Turbomachinery cover
Interactive book · Semester 4-7

Turbomachinery

Draw the velocity triangles, spin a Pelton wheel, push a pump into cavitation and walk a compressor up to surge

25 chapters, 25 interactives, few words. This book covers the turbomachines course of mechanical engineering from the rotor up: classification and similarity, stagnation states and polytropic efficiency, Euler's equation, degree of reaction and utilisation factor, jets on vanes, radial and axial stages, compounded and reaction steam turbines, hydro plants with Pelton, Francis and Kaplan turbines, draft tubes and cavitation, centrifugal pumps, NPSH, pumps in series and parallel, fans, centrifugal and axial compressors, surge and gas turbine stages. One centrifugal pump runs through the pump chapters and real Indian hydro stations through the turbine chapters, so the numbers describe machines you could visit.

  • AUCAE353Turbo Machines
  • VTUBME502Turbomachines
  • SPPU402043Turbomachinery
  • MUMEC602Turbo Machinery
  • KTUPEMET411Turbo Machinery

By Skill-Lync · Updated 22 Nov 2026 · 25 chapters · 25 interactives · free, no sign-up

Chapter 1

What a turbomachine is: classification, parts and positive-displacement machines

Which syllabus covers thisAU CAE353 Unit 1VTU BME502 Module 1SPPU 402043 Unit 1MU MEC602 Module 1; Module 3 (positive displacement pumps)KTU PEMET411 Module 1

What makes a machine a turbomachine?

A turbomachine exchanges energy with a fluid that flows through it without stopping, by changing the fluid's angular momentum as it passes a spinning row of blades. There is no sealed chamber and no valve. The book's pump (assumed numbers in the class of a 1450 rpm end-suction pump) shows the scale: its 260 mm impeller has a tip speed of 19.7 m/s, and at its best point it lifts 26.8 L/s through 23 m, putting 6.05 kW into the water for 8 kW at the shaft. If the blades push the fluid it is a pump, fan or compressor; if the fluid pushes the blades, a turbine.

What are the parts, and what does each do?

Every turbomachine has a rotor (impeller, runner or blade row) on a shaft, where all the energy exchange happens; a stator (guide vanes, nozzles, diffuser or volute) that only turns the flow and speeds it up or slows it down, doing no work; and a casing that holds the pressure and guides the fluid in and out, plus bearings and a shaft seal. Water leaves the book's pump impeller with a whirl speed of about 12.4 m/s; the volute slows it and turns that kinetic energy into pressure. In a Francis turbine the guide vanes do the reverse: they speed the water up and give it whirl before the runner.

How are turbomachines classified?

By the direction of energy flow: power-absorbing machines (pumps, fans, blowers, compressors) and power-producing ones (hydraulic, steam, gas and wind turbines). By the path through the rotor: axial, radial or mixed flow. By the fluid: hydraulic machines treat it as incompressible, thermal machines must track its density. By where the pressure falls: in an impulse machine all of it falls in the stationary nozzles, in a reaction machine part falls in the rotor. India runs all of them: Idukki's six-jet Pelton wheels under 660 m, Teesta Low Dam III's Kaplans under 21.34 m, and BHEL's 500 MW steam turbine with 17 reaction stages in its high-pressure cylinder.

How is a turbomachine different from a positive-displacement machine?

A positive-displacement machine traps a fixed volume and pushes it out: its flow is set by speed, its pressure by whatever the system demands. A turbomachine is the other way round. Shut the discharge valve in the interactive: the book's pump settles at its shut-off head of 28.2 m and its shaft power falls from 8 to 1.6 kW. The model gear pump (1.2 L per revolution) keeps pushing; its pressure climbs until the relief valve opens, settling at 58.9 m with 26.6 L/s going back to suction and 19.7 kW turning into heat. A positive-displacement pump must never run without a relief valve.

Turbomachine or positive-displacement machine: head against flow for a centrifugal pump and a gear pump, and what a closed valve does to each: interactive loads as you scroll

What does a reciprocating pump add: slip, indicator diagram and air vessel?

A reciprocating pump is positive displacement too. A single-acting 150 mm piston with a 300 mm stroke at 40 rpm sweeps 3.534 L/s; if 3.4 L/s arrives, the slip is 3.8 %. Its indicator diagram (cylinder pressure over one stroke) would be a rectangle between the suction and delivery heads, but the water in the pipes must be accelerated and stopped every stroke. With 8 m of 100 mm suction pipe the acceleration head at the start of the stroke is 4.83 m, which tilts the rectangle and risks separation. An air vessel next to the cylinder leaves only about 1 m of pipe to accelerate: 0.6 m.

Predict, then check

Both pumps run at 1450 rpm and the discharge valve is shut fully. Which one now draws MORE shaft power than with the valve open?

Worked examples

Energy per kilogram, water power and efficiency of the book's pump · The book's pump: impeller D2 = 260 mm at 1450 rpm; best point 26.85 L/s at 22.98 m with 8.01 kW at the shaft; ρ = 1000 kg/m³, g = 9.81 m/s²
  1. Tip speed U2 = πD2N/60 = π × 0.26 × 1450/60 = 19.74 m/s
  2. Energy given to each kg of water: gH = 9.81 × 22.98 = 225.5 J/kg
  3. Upper limit for radial blades with no slip and no loss: U2²/g = 19.74²/9.81 = 39.7 m; the real pump makes 23 m (backward-curved blades, slip and friction)
  4. Water power = ρgQH = 1000 × 9.81 × 0.02685 × 22.98 = 6.05 kW
  5. Overall efficiency = 6.05/8.01 = 75.6 %

Answer: U2 ≈ 19.7 m/s, 225 J/kg, 6.05 kW into the water, η ≈ 75.6 %

A gear pump against 40 m, then against a shut valve · Model external gear pump: 1.2 L per revolution at 1450 rpm, slip 0.04 L/s per metre of head, hydro-mechanical efficiency 0.85; relief valve set at 50 m, passing 3 L/s per metre above that; water, ρ = 1000
  1. Theoretical flow Qth = 1.2 × 1450/60 = 29 L/s
  2. At 40 m: slip = 0.04 × 40 = 1.6 L/s, delivered Q = 27.4 L/s, volumetric efficiency 94.5 %
  3. Shaft power = ρgQthH/ηhm = 1000 × 9.81 × 0.029 × 40/0.85 = 13.39 kW; water power = 1000 × 9.81 × 0.0274 × 40 = 10.75 kW; overall η = 80.3 %
  4. Valve shut, Q = 0: 29.0 − 0.04H − 3(H − 50) = 0, so H = (29.0 + 150)/3.04 = 58.88 m
  5. Relief flow 3 × 8.88 = 26.64 L/s; shaft power 1000 × 9.81 × 0.029 × 58.88/0.85 = 19.71 kW, all of it heat

Answer: 27.4 L/s at 80.3 % overall; with the valve shut it holds 58.9 m and burns 19.7 kW

Slip and acceleration head of a reciprocating pump, with and without an air vessel · Single-acting reciprocating pump (original): bore 150 mm, stroke 300 mm, 40 rpm; measured delivery 3.4 L/s; suction pipe 8 m long, 100 mm bore; then an air vessel fitted 1 m from the cylinder
  1. Swept flow = (π/4) × 0.15² × 0.3 × 40/60 = 3.534 L/s
  2. Slip = (3.534 − 3.4)/3.534 = 3.8 %; coefficient of discharge 0.962
  3. ω = 2π × 40/60 = 4.189 rad/s, crank radius r = 0.15 m, area ratio A/a = (150/100)² = 2.25
  4. Acceleration head at the stroke ends ha = (L/g)(A/a)ω²r = (8/9.81) × 2.25 × 4.189² × 0.15 = 4.83 m
  5. With the air vessel only 1 m accelerates: ha = 0.6 m; beyond the vessel the pipe flows steadily at Q/a = 0.45 m/s

Answer: Slip 3.8 %; the acceleration head drops from 4.83 m to 0.6 m with the air vessel

Chapter 2

Dimensionless groups, unit quantities, specific speed and model testing

Which syllabus covers thisAU CAE353 Unit 1VTU BME502 Module 1SPPU 402043 Unit 1 (unit quantities); Unit 3 (specific speed)MU MEC602 Module 1KTU not in syllabus

Which dimensionless groups describe a turbomachine?

Four groups carry almost everything: flow coefficient φ = Q/(ND³), head coefficient ψ = gH/(N²D²), power coefficient Π = P/(ρN³D⁵) and the Reynolds number, with N in rev/s. For the book's pump at its best point: φ = 0.0632, ψ = 5.71, Π = 0.478. Any geometrically similar pump, at any speed, that runs at the same φ has the same ψ, Π and efficiency. That is where the affinity laws come from: at 1200 rpm the same pump gives 22.2 L/s at 15.7 m for 4.54 kW. Dimensional analysis itself is in the Fluid Mechanics book.

What are unit speed, unit discharge and unit power for?

They scale one turbine's speed, flow and power to a head of 1 m: Nu = N/√H, Qu = Q/√H, Pu = P/H^1.5. One Teesta-V unit (NHPC, Sikkim: 197 m, 93.67 m³/s, 214.3 rpm, 170 MW) has Nu = 15.27, Qu = 6.673 and Pu = 61.48 kW. If the reservoir drew down from 579 m to 568 m and the head fell to 186 m, the same flow pattern would need 208.2 rpm and pass 91 m³/s for 156 MW. The generator is locked to 214.3 rpm by the 50 Hz grid, so the turbine runs slightly off its best point instead.

What is specific speed and why does it pick the machine type?

Specific speed folds speed, power (or flow) and head into one number that stays the same for every size of one design: Ns = N√P/H^1.25 for turbines (rpm, kW, m) and Ns = N√Q/H^0.75 for pumps (rpm, m³/s, m). It sorts shapes: Idukki's six-jet Pelton wheels give 40 (about 17 per jet), a Teesta-V Francis 120, a Teesta Low Dam III Kaplan 540, and the book's pump 22.6. Low Ns means high head and little flow: narrow, large runners. The Fluid Mechanics book's turbine interactive picks a type from it; here it is the number model and prototype share.

How do you test a 170 MW turbine in a laboratory?

Build a homologous model (every dimension scaled, the same blade angles) and run it at the same φ and ψ. A 1:10 model of the Teesta-V runner (4.02 m, derived) is 402 mm across. A lab cannot supply 197 m, so test at 30 m: equal ψ gives Nm = Np λ √(Hm/Hp) = 836 rpm, equal φ gives Qm = 0.366 m³/s, and at 90 % efficiency the model makes 96.8 kW. Its N11 = ND/√H = 61.4 equals the prototype's, so the test maps straight onto the real machine. IEC 60193 sets the rules for such tests of machines above 5 MW or 3 m.

Testing a model: scale ratio and test head to the prototype's speed, flow and power, unit quantities and Moody's efficiency step-up: interactive loads as you scroll

Why is the big machine more efficient than its model?

Friction depends on the Reynolds number, Re = uD/ν, and it is not kept equal: the 1:10 model at 30 m runs at Re = 7.1 million, the prototype at 182 million. Relative friction is smaller in the big machine. Moody's formula 1 − ηp = (1 − ηm)(Dm/Dp)^(1/5) turns the model's 90 % into 93.69 %; Hutton's, which treats 30 % of the loss as not scalable, gives 93.34 %. The catch: IEC 62097 (2009) says the step-up formulas of IEC 60193 and earlier standards overstated the gain, so it steps up each component's losses separately, roughness included.

Predict, then check

A 1:10 model of the Teesta-V Francis runner (prototype 214.3 rpm at 197 m) is tested under 30 m. To stay similar it must run at about:

Worked examples

Speed, flow and power of a Teesta-V model, and its stepped-up efficiency · Prototype: one Teesta-V unit, 197 m, 93.67 m³/s, 214.3 rpm, runner inlet diameter 4.02 m (derived). Model 1:8 tested at 25 m; measured efficiency 91 %; ν = 1.0 × 10⁻⁶ m²/s, ρ = 1000
  1. Dm = 4.022/8 = 0.503 m
  2. Equal ψ: Nm = 214.3 × 8 × √(25/197) = 610.7 rpm
  3. Equal φ: Qm = 93.67/64 × √(25/197) = 0.5214 m³/s
  4. Model power = 1000 × 9.81 × 0.5214 × 25 × 0.91 = 116.4 kW
  5. Moody: ηp = 1 − 0.09 × (1/8)^0.2 = 94.06 %
  6. Hutton: Re_p/Re_m = 22.46; 0.3 + 0.7 × (1/22.46)^0.2 = 0.6757; ηp = 1 − 0.09 × 0.6757 = 93.92 %
  7. Prototype power with Moody's η: 1000 × 9.81 × 93.67 × 197 × 0.9406 = 170.3 MW, close to the 170 MW rating

Answer: 611 rpm, 0.521 m³/s, 116 kW; prototype η 94.1 % (Moody) or 93.9 % (Hutton)

Unit quantities of a Teesta Low Dam III Kaplan unit at a lower head · One unit (NHPC): 33 MW at a rated head of 21.34 m, 693.6/4 = 173.4 m³/s, 136.36 rpm. The head falls to 18 m and the turbine keeps the same flow pattern (same unit quantities)
  1. Nu = 136.36/√21.34 = 29.52; Qu = 173.4/√21.34 = 37.54; Pu = 33,000/21.34^1.5 = 334.8 kW
  2. At 18 m: N = 29.52 × √18 = 125.2 rpm
  3. Q = 37.54 × √18 = 159.3 m³/s
  4. P = 334.8 × 18^1.5 = 25,564 kW
  5. Efficiency unchanged: at rating 33 MW/(ρgQH) = 33 × 10⁶/(1000 × 9.81 × 173.4 × 21.34) = 90.9 %

Answer: About 125 rpm, 159 m³/s and 25.6 MW; the real machine stays at 136.36 rpm and moves its blades instead

Scaling the book's pump up 2:1 and slowing it to 960 rpm · The book's pump at its best point: 26.85 L/s, 22.98 m, 8.01 kW, η 75.6 %, D2 = 260 mm, 1450 rpm. A geometrically similar pump twice the size (520 mm) runs at 960 rpm
  1. Q2 = Q1 (N2/N1)(D2/D1)³ = 26.85 × (960/1450) × 8 = 142.2 L/s
  2. H2 = H1 (N2/N1)²(D2/D1)² = 22.98 × (960/1450)² × 4 = 40.3 m
  3. P2 = P1 (N2/N1)³(D2/D1)⁵ = 8.01 × (960/1450)³ × 32 = 74.4 kW at the same efficiency
  4. Check: Ns = 1450 √0.02685/22.98^0.75 = 22.63 and 960 √0.1422/40.3^0.75 = 22.63: one design
  5. Moody step-up for the bigger pump: η = 1 − 0.244 × (0.5)^0.2 = 78.7 %, so the shaft power is nearer ρgQH/η = 71.4 kW

Answer: 142 L/s at 40.3 m; about 71 kW with the stepped-up 78.7 %

Chapter 3

Thermodynamics of turbomachines: static and stagnation states, total-to-total and total-to-static efficiency

Which syllabus covers thisAU CAE353 Unit 1 (efficiencies)VTU BME502 Module 1SPPU not in syllabusMU MEC602 Module 1 (review)KTU PEMET411 Module 1

What does the first law say about a turbomachine?

For steady flow through a machine, per kg: q − w = (h2 − h1) + (C2² − C1²)/2 + g(z2 − z1). The fluid passes so fast that little heat escapes (adiabatic), and for gases the height term is negligible, so the work is the change of stagnation enthalpy h0 = h + C²/2: w = h01 − h02. The interactive's turbine stage drops T0 from 1200 K by 189.7 K: w = 1.005 × 189.7 = 190.7 kJ/kg, or 1.91 MW for 10 kg/s of gas. In a pump the useful part of the same energy is the rise of p/ρ + C²/2 + gz, which is gH per kg.

What is the difference between static and stagnation states?

The static state is what the moving fluid itself has: the temperature and pressure an instrument travelling with it would read. The stagnation state is what it would reach if brought to rest with no loss: T0 = T + C²/(2cp) and p0/p = (T0/T)^(γ/(γ − 1)). At the interactive's compressor inlet, air with stagnation 288.15 K and 101.325 kPa (ISA sea level) moving at 150 m/s has T = 276.96 K and p = 88.2 kPa: the static pressure is 12.9 % lower, at Mach 0.45. At 30 m/s the temperature gap would be only 0.45 K.

How does the second law show up on the h-s diagram?

An ideal (reversible, adiabatic) stage moves straight up or down at constant entropy. Friction, mixing and shocks raise the entropy, so the real exit sits to the right of the ideal one on the same pressure line. In the interactive's compressor stage (total pressure ratio 1.8, ηtt 88 %) entropy rises by 21 J/kg K and the work grows from the ideal 53 to 60.2 kJ/kg. Pressure lines slope upward at dh/ds = T and spread apart, so the same entropy rise costs more when the gas is hotter. The Engineering Thermodynamics book shows one process this way on T-s.

What are total-to-total and total-to-static efficiency?

Both divide the actual work by an ideal work between the same pressures; they differ in what counts as useful at the exit. Total-to-total compares with the isentropic change to the exit STAGNATION pressure, so the exit kinetic energy is kept. Total-to-static compares with the change to the exit STATIC pressure, so the exit kinetic energy counts as lost. In the interactive's turbine stage at ηtt = 88 % and C2 = 250 m/s, ηts = 77.2 %; the compressor stage with C2 = 120 m/s gives ηts = 76.3 %. ηts is never higher than ηtt.

Static and stagnation states on the h-s diagram: kinetic energy, total-to-total and total-to-static efficiency: interactive loads as you scroll

When is each efficiency the right one?

Use total-to-static when the exit kinetic energy is thrown away: the last stage of a steam turbine before the condenser, or a fan blowing into a room. Use total-to-total when the next stage, a diffuser or a nozzle uses it: intermediate stages, or a gas turbine feeding a jet nozzle. Exit velocity alone moves ηts: in the interactive's turbine, raising C2 from 100 to 300 m/s drops ηts from 86.1 to 73.2 % while ηtt stays at 88 %. That is why last stages get tall blades and a small leaving velocity.

Predict, then check

A turbine stage keeps its pressures, its work and ηtt = 88 %, but its exit velocity rises from 100 to 300 m/s. What happens?

