is double the original spec — that's another 1-1.5% on heat rate before anything else shows up.
Sometimes — but only if the original rotor was conservatively stressed. The retrofit gives you 1.5-3% on cylinder efficiency from better aerofoil profile and controlled-vortex stacking, but it usually raises stage reaction at the root by 5-10%, which raises axial thrust. If the original dummy piston was sized tight, you trip the thrust bearing on the first hot start.
The decision rule: pull the OEM thrust margin number from the original commissioning file. Below 25% margin, don't retrofit without re-sizing the dummy piston. Above 40% margin, retrofit is usually clean.
That pattern means flow separation on the suction surface, not droplet impact. It happens when the stage is running at part-load with too low a volumetric flow — incidence angle goes negative, the suction surface stalls, and recirculating wet steam droplets pile up on the trailing edge.
Check your minimum-load running hours. Modern combined-cycle plant cycling below 40% load racks up this damage fast — 5,000 hours at low load can equal 100,000 hours of base-load wear on the same blade. The fix is a minimum-flow operating limit, not a blade-material change.
Worse, and you have to plan for it. On full load rejection the steam keeps flowing for several seconds while valves close, and reaction blading continues to extract work — but with no electrical load the rotor accelerates fast. Overspeed of 8-12% in 200 ms is typical before the trip valves shut.
Impulse stages decelerate quicker because the pressure drop is in the nozzle and stops the moment upstream pressure falls. This is why Parsons machines always carry an independent mechanical or electronic overspeed trip set at 110-112% — not a luxury, a survival requirement.
Yes — first-stage shell pressure relative to throttle pressure at a known load. On a healthy machine the ratio holds within ±1% of the commissioning number across the load range. If first-stage pressure drifts up at constant flow, dummy-piston or HP gland labyrinths are leaking and bypassing the blading.
Rule of thumb: 3% rise in first-stage pressure ratio at rated flow corresponds to roughly 1.5% extra steam bypassing the HP cylinder. That's worth a casing lift at the next outage.
References & Further Reading
- Wikipedia contributors. Steam turbine. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