Worked examples

Static temperature, pressure and Mach number at a compressor intake · Air at a compressor inlet: stagnation 288.15 K and 101.325 kPa (ISA sea level), velocity 180 m/s; cp = 1005 J/kg K, γ = 1.4, R = 287 J/kg K
  1. T = T0 − C²/(2cp) = 288.15 − 180²/2010 = 272.03 K
  2. p = p0 (T/T0)^(γ/(γ − 1)) = 101.325 × (272.03/288.15)^3.5 = 82.83 kPa
  3. Speed of sound a = √(γRT) = √(1.4 × 287 × 272.03) = 330.6 m/s; M = 180/330.6 = 0.544
  4. Density ρ = p/(RT) = 82835/(287 × 272.03) = 1.061 kg/m³, against 1.225 kg/m³ at rest

Answer: T ≈ 272 K, p ≈ 82.8 kPa, M ≈ 0.54; the air is 13.4 % less dense than at rest

Total-to-total and total-to-static efficiency of a compressor stage · Air enters at 100 kPa, 300 K (stagnation, negligible velocity); total pressure ratio 1.6; ηtt = 0.86; exit velocity 140 m/s; cp = 1005 J/kg K, γ = 1.4
  1. T02s = 300 × 1.6^(0.4/1.4) = 343.12 K
  2. T02 = 300 + (343.12 − 300)/0.86 = 350.14 K; work w = 1.005 × 50.14 = 50.39 kJ/kg
  3. Exit static T2 = 350.14 − 140²/2010 = 340.38 K; p2 = 160 × (340.38/350.14)^3.5 = 144.94 kPa
  4. Isentropic to the static exit pressure: T2s = 300 × (144.94/100)^(0.4/1.4) = 333.56 K
  5. ηts = (333.56 − 300)/(350.14 − 300) = 66.9 %

Answer: w ≈ 50.4 kJ/kg; ηts ≈ 66.9 % against ηtt = 86 %, because the exit's 9.8 kJ/kg of kinetic energy counts as lost

Power, total-to-static efficiency and leaving loss of a gas turbine stage · Gas (treated as air) enters at 1100 K, 450 kPa (stagnation); total pressure ratio p01/p02 = 2.2; ηtt = 0.90; exit velocity 220 m/s; 12 kg/s; cp = 1005 J/kg K, γ = 1.4
  1. T02s = 1100 × (1/2.2)^(0.4/1.4) = 878.13 K
  2. T02 = 1100 − 0.90 × (221.87) = 900.31 K; w = 1.005 × 199.69 = 200.68 kJ/kg; power = 12 × 200.68 = 2.408 MW
  3. T2 = 900.31 − 220²/2010 = 876.24 K; p2 = 204.55 × (876.24/900.31)^3.5 = 186.03 kPa
  4. T2s = 1100 × (186.03/450)^(0.4/1.4) = 854.64 K; ηts = 199.69/245.36 = 81.4 %
  5. Leaving kinetic energy 220²/2 = 24.2 kJ/kg = 12.1 % of the work

Answer: 2.41 MW; ηts ≈ 81.4 %; the leaving loss is 12.1 % of the work if no stage follows

Chapter 4

Stage, overall and polytropic efficiency, reheat factor and preheat factor

Which syllabus covers thisAU CAE353 Unit 1 (stage efficiencies)VTU BME502 Module 1SPPU not in syllabusMU not in syllabusKTU PEMET411 Module 1 (preheat factor)

How do stage efficiency and overall efficiency differ?

Stage efficiency is the isentropic efficiency of one stage across its own pressure ratio; overall efficiency is the same idea across the whole machine. They are not equal even when every stage is identical. The interactive's gas turbine expands air from 1400 K over a pressure ratio of 10 in 5 equal stages of 85 % each, a ratio of 1.585 per stage. Overall it reaches 88.18 %, better than any of its stages. The same 5 stages compressing air from 288.15 K over a ratio of 10 reach only 80.93 %.

Why is the reheat factor greater than 1?

Friction in one stage turns some work back into heat, so the next stage starts hotter than an ideal expansion would leave it. Constant-pressure lines slope at dh/ds = T and spread apart, so a hotter start means a bigger isentropic drop to the next pressure. Add the stage drops and they exceed the single isentropic drop of the whole machine: the reheat factor RF = 1.0374 for the 5-stage default, and overall η = RF × stage η = 88.18 %. With 20 stages RF = 1.0434; the limit is 88.86 %.

Many small stages: stage, overall and polytropic efficiency, and why the reheat factor beats 1 in a turbine: interactive loads as you scroll

What happens in a multistage compressor?

The same fan-out works against you. Friction heats the air, so each following stage must compress hotter air, which costs more work for the same pressure ratio. The sum of the stage isentropic rises exceeds the overall isentropic rise; that ratio is the preheat factor, 1.0503 for 5 stages of 85 % over a pressure ratio of 10, and the overall efficiency is stage η divided by it: 80.93 %. With many tiny stages it falls towards 79.66 %. A high-ratio compressor is always less efficient overall than its stages.

What is polytropic efficiency and why do designers quote it?

It is the efficiency of an infinitely small stage, held constant along the whole compression or expansion. It does not depend on the pressure ratio, so it compares machines fairly: a compressor with ηp = 88 % has an overall efficiency of 85.5 % at a pressure ratio of 4 but only 81.4 % at 30. Every finite stage has its own ηp too: a stage of 85 % across a ratio of 1.585 has ηp = 84.14 % in a turbine and 85.94 % in a compressor, and for a perfect gas a machine built from such stages has exactly that ηp.

How big is the reheat factor in a steam turbine?

Steam behaves the same way, and Indian textbooks read it off the h-s (Mollier) chart. In the interactive, expand steam from 100 bar and 500 °C to 1 bar in 8 equal-ratio stages of 85 %: RF = 1.0373, overall η = 88.17 %, and the exhaust leaves wet at dryness 0.926. Real turbines have many stages: BHEL's 500 MW unit has 17 reaction stages in its high-pressure cylinder alone. The reheat factor is not reheating, which sends steam back to the boiler between cylinders (see the Thermal Engineering book).

Predict, then check

Ten identical stages, each 85 % efficient, expand air from 1400 K over a pressure ratio of 10. The overall efficiency is:

Worked examples

Reheat factor of a three-stage air turbine · Air (cp 1005 J/kg K, γ 1.4) enters a 3-stage turbine at 1300 K; overall pressure ratio 8, split equally; every stage 87 % efficient
  1. Stage pressure ratio 8^(1/3) = 2; isentropic temperature ratio 2^(−0.2857) = 0.8203
  2. Stage 1: T1s = 1300 × 0.8203 = 1066.4 K; T1 = 1300 − 0.87 × 233.6 = 1096.8 K
  3. Stage 2: T2s = 899.7 K; T2 = 1096.8 − 0.87 × 197.1 = 925.4 K
  4. Stage 3: T3s = 759.1 K; T3 = 780.7 K
  5. Σ stage isentropic drops = 1.005 × (233.6 + 197.1 + 166.3) = 599.9 kJ/kg
  6. Overall isentropic drop = 1.005 × (1300 − 717.7) = 585.3 kJ/kg; RF = 599.9/585.3 = 1.025
  7. Overall η = 1.005 × (1300 − 780.7)/585.3 = 89.17 % = 1.025 × 87 %

Answer: RF ≈ 1.025; overall efficiency ≈ 89.2 %, exit 781 K

Preheat factor and polytropic efficiency of a four-stage compressor · Air at 300 K is compressed in 4 stages of equal pressure ratio to an overall ratio of 6; each stage 84 % efficient; cp 1005 J/kg K, γ 1.4
  1. Stage ratio 6^(1/4) = 1.5651
  2. Stage temperatures (T rises by ΔTs/0.84 each time): 348.8 K, 405.4 K, 471.4 K, 548 K
  3. Actual work = 1.005 × (548 − 300) = 249.2 kJ/kg; ideal overall = 1.005 × (500.6 − 300) = 201.6 kJ/kg
  4. Overall η = 201.6/249.2 = 80.88 %; Σ stage ideal rises = 209.3 kJ/kg, PF = 1.0386, and 84/1.0386 = 80.88 %
  5. Polytropic efficiency ηp = (γ − 1)/γ × ln r / ln(1 + (r^0.2857 − 1)/0.84) = 84.98 %, the same for one stage and the whole machine
  6. With infinitely many stages of 84 %: η = (6^0.2857 − 1)/(6^(0.2857/0.84) − 1) = 79.64 %

Answer: Overall 80.9 %, PF 1.039, ηp 85 %

Reheat factor of a three-stage steam turbine · Steam at 40 bar, 400 °C expands to 2 bar in 3 stages of equal pressure ratio; each stage 80 % efficient; IAPWS-IF97 properties
  1. Stage ratio 20^(1/3) = 2.714; stage exits at 14.74, 5.43 and 2 bar
  2. Inlet h = 3214.4 kJ/kg. Stage 1: isentropic to 2951.6, actual h = 3214.4 − 0.8 × 262.7 = 3004.2 kJ/kg
  3. Stage 2 (from the new, higher entropy): isentropic drop 220, h = 2828.2 kJ/kg
  4. Stage 3: isentropic drop 185.6, h = 2679.7 kJ/kg, dryness 0.988
  5. Σ stage drops = 668.4 kJ/kg; one isentropic drop from 40 bar to 2 bar = 648 kJ/kg; RF = 1.0314
  6. Overall η = (3214.4 − 2679.7)/648 = 82.51 % = 1.0314 × 80 %

Answer: RF ≈ 1.031; overall efficiency ≈ 82.5 %

Chapter 5

Euler's turbine equation and its three energy components

Which syllabus covers thisAU CAE353 Unit 1VTU BME502 Module 2SPPU 402043 Units 1, 4 (Euler work)MU MEC602 Module 1KTU PEMET411 Module 1

Where does a rotor's torque come from?

Fluid carries angular momentum about the shaft: r × Ct per kg, where Ct is the whirl, the tangential part of its velocity. A rotor can change that only by pushing on the fluid, so the torque is T = ṁ(r2Ct2 − r1Ct1). In the book's pump at the interactive's default, 28.7 L/s enters with no whirl and leaves the 130 mm tip with Ct2 = 15.45 m/s: T = 57.7 N·m. Times ω = 151.8 rad/s that is 8.76 kW, the ideal power before slip and losses (chapter 18).

What is Euler's turbomachine equation?

Divide that power by the mass flow and the radius turns into blade speed, U = ωr: energy per kg e = U2Ct2 − U1Ct1. A pump, fan or compressor adds e; a turbine takes w = U1Ct1 − U2Ct2. Losses, the fluid and the blade count do not appear, only the velocity triangles. For water e/g is the head: the book's pump in the interactive gives 305 J/kg, a Euler head of 31.1 m. The same rotor in air adds the same J/kg but about 830 times less pressure, because air is that much lighter.

Euler's equation in parts: the dynamic, centrifugal and relative-velocity shares of the energy a rotor exchanges: interactive loads as you scroll

What do the three components mean?

Expand the triangles with the cosine rule and e splits into three parts. ½(C2² − C1²) is kinetic energy gained: the fluid leaves the impeller fast, and the volute or diffuser must slow it down to turn it into pressure. ½(U2² − U1²) is the centrifugal part: fluid spun at a larger radius presses outwards, like water on the wall of a spinning bucket. ½(W1² − W2²) is diffusion in the blade passages: the relative flow slows, so its pressure rises. In the book's pump: 12.2 m + 16.3 m + 2.6 m = 31.1 m.

Why do centrifugal machines make pressure almost for free?

The last two parts raise static pressure inside the rotor. The centrifugal part needs no slowing of the flow at all, so it costs little in losses, and 52 % of the book's pump head comes from it. An axial stage has U1 = U2, so that part is zero and all the rotor's pressure rise must come from slowing W, which boundary layers tolerate only so far (the de Haller limit, chapter 9). The interactive's axial stage adds 17.5 kJ/kg, a stagnation temperature rise of 17.4 K, and 56 % of that is rotor static rise, all of it from diffusion.

How does the same equation read for a turbine?

Turn the flow round. Guide vanes give the water whirl Ct1 before the runner, and the runner takes it out: w = U1Ct1 − U2Ct2. Good runners leave almost no whirl (Ct2 ≈ 0), so w ≈ U1Ct1 and the leaving water keeps only its through-flow speed. In the interactive's inward-flow turbine, a 27° blade outlet angle leaves Ct2 = 0.23 m/s: w = 560 J/kg, so every kg gives up 57 m of head. On a site with 65 m net head that is a hydraulic efficiency of 87.8 %. Chapters 13 and 15 apply this to real stations.

Predict, then check

In an axial compressor stage, which part of Euler's equation is zero?

Worked examples

Torque and power of a blower impeller from its outlet triangle · Radial blower impeller, D2 = 0.6 m, 1440 rpm, outlet blade angle β2 = 40° (from the tangent, backward), radial velocity at outlet 12 m/s, no inlet whirl, air 2.4 kg/s at ρ 1.2 kg/m³, infinite blades
  1. U2 = π × 0.6 × 1440/60 = 45.24 m/s
  2. Ct2 = U2 − Cm2/tan β2 = 45.24 − 12/tan 40° = 30.94 m/s
  3. e = U2Ct2 = 45.24 × 30.94 = 1,400 J/kg
  4. Torque T = ṁ r2 Ct2 = 2.4 × 0.3 × 30.94 = 22.28 N·m; ω = 2π × 1440/60 = 150.8 rad/s
  5. Power = Tω = 3.36 kW (check: ṁe = 2.4 × 1,400 = 3.36 kW)
  6. Ideal total pressure rise ρe = 1.2 × 1,400 = 1,680 Pa

Answer: e ≈ 1,400 J/kg, torque ≈ 22.3 N·m, power ≈ 3.36 kW, ideal pressure rise ≈ 1,680 Pa

The three energy components of a pump impeller · Pump impeller D1 = 150 mm, D2 = 350 mm, 1450 rpm, β2 = 25°, meridional velocity 2.2 m/s at inlet and outlet, no inlet whirl, water
  1. U1 = π × 0.15 × 1450/60 = 11.388 m/s; U2 = π × 0.35 × 1450/60 = 26.573 m/s
  2. Inlet: C1 = 2.2 m/s (no whirl); W1 = √(2.2² + 11.388²) = 11.599 m/s
  3. Outlet: Ct2 = 26.573 − 2.2/tan 25° = 21.855 m/s; C2 = √(2.2² + 21.855²) = 21.965 m/s; W2 = 2.2/sin 25° = 5.206 m/s
  4. ½(C2² − C1²) = 238.8 J/kg; ½(U2² − U1²) = 288.2 J/kg; ½(W1² − W2²) = 53.7 J/kg
  5. Sum = 580.7 J/kg = U2Ct2 = 26.573 × 21.855 ✓; head = 580.7/9.81 = 59.2 m
  6. Degree of reaction = static/total = (288.2 + 53.7)/580.7 = 0.589

Answer: Euler head ≈ 59.2 m: 24.3 m dynamic, 29.4 m centrifugal, 5.5 m relative; R ≈ 0.59

Head used by an inward-flow water turbine runner · Runner D1 = 1.2 m (inlet), D2 = 0.6 m (outlet), 375 rpm, guide vane angle 15° from the tangent, meridional velocity 4 m/s at inlet and outlet, radial (whirl-free) exit, net head 40 m
  1. U1 = π × 1.2 × 375/60 = 23.562 m/s; U2 = 11.781 m/s
  2. Ct1 = 4/tan 15° = 14.928 m/s; Ct2 = 0
  3. w = U1Ct1 = 23.562 × 14.928 = 351.7 J/kg; head used = 35.85 m; hydraulic efficiency = 35.85/40 = 89.6 %
  4. Inlet blade angle: W1t = Ct1 − U1 = −8.634 m/s, so the blade leans back at 24.9° from −U, i.e. θ = 155.1° from the direction of motion
  5. Outlet blade angle for zero whirl: tan β2 = 4/11.781, β2 = 18.8°
  6. Components (turbine signs): dynamic 111.4, centrifugal 208.2, relative 32.1 J/kg; R = 0.683

Answer: 35.9 m of the 40 m used (η_h ≈ 90 %); blade angles θ ≈ 155° at inlet and 18.8° at outlet; R ≈ 0.68

Chapter 6

Degree of reaction and utilisation factor

Which syllabus covers thisAU CAE353 Unit 1VTU BME502 Module 2SPPU 402043 Units 1, 2 (degree of reaction)MU MEC602 Module 5 (degree of reaction)KTU PEMET411 Module 2

What is the degree of reaction?

The degree of reaction R is the share of a stage's energy change that happens as static pressure (enthalpy) change inside the rotor. From chapter 5, R = (centrifugal part + relative part)/e. R = 0 is a pure impulse rotor: the nozzles do all the pressure drop and the blades only turn the jet, as in a Pelton wheel. R = 0.5 splits it equally between stator and rotor, as in a Parsons turbine. The book's pump has R = 0.61 at the chapter 5 default: 61 % of its head is made as pressure inside the impeller.

How does R shape the velocity triangles of an axial stage?

With the same blade speed U at inlet and outlet and the same axial velocity, the rotor's static change is ½(W2² − W1²) and the algebra collapses to R = 1 − (Ct1 + Ct2)/(2U), whirl counted positive along U. So R fixes the exit whirl: Ct2 = 2U(1 − R) − Ct1. At a nozzle angle α1 = 20° (from the wheel plane) and U/V1 = 0.6, R = 0 gives Ct2 = 0.26 V1, R = 0.25 gives −0.04 V1 and R = 0.5 gives −0.34 V1. The 50 % stage has mirror-image triangles: blade outlet angle equal to the nozzle angle.

What is the utilisation factor?

Not all the energy offered to a rotor can become work: the kinetic energy still in the fluid at exit, ½V2², leaves the stage (V is the absolute speed, C in chapter 5). The utilisation factor ε = work/(work + ½V2²) measures the share used. Writing the work through R gives ε = (V1² − V2²)/(V1² − R V2²). It ignores blade friction; the Thermal Engineering book's impulse-stage interactive adds friction to get blade efficiency. At the interactive's default (R 0.25, α1 20°, U/V1 0.6), ε = 90.8 %.

Degree of reaction and utilisation factor: velocity triangles for R = 0 to 1 and the best speed ratio for each: interactive loads as you scroll

Which blade speed gives the most?

ε is largest when V2 is smallest, and with a fixed axial velocity that means no exit whirl: the fluid leaves axially. Setting Ct2 = 0 gives the best blade speed ratio U/V1 = cos α1/(2(1 − R)) and ε_max = cos²α1/(1 − R sin²α1). At α1 = 20°: impulse (R = 0) is best at 0.47 with ε_max = 88.3 %; R = 0.25 at 0.626, 91 %; R = 0.5 at 0.94, 93.8 %. More reaction raises the ceiling, but every step asks for a faster blade.

If more reaction is better, why are impulse stages still built?

The faster blade is the price. With steam at V1 = 1,000 m/s and α1 = 20°, an impulse stage peaks at U = 470 m/s while a 50 % stage wants 940 m/s, and blade stress grows with U². A reaction rotor also has a pressure drop across it, so fluid leaks past the blade tips and pushes the rotor along its axis. At a fixed blade speed the best R depends on U/V1: the ε-against-R view shows that at U/V1 = 0.6 the best R is 0.27 (ε 90.9 %). Hydraulic turbines follow the same logic: impulse Pelton wheels for high heads, reaction Francis and Kaplan runners for lower heads.

Predict, then check

An impulse stage (α1 = 20°) runs near its best, U/V1 = 0.47. It is rebuilt as a 50 % reaction stage but keeps the same blade speed ratio. Its utilisation factor:

Worked examples

Utilisation factor of an impulse stage from its triangles · Axial impulse stage (R = 0), V1 = 500 m/s at α1 = 18° from the wheel plane, U = 200 m/s, constant axial velocity, no blade friction
  1. Vw1 = 500 cos 18° = 475.53 m/s; Va = 500 sin 18° = 154.51 m/s
  2. R = 0: Vw2 = 2U − Vw1 = 400 − 475.53 = −75.53 m/s (against the blade motion)
  3. V2 = √(154.51² + 75.53²) = 171.98 m/s
  4. Work E = U(Vw1 − Vw2) = 200 × 551.06 = 110,211 J/kg
  5. ε = E/(E + V2²/2) = 110,211/(110,211 + 14,789) = 0.8817 (same as (V1² − V2²)/V1² for R = 0)
  6. Best blade speed U = V1 cos α1/2 = 237.8 m/s, where ε_max = cos²18° = 0.9045

Answer: ε ≈ 88.2 %; running at 238 m/s would raise it to 90.5 %

A 30 % reaction gas turbine stage · R = 0.3, V1 = 420 m/s at α1 = 22°, U = 230 m/s, constant axial velocity, U1 = U2
  1. Vw1 = 420 cos 22° = 389.42 m/s; Va = 157.33 m/s
  2. Vw2 = 2U(1 − R) − Vw1 = 2 × 230 × 0.7 − 389.42 = −67.42 m/s
  3. V2 = 171.17 m/s; E = 230 × (389.42 − (−67.42)) = 105,072 J/kg
  4. ε = (V1² − V2²)/(V1² − R V2²) = (176,400 − 29,299)/(176,400 − 0.3 × 29,299) = 0.8776
  5. Best U/V1 = cos 22°/(2 × 0.7) = 0.6623, i.e. U = 278.2 m/s; ε_max = cos²22°/(1 − 0.3 sin²22°) = 0.8975

Answer: ε ≈ 87.8 % now; 89.7 % at U ≈ 278 m/s

How much reaction for a 92 % ceiling? · Nozzle angle α1 = 20°, steam leaves the nozzles at 400 m/s; the designer wants ε_max ≥ 0.92
  1. cos²α1/(1 − R sin²α1) ≥ 0.92 → R ≥ (1 − cos²20°/0.92)/sin²20°
  2. cos²20° = 0.883, sin²20° = 0.117: R ≥ 0.3436
  3. Best U/V1 = cos 20°/(2(1 − 0.344)) = 0.7158, so U = 400 × 0.7158 = 286.3 m/s
  4. An impulse stage (ε_max = 88.3 %) would run at only 187.9 m/s

Answer: R ≈ 0.34 or more, with the blade at about 286 m/s instead of 188 m/s

Chapter 7

Impact of jets: force and work on fixed and moving flat, inclined and curved vanes

Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU 402043 Unit 1MU MEC602 Module 2 (impulse theory)KTU not in syllabus

How does a water jet push on a plate?

By Newton's second law in its momentum form: the force on the water equals the mass flow times its change of velocity, and the plate feels the opposite. A 50 mm jet at 20 m/s carries ṁ = ρaV = 39.27 kg/s. On a fixed flat plate square to the jet the water leaves sideways with no forward speed, so its whole forward momentum is destroyed: F = ρaV² = 785.4 N, about the weight of 80 kg. Water is 1000 kg/m³, so even modest jets push hard.

What changes when the plate moves away from the jet?

Two things. The water hits at the relative speed V − u, and only ρa(V − u) kg/s reaches the plate each second, because the plate runs away and the jet between nozzle and plate keeps getting longer. So F = ρa(V − u)², power Fu, and the efficiency against the jet's kinetic power is 2u(V − u)²/V³. It peaks at u = V/3 at 8/27 = 29.6 %. The interactive's default (50 mm, 20 m/s, u/V 0.33) gives 353 N and 2.33 kW.

A jet striking vanes: force, work and efficiency on flat, inclined and curved vanes, one vane against a wheel of vanes: interactive loads as you scroll

What about a plate set at an angle to the jet?

A smooth plate can only push square to its own surface, so the force is normal to the plate: Fn = ρaV² sin θ for a fixed plate at angle θ to the jet. Its component along the jet is ρaV² sin²θ. The water spreads along the plate in two sheets; momentum along the frictionless plate is kept, so (1 + cos θ)/2 of the flow goes the downstream way. For our 50 mm, 20 m/s jet at θ = 30°: Fn = 392.7 N, 196.3 N along the jet, and the flow splits 93.3 % to 6.7 %.

Why do curved vanes do better than flat plates?

A curved vane turns the relative jet through an angle δ instead of stopping it, and a jet sent back the way it came changes its momentum by up to twice as much. With friction the relative speed falls to k(V − u), so the force carries the factor (1 − k cos δ): 1 for a flat plate (δ = 90°), 1.92 for a 165° bucket with k = 0.95. One such vane peaks at 56.8 %, still at u = V/3. A one-sided vane also feels a side force, 193 N for our jet on a fixed 165° vane; splitting the jet into two mirror halves cancels it.

How does a wheel of vanes catch all the water?

Put vanes on a wheel and a new one enters the jet as the last one leaves, so no water is left behind: ṁ = ρaV, not ρa(V − u). Then F = ρaV(V − u)(1 − k cos δ) and the efficiency (1 − k cos δ)·2u(V − u)/V² peaks at u = V/2. Flat plates on a wheel reach 50 %; 165° buckets with k = 0.95 reach 95.9 %, and ideal 180° vanes 100 %, because the water leaves with no speed. That is the Pelton wheel of chapter 14, like the six-jet wheels at Idukki under 660 m of head, which run their buckets near half the jet speed.

Predict, then check

One flat plate moving away from a jet peaks at 29.6 % efficiency. A wheel of the same flat plates, at its own best speed, reaches:

Worked examples

A jet on a fixed and on a moving flat plate · Water jet 75 mm in diameter at 25 m/s hits a flat plate square to it. Case 1: plate fixed. Case 2: one plate moving away along the jet at 8 m/s
  1. a = π × 0.075²/4 = 4.4179 × 10⁻³ m²
  2. Fixed: F = ρaV² = 1000 × 0.004418 × 25² = 2,761 N
  3. Moving: mass reaching the plate ρa(V − u) = 75.1 kg/s; F = ρa(V − u)² = 1276.8 N
  4. Power = Fu = 1276.8 × 8 = 10.214 kW; jet power ½ρaV³ = 34.515 kW
  5. η = 2u(V − u)²/V³ = 2 × 8 × 17²/25³ = 0.2959
  6. Best speed V/3 = 8.33 m/s gives η = 8/27 = 0.2963 and 10.227 kW

Answer: 2,761 N on the fixed plate; 1277 N, 10.21 kW and η ≈ 29.6 % on the moving plate

A wheel of curved vanes · Jet 40 mm in diameter at 30 m/s drives a wheel of symmetric curved vanes moving at 14 m/s. Each vane turns the relative jet through 160°; friction leaves 90 % of the relative speed (k = 0.9)
  1. a = 1.2566 × 10⁻³ m²; every kg is caught: ṁ = ρaV = 37.7 kg/s
  2. Relative speed at inlet 30 − 14 = 16 m/s; factor 1 − k cos 160° = 1 + 0.9 × 0.9397 = 1.8457
  3. F = ṁ(V − u)(1 − k cos δ) = 37.7 × 16 × 1.8457 = 1113.3 N; P = Fu = 15.586 kW
  4. Jet power ½ṁV² = 16.965 kW, so η = 0.9188
  5. Check by energy: leaving KE 461 W + friction 917 W + shaft 15586 W = 16965 W ✓
  6. At the best speed u = V/2 = 15 m/s, η = (1 − k cos δ)/2 = 0.9229

Answer: F ≈ 1113 N, P ≈ 15.59 kW, η ≈ 91.9 % (best 92.3 % at 15 m/s)

A jet at an angle on a moving curved vane: vane angles and work · Jet at 30 m/s, 25° to the direction of motion of a series of curved vanes moving at 12 m/s. Water must enter without shock; it leaves the vane tip at 20° to the direction opposite to the motion; no friction (relative speed unchanged). Angles from the direction of motion
  1. Vw1 = 30 cos 25° = 27.189 m/s; Vf1 = 30 sin 25° = 12.679 m/s
  2. Relative velocity at inlet: Vr1 = √((27.189 − 12)² + 12.679²) = 19.785 m/s; vane inlet angle θ = tan⁻¹(12.679/15.189) = 39.85°
  3. Outlet: Vr2 = 19.785 m/s; its backward component 18.592 m/s exceeds u, so Vw2 = 18.592 − 12 = 6.592 m/s against the motion
  4. Work per kg = u(Vw1 + Vw2) = 12 × (27.189 + 6.592) = 405.4 J/kg
  5. Exit speed V2 = √(6.592² + 6.767²) = 9.447 m/s; check: 405.4 + 9.447²/2 = 450 = 30²/2 ✓
  6. Efficiency of the series of vanes = 2 × 405.4/30² = 0.9008

Answer: Vane inlet angle ≈ 39.9°, work ≈ 405 J/kg, efficiency ≈ 90.1 %

Chapter 8

Radial pumps and compressors: blade discharge angle, energy transfer and the head-flow line

Which syllabus covers thisAU CAE353 Units 2, 3VTU BME502 Module 2SPPU 402043 Unit 3 (effect of outlet blade angle)MU not in syllabusKTU PEMET411 Module 2

What are backward, radial and forward-curved blades?

Measure the blade outlet angle β2 between the blade tip and the tangent pointing backwards, against the rotation (the pump convention used in this chapter). Below 90° the blade tip leans back: backward-curved. At 90° it points straight out: radial. Above 90° it leans forward into the rotation: forward-curved. KSB's pump lexicon gives 17° to 40° as the usual range for pump impellers and notes that forward-curved vanes give extremely high pressure coefficients. The book's pump uses 22.5°.

How does β2 set the theoretical head-flow line?

With no inlet whirl, Euler's equation is H = U2Ct2/g, and the outlet triangle gives Ct2 = U2 − Cm2 cot β2 with Cm2 = Q/A2. So the ideal head is a straight line in Q. At zero flow every impeller gives U2²/g = 39.7 m (the book's pump, U2 = 19.74 m/s). At 26.8 L/s: 31.7 m with 22.5° blades, 39.7 m radial, 43.7 m forward at 140°. The real book's pump gives 23 m there: slip (Wiesner σ = 0.84) and losses take the rest (chapter 18).

Backward, radial or forward blades: the outlet triangle, theoretical head against flow, power and the degree of reaction: interactive loads as you scroll

Why do forward-curved blades overload the motor?

Theoretical power is ρgQH. For forward blades H rises with Q, so power rises faster than flow and never peaks: open the valve and the motor is overloaded. For backward blades the falling head line makes power peak at Q = A2U2 tan β2/2, 66 L/s for the book's impeller (12.9 kW), and fall after it: a motor sized for the peak can never be overloaded. At 60 L/s the 22.5° impeller needs 12.8 kW of ideal power and a 140° one 28.6 kW.

How does β2 change the degree of reaction?

With no inlet whirl and equal meridional speeds, the dynamic part is Ct2²/2 and R = 1 − Ct2/(2U2) = ½(1 + φ cot β2), where φ = Cm2/U2 = 0.084 at 26.8 L/s. Radial blades always give R = 0.5. Backward blades give more: 0.6 at 22.5°. Forward blades give less: 0.45 at 140°, so more than half their head leaves the impeller as velocity that the volute must slow down, and slowing a flow is where losses grow. That is a second reason pumps use backward blades.

So why do some fans use forward or radial blades?

Because head per size is not the only goal. Forward-curved blowers make the most pressure from a small, slow wheel and are used where space is tight and a backward-curved fan would be too noisy, though they are less efficient. Radial blowers suit medium speeds, low volumes and high pressures, as in vacuum cleaners and pneumatic conveying. Backward-curved fans are the most energy efficient, so large, long-running fans use them. Pumps, which run for years at high power, almost always use backward blades of 17° to 40°.

Predict, then check

The book's impeller is rebuilt with forward-curved blades (β2 = 140°) and run at the same 26.8 L/s and 1450 rpm. Its theoretical (Euler) head:

Worked examples

Theoretical head, reaction and power of a backward-curved impeller · Pump impeller D2 = 200 mm, outlet width b2 = 12 mm, 2900 rpm, β2 = 30°, Q = 20 L/s, no inlet whirl, infinite blades, blade thickness neglected, water
  1. U2 = π × 0.2 × 2900/60 = 30.369 m/s; A2 = π × 0.2 × 0.012 = 7.5398 × 10⁻³ m²
  2. Cm2 = 0.020/0.00754 = 2.653 m/s
  3. Ct2 = U2 − Cm2 cot 30° = 30.369 − 2.653 × 1.7321 = 25.774 m/s
  4. Hth = U2Ct2/g = 30.369 × 25.774/9.81 = 79.79 m
  5. R = 1 − Ct2/(2U2) = 1 − 25.774/60.737 = 0.5756
  6. Theoretical power ρgQHth = 1000 × 9.81 × 0.020 × 79.79 = 15.65 kW

Answer: Hth ≈ 79.8 m, R ≈ 0.576, ideal power ≈ 15.7 kW

Same impeller, three blade angles, two flows · The impeller of the previous example with β2 = 30°, 90° and 120°, at 20 and 30 L/s
  1. Cm2 = 2.653 m/s at 20 L/s and 3.979 m/s at 30 L/s; cot 30° = 1.7321, cot 90° = 0, cot 120° = −0.5774
  2. β2 30°: Hth = 79.8 → 72.7 m; power 15.7 → 21.4 kW
  3. β2 90°: Hth = 94 m at both flows; power 18.4 → 27.7 kW
  4. β2 120°: Hth = 98.8 → 101.1 m; power 19.4 → 29.8 kW
  5. For 50 % more flow, power grows by 37 % (30°), 50 % (90°) and 54 % (120°)

Answer: Head falls with flow only for backward blades; the forward-curved impeller's power grows 54 % for a 50 % flow rise, against 37 % for 30° blades

Why a compact blower uses forward-curved blades · Centrifugal fan wheel D2 = 0.4 m, outlet width 0.15 m, 960 rpm, 2.0 m³/s of air (ρ 1.2 kg/m³), no inlet whirl, infinite blades. Compare β2 = 140° (forward) with β2 = 35° (backward)
  1. U2 = π × 0.4 × 960/60 = 20.106 m/s; A2 = π × 0.4 × 0.15 = 0.1885 m²; Cm2 = 2.0/0.1885 = 10.61 m/s
  2. Forward: Ct2 = 20.106 − 10.61 × cot 140° = 32.751 m/s; ideal pressure rise ρU2Ct2 = 790 Pa; R = 0.186
  3. Backward: Ct2 = 20.106 − 10.61 × cot 35° = 4.953 m/s; ρU2Ct2 = 120 Pa; R = 0.877
  4. Ideal air power QΔp: 1.58 kW forward, 0.24 kW backward

Answer: At the same size and speed the forward-curved wheel gives about 6.6 times the pressure (790 against 120 Pa), but with R ≈ 0.19 most of it is velocity to be diffused, which is why it is less efficient

Chapter 9

Axial pumps and compressors: stage velocity triangles, flow and loading coefficients and reaction

Which syllabus covers thisAU CAE353 Unit 4VTU BME502 Module 2SPPU 402043 Unit 4MU not in syllabusKTU PEMET411 Module 2

What does each row of an axial compressor stage do?

A stage is one row of moving blades (the rotor) and one row of fixed blades (the stator). The rotor turns the air in the direction of rotation, so it gives energy w = U ΔCt (Euler's equation, chapter 5), and part of it appears at once as a pressure rise. The stator turns the air back towards the axis and slows it, turning the extra kinetic energy into more pressure, and sets it up for the next rotor. A stage that adds 26.9 K at 300 m/s gives a pressure ratio of only about 1.33, which is why a pressure ratio of 6 takes 7 such stages (example 2). Angles in this chapter are measured from the axial direction.

What do the flow coefficient, stage loading and reaction tell a designer?

Three numbers fix the triangles at the mean radius: flow coefficient φ = Ca/U (how fast the air goes through compared with the blades), stage loading ψ = Δh0/U² = ΔCt/U (how much work per blade speed squared) and reaction R (the rotor's share of the static pressure rise). The interactive's default, φ = 0.5, ψ = 0.3, R = 0.5 at U = 300 m/s, gives β1 = 52.4° → β2 = 35°, ΔT0 = 26.9 K and W2/W1 = 0.74. An MIT study of repeating stages put peak efficiency at φ 0.4-0.6 and ψ 0.2-0.3.

An axial compressor stage: rotor and stator rows, velocity triangles at mean radius, flow and loading coefficients and reaction: interactive loads as you scroll

Why is the 50 % reaction stage the usual choice?

At R = 0.5 the triangles are mirror images: α1 = β2 and α2 = β1, so rotor and stator turn the flow by the same angle and slow it by the same ratio. Neither row is overloaded, and the same blade shape serves both. Move R to 0 and the rotor only turns the air with no pressure rise (W2 = W1, de Haller ratio 1.00), leaving the stator to do all the diffusing: at φ 0.5, ψ 0.3 its C3/C2 falls to 0.79. A high reaction does the reverse. Real compressors stay near 0.5 at the mean radius.

How much can a blade row slow the flow before it stalls?

The air on a blade's suction side has to climb a pressure rise; slow it too much and the boundary layer separates and the row stalls. De Haller's rule of thumb: keep W2/W1 in a rotor (and C3/C2 in a stator) at or above 0.72. At φ = 0.5 and R = 0.5 that caps the loading at ψ = 0.335. It is a soft limit: published studies list many stages running well below 0.72 far from surge, and the MIT study found ratios near 0.75 at best efficiency. Designers now also check a diffusion factor that includes the blade spacing.

Is an axial-flow pump or fan the same machine?

Yes: the same triangles, with water and head instead of air and temperature. Head = ηh U ΔCt/g. With no inlet whirl the reaction is 1 − Ct2/(2U), which is high (about 0.94 in this chapter's pump example), so most of the pressure rises in the rotor. Axial pumps and fans move large flows at low head: drainage and lift irrigation, condenser cooling water in power stations, cooling towers and tunnel ventilation. They have high specific speed, and their power rises as the flow falls, so they are started with the valve open.

Predict, then check

An axial compressor stage runs at φ = 0.5 and R = 0.5. You raise the stage loading ψ from 0.3 to 0.5. What happens to the rotor's de Haller ratio W2/W1?

Worked examples

One compressor stage from its blade angles · Mean diameter 0.5 m at 10,500 rpm, axial velocity 150 m/s (constant), air arrives at α1 = 15° from the previous stator, rotor exit relative angle β2 = 30° (both from axial), inlet T01 = 300 K, stage efficiency 0.88 (assumed); cp = 1005 J/kg K
  1. U = π × 0.5 × 10,500/60 = 274.9 m/s; φ = 150/274.9 = 0.546
  2. Ct1 = 150 tan 15° = 40.19 m/s; tan β1 = (274.9 − 40.19)/150, β1 = 57.42°
  3. Ct2 = U − Ca tan β2 = 274.9 − 150 tan 30° = 188.29 m/s, so α2 = 51.46°
  4. w = U (Ct2 − Ct1) = 274.9 × 148.09 = 40,710 J/kg; ψ = 0.539
  5. ΔT0 = 40,710/1005 = 40.51 K; pressure ratio = (1 + 0.88 × 40.51/300)^3.5 = 1.4814
  6. W1 = 150/cos 57.42° = 278.5 m/s, W2 = 150/cos 30° = 173.2 m/s: de Haller 0.622
  7. Reaction = (W1² − W2²)/(2w) = 0.584 (= φ(tan β1 + tan β2)/2)
  8. W2/W1 = 0.62 is below de Haller's 0.72: the rotor turns the flow too much and is likely to stall; a smaller β1 − β2 (less loading) or more axial velocity would fix it

Answer: ΔT0 = 40.5 K, pressure ratio 1.481, R = 0.58, W2/W1 = 0.62 (below 0.72)

Designing a symmetric stage and counting stages · φ = 0.55, ψ = 0.35, R = 0.5, U = 320 m/s at every stage; inlet 288.15 K; stage efficiency 0.88 (assumed); overall pressure ratio wanted 6; no work-done factor (chapter 24 adds it)
  1. tan β1 = (R + ψ/2)/φ = 1.2273, β1 = 50.83° = α2; tan β2 = (R − ψ/2)/φ = 0.5909, β2 = 30.58° = α1
  2. de Haller W2/W1 = cos β1/cos β2 = 0.734, the same in the stator
  3. w = ψU² = 0.35 × 320² = 35,840 J/kg, ΔT0 = 35.66 K per stage
  4. Stage 1: (1 + 0.88 × 35.66/288.15)^3.5 = 1.4359; each later stage starts hotter, so its ratio is smaller
  5. Multiplying stage by stage: after 6 stages 5.585, after 7 stages 6.905 (last stage ratio 1.2364)

Answer: Angles 50.8° and 30.6°, W2/W1 = 0.73, 7 stages for a pressure ratio of 6

An axial-flow (propeller) pump · Tip diameter 0.45 m, hub 0.25 m, 1450 rpm, axial velocity 5 m/s, no inlet whirl, rotor exit relative angle 78° from axial at the mean radius, hydraulic efficiency 0.85 (assumed); ρ = 1000 kg/m³, g = 9.81 m/s²
  1. Mean diameter 0.35 m, U = π × 0.35 × 1450/60 = 26.57 m/s
  2. Inlet: Ct1 = 0, so β1 = tan⁻¹(26.57/5) = 79.34°
  3. Exit: Ct2 = U − Ca tan 78° = 26.57 − 23.52 = 3.05 m/s
  4. Euler head U Ct2/g = 8.26 m; actual head 0.85 × 8.26 = 7.02 m
  5. Q = 5 × π/4 (0.45² − 0.25²) = 0.55 m³/s; water power ρgQH = 37.9 kW
  6. Reaction 1 − Ct2/(2U) = 0.943; W2/W1 = cos β1/cos β2 = 0.889

Answer: Head 7.02 m at 0.55 m³/s (37.9 kW to the water), reaction 0.94

Chapter 10

Steam turbines: compounding and the velocity-compounded (Curtis) wheel

Which syllabus covers thisAU CAE353 Unit 5 (compounding)VTU BME502 Module 3SPPU 402043 Unit 2MU MEC602 Module 5KTU not in syllabus

Why does a steam turbine need compounding?

Expand steam from 30 bar, 350 °C to 0.1 bar in one nozzle and the isentropic drop is 980 kJ/kg, a jet of 1,400 m/s. One row of blades works best at U = V1 cos α/2, about 666 m/s for α = 18°: a wheel 4.2 m across at 3,000 rpm, or a small wheel at a speed no generator wants, with huge leaving and friction losses. Compounding splits the job so each row meets a slower jet. Angles in chapters 10-12 are from the plane of the wheel (the direction of blade motion).

What are velocity, pressure and reaction compounding?

Velocity compounding (Curtis): one nozzle takes the whole pressure drop, then two or three moving rows on one wheel take the jet's speed in turn, with fixed guide blades between them. Pressure compounding (Rateau): the drop is split into several nozzle-plus-moving-row stages in series, each a small impulse stage. Pressure-velocity compounding uses Curtis wheels in a pressure-compounded series. Reaction (Parsons) blading spreads the drop over many stages with pressure falling in every row (chapter 11). Curtis patented his turbines in 1896 and sold the rights to GE in 1901; by 1905 designers put Curtis wheels at the high-pressure end followed by reaction stages.

How does a two-row Curtis wheel share the work?

The first moving row takes the jet at V1 and sends it out backwards; a guide row turns it forwards again; the second row takes most of what is left. With frictionless blades the best blade speed falls to U = V1 cos α/(2n) for n rows, the maximum efficiency stays cos²α, and the rows' work splits 3:1 for two rows and 5:3:1 for three. With a blade coefficient k = 0.88 and α = 18° (the interactive's default) the best is U/V1 = 0.231 at 73.3 %, with an 85 : 15 split. The Thermal Engineering book has the single-row stage.

A velocity-compounded Curtis wheel: two rows of moving blades, the work each row does and the best blade speed: interactive loads as you scroll

Why do Curtis wheels stop at two rows?

Every row loses a share of the steam's speed to friction, and later rows see slow steam that carries little energy. With k = 0.88 and α = 18° the best diagram efficiency is 85 % for one row, 73.3 % for two and 63.1 % for three, and at its best the third row does −1.3 % of the work. So a Curtis wheel is used where a big drop must be taken in a short, cheap stage: the first (control) stage of industrial turbines, whose nozzle groups also do the governing (chapter 12).

What does pressure compounding buy instead?

Split an isentropic drop into n equal stages and each jet is V1/√n, so the best blade speed falls by √n while each stage keeps the high single-row efficiency. In this chapter's third example a 15 bar to 1.5 bar drop needs U = 429 m/s with one row, 214 m/s with a two-row Curtis wheel and 247 m/s with three Rateau stages. Curtis gets the speed down with fewer, shorter stages; Rateau keeps the efficiency but needs more stages, diaphragms and shaft length.

Predict, then check

An ideal two-row Curtis wheel (no blade friction, symmetric blades) runs at its best blade speed. What share of the total work does the second moving row do?

Worked examples

A two-row Curtis wheel, row by row · Jet V1 = 900 m/s at α = 17°, blade speed 190 m/s, symmetric moving and guide blades, velocity coefficient k = 0.9 in every row, steam flow 5 kg/s
  1. Row 1 in: Vw1 = 900 cos 17° = 860.7, Vf1 = 263.1 m/s; W1 = 720.4 m/s at β1 = 21.42°
  2. Row 1 out: W2 = 0.9 W1 = 648.4 m/s at 21.42° backwards; V2 = 476.6 m/s, leaving backwards at 29.79°
  3. Row 1 work U ΔVw = 190 × 1274.3 = 242,113 J/kg
  4. Guide row: V3 = 0.9 × 476.6 = 428.9 m/s, turned forwards at the same angle
  5. Row 2: W3 = 280.4 m/s at β = 49.47°; work = 190 × 346.3 = 65,791 J/kg; leaves at 193.6 m/s
  6. Diagram efficiency = 2 × 307,904/900² = 76.03 %; leaving loss 4.63 %, friction 19.35 %
  7. Power = 5 × 307,904/1000 = 1539.5 kW

Answer: η = 76 %, rows 242,113 and 65,791 J/kg, 1,540 kW

Best speed of an ideal two-row wheel · Jet 1000 m/s at α = 16°, frictionless symmetric blades (k = 1)
  1. Best U = V1 cos α/4 = 1000 × 0.9613/4 = 240.3 m/s
  2. η max = cos²16° = 92.4 %, work = 0.924 × 1000²/2 = 462,012 J/kg
  3. Row 1 does 346,509 J/kg, row 2 115,503 J/kg: 3 : 1
  4. One row at the same 240 m/s: η = 4ρ(cos α − ρ) with ρ = 0.2403 gives 69.3 %
  5. One row would need U = V1 cos α/2 = 481 m/s to reach the same 92.4 %

Answer: U = 240 m/s, η = 92.4 %, split 3 : 1; a single row at that speed manages 69 %

Curtis or Rateau for one pressure drop · Steam at 15 bar, 250 °C expands to 1.5 bar; nozzle efficiency 0.95 (assumed); α = 18°; IAPWS-IF97; compare the best blade speeds, and the mean diameter each needs at 3,000 rpm
  1. h1 = 2,924.0 kJ/kg, s1 = 6.7111 kJ/kg K; at 1.5 bar with s1: x = 0.9116, h = 2,496.3 kJ/kg
  2. Δhs = 427.6 kJ/kg; V1 = √(2 × 0.95 × 427.6 × 1000) = 901 m/s
  3. One row: U = V1 cos 18°/2 = 429 m/s, D = 2.73 m
  4. Two-row Curtis: U = V1 cos 18°/4 = 214 m/s, D = 1.36 m
  5. Three Rateau stages (a third of Δhs each, reheat ignored): jet 520 m/s, U = 247 m/s, D = 1.58 m

Answer: 429, 214 and 247 m/s: the Curtis wheel gets the speed lowest in one wheel

Chapter 11

Reaction steam turbines: Parsons stages, blade height and the number of stages

Which syllabus covers thisAU CAE353 Unit 5 (reaction stages)VTU BME502 Module 3SPPU 402043 Unit 2MU MEC602 Module 5KTU not in syllabus

What makes a Parsons stage different from an impulse stage?

In an impulse stage all the pressure falls in the nozzles and the moving blades only turn the steam. In a reaction stage the pressure falls in both rows: the fixed blades are nozzles, and the moving blade passages are nozzles too, so the steam speeds up relative to the blades and pushes back on them. Parsons' 50 % reaction stage uses the same blade shape in both rows (fixed exit angle = moving exit angle = α), so V1 = W2 and W1 = V2. Charles Parsons built the first reaction turbine in 1884; it drove a 7.5 kW dynamo.

How much work does one Parsons stage do?

The whirl change is V1 cos α + (V1 cos α − U), so w = U (2V1 cos α − U) per kg. For a given steam speed V1 it peaks at U = V1 cos α, which is also the best-efficiency speed ratio, and there w = U². That is why a Parsons turbine needs many stages: at U = 150 m/s each stage does only 22.5 kJ/kg. Its diagram efficiency, 2cos²α/(1 + cos²α), is 95 % at α = 18°; the Thermal Engineering book compares it with the impulse stage.

Why do the blades get taller along the turbine?

Every stage passes the same mass flow, but each kilogram takes up more room as the pressure falls. Continuity at the mean diameter, ṁ v = π D h V1 sin α, so blade height h grows with the specific volume v. The interactive marches the HP cylinder of a BHEL 500 MW turbine (166.7 bar, 536 °C, 420 kg/s assumed) to 44 bar: at D = 0.75 m, α = 20°, U/V1 = 0.7 it needs 14.7 stages against BHEL's 17, and the blades grow from 66 to 180 mm.

Reaction staging in a steam turbine: Parsons stages, growing blade height as steam expands, and the number of stages: interactive loads as you scroll

How do you count the stages a section needs?

Each stage takes w of actual enthalpy, so N = actual drop/w. The actual drop is the stage efficiency times the sum of the stage isentropic drops, and that sum is the reheat factor times the overall isentropic drop (chapter 4), so N = RF η Δhs/w. For the HP default, Δhs = 382.1 kJ/kg, η = 88.6 % (Soderberg's blade-loss correlation), RF = 1.0157 and w = 23.38 kJ/kg. Halve the diameter and w falls four times, so the stage count roughly quadruples.

Why can't the low-pressure end stay a simple Parsons drum?

The specific volume grows from 0.434 m³/kg at 6 bar to 13.6 m³/kg at 0.1 bar. With 150 kg/s per flow at D = 1.9 m the constant-diameter 50 % model asks for a last blade 2.3 m long, longer than the radius. Real LP cylinders are double flow (BHEL's has 6 reaction stages per flow), the annulus flares outwards, and the long last blades are twisted so the reaction rises from hub to tip. The last rows also run in wet steam (chapter 12).

Predict, then check

A Parsons section runs at a fixed speed ratio U/V1 and blade angle. You halve its mean blade diameter. Roughly how many stages does it now need for the same pressure drop?

Worked examples

One Parsons stage and its blade height · Mean blade diameter 0.9 m at 3,000 rpm, blade exit angle α = 20° in both rows, U/V1 = 0.75, 60 kg/s of steam at 20 bar, 320 °C (IAPWS-IF97) in this stage
  1. U = π × 0.9 × 3000/60 = 141.37 m/s; V1 = U/0.75 = 188.5 m/s
  2. w = U (2V1 cos 20° − U) = 141.37 × (354.26 − 141.37) = 30,096 J/kg
  3. Relative inlet: W1 = 73.72 m/s at β1 = 60.99° (the blade inlet angle); W2 = V1 = 188.5 m/s
  4. Axial velocity V1 sin 20° = 64.47 m/s; v at 20 bar, 320 °C = 0.13082 m³/kg
  5. h = ṁ v/(π D V1 sin α) = 60 × 0.13082/(π × 0.9 × 64.47) = 43.1 mm

Answer: 30.1 kJ/kg per stage, blade inlet angle 61°, blade height 43 mm

How many stages from 40 bar to 8 bar · Steam 40 bar, 450 °C expands to 8 bar in identical Parsons stages, each doing 30 kJ/kg with a stage efficiency of 0.88 (assumed); IAPWS-IF97
  1. h1 = 3,331.0 kJ/kg, s1 = 6.9383 kJ/kg K; overall Δhs to 8 bar = 432.6 kJ/kg
  2. Ignoring reheat: N ≈ 0.88 × 432.6/30 = 12.69
  3. Stage by stage (each takes 30/0.88 = 34.09 kJ/kg isentropic from its own inlet): the first stage ends at 35.91 bar
  4. The pressure lines diverge, so the stage drops add up to RF = 1.0198 times Δhs
  5. N = RF × 0.88 × Δhs/30 = 12.94, so 13 stages (the last one lightly loaded)

Answer: 12.94 stages of work, built as 13; exit 246.4 °C at 8 bar

Best speed ratio and the stage count it gives · Blade exit angle 18°, blade speed 150 m/s, the section must deliver 300 kJ/kg of actual work
  1. The diagram efficiency 2ρ(2cos α − ρ)/(1 − ρ² + 2ρ cos α) is largest at ρ = U/V1 = cos 18° = 0.9511 (a golden-section search agrees)
  2. So V1 = 150/0.9511 = 157.7 m/s and w = U² = 22,500 J/kg
  3. At U/V1 = 0.7 instead, w = 38,639 J/kg: more work per stage, but the diagram efficiency drops to 91.4 %
  4. Stages at the best ratio = 300,000/22,500 = 13.33, so 14
  5. Diagram efficiency at the best ratio 2cos²18°/(1 + cos²18°) = 94.99 %

Answer: U/V1 = 0.951, 22.5 kJ/kg per stage, 14 stages

Chapter 12

Steam turbine losses, wetness and governing: throttle and nozzle control

Which syllabus covers thisAU CAE353 Unit 5 (losses)VTU not in syllabusSPPU 402043 Unit 2 (losses, governing)MU not in syllabusKTU not in syllabus

Where does a steam turbine's isentropic drop go?

Into work and a handful of losses: friction in the nozzles and moving blades, disc friction and windage (the spinning discs churn steam), leakage past blade tips and glands, wetness in the last stages, and the leaving loss, the kinetic energy the steam still has when it leaves the last row. In the interactive's turbine (60 bar, 480 °C to 0.1 bar, assumed loss fractions) the drop of 1,215 kJ/kg gives 1,047 kJ/kg of work (86.2 %). Friction heat is partly recovered by later stages, a reheat credit of 45.8 kJ/kg.

Steam turbine losses and governing: wetness, leakage and exit loss, throttle against nozzle control at part load: interactive loads as you scroll

Why is wet steam a problem in the last stages?

Below the saturation line water droplets form. They move slower than the steam, hit the backs of the moving blades and brake them, and they erode the leading edges of long LP blades. Baumann's rule (1921): each 1 % of average wetness in a stage costs about 1 % of stage efficiency; a later study puts it nearer 0.5 %. In the default turbine the exhaust is 11.5 % wet and wetness costs 31.7 kJ/kg stage by stage (34.6 by the hand rule on the average wetness). Reheating the steam keeps it drier.

Why doesn't a lower condenser pressure always pay?

A lower exhaust pressure adds isentropic drop, but saturated steam takes up 2.4 times as much room at 0.04 bar as at 0.1 bar. The last stage's annulus is fixed, so the steam must leave faster and its kinetic energy is lost in the condenser. In the default turbine the exit velocity is 195 m/s at 0.1 bar (1.6 % of the drop) but 451 m/s at 0.04 bar (7.7 %), and the wetter steam costs more too. Last-stage blades are sized for the condenser pressure the site will see.

How is a turbine's steam flow cut at part load?

Throttle governing closes a valve at the inlet: the steam keeps its enthalpy but loses pressure, so each kilogram does less work. Steam flow against load is then almost a straight line, the Willans line ṁ = aL + C, C being the no-load flow. Nozzle governing shuts whole groups of first-stage nozzles and throttles only the one partly open group, so less steam is throttled. In the default turbine at 60 % load: 3.81 kg/kWh throttle against 3.68 kg/kWh nozzle, with a no-load flow of 4.7 % of rated.

What other governing and protection do turbines use?

Bypass governing opens a valve that feeds fresh steam into later stages for short overloads, when the first-stage nozzles are already full. Large power-station units often use sliding pressure: the boiler pressure is lowered with load so the turbine valves stay wide open and throttling loss stays small. Every turbine also has an emergency governor (overspeed trip) that shuts the stop and control valves at once, independent of the speed governor.

Predict, then check

A throttle-governed turbine follows a Willans line ṁ = aL + C with a no-load flow C greater than zero. As the load L falls, what happens to the steam rate ṁ/L (kg/kWh)?

Worked examples

Steam rate from a Willans line · A throttle-governed 6 MW turbine-generator uses 2,400 kg/h of steam at no load and 26,000 kg/h at full load (test values, assumed); find the steam at 3.5 MW and the steam rates
  1. a = (26,000 − 2,400)/6,000 = 3.9333 kg/kWh
  2. At 3.5 MW: ṁ = 3.9333 × 3,500 + 2,400 = 16,167 kg/h
  3. Steam rate 16,167/3,500 = 4.619 kg/kWh, against 4.333 at full load
  4. At 1.5 MW it is 5.533 kg/kWh: the no-load flow weighs more as the load falls

Answer: 16,167 kg/h at 3.5 MW, 4.62 kg/kWh

Baumann's rule on one wet stage · Dry saturated steam at 0.5 bar enters a stage that expands it to 0.2 bar; the stage efficiency with dry steam would be 0.85; Baumann factor 1 (1 % efficiency per 1 % average wetness); IAPWS-IF97
  1. Inlet h = 2,645.2 kJ/kg, s = 7.593 kJ/kg K; at 0.2 bar with that s: h = 2,504.2 kJ/kg
  2. Isentropic drop 140.98 kJ/kg; with η = 0.85 the exit would be x = 0.9646
  3. Average wetness = (0 + exit wetness)/2, and η = 0.85 (1 − average wetness); the exit wetness depends on η, so iterate
  4. Converged: exit x = 0.9654, average wetness 1.73 %, η = 0.8353
  5. Wetness loss = (0.85 − 0.8353) × 140.98 = 2.07 kJ/kg

Answer: Stage efficiency 83.5 % instead of 85 %, 2.1 kJ/kg lost to wetness

The leaving loss at two condenser pressures · 40 kg/s leaves a last stage of annulus area 3.2 m², at 0.08 bar with x = 0.9; then at 0.05 bar with x = 0.88 (same annulus); IAPWS-IF97 saturation volumes
  1. 0.08 bar: v = vf + x (vg − vf) = 16.29 m³/kg, V = 40 × 16.29/3.2 = 203.6 m/s
  2. Leaving loss V²/2 = 20.73 kJ/kg, 829 kW
  3. 0.05 bar: v = 24.804 m³/kg, V = 310.1 m/s, loss 48.07 kJ/kg, 1923 kW

Answer: 20.7 kJ/kg at 0.08 bar, 48.1 kJ/kg at 0.05 bar: 2.32 times more

Chapter 13

Hydroelectric plants: heads, efficiencies and power through the penstock

Which syllabus covers thisAU not in syllabusVTU BME502 Module 4 (efficiencies)SPPU 402043 Unit 1 (hydro power plant)MU MEC602 Module 2KTU PEMET411 Module 3

What are the parts of a hydroelectric plant, from reservoir to river?

A dam or barrage stores water and raises its level. An intake with screens lets it into a head race tunnel or straight into a penstock, the steel pipe that carries it down to the turbines. A surge tank, a tall shaft near the end of a long tunnel, absorbs the pressure surge when the gates close quickly. The power house holds the turbines and generators, and the tailrace returns the water to the river. Idukki in Kerala puts six 130 MW units in a cavern 141 m long, 20 m wide and 34.5 m high; Teesta Low Dam III feeds each of its four 33 MW units through a penstock only 44 m long.

Gross head, net head: what is the difference?

Gross head is the height between the reservoir surface and the tailwater. Net head is what reaches the turbine after friction in the tunnel, penstock, bends and valves. Chutak in Ladakh has a gross head of 63.4 m but its units are rated at a net 52 m: 11.4 m, or 18 %, is spent pushing 24 m³/s per unit through a 10 km head race tunnel. Teesta-V's full reservoir (579 m) sits 219 m above its highest tailwater (360 m), against a rated head of 197 m. The reservoir level also moves between full and minimum draw-down, so the net head is a range, not one number.

Why is maximum power through a pipe not the best design?

Power is flow times net head. More flow gives more power until friction, which grows as the flow squared, eats the head: with a constant friction factor the jet power peaks when the friction loss is one third of the gross head. In the Idukki-scale default (assumed 2.5 km of 2 m pipe, 700 m gross head) 218 mm nozzles pass 25 m³/s and lose only 5.5 % of the head; 375 mm nozzles would give the most power, 239 MW instead of 136 MW, but waste a third of the water's energy for ever. USBR guidance keeps long-penstock losses near 3 to 10 % of the head.

Power through a penstock: gross and net head, nozzle size, the maximum-power condition and the chain of turbine efficiencies: interactive loads as you scroll

Which efficiencies stand between the water and the grid?

Hydraulic efficiency is the share of the net head the runner turns into work: losses in the nozzle, casing, vanes, runner and leaving velocity. Volumetric efficiency counts water that slips past the runner without working. Mechanical efficiency covers bearings, seals and windage. Their product is the turbine's overall efficiency; the generator takes about 2 % more. Published unit data imply 93.9 % from water to wire at Teesta-V and 90.9 % at Teesta Low Dam III. Idukki's rounded 25 m³/s implies only 80 %, so that published flow is likely a rounded maximum.

How are hydraulic turbines classified?

By how they use the head: impulse turbines (Pelton) turn it all into a jet at atmospheric pressure; reaction turbines (Francis, Kaplan) run full of water under pressure. By head: high, medium, low. By flow through the runner: tangential, radial-mixed, axial. Specific speed Ns = N√P/H^1.25 (rpm, kW, m) ties them together: Idukki's Pelton 16.5 per jet, Teesta-V's Francis 120, Chutak's Francis 250, Teesta Low Dam III's Kaplan 540. The Fluid Mechanics book has an interactive that picks the turbine type from head, flow and speed.

Predict, then check

At the same flow and friction factor, you double the penstock diameter. The friction loss falls by about…

Worked examples

Chutak: gross head, net head and efficiency · Chutak (NHPC, Ladakh): gross head 63.4 m, rated net head 52 m, 24.05 m³/s and 11 MW per unit; ρ = 1000 kg/m³, g = 9.81 m/s².
  1. Head lost before the turbine = 63.4 − 52 = 11.4 m (18 % of gross).
  2. Water power at the net head = ρgQH = 1000 × 9.81 × 24.05 × 52 = 12.27 MW.
  3. Overall unit efficiency (turbine × generator) = 11 / 12.27 = 89.7 %.
  4. Water power at the gross head = 1000 × 9.81 × 24.05 × 63.4 = 14.96 MW, so the plant turns 73.5 % of the gross head into electricity.

Answer: Net head 52 m (11.4 m lost); 12.27 MW of water power; η ≈ 89.7 % on net head, 73.5 % on gross.

The nozzle that gives maximum power from a pipe · Original small-hydro scheme: gross head 120 m, penstock 400 m long, D = 0.6 m, friction factor f = 0.014 (taken constant), 2 jets, Cv = 0.98.
  1. Pipe constant: hf = K Q², K = fL/(2gDA²); A = π × 0.6²/4 = 0.2827 m², K = 0.014 × 400 / (2 × 9.81 × 0.6 × 0.2827²) = 5.95 s²/m⁵.
  2. Jet power ∝ Q (Hg − KQ²) is largest when KQ² = Hg/3 = 40 m: Q = √(40 / 5.95) = 2.593 m³/s, net head 80 m.
  3. Nozzle for that flow: d = (D⁵/(2fL))^(1/4) / √(n Cv) = (0.6⁵ / (2 × 0.014 × 400))^(1/4) / √(2 × 0.98) = 206 mm per jet.
  4. Maximum jet power = ρgQ Cv² Hn = 1000 × 9.81 × 2.593 × 0.98² × 80 = 1.954 MW.
  5. With 80 mm nozzles instead: Q = √(Hg / (K + 1/(2g Cv² a²))) = 0.475 m³/s, hf = 1.34 m (1.1 % of the head), jet power 531 kW.

Answer: Best nozzle ≈ 206 mm per jet for 1954 kW at hf = 40 m; 80 mm nozzles give 531 kW but lose only 1.1 % of the head.

Idukki: water power, specific speed and the efficiency chain · Idukki unit (KSEB): net head 660 m, 25 m³/s, 130 MW, 375 rpm, six jets. Assumed chain: Cv 0.98, bucket efficiency at u/V1 0.47 with 165° and k 0.9, volumetric, mechanical and generator efficiencies 0.98 each.
  1. Water power = 1000 × 9.81 × 25 × 660 = 161.9 MW; published 130 MW means 80.3 % overall.
  2. Bucket efficiency = 2 × 0.47 × 0.53 × (1 + 0.9 cos 15°) = 0.9313; chain = 0.98² × 0.9313 × 0.98³ = 0.8418.
  3. That chain would give 136.3 MW from 25 m³/s, or need only 23.9 m³/s for 130 MW.
  4. Ns = N√P/H^1.25 = 375 × √130,000 / 660^1.25 = 40.4 for the wheel; per jet √(130,000/6) gives 16.5.
  5. A 50 Hz generator at 375 rpm has 120 × 50 / 375 = 16 poles.

Answer: 161.9 MW of water power, 80 % implied efficiency (the assumed chain gives 84.2 %), Ns ≈ 16.5 per jet.

Chapter 14

The Pelton wheel: velocity triangles, best speed ratio and design

Which syllabus covers thisAU not in syllabusVTU BME502 Module 4SPPU 402043 Unit 1MU MEC602 Module 2KTU PEMET411 Module 3

What are the parts of a Pelton turbine?

A nozzle turns the whole net head into a free jet at atmospheric pressure. A spear (needle) inside the nozzle moves to change the jet area and so the flow. The jet hits buckets round the wheel; a splitter ridge divides it into two halves that turn back almost 180° and fall into the tailrace. A deflector can swing into the jet to cut power at once without closing the spear fast. Big wheels take 4 or 6 jets (NPTEL); Idukki's vertical wheels have six. BHEL's Pykara Ultimate Stage wheels in Tamil Nadu run under 1,039 m, the highest head in India.

What do the Pelton velocity triangles look like?

Angles here are measured from the direction of bucket motion. At inlet the jet V1 travels along the bucket's motion, so the triangle is flat: relative velocity W1 = V1 − u. The bucket turns the relative jet through its deflection (165° is typical, so the outlet angle φ = 15°) and friction leaves W2 = k W1. The work per kg is u(Vw1 − Vw2) = u(V1 − u)(1 + k cos φ). In the default state (u/V1 0.47, k 0.9) the bucket efficiency is 93.1 % and the water leaves at only 13.8 m/s from a 111.5 m/s jet.

A Pelton wheel: jet and bucket velocity triangles, efficiency against bucket speed, and sizing the wheel, jets and buckets: interactive loads as you scroll

What is the best bucket speed, and why do real wheels run a little slower?

Hydraulic efficiency 2(u/V1)(1 − u/V1)(1 + k cos φ) is a parabola with its top at u/V1 = 0.5: the bucket moves at half the jet speed and, for a perfect 180° bucket, the water drops off with no speed left. The top value is (1 + k cos φ)/2, 93.5 % for k 0.9 and φ 15°. Real wheels run at about 0.43 to 0.48 of √(2gH) (IIT ISM) because friction, windage and the jet's spread favour a slightly slower bucket; ESHA's statistics put the bucket circle at 0.68√H/n, which is u = 0.482√(2gH).

How is a Pelton wheel sized?

Jet speed V1 = Cv√(2gH) with Cv about 0.98. Pick u/V1 near 0.46 and the generator speed N, then the bucket circle D = 60u/(πN). Each jet carries Q/n, so its diameter d follows from Q/n = (π/4)d²V1. The jet ratio m = D/d should stay between about 6 and 35, near 12 in most designs; buckets Z = 15 + m/2 (Tygun's rule), bucket width 3d to 4d. For Idukki at 0.47 this gives D 2.67 m, d 218 mm, m 12.2 and 22 buckets.

How is a Pelton turbine governed?

The governor holds the speed, and so the 50 Hz frequency, as the load changes. It moves the spear to change the jet area. Closing a long penstock quickly would cause water hammer, so for a sudden load drop the deflector swings into the jet first and the spear then closes slowly. A braking jet on the back of the buckets stops the wheel. If all control fails the wheel races to its runaway speed: Idukki's is 625 rpm against a normal 375 rpm, 1.67 times, and the generator must survive it.

Predict, then check

An ideal bucket (no friction, turns the jet a full 180°) moves at exactly half the jet speed. How fast does the water leave it?

Worked examples

Sizing Idukki's wheel · Idukki unit (KSEB): net head 660 m, 25 m³/s, 6 jets, 375 rpm. Take Cv = 0.98 and u/V1 = 0.46.
  1. Jet speed V1 = 0.98 √(2 × 9.81 × 660) = 111.52 m/s; bucket speed u = 0.46 × 111.52 = 51.3 m/s.
  2. Bucket circle D = 60u/(πN) = 60 × 51.3 / (π × 375) = 2.613 m (ESHA's statistical 0.68√H/n gives 2.8 m).
  3. Flow per jet = 25/6 = 4.167 m³/s; jet diameter d = √(4 × 4.167 / (π × 111.52)) = 218 mm.
  4. Jet ratio m = D/d = 2.613 / 0.2181 = 12; buckets Z = 15 + m/2 = 21 → 21.
  5. Bucket width 3d to 4d = 654 to 872 mm.

Answer: D ≈ 2.61 m, jets of 218 mm, m ≈ 12, 21 buckets.

Force, power and efficiency of a bucket wheel · Original test wheel: jet 150 mm diameter at 80 m/s, buckets at 38 m/s, deflection 165°, bucket friction k = 0.88; ρ = 1000 kg/m³.
  1. Flow Q = π × 0.15²/4 × 80 = 1.4137 m³/s; relative inlet W1 = 80 − 38 = 42 m/s, outlet W2 = 0.88 × 42 = 36.96 m/s.
  2. Force along the motion = ρQ W1 (1 + k cos 15°) = 1000 × 1.4137 × 42 × 1.85 = 109.85 kN.
  3. Power = F u = 109.85 × 38 = 4174.2 kW.
  4. Jet power = ½ ρ Q V1² = 4523.9 kW; efficiency = 92.3 %.
  5. Check by the formula: 2 × 0.475 × 0.525 × (1 + 0.88 cos 15°) = 0.9227; water leaves at 9.84 m/s.

Answer: F ≈ 109.85 kN, P ≈ 4174 kW, η ≈ 92.3 %.

How many jets does a site need? · Original site: net head 450 m, 10 m³/s, generator 500 rpm. ESHA: one nozzle works for nQE = n√Q/E^0.75 up to 0.025 (n in rev/s, E = gH).
  1. E = 9.81 × 450 = 4414.5 J/kg; n = 500/60 = 8.3333 rev/s.
  2. Whole-machine nQE = 8.3333 × √10 / 4414.5^0.75 = 0.0487.
  3. nQE grows with √(number of jets), so jets ≥ (0.0487 / 0.025)² = 3.79 → 4 jets.
  4. Per jet nQE = 0.0243 ≤ 0.025. Jet speed 0.98√(2g × 450) = 92.1 m/s; jet diameter √(4 × 2.5 / (π × 92.1)) = 186 mm.
  5. At u/V1 = 0.46 the wheel is D = 60 × 0.46 × 92.1 / (π × 500) = 1.62 m, m = 8.7.

Answer: 4 jets of about 186 mm on a 1.62 m wheel.

Chapter 15

The Francis turbine: velocity triangles, design and part load

Which syllabus covers thisAU not in syllabusVTU BME502 Module 4SPPU 402043 Unit 1MU MEC602 Module 2KTU PEMET411 Module 3

How does water flow through a Francis turbine?

A spiral casing, shrinking as it goes round, feeds water evenly to fixed stay vanes and then to the wicket (guide) gates, which set the whirl and the flow. The water enters the runner radially inwards, turns, and leaves axially into the draft tube: a mixed-flow reaction turbine running full under pressure. It covers medium heads: Teesta-V's three 170 MW units run at 197 m, Nathpa Jhakri's six 250 MW units at 426 m (SJVN; Wikipedia gives 428 m), and BHEL's Kameng units in Arunachal Pradesh at 531 m, the highest-head Francis BHEL lists.

What do the Francis inlet and outlet triangles tell you?

Angles are measured from the direction of blade motion. The guide vane angle α1 sets the absolute inlet velocity; the runner vane angle θ must match the relative velocity there, or the water hits the blade (shock). At the design point the water leaves with no whirl, so Euler's work is just Vw1 u1 and ηh = Vw1 u1/(gH). For Teesta-V with the default Ku 0.73, B1/D1 0.12 and ηh 0.96: α1 = 21.4°, θ = 105.7° (more than 90° because Vw1 is less than u1) and outlet φ = 15.6°.

A Francis runner: guide vane and runner blade angles from head, flow and speed, and why efficiency falls at part load: interactive loads as you scroll

How do head, flow and speed fix the runner's size?

The speed ratio Ku = u1/√(2gH) and the generator speed fix the inlet diameter D1 = 60u1/(πN). The flow ratio Kf = Vf1/√(2gH) and continuity Q = πD1B1Vf1 × (vane blockage, here 0.95) fix the inlet width. ESHA's statistics of over two hundred runners give the outlet diameter. Teesta-V: u1 = 45.4 m/s, D1 = 4.04 m, B1 = 0.485 m, Kf = 0.257, outlet D3 = 3.37 m. Where a real dimension is not published, these are derived numbers, not NHPC's drawings.

Why does a Francis turbine lose efficiency at part load?

To cut flow the wicket gates close, but the runner blades are fixed. The water now meets the blades at the wrong angle (shock loss) and leaves with whirl that the draft tube cannot recover; at low loads that swirl forms a vortex rope that shakes the machine. In the book's simple model, half-open gates at Teesta-V pass 55.7 m³/s at 90.6 % hydraulic efficiency, against 96 % at full gate, while faster Chutak drops to 79.3 %. ESHA says Francis turbines work well from about half of rated flow and may become unstable below 40 %.

Why does a low-head Francis runner look so different from a high-head one?

Specific speed sets the shape. High-head Teesta-V (nQE 0.119, Ns 120) has a wide, narrow runner: outlet only 0.83 of the inlet diameter, the water mostly radial. Low-head Chutak (nQE 0.254, Ns 250) must swallow more flow per unit size, so its blades are taller and its outlet is 1.29 times its inlet: the flow turns axial early. Push Ns higher still and the Francis becomes a propeller, the Kaplan of chapter 16. ESHA's Francis range is nQE 0.05 to 0.33.

Predict, then check

A Francis turbine running at fixed speed closes its guide vanes to 60 %. How does the water leave the runner?

Worked examples

Teesta-V runner inlet · Teesta-V unit (NHPC): H = 197 m, Q = 281/3 = 93.67 m³/s, N = 214.3 rpm. Assume Ku = 0.73, B1/D1 = 0.12, vane blockage 0.95, ηh = 0.96, radial exit.
  1. √(2gH) = √(2 × 9.81 × 197) = 62.17 m/s; u1 = 0.73 × that = 45.38 m/s.
  2. D1 = 60u1/(πN) = 60 × 45.38 / (π × 214.3) = 4.045 m; B1 = 0.12 × D1 = 0.485 m.
  3. Vf1 = Q/(π D1 B1 × 0.95) = 93.67 / 5.859 = 15.99 m/s.
  4. Vw1 = ηh gH/u1 = 0.96 × 9.81 × 197 / 45.38 = 40.88 m/s.
  5. α1 = tan⁻¹(Vf1/Vw1) = 21.36°; θ = 180° − tan⁻¹(Vf1/(u1 − Vw1)) = 105.74°.

Answer: D1 ≈ 4.04 m, B1 ≈ 0.49 m, guide vane α1 ≈ 21.4°, inlet vane θ ≈ 105.7°.

Teesta-V runner outlet · Same unit, D1 = 4.045 m. ESHA's statistical outlet: D3/D1 = 1/(0.4 + 0.095/nQE), nQE = n√Q/(gH)^0.75 with n in rev/s.
  1. nQE = (214.3/60) × √93.67 / (9.81 × 197)^0.75 = 0.1186.
  2. D3/D1 = 1/(0.4 + 0.095/0.1186) = 0.8326; D3 = 3.368 m.
  3. u2 = π D3 N/60 = 37.79 m/s; Vf2 = Q/(πD3²/4) = 10.52 m/s.
  4. Outlet vane angle φ = tan⁻¹(Vf2/u2) = 15.55°.
  5. Leaving kinetic head Vf2²/2g = 5.64 m = 2.9 % of H: the draft tube recovers most of it.

Answer: D3 ≈ 3.37 m, φ ≈ 15.6°, exit kinetic head ≈ 5.6 m.

Power from a runner with given angles · Original inward-flow runner: D1 = 1.2 m, B1 = 0.15 m, 500 rpm, guide vane angle 18°, inlet vane angle 100° (from the direction of motion), radial exit, net head 110 m; vane thickness neglected.
  1. u1 = π × 1.2 × 500/60 = 31.416 m/s.
  2. From the triangle: Vf1 = Vw1 tan 18° and Vf1 = (u1 − Vw1) tan 80°, so Vw1 = u1 tan 80°/(tan 18° + tan 80°) = 29.714 m/s.
  3. Vf1 = 29.714 × tan 18° = 9.655 m/s; Q = π D1 B1 Vf1 = 5.46 m³/s.
  4. ηh = Vw1 u1/(gH) = 29.714 × 31.416 / (9.81 × 110) = 86.5 %.
  5. Power to the runner = ρ Q Vw1 u1 = 1000 × 5.46 × 933.5 = 5.1 MW.

Answer: Q ≈ 5.46 m³/s, ηh ≈ 86.5 %, runner power ≈ 5.1 MW.

Chapter 16

Kaplan and propeller turbines: blade angles from hub to tip

Which syllabus covers thisAU not in syllabusVTU BME502 Module 4SPPU 402043 Unit 1MU MEC602 Module 2KTU PEMET411 Module 3

What makes a Kaplan turbine different from a propeller turbine?

Both are axial-flow reaction turbines for low heads and big flows: a hub with four to eight blades in a pipe, wicket gates upstream, a draft tube below. In a propeller turbine the blades are fixed. In a Kaplan a servomotor in the hub turns every blade about its own axis, so both gates and blades follow the load: 'double regulated'. Teesta Low Dam III's four 33 MW Kaplans run under 21.34 m; BHEL's largest Kaplans, at Polavaram, are 12 × 80 MW with 6.85 m runners at up to 31 m, and its highest-head Kaplan, at Dudhganga, works at 62 m.

Why must a Kaplan blade twist from hub to tip?

Blade speed u = ωr grows with radius, but every radius should do the same work u Vw1 = ηh gH. That is the free vortex: r Vw1 is constant, so the whirl is strong at the hub and weak at the tip, while the axial velocity is the same everywhere. Angles here are from the plane of rotation. At Teesta Low Dam III (default hub ratio 0.42, ηh 0.94) the inlet blade angle falls from 73° at the hub to 17° at the tip, a twist of 56°: the tip blade lies almost flat.

A Kaplan runner from hub to tip: blade speed, whirl and blade angle at each radius, and adjustable against fixed blades: interactive loads as you scroll

How do flow, hub and speed set the runner?

Water passes through the annulus between hub and tip, so Q = (π/4)(Do² − Db²) Vf. ESHA's statistics suggest a hub ratio Db/Do = 0.25 + 0.0951/nQE, 0.425 for Teesta Low Dam III, and a runner of 4.76 m against the real 5.2 m (ANDRITZ). With a 0.42 hub the axial velocity is 9.91 m/s (Kf = 0.484) and the tip moves at 37.1 m/s, 1.81 times √(2gH): Kaplan tips run faster than the water.

Why does a Kaplan keep its efficiency at part load when a propeller does not?

A fixed-blade propeller is designed for one flow: away from it the water meets the blades at the wrong angle and leaves swirling. A Kaplan re-sets its blades for each flow, so its efficiency curve is the upper envelope of a whole family of propeller curves. In the book's model shapes, at half flow the Kaplan gives 92 % and the propeller 63 %. ESHA: double-regulated Kaplans work from about one fifth of rated flow, propellers only close to rated flow; best efficiency about 93 %.

What comes with a high specific speed?

Teesta Low Dam III has Ns 540 against 120 for Teesta-V's Francis: a small, fast runner for its flow, with few blades. The price is a low pressure under the blades and a big leaving velocity. Its critical Thoma number by ESHA's correlation is about 0.63, ten times a high-head Francis's, so it must sit below the tailwater, and its draft tube recovers a large share of the head. Chapter 17 does that calculation.

Predict, then check

In a free-vortex Kaplan runner, where is the inlet blade angle (measured from the plane of rotation) largest?

Worked examples

Teesta Low Dam III: flow velocity and speed ratios · Teesta Low Dam III unit (NHPC, ANDRITZ): H = 21.34 m, Q = 693.6/4 = 173.4 m³/s, N = 136.36 rpm, runner 5.2 m, 33 MW. Take hub ratio 0.42.
  1. Hub Db = 0.42 × 5.2 = 2.184 m; annulus A = (π/4)(5.2² − 2.184²) = 17.491 m².
  2. Vf = Q/A = 173.4 / 17.491 = 9.914 m/s; Kf = Vf/√(2gH) = 0.484.
  3. Tip speed = π × 5.2 × 136.36/60 = 37.13 m/s; Ku = 37.13 / √(2 × 9.81 × 21.34) = 1.814.
  4. Ns = 136.36 √33,000 / 21.34^1.25 = 540.

Answer: Vf ≈ 9.91 m/s, Ku ≈ 1.81, Kf ≈ 0.48, Ns ≈ 540.

Blade angles at hub and tip · Same unit, free vortex with whirl-free exit, ηh = 0.94 (assumed), Vf = 9.914 m/s, hub 2.184 m, tip 5.2 m.
  1. Hub: u = 15.59 m/s, Vw1 = ηh gH/u = 0.94 × 9.81 × 21.34 / 15.59 = 12.62 m/s.
  2. Hub inlet angle θ = tan⁻¹(Vf/(u − Vw1)) = 73.3°; outlet φ = tan⁻¹(Vf/u) = 32.4°.
  3. Tip: u = 37.13 m/s, Vw1 = 5.3 m/s, θ = 17.3°, φ = 15°.
  4. Twist of the inlet edge = 73.3 − 17.3 = 56°.

Answer: θ ≈ 73° at the hub and 17° at the tip, a twist of about 56°.

Speed and flow of a small Kaplan · Original low-head site: H = 12 m, 5 MW, overall efficiency 0.88, runner 3.0 m, hub 1.1 m, speed ratio Ku = 1.9 at the tip.
  1. Flow Q = P/(ρgHη) = 5×10⁶ / (1000 × 9.81 × 12 × 0.88) = 48.27 m³/s.
  2. Annulus = (π/4)(3.0² − 1.1²) = 6.118 m²; Vf = 7.89 m/s (Kf = 0.514).
  3. Tip speed = 1.9 √(2 × 9.81 × 12) = 29.15 m/s; N = 60u/(πD) = 185.6 rpm.
  4. Ns = N√P/H^1.25 = 185.6 × √5000 / 12^1.25 = 588: inside the Kaplan range.

Answer: Q ≈ 48.3 m³/s, Vf ≈ 7.89 m/s, N ≈ 186 rpm, Ns ≈ 588.

Chapter 17

Draft tubes, cavitation and the Thoma number, governing and turbine characteristics

Which syllabus covers thisAU not in syllabusVTU BME502 Module 4 (draft tubes)SPPU 402043 Unit 1 (draft tube, governing, cavitation)MU MEC602 Module 2 (draft tube, characteristics)KTU PEMET411 Module 3

What does a draft tube do?

Water leaves a reaction runner fast: 10.6 m/s at Teesta-V, a kinetic head of 5.8 m or 2.9 % of the head; 3.4 m or 16 % at low-head Teesta Low Dam III. A draft tube, a pipe that widens slowly from the runner to below the tailwater, turns most of that speed back into pressure and lets the runner sit above or below the tailwater without losing the height. Conical tubes reach about 90 %, simple elbows about 60 %; ESHA's best cone half-angle is 7°.

Draft tube and cavitation: pressure at the runner exit, the Thoma number and the highest safe setting of the turbine: interactive loads as you scroll

How low is the pressure at the runner exit?

Bernoulli from the runner exit (height Hs above the tailwater) to the tube outlet gives the absolute pressure head p2/ρg = Ha − Hs − ηd V2²/2g, with ηd = (V2² − V3² − 2g hd)/V2². The faster the exit and the better the tube, the lower p2: the tube 'sucks'. In the default state (Teesta-V, cone, area ratio 4, runner 4 m below the tailwater, 360 m altitude) p2 = 8.71 m absolute, still far above the 0.32 m vapour pressure at 25 °C.

What is the Thoma number and how does it set the turbine's height?

Cavitation starts on the blades, where local pressure is lower than the average p2. Thoma's σ = (Ha − Hv − Hs)/H measures the margin; model tests give the critical σc below which blades pit. ESHA's statistical fits: σc = 1.2715 nQE^1.41 (Francis), 1.5241 nQE^1.46 (Kaplan). So Hs,max = Ha − Hv − σc H: −2.8 m for Teesta-V and −3.6 m for Teesta Low Dam III, both below the tailwater. USBR adds a 0.3 m margin for weather and runner differences.

What are a turbine's characteristic curves?

Main characteristics are measured at constant head while the speed is varied, plotted per unit head: unit speed Nu = N/√H, unit flow Qu = Q/√H, unit power Pu = P/H^1.5 (Teesta-V: 15.27, 6.67, 61.5 kW). Operating characteristics are at constant speed: efficiency against load, flat for Pelton and Kaplan, peaked for Francis and propeller. Iso-efficiency 'hill' (muschel) curves combine both on one chart and show the best operating zone.

How are hydraulic turbines governed?

A synchronous generator must turn at 120f/poles: 28 poles give Teesta-V 214.3 rpm, 44 poles give Teesta Low Dam III 136.36 rpm. When load changes, the governor senses speed and moves the flow control through an oil servomotor: wicket gates on a Francis, gates and blades together on a Kaplan, spear and deflector on a Pelton. A surge tank and slow closure keep the water hammer in the penstock within limits.

Predict, then check

The same Francis turbine is installed at 2,700 m altitude instead of at sea level. Compared with sea level, the runner must sit…

Worked examples

Teesta Low Dam III: how high can the runner sit? · Teesta Low Dam III (NHPC): H = 21.34 m, Q = 173.4 m³/s per unit, 136.36 rpm, maximum tailwater 184.24 m; water at 25 °C.
  1. nQE = (136.36/60) × √173.4 / (9.81 × 21.34)^0.75 = 0.5438.
  2. ESHA Kaplan σc = 1.5241 × 0.5438^1.46 = 0.6262.
  3. Ha at 184 m (ISA) = 99.13 kPa = 10.105 m; Hv at 25 °C = 3.17 kPa = 0.323 m.
  4. Hs,max = Ha − Hv − σc H = 10.105 − 0.323 − 0.6262 × 21.34 = −3.58 m.

Answer: Hs,max ≈ −3.6 m: the runner must be at least 3.6 m below the tailwater (plus USBR's 0.3 m margin).

What a conical draft tube buys at Teesta-V · Teesta-V: Q = 93.67 m³/s, runner exit D3 = 3.349 m (ESHA estimate), conical tube with outlet area 4 times the inlet and ηd = 0.90, runner 4 m below the tailwater, plant at 360 m, water 25 °C.
  1. V2 = Q/(πD3²/4) = 10.64 m/s; V2²/2g = 5.765 m. V3 = V2/4 = 2.659 m/s.
  2. Tube loss hd = V2²/2g − V3²/2g − ηd V2²/2g = 5.765 − 0.36 − 5.189 = 0.216 m.
  3. Ha at 360 m = 9.895 m; p2/ρg = Ha − Hs − ηd V2²/2g = 9.895 + 4 − 5.189 = 8.71 m absolute.
  4. Without a tube the exit kinetic head 5.77 m is lost; with it only 0.577 m is (tube loss plus outlet velocity).
  5. σ plant = (Ha − Hv − Hs)/H = (9.895 − 0.323 + 4)/197 = 0.0689 > σc 0.0629: safe.

Answer: p2 ≈ 8.71 m abs, 5.19 m of the 5.77 m exit head recovered, no cavitation at Hs = −4 m.

Chutak: the price of altitude · Chutak (NHPC, Ladakh): H = 52 m, Q = 24.05 m³/s, 333.33 rpm, about 2,676 m above sea level (Kargil town; the power house elevation is not published); water at 10 °C.
  1. nQE = (333.33/60) × √24.05 / (9.81 × 52)^0.75 = 0.2538; σc = 1.2715 × 0.2538^1.41 = 0.1839; σc H = 9.57 m.
  2. Ha at 2,676 m = 73.05 kPa = 7.446 m (sea level 10.329 m); Hv at 10 °C = 0.125 m.
  3. Hs,max at Chutak = 7.446 − 0.125 − 9.57 = −2.24 m.
  4. The same machine at sea level: 10.329 − 0.125 − 9.57 = 0.64 m.

Answer: Hs,max ≈ −2.2 m at Chutak against 0.6 m at sea level: 2.9 m deeper excavation because of thin air.

Chapter 18

Centrifugal pumps: heads, efficiencies, slip and the minimum starting speed

Which syllabus covers thisAU not in syllabusVTU BME502 Module 5SPPU 402043 Unit 3MU MEC602 Module 3KTU PEMET411 Module 4

What are the parts of a centrifugal pump, and what does each do?

The impeller, a wheel of curved blades, spins the water and throws it outwards, raising its speed and its pressure. The volute (a spiral casing whose area grows around the impeller) or a ring of diffuser vanes slows the water and turns most of that speed into pressure. A shaft seal (gland packing or a mechanical seal) stops leaks where the shaft leaves the casing; bearings carry the shaft. The book's pump has a 260 mm impeller with 7 backward-curved blades at 22.5°, 22 mm wide at the tip, at 1450 rpm (assumed numbers). A real Kirloskar KS-1022+ monobloc (7.5 kW, 1430 rpm) is in the same class: 29.0 L/s against 18 m.

Suction, delivery, static and manometric head: what is the difference?

Suction lift hs is the height of the pump above the sump's water level, and delivery height hd is the height from the pump up to the outlet level. Their sum is the static head Hst, the height the water is actually raised. The pump must also overcome friction in both pipes and give the water its exit velocity head. That total is the manometric head Hm, which is what you read as the delivery gauge minus the suction gauge (in metres of water). The book's pump at its best point gives Hm = 23.0 m at 26.8 L/s. If the static head is 18 m, about 5 m of that goes on friction and exit velocity.

Why does the real head-flow curve sit so far below Euler's line?

Take the book's pump apart at 26.8 L/s and 1450 rpm. Euler's equation with infinitely many blades gives H∞ = 31.25 m. With only 7 blades the water leaves with less swirl than the blades suggest. This is slip: Wiesner's slip factor σ = 0.842, and it takes away 6.29 m, leaving Hth = 24.96 m. Friction and incidence losses take 1.95 m more, leaving H = 23.00 m. Slip costs head but no power: the blades simply do less work. Meanwhile 1.4 L/s leaks from the volute back to the eye and is pumped again.

Inside a centrifugal pump's curve: Euler head, slip, hydraulic losses and leakage, and the minimum speed to start pumping: interactive loads as you scroll

What are manometric, volumetric, mechanical and overall efficiency?

Manometric (hydraulic) efficiency is the head delivered divided by the head the blades gave: 23.00/24.96 = 92.2 %. Textbooks that ignore slip divide by H∞ = Vw2 u2/g instead, which gives a lower 73.6 %. Volumetric efficiency is the delivered flow over the flow through the impeller: 95 %. Mechanical efficiency is the power reaching the water over the shaft power: 86.3 % (bearings, seal and disc friction cost 1.1 kW). Their product is the overall efficiency, 75.6 %. Of the 8.00 kW at the shaft, 6.05 kW reaches the water, 0.51 kW is lost to friction and incidence, and 0.34 kW goes to pumping the leakage.

Why does a pump need a minimum speed before it delivers anything?

With the delivery valve shut, the water inside the impeller just spins with it, like a forced vortex. Its pressure rises from eye to tip by (u2² − u1²)/2g, which is 16.3 m for the book's pump at 1450 rpm. The textbook rule: the pump delivers only once this head reaches the manometric head, which at zero flow is just the static head. Against 18 m that needs N ≥ 1,523 rpm. The rule is cautious. Water also leaves the tip swirling, worth up to u2²/2g more, and the volute recovers part of it, so the model's shut-off head is 28.2 m and it starts delivering at about 1,158 rpm.

Predict, then check

The book's pump shuts off at 28.2 m at 1450 rpm. A drive slows it to 1160 rpm (80 %). Its shut-off head becomes about:

Worked examples

Manometric efficiency from gauge readings · An original test: impeller 250 mm, outlet width 20 mm (blade thickness neglected), β2 = 30° from the tangent, 1440 rpm, flow 25 L/s with no inlet whirl. Suction gauge −4.0 m, delivery gauge +18.0 m of water at the same level, equal pipe sizes. Shaft power 8.4 kW
  1. u2 = π × 0.25 × 1440/60 = 18.85 m/s; flow velocity Vf2 = Q/(π D2 b2) = 0.025/(π × 0.25 × 0.02) = 1.592 m/s
  2. Vw2 = u2 − Vf2 cot 30° = 18.85 − 1.592 × 1.732 = 16.09 m/s; Euler head u2 Vw2/g = 30.92 m
  3. Manometric head = 18.0 − (−4.0) = 22 m (equal velocities, same level)
  4. Manometric efficiency = 22/30.92 = 71.1 %
  5. Water power = 1000 × 9.81 × 0.025 × 22 = 5396 W; overall efficiency = 5.396/8.4 = 64.2 %
  6. Check with slip (Wiesner, 7 blades): σ = 0.819, Hth = 24.36 m, so the blades really gave less and the hydraulic losses are only 2.36 m (η = 90.3 %)

Answer: Manometric efficiency ≈ 71 % (no-slip textbook basis), overall ≈ 64 %

Minimum starting speed of a pump · An original pump: impeller outer diameter 400 mm, eye diameter 200 mm, manometric head 30 m. Use the forced-vortex rule (u2² − u1²)/2g ≥ Hm
  1. (π N/60)² (D2² − D1²)/(2g) = Hm → N = (60/π) √(2g Hm/(D2² − D1²))
  2. N = (60/π) √(2 × 9.81 × 30/(0.16 − 0.04)) = 19.099 × √4905 = 1337.6 rpm
  3. Check at 1450 rpm: u2 = 30.37, u1 = 15.18 m/s, forced-vortex head = 35.25 m, more than 30 m
  4. So a 4-pole motor (about 1450 rpm) starts this pump; a 6-pole motor (under 1000 rpm) would not

Answer: Minimum starting speed ≈ 1,338 rpm; at 1450 rpm the pump starts

The book's pump at its duty: from Euler's head to the shaft · The book's pump (assumed): D2 260 mm, b2 22 mm, blockage 0.90, β2 22.5°, Z = 7, 1450 rpm, Q = 26.8 L/s. Model: leakage QL = 0.00028 √Hth, friction 2600 Q², incidence 4900 (Q − 0.031)², mechanical loss 1.1 kW
  1. U2 = π × 0.26 × 1450/60 = 19.74 m/s; outlet area π × 0.26 × 0.022 × 0.90 = 0.01617 m²
  2. Leakage converges in three passes to QL = 1.399 L/s, so the impeller passes 28.2 L/s and Cm2 = 1.744 m/s
  3. H∞ = U2 (U2 − Cm2 cot 22.5°)/g = 19.74 × (19.74 − 4.209)/9.81 = 31.25 m
  4. Wiesner σ = 1 − √(sin 22.5°)/7^0.7 = 0.8416; slip head (1 − σ) U2²/g = 6.29 m; Hth = 24.96 m
  5. Losses: friction 1.867 m + incidence 0.086 m; H = 23.00 m
  6. Shaft power = ρg(Q + QL)Hth + 1.1 = 6.904 + 1.1 = 8.00 kW; water power ρgQH = 6.048 kW
  7. η = 6.048/8.004 = 75.6 % = 0.922 × 0.95 × 0.863

Answer: H ≈ 23.0 m, shaft power ≈ 8.00 kW, η ≈ 75.6 %

Chapter 19

Suction lift, NPSH, cavitation and priming

Which syllabus covers thisAU not in syllabusVTU BME502 Module 5SPPU 402043 Unit 3 (priming; NPSH and Thoma in term work)MU MEC602 Module 3KTU PEMET411 Module 4

What is NPSH available, and how much does the book's pump get?

NPSH (net positive suction head) available is how far the total head at the pump inlet sits above the vapour pressure of the water, in metres. Start with the atmosphere pushing on the sump: Ha = 10.33 m at sea level. Then subtract the vapour pressure head (Hv = 0.32 m at 25 °C), the suction lift (4 m) and the suction losses (0.95 m of pipe friction plus 1.07 m in the foot valve and bend at 26.8 L/s in a 100 mm pipe). That leaves NPSHa = 3.98 m against the pump's NPSH3 of 2.38 m, a margin of 1.60 m. Above 32.7 L/s the pump cavitates.

Suction side of a pump: NPSH available against required, with water temperature, altitude, suction lift and pipe losses: interactive loads as you scroll

What is cavitation, and what is NPSH3?

As water enters the impeller eye it speeds up and its pressure drops. If the pressure falls to the vapour pressure, bubbles of vapour form. A little further along the blade the pressure is higher again, and the bubbles collapse violently against the metal. The result is pitting, noise, vibration and a fall in head. The Hydraulic Institute defines a pump's required NPSH as the NPSH that causes a 3 % head drop (NPSH3), a rule in use since 1932, so some cavitation is already present at NPSH3. Materials matter: relative to cast iron, HI rates carbon steel 3.4 times, CF8M stainless steel 14 times and aluminium bronze 28 times as resistant to cavitation damage.

How do hot water and a hill site change the suction side?

Warm water boils at a higher pressure: the vapour pressure head is 0.32 m at 25 °C, 4.83 m at 80 °C and 7.15 m at 90 °C. Altitude removes atmosphere: Ha is 10.33 m at sea level, 8.10 m at 2,000 m. For the book's pump at its best point, with the HI margin, the highest suction lift is 4.54 m at sea level in 25 °C water. At 2,000 m it falls to 2.53 m. With 80 °C water only 0.47 m is left, and at 90 °C the pump needs a flooded suction, the tank surface 1.8 m above it. No pump lifts water by suction more than the atmosphere's 10.3 m.

How much margin above NPSH3 is enough?

ANSI/HI 9.6.1-2017 sets the margin as NPSHA − NPSH3 and the margin ratio as NPSHA/NPSH3. For general industrial pumps it suggests a ratio of 1.05 with at least 0.6 m in the preferred operating region near the best point, and 1.1 with at least 1.0 m over the wider allowable region. For municipal water pumps it says 1.0 m within the preferred region is adequate. The interactive uses the stricter 1.1 / 1.0 m rule. The book's pump has suction specific speed S = N√Q/NPSH3^0.75 = 124 (6,388 in US units), under the 165 limit often used for water pumps. Its eye tip speed of 8.35 m/s is 'low' by HI (under 15 m/s).

Why must a centrifugal pump be primed, and what does a foot valve do?

An impeller makes a head in metres of whatever fluid fills it. Full of air, the book's pump still makes its 28.2 m at shut-off, but that is 28.2 m of air. Air at 1.204 kg/m³ is about 830 times lighter than water, so that head equals only 34 mm of water: far too little to lift water up the suction pipe. Priming means filling the casing and the suction pipe with water before starting. A foot valve (a non-return valve with a strainer at the bottom of the suction pipe) holds that water in when the pump stops. Self-priming pumps, a flooded suction or a submersible pump avoid the problem.

Predict, then check

The book's pump lifts 25 °C water 4 m at 26.8 L/s (NPSHa 3.98 m, NPSH3 2.38 m). The water is warmed to 70 °C, nothing else changes. Now:

Worked examples

NPSH available at a hill site · An original installation: 15 L/s lifted 3.5 m from an open sump at 900 m altitude, water at 30 °C, 80 mm commercial steel suction pipe (ε 0.045 mm) 7 m long, foot valve + bend K = 2.0 in all. Catalogue NPSH3 = 2.6 m at this flow. ρ = 1000, ν = 1.0 × 10⁻⁶ m²/s
  1. V = 0.015/(π × 0.08²/4) = 2.984 m/s; Re = 238,732; Swamee-Jain f = 0.019
  2. hf = (f L/D + K) V²/2g = (1.661 + 2.0) × 0.4539 = 1.662 m
  3. Atmosphere at 900 m (ISA): 90.97 kPa → Ha = 9.273 m; vapour pressure at 30 °C: 4.247 kPa → Hv = 0.433 m
  4. NPSHa = 9.273 − 0.433 − 3.5 − 1.662 = 3.68 m
  5. HI (allowable region): need max(1.1 × 2.6, 2.6 + 1.0) = 3.6 m; margin = 1.08 m

Answer: NPSHa ≈ 3.68 m: above NPSH3 by 1.08 m, meets the HI 1.0 m margin

Highest suction lift with the HI margin · An original pump with NPSH3 = 3.0 m at its duty, water at 20 °C, plant at 1,500 m, suction losses 1.2 m at the duty flow
  1. Ha = 84.56 kPa → 8.619 m; Hv at 20 °C = 2.339 kPa → 0.238 m
  2. Required NPSHa (HI allowable region) = max(1.1 × 3.0, 3.0 + 1.0) = 4 m
  3. z_max = Ha − Hv − hf − NPSHa,req = 8.619 − 0.238 − 1.2 − 4 = 3.18 m
  4. Near the best point only (HI preferred region: 1.05, 0.6 m): need 3.6 m and z_max = 3.58 m

Answer: Mount the pump no more than about 3.2 m above the water (3.6 m if it always runs near its best point)

Doubling a pump's speed on the same tank · The book's pump is run at 2900 rpm instead of 1450 rpm (same impeller). It draws from a tank through 5 m of 150 mm pipe (K 1.8), 25 °C water, sea level
  1. Affinity: best point moves to Q = 53.7 L/s, H = 91.9 m; NPSH3 ∝ N²: 2.39 × 4 = 9.54 m
  2. HI need = max(1.1 × 9.54, 9.54 + 1.0) = 10.54 m
  3. With the pump level with the tank surface: V = 3.04 m/s, hf = 1.105 m, NPSHa = 10.329 − 0.323 − 1.105 = 8.9 m
  4. Shortfall 10.54 − 8.9 = 1.64 m: the tank surface must stand that far above the pump
  5. At 1450 rpm the same line gives NPSHa = 9.73 m against NPSH3 = 2.39 m: a lift of several metres was fine

Answer: At 2900 rpm the pump needs a flooded suction of about 1.6 m; speeding a pump up quadruples its NPSH3

Chapter 20

Pumps in series and parallel, multistage pumps and specific speed of pumps

Which syllabus covers thisAU not in syllabusVTU BME502 Module 5SPPU 402043 Unit 3MU MEC602 Module 3 (operating point)KTU PEMET411 Module 4

What sets where a pump runs?

A pump runs where its head-flow curve meets the system curve, H = Hst + K Q²: the static lift plus friction that grows with the square of the flow. The Fluid Mechanics book has an interactive of one pump on one system. Put the book's pump on a 15 m static lift with 5 m of pipe friction at 30 L/s. It settles at 32.1 L/s and 20.7 m, η 74.3 %, drawing 8.77 kW. Ask for more flow and there are three choices: a bigger pump, a faster one, or a second pump in parallel or in series.

Two pumps: parallel or series?

In parallel, flows add at the same head. In series, heads add at the same flow. On the system above, two book pumps in parallel give 42.4 L/s (+32 %), each at 21.2 L/s and η 74.1 %, for 14 kW. In series they give 45.2 L/s (+41 %), but each pump runs far right of its best point at η 58.6 %, for 19.9 kW. On a friction-heavy system (1 m static, 25 m friction at 30 L/s), parallel adds only 9 % and series 30 %. Doubling the pumps never doubles the flow, because friction grows as Q².

Two pumps in series or parallel on a system curve: the operating point, each pump's share and when combining helps: interactive loads as you scroll

What happens when the two pumps are not the same?

Slow pump B to 1200 rpm and its shut-off head falls to 19.3 m. In parallel on the 15 m system, the common head is 20.7 m, higher than B can reach. B's non-return valve stays shut, B delivers nothing and churns, wasting 0.92 kW and heating the water trapped in its casing. In series, a weak pump pushed past its own run-out makes negative head and acts as a brake. Mixed pumps in parallel need matched shut-off heads, or controls that stop the weaker one.

What are multistage pumps, and what does specific speed tell you?

A multistage pump puts several impellers in series on one shaft. Borewell submersibles and boiler feed pumps work this way, so each stage stays a good shape. Pump specific speed Ns = N√Q/H^0.75 (rpm, m³/s, m) sets that shape. The book's pump has Ns = 22.6. To lift 26.8 L/s through 92 m at 1450 rpm in one stage would mean Ns = 8.0: a very large, narrow impeller with high disc friction. Four stages of 23 m keep Ns at 22.6. Low Ns means narrow radial impellers (high head, low flow). Rising Ns goes to wider mixed-flow, then axial propellers (high flow, low head).

Do real catalogue pumps behave like this?

Kirloskar's KS-1022+ monobloc (7.5 kW, 1430 rpm) lists 36.0, 33.0, 29.0, 24.2 and 17.5 L/s at 14, 16, 18, 20 and 22 m. A parabola H = 24.65 − 0.00795 q² (q in L/s) through the 16 m and 20 m points reproduces the others within 0.8 L/s. That is the same falling, bowed shape as the book's pump, which delivers 26.8 L/s at 23 m. Two such pumps on a 10 m static lift with 4 m friction at 30 L/s give 34.4 L/s alone, 47.7 L/s in parallel and 44 L/s in series (fit values).

Predict, then check

Two of the book's pumps in parallel on a 15 m static lift with 5 m friction at 30 L/s. One pump alone gives 32.1 L/s. The pair gives about:

Worked examples

Single, parallel and series operating points · Original pumps, each H = 30 − 4000 Q² (Q in m³/s), η = 0.8 (2x − x²) with x = Q/0.04. System H = 12 + 6000 Q²
  1. One pump: 30 − 4000 Q² = 12 + 6000 Q² → Q = √(18/10,000) = 0.0424 m³/s, H = 22.8 m
  2. Parallel (each pump Q/2): 30 − 1000 Q² = 12 + 6000 Q² → Q = √(18/7000) = 0.0507 m³/s, H = 27.43 m
  3. Series: 60 − 8000 Q² = 12 + 6000 Q² → Q = √(48/14,000) = 0.0586 m³/s, H = 32.57 m
  4. Efficiencies: one pump 79.7 %; parallel each at 0.0254 m³/s: 69.3 %; series each at 0.0586 m³/s: 62.8 %
  5. Power ρgQH/η: one 11.91 kW; parallel 19.7 kW; series 29.8 kW

Answer: Parallel 50.7 L/s, series 58.6 L/s (one pump 42.4 L/s); series gives more flow here but at 63 % efficiency

Two real catalogue pumps from a curve fit · Kirloskar KS-1022+ catalogue: 33.0 L/s at 16 m and 24.2 L/s at 20 m (1430 rpm). Original system: 10 m static, 4 m friction at 30 L/s
  1. Fit H = A − B q²: B = (20 − 16)/(33.0² − 24.2²) = 0.007947 m/(L/s)²; A = 16 + B × 33.0² = 24.654 m
  2. Check: at 14 m the fit gives 36.6 L/s (catalogue 36.0); at 22 m 18.3 (catalogue 17.5)
  3. System K = 4/30² = 0.00444 m/(L/s)². One pump: q = √((A − 10)/(B + K)) = 34.39 L/s at 15.26 m
  4. Parallel: q = √((A − 10)/(B/4 + K)) = 47.73 L/s at 20.13 m (23.87 L/s each)
  5. Series: q = √((2A − 10)/(2B + K)) = 43.96 L/s at 18.59 m

Answer: One pump ≈ 34.4 L/s; parallel ≈ 47.7 L/s; series ≈ 44 L/s

How many stages, and the specific speed of each · 26.8 L/s must be lifted through 92 m (manometric). Stages like the book's pump give 23.0 m each at 26.8 L/s and 1450 rpm
  1. Stages = 92/23.0 = 4 → 4 stages on one shaft
  2. Ns per stage = 1450 √0.0268/23^0.75 = 22.6
  3. One stage doing all 92 m at 1450 rpm: Ns = 1450 √0.0268/92^0.75 = 8, a quarter as large: a narrow, inefficient impeller
  4. Two stages of 46 m at 2900 rpm: Ns = 2900 √0.0268/46^0.75 = 26.9, close to the book's pump again

Answer: 4 stages at 1450 rpm (Ns ≈ 23 each) or 2 stages at 2900 rpm (Ns ≈ 27)

Chapter 21

Fans and blowers: types, fan laws, characteristics and noise

Which syllabus covers thisAU CAE353 Unit 2VTU BME502 practicals (fans, blowers)SPPU not in syllabusMU not in syllabusKTU PEMET411 Module 4

Fan, blower or compressor: where is the line?

By pressure. BEE's guide follows ASME's specific ratio, discharge pressure over suction pressure. A fan has a specific ratio up to 1.11 (up to 1136 mm water gauge), a blower 1.11 to 1.20 (1136-2066 mmWg), and a compressor more than 1.20. The fans in this chapter make 1000 Pa static, a ratio of 1.01, so the air hardly compresses. ASHRAE treats fan air as incompressible below 10 inches of water (about 2.5 kPa). Air at 20 °C weighs 1.204 kg/m³. Fans range from a ceiling fan (a propeller) through air-handling unit (AHU) fans to a boiler's forced draught (FD) and induced draught (ID) fans.

How do forward-curved, backward-curved, radial and axial fans differ?

BEE gives these peak efficiency ranges: airfoil and backward-curved 79-83 %, radial 69-75 %, forward-curved 60-65 %, vaneaxial 78-85 %. The interactive sizes one illustrative fan of each type for the same duty, 10 m³/s at 1000 Pa. At its best point the backward-curved fan needs 12.9 kW, the vaneaxial 13.8 kW, the radial 14.8 kW and the forward-curved 17.1 kW. Forward-curved fans are small, slow and cheap, so they suit packaged HVAC units. Radial fans are rugged and handle dusty gas. Airfoil blades are the most efficient but need clean air. Vaneaxial fans use guide vanes to straighten the swirl.

Fan types compared: forward-curved, backward-curved, radial and axial pressure and power curves, the fan laws and noise: interactive loads as you scroll

What are the fan laws, and why do they matter for hot gas?

For one fan on one duct, flow goes as speed, pressure as density × speed², and power as density × speed³. BEE's numbers: 10 % slower gives 19 % less pressure and 27 % less power. The backward-curved fan at 80 % speed moves 8 m³/s at 640 Pa and needs 6.62 kW, about half. Density matters too. At 150 °C, gas density is 0.693 of its 20 °C value, so pressure and power fall in proportion, while the flow in m³/s stays the same. The Electrical Drives book shows the energy a variable-speed drive saves over a damper.

What do the characteristic curves warn you about?

Backward-curved fans are 'non-overloading': power peaks near the best point, then falls, so a motor sized for that peak is never overloaded. In the interactive the peak is at 60 % of free delivery (ASHRAE: 50-60 % for airfoil fans). Forward-curved and radial fans draw more power the more air they move, so a duct with less resistance than expected overloads the motor. Forward-curved and axial pressure curves dip left of their peak, where the fan can hunt between flows: ASHRAE says avoid it. Total pressure is static plus velocity pressure ½ρV²: the backward-curved fan adds 47 Pa to its 1000 Pa static, so its total efficiency (81 %) beats its static efficiency (77.4 %).

How does fan noise change with speed?

Sound power level is L = 10 log10(W/10⁻¹² W) in dB. For the same fan at another speed, a rough rule puts the change at 50 log10(N2/N1) dB, plus 20 log10(ρ2/ρ1). Doubling the speed adds 15.1 dB, and 80 % speed takes off 4.8 dB. Two equal fans add only 10 log10 2 = 3 dB. These are estimates. Sources disagree on how size enters (20 or 70 log10 of the diameter ratio), and ASHRAE warns that tip speed alone is no basis for comparing different fans. Aerovent advises low-rpm fans and less system resistance, since sound output rises with static pressure.

Predict, then check

The backward-curved fan is slowed to 80 % speed on the same duct. Its shaft power becomes about:

Worked examples

Fan laws for an air-handling unit · An original AHU fan: 6.0 m³/s at 750 Pa static, 7.2 kW at 960 rpm. Demand falls to 4.8 m³/s on the same ducts (no dampers moved)
  1. N2 = 960 × 4.8/6.0 = 768 rpm
  2. p2 = 750 × (768/960)² = 480 Pa
  3. P2 = 7.2 × (768/960)³ = 3.686 kW, saving 3.51 kW (48.8 %)
  4. Noise change ≈ 50 log10(0.8) = −4.85 dB

Answer: 768 rpm, 480 Pa, 3.69 kW, about 4.8 dB quieter

Cold start of a boiler ID fan · An original ID fan rated 40 m³/s at 2.2 kPa and 112 kW shaft when handling flue gas at 160 °C (treated as air, R = 287, 101.325 kPa). It is started on 30 °C air at the same speed and damper setting
  1. ρ at 160 °C = 101,325/(287 × 433.15) = 0.8151 kg/m³; at 30 °C: 1.1646 kg/m³; ratio 1.4288
  2. Same speed, same duct: volume flow unchanged at 40 m³/s; mass flow 32.6 → 46.6 kg/s
  3. Pressure ∝ ρ: 2.2 × 1.4288 = 3.143 kPa; power ∝ ρ: 112 × 1.4288 = 160 kW
  4. A motor sized for 112 kW would be 43 % overloaded, so ID fans start with the inlet damper closed or on a variable-speed drive

Answer: Cold-air power ≈ 160 kW, 43 % above the hot rating

Static and total pressure, efficiency and specific ratio · An original fan test: 3.5 m³/s, static pressure rise 1250 Pa, outlet duct 0.7 m × 0.5 m, inlet velocity pressure negligible, shaft power 6.4 kW, air ρ = 1.2 kg/m³
  1. Outlet velocity = 3.5/0.35 = 10 m/s; velocity pressure ½ρV² = 60 Pa
  2. Fan total pressure = 1250 + 60 = 1310 Pa (127.4 mm water gauge static)
  3. Total efficiency = Q pt/P = 3.5 × 1310/6400 = 71.6 %; static efficiency = 3.5 × 1250/6400 = 68.4 %
  4. Specific ratio = (101,325 + 1310)/101,325 = 1.0129, below 1.11: a fan by the ASME bands

Answer: pt = 1310 Pa, ηtotal ≈ 71.6 %, ηstatic ≈ 68.4 %; it is a fan

Chapter 22

Centrifugal compressors: slip factor, power input factor and pressure ratio

Which syllabus covers thisAU CAE353 Unit 3VTU BME502 Module 5SPPU 402043 Unit 4MU MEC602 Module 4KTU PEMET411 Module 4

What are the parts of a centrifugal compressor stage, and what does each do?

Air enters axially through the inducer, the curved front of the impeller blades that turns it without a shock. The impeller flings it outwards: the book's stage has a 54 mm inducer and a 67 mm wheel, sizes from a Garrett G25-660 turbocharger that may spin at 165,000 rpm, a tip speed of 579 m/s. A diffuser (a ring with or without vanes) slows the fast air leaving the wheel and turns speed into pressure, and the volute, a growing spiral, collects it into the outlet pipe. In the book's stage about 70 % of the static pressure rise happens inside the impeller (its degree of reaction); the diffuser makes the rest.

How much work does the impeller put into each kilogram of air?

Euler's equation gives the work as U2 Cθ2 − U1 Cθ1 (Cθ is whirl, the tangential part of the air's velocity). With no whirl at the inlet and radial blades, ideal air would leave with Cθ2 = U2, so the work is U2². Real air leaves with less whirl (slip), backswept blades remove some on purpose, and disc friction adds a little. In the interactive's default stage (500 m/s tip speed, 12 blades, 40° backsweep) the air is heated by ΔT0 = 164.7 K and, at an assumed 76 % isentropic efficiency, leaves at 3.53 times the inlet pressure.

A centrifugal compressor stage: tip speed, slip and power input factor, pressure ratio and the inducer's relative Mach number: interactive loads as you scroll

What is the slip factor, and why do the formulas disagree?

Between two blades the air lags the rotation (a relative eddy), so it leaves with less whirl than the blade angle asks for. The slip factor σ is the whirl that survives as a share of the ideal. For 12 radial blades Stodola's formula gives 0.738, Stanitz's (valid for blade angles 45-90° from the tangent) 0.835 and Wiesner's 0.824. They are fits to different data, and none is exact: a Purdue study found Wiesner within ±0.05 for only 52 % of high-speed compressor cases. More blades mean less slip; in the interactive, doubling 12 blades to 24 lifts the pressure ratio from 3.53 to 3.89.

What does the power input factor add, and why is it not the same as slip?

Slip lowers the work the blades put into the air. The power input factor ψp (also called work factor) goes the other way: the shaft must also drive disc friction on the back of the wheel and windage, which heat the air without raising its pressure. Course notes give 1.035-1.04 as typical. So the drive sees ψp σ U2² per kg: at the default, 165.5 kJ/kg, of which 6.4 kJ/kg is the 4 % extra. Raising ψp in the interactive raises ΔT0 but, at the same isentropic efficiency, the extra heat is counted as useful; in reality it is a loss.

Why do turbocharger compressors use inlet guide vanes and backswept blades?

At the inducer tip the blade meets the air at the relative velocity W1, which combines the axial inflow with the blade speed. In the default stage that is Mach 1.29: supersonic, so shocks and losses. Guide vanes that pre-swirl the air with the rotation (20° here) cut it to 1.14 but also cut the work (pressure ratio 3.16). Backsweep trades work for a wider, more stable map: with radial blades the same speed gives 4.77, but their flatter characteristic leaves less room before surge. Garrett quotes 79 % peak efficiency for the G25-660 compressor; the Automobile book's turbocharging interactive shows what that boost does to the engine.

Predict, then check

In the book's stage at 500 m/s, the number of blades doubles from 12 to 24 (everything else fixed). The pressure ratio:

Worked examples

Pressure ratio of a radial-bladed industrial impeller · Impeller tip diameter 0.30 m at 18,000 rpm, 19 radial blades, no prewhirl, power input factor 1.04, isentropic efficiency 0.80, inlet 288.15 K, 2.5 kg/s
  1. U2 = π × 0.30 × 18,000/60 = 282.7 m/s
  2. Stanitz slip factor σ = 1 − 0.63π/19 = 0.8958
  3. ΔT0 = ψp σ U2²/cp = 1.04 × 0.8958 × 282.7²/1,005 = 74.1 K
  4. PR = (1 + 0.80 × 74.1/288.15)^3.5 = 1.925
  5. Power = ṁ cp ΔT0 = 2.5 × 1.005 × 74.1 = 186 kW
  6. Check with Wiesner: σ = 1 − 1/19^0.7 = 0.8727, ΔT0 = 72.2 K: the correlations differ by 1.9 K

Answer: PR ≈ 1.92, ΔT0 ≈ 74 K, about 186 kW of shaft power

The book's backswept turbocharger stage at 150,000 rpm · Exducer 67 mm, 12 blades, backsweep 40° from radial, Cm2 = 0.25 U2, Wiesner slip, ψp 1.04, efficiency 0.76, no prewhirl, inlet 288.15 K
  1. U2 = π × 0.067 × 150,000/60 = 526.2 m/s; Cm2 = 131.6 m/s
  2. Whirl without slip Cθ2∞ = U2 − Cm2 tan 40° = 415.8 m/s
  3. Wiesner slip velocity = U2 √(cos 40°)/12^0.7 = 80.9 m/s, so Cθ2 = 334.9 m/s
  4. ΔT0 = 1.04 × 526.2 × 334.9/1,005 = 182.4 K
  5. PR = (1 + 0.76 × 182.4/288.15)^3.5 = 3.954; power per kg/s = 183.3 kW

Answer: PR ≈ 3.95 with ΔT0 ≈ 182 K

Inducer tip Mach number with and without inlet guide vanes · The same stage at 150,000 rpm: inducer tip 54 mm, hub 18 mm, axial inflow Cx1 = 0.30 U2 = 157.9 m/s; then guide vanes give 25° prewhirl with the rotation
  1. U1 tip = π × 0.054 × 150,000/60 = 424.1 m/s
  2. No IGV: T1 = 288.15 − 157.9²/2,010 = 275.8 K, a1 = √(1.4 × 287 × T1) = 332.9 m/s; W1 = √(157.9² + 424.1²) = 452.5 m/s; M = 1.36
  3. With 25° prewhirl: Cθ1 = 157.9 tan 25° = 73.6 m/s; W1 = √(157.9² + (424.1 − 73.6)²) = 384.4 m/s; T1 = 273.1 K; M = 1.161
  4. Work lost at the rms inducer radius (U1 = 316.1 m/s): 1.04 × 316.1 × 73.6 = 24.2 kJ/kg; PR falls from 3.954 to 3.391

Answer: Prewhirl cuts the tip relative Mach number from 1.36 to 1.16 for a 13.2 % loss of work

Chapter 23

Compressor characteristics: surge, stall and choking

Which syllabus covers thisAU CAE353 Units 3, 4 (stall, surge)VTU BME502 Module 5 (surging)SPPU 402043 Unit 4 (surging, choking in term work)MU MEC602 Module 4KTU PEMET411 Module 4

How do you read a compressor map?

A map plots pressure ratio against flow, with one line per speed. Both are 'corrected' to the standard sea-level day (288.15 K, 101.325 kPa) so one map serves a hot Chennai afternoon and a cold test cell. Along each speed line the pressure rises as the flow is throttled, up to a peak. Joining the peaks gives the surge line; the right-hand ends, where the flow cannot grow any more, give the choke line. Closed loops of equal efficiency, the islands, sit between them. The interactive's map is ORIGINAL and illustrative, not any real compressor: at 100 % speed its surge point is 0.25 kg/s at PR 3.38.

Why does a compressor surge when the flow is throttled past the peak?

Left of the peak the speed line rises with flow. If the flow dips, the compressor's pressure drops while the plenum downstream (pipes, a combustor, a receiver) still holds pressure, so the flow dips further. Greitzer's model shows when this runs away: a small plenum stays steady a little past the peak, a big one (Greitzer B above a critical value) breaks into a cycle. In the interactive at 18 % throttle the point sits past the peak; with B = 2 the flow swings from −0.088 kg/s (reversed) to 0.353 kg/s and back, the deep surge of loud bangs; with B = 0.6 it stays put.

A compressor map: speed lines, the surge line and choking, and an operating point pushed towards surge: interactive loads as you scroll

What is rotating stall, and how is it different from surge?

Near the peak one or more blade passages can stall (the flow separates) while their neighbours carry more flow. The stalled patch, a stall cell, moves from passage to passage: MIT's propulsion notes say cells turn at about 0.4-0.6 of rotor speed when seen from outside. The average flow stays positive, but the pressure rise drops, the blades are shaken at a new frequency, and the stall is hard to clear. Surge is the whole system oscillating, with flow reversing. Which one appears depends on Greitzer's B: for a single stage, values below about 0.8 tend to give rotating stall, higher values surge. The 1-D model in the interactive cannot show rotating stall; the cell drawn is a picture.

What is choking, and why are high-speed lines almost vertical?

Open the throttle and the flow grows until the narrowest passage, the throat at the inducer or the diffuser, reaches the speed of sound. Beyond that, lowering the outlet pressure cannot pull more air through: the speed line drops straight down. At high speed the air is already close to sonic in the inducer, so the lines are short and steep; at low speed they are flatter. In the interactive at 60 % speed and a wide-open throttle the point sits on the vertical part at 0.228 kg/s, PR 1.29, efficiency 52 %: choked flow is inefficient flow.

What is surge margin, and how do plants protect a compressor?

There are several definitions. A common one compares flows at the same speed: SM = (ṁ − ṁsurge)/ṁ. Another, turndown, compares them at the same head. The interactive's default point has SM = 15.7 %. Industrial compressors carry an anti-surge recycle valve: when the controller sees the point reach a control line, typically 10 % of flow from surge, it opens a bypass from outlet to inlet so the compressor keeps enough flow even when the process wants less. At 18 % throttle that recycle is 0.063 kg/s, wasted work that buys safety.

Predict, then check

The throttle is closed until the operating point sits just past the peak of its speed line. Which system surges?

Worked examples

Corrected flow and speed from a hot test · A turbocharger compressor tested at 40 °C inlet (313.15 K) and 99.0 kPa passes 0.30 kg/s at 140,000 rpm
  1. θ = 313.15/288.15 = 1.0868; δ = 99.0/101.325 = 0.9771
  2. ṁc = ṁ √θ/δ = 0.30 × 1.0425/0.9771 = 0.3201 kg/s
  3. Nc = N/√θ = 140,000/1.0425 = 134,295 rpm

Answer: Plot it on the map at 0.32 kg/s and 134,295 rpm (corrected)

Surge margin two ways on the book's map · Book's map at 100 % speed, throttle 35 %: operating point 0.2967 kg/s at PR 3.224. The surge line is the locus of speed-line peaks, ṁ = 0.25 N and PR = 1 + 2.376 N² (N = fraction of design speed)
  1. Same speed: surge flow at N = 1 is 0.25 kg/s, so SM = (0.2967 − 0.25)/0.2967 = 15.7 %
  2. Same head (turndown): the surge line reaches PR 3.224 at N = √((3.224 − 1)/2.376) = 0.9674, where ṁ = 0.2418 kg/s
  3. TD = (0.2967 − 0.2418)/0.2967 = 18.5 %

Answer: SM = 15.7 % at constant speed but 18.5 % at constant head: always say which definition

Will it surge? Greitzer's stability test · Book's map at 100 % speed, throttle 18 % (γ = 0.534 in φT = γ√ψ). The operating point is φ = 0.4269, ψ = 0.6392, left of the peak at φ = 0.5
  1. Slope of the characteristic dψc/dφ = (H/W)(1.5 − 1.5x²), x = φ/W − 1 = 0.7077: 0.5391 (positive: past the peak)
  2. Throttle term dφT/dψ = γ/(2√ψ) = 0.334
  3. Unstable when B × 0.5391 > 0.334/B, i.e. B > √(0.334/0.5391) = 0.787
  4. B = 0.6: stable. B = 2: surge, with a cycle period of 8.79 in Greitzer's time units and the flow reversing to −0.088 kg/s

Answer: Critical B ≈ 0.79: the same compressor and throttle are safe with a small plenum and surge with a big one

Chapter 24

Axial-flow compressors: work-done factor, stage count and free vortex design

Which syllabus covers thisAU CAE353 Unit 4VTU BME502 Module 2 (axial compressors)SPPU 402043 Unit 4MU not in syllabusKTU PEMET411 Module 2

How is an axial compressor built, and why does it need so many stages?

A drum or a stack of discs carries rows of rotor blades; between them, rows of stator blades are fixed to the casing. One rotor row and the stator after it make a stage; inlet guide vanes may set the first whirl. The air moves along the axis, so each stage can only diffuse it a little before the boundary layers separate. In the interactive's default stage (tip speed 400 m/s, hub-tip ratio 0.6) each stage does about 30.6 K of ideal work (before the work-done factor), so a pressure ratio of 8 needs 11 stages. Chapter 9 covers one stage at its mean radius; this chapter follows the blade from hub to tip.

What is the work-done factor, and why does it fall with the number of stages?

Stage by stage the slow boundary layers on the hub and casing grow, so the axial velocity peaks in the middle of the annulus. Where the axial velocity is higher, the blades turn the air less and do less work. The work-done factor λ scales the ideal stage work down. Commonly quoted values are 0.98, 0.93 and 0.88 for the first three stages and 0.83 after; their average over 4 stages is 0.905 and over 12 stages 0.855, close to Howell and Bonham's 0.9 for four stages and Howell's 0.86 for many. Sources quote different values: treat λ as experience, not physics.

What does a free vortex design do from hub to tip?

Blade speed grows with radius, so the blade cannot have the same angles everywhere. A free vortex keeps r Cθ constant at inlet and outlet: whirl falls as 1/r while U grows as r, so U ΔCθ, the work, is the same at every radius, and with uniform axial velocity the flow is in radial equilibrium. The price is twist and reaction: R = 1 − (1 − Rm)(rm/r)². In the default (ν 0.6, Rm 0.5) R is 0.111 at the hub and 0.68 at the tip, the stagger changes by 49.1° and the hub is the most loaded section (de Haller 0.76).

Axial compressor design: free vortex against forced vortex blades from hub to tip, reaction, twist and the work-done factor: interactive loads as you scroll

Why not use a forced vortex or a constant-reaction design?

A forced vortex (Cθ proportional to r, like a solid body) keeps the reaction at the mean value everywhere and cuts the twist to 13.7° from 49.1° in the default. But the work now grows as r², and the flow is no longer in radial equilibrium: to balance the pressure across the annulus the outlet axial velocity would have to fall from 210 m/s at the hub to 137 m/s at the tip. The first-power law (Cθ = a r ∓ b/r) keeps constant work and constant reaction with uniform axial velocity, but it too needs the axial velocity to vary. Real blades are designed with full radial equilibrium and CFD.

What limits a stage, and how do axial compressors stall and surge?

Stage losses come from blade profiles, the end walls (annulus), secondary flows in the corners and leakage over the blade tips. The rotor may only slow the relative flow so much: de Haller's rule keeps W2/W1 above 0.72, and the diffusion factor stays below about 0.5. Raise the default loading from 0.30 to 0.40 and the hub's ratio falls to 0.70. A low hub-tip ratio of 0.45 makes the hub reaction −0.3, so the rotor hub accelerates the flow. At low flow the incidence rises, rotating stall appears (the front stages are the first to stall at low speed) and the whole machine can surge, as in chapter 23.

Predict, then check

A free vortex stage with mean reaction 0.5 has its hub-tip ratio lowered from 0.6 to 0.45. The reaction at the hub:

Worked examples

Free vortex reaction at hub and tip · Hub-tip ratio 0.5, mean radius at (rh + rt)/2, mean reaction 0.5, free vortex with uniform axial velocity
  1. rm/rt = 0.75, rh/rt = 0.5
  2. R hub = 1 − (1 − 0.5)(0.75/0.5)² = −0.125
  3. R tip = 1 − (1 − 0.5)(0.75/1)² = 0.7188
  4. For R hub ≥ 0 the mean reaction must be at least 1 − (0.5/0.75)² = 0.5556

Answer: R hub = −0.125 (negative), R tip = 0.719; a mean reaction of 0.56 or more keeps the hub positive

Number of stages with the work-done factor · Overall pressure ratio 6 at polytropic efficiency 0.90, inlet 288.15 K, mean blade speed 300 m/s, stage loading ψ = 0.35 at every stage
  1. Required T02/T01 = 6^(0.2857/0.90) = 1.7662, so ΔT0 total = 220.8 K
  2. Ideal stage ΔT0 = ψ Um²/cp = 0.35 × 300²/1,005 = 31.34 K
  3. With λ = 0.98, 0.93, 0.88, 0.83, 0.83, …: cumulative ΔT0 = 30.7, 59.9, 87.4, 113.5, 139.5, 165.5, 191.5, 217.5, 243.5 K
  4. With one mean λ = 0.86: n = 220.8/(0.86 × 31.34) = 8.19 → 9

Answer: 9 stages. Ignoring λ, 8 stages look enough, but with λ they give only 217.5 K, a pressure ratio of 5.88 instead of 6

One stage at its mean radius with λ · Second stage of an axial compressor: U = 280 m/s, Cx = 160 m/s, rotor relative angles β1 = 52°, β2 = 33° from axial, λ = 0.93, stage isentropic efficiency 0.88, inlet T0 = 288.15 K
  1. ΔT0 = λ U Cx (tan β1 − tan β2)/cp = 0.93 × 280 × 160 × (1.2799 − 0.6494)/1,005 = 26.14 K
  2. Stage PR = (1 + 0.88 × 26.14/288.15)^3.5 = 1.3084
  3. de Haller W2/W1 = cos 52°/cos 33° = 0.734 (≥ 0.72: acceptable)
  4. Reaction with uniform Cx: R = (Cx/2U)(tan β1 + tan β2) = 0.551

Answer: ΔT0 ≈ 26.1 K and stage PR ≈ 1.308, with de Haller 0.73

Chapter 25

Axial and radial gas turbines: stage loading, losses and the Smith chart

Which syllabus covers thisAU CAE353 Unit 5VTU not in syllabusSPPU not in syllabusMU not in syllabusKTU not in syllabus

How is an axial gas turbine stage described by three numbers?

A stage is a row of nozzle guide vanes (stator) that accelerates and turns the hot gas, then a rotor row that takes the whirl out. Divide everything by the blade speed U and three numbers fix the velocity triangles: flow coefficient φ = Cx/U, stage loading ψ = Δh0/U² (work per kg over U²) and reaction R, the share of the static enthalpy drop that happens in the rotor. For a repeating stage (exit velocity equal to inlet velocity) the whirl leaving the nozzles is Cθ2/U = 1 − R + ψ/2 and leaving the rotor Cθ3/U = 1 − R − ψ/2. Angles here are measured from the axial direction.

Impulse or 50 % reaction: what changes?

With no exit whirl (Cθ3 = 0) the formulas give ψ = 2(1 − R): an impulse stage (R = 0) does twice the work per U² of a 50 % reaction stage (ψ = 1). So an impulse design needs fewer stages for the same enthalpy drop. The price is loss: at φ = 0.6 Soderberg's correlation gives ηtt = 89.2 % for the impulse stage against 92.3 % for the reaction one, because the impulse rotor turns the gas through 118° instead of 59°. The Thermal Engineering book has steam-turbine impulse and reaction stage interactives.

Where are the losses in a turbine stage, and what does Soderberg's correlation do?

Losses come from the blade profile (boundary layers and the trailing-edge wake), secondary flows (vortices in the passage corners), leakage over the rotor tips and the kinetic energy thrown away at the exit. Soderberg lumped profile and secondary loss of each row into one coefficient that depends mainly on how far the row turns the flow, ε, for blades three times as tall as their axial chord. A row turning 60° loses 6.16 % of its exit kinetic energy; one turning 120° loses 12.64 %. Tip clearance and Mach effects are left out: published studies call it oversimplified but fine for first design.

What does the Smith chart show?

Plot stage loading ψ against flow coefficient φ and mark efficiency at each point: that is a Smith chart, first drawn from measured turbines in 1965. The interactive computes one from Soderberg's losses. At the default (φ 0.6, ψ 1.4, R 0.5) ηtt = 90.7 %, but total-to-static efficiency, which counts the exit kinetic energy as lost, is only 80.3 %. Efficiency falls as loading rises: at φ 0.6, ψ = 1 gives 92.3 % and ψ = 2 gives 87.4 %. The dashed line marks the best φ at each loading (0.99 at ψ = 2). Designers trade a few points for fewer stages.

Gas turbine stages on the Smith chart: flow and loading coefficients, Soderberg losses, efficiency and the radial turbine: interactive loads as you scroll

Why do turbochargers use radial inflow turbines, and how fast must the wheel spin?

For small flows a radial inflow turbine is cheaper and tougher than an axial one: one casting takes a large enthalpy drop. In the nominal design the gas meets radial blades with whirl equal to the tip speed and leaves without whirl, so the work is U2²; dividing by the spouting velocity c0 = √(2Δh0s) gives U2/c0 = √(ηts/2), 0.707 at best. NASA measured ηts 0.84 and ηtt 0.90 at 0.697, because real rotors run with negative incidence (whirl about 0.85 U2: 0.703). Garrett's G25-660 turbine wheel (54 mm) at 165,000 rpm has U2 = 467 m/s and uses 220 kJ/kg. Big gas turbines split the expansion between a gas generator turbine and a free power turbine.

Predict, then check

At φ = 0.6 and 50 % reaction, the stage loading ψ is doubled from 1 to 2. The total-to-total efficiency (Soderberg):

Worked examples

Velocity triangles of a repeating 50 % reaction stage · Mean blade speed 340 m/s, φ = 0.6, ψ = 1.5, R = 0.5, repeating stage (C3 = C1), angles from axial, whirl positive with the rotation
  1. Cx = 0.6 × 340 = 204 m/s; work = ψU² = 1.5 × 340² = 173,400 J/kg = 173.4 kJ/kg
  2. Cθ2/U = 1 − 0.5 + 0.75 = 1.25 → α2 = atan(1.25/0.6) = 64.4°
  3. Cθ3/U = 1 − 0.5 − 0.75 = −0.25 → α3 = α1 = −22.6°
  4. Rotor: β2 = atan((1.25 − 1)/0.6) = 22.6°, β3 = atan((−0.25 − 1)/0.6) = −64.4°
  5. Deflections: nozzle 87°, rotor 87° (equal: a 50 % stage is symmetric)

Answer: Nozzle exit angle 64.4°, rotor exit angle −64.4°, 173.4 kJ/kg per stage

Soderberg efficiency of that stage · The stage above, blades of height 3 times the axial chord (Soderberg's nominal case), Re 10⁵
  1. ζ = 0.04 [1 + 1.5 (87/100)²] = 0.0854 for each row
  2. c2 = U √(0.6² + 1.25²) = 471.4 m/s; w3 = U √(0.6² + (−1.25)²) = 471.4 m/s
  3. Nozzle loss ζ c2²/2 = 9.49 kJ/kg; rotor loss ζ w3²/2 = 9.49 kJ/kg
  4. ηtt = 173.4/(173.4 + 9.49 + 9.49) = 0.9014
  5. Exit kinetic energy c1²/2 = 24.42 kJ/kg; ηts = 0.7998

Answer: ηtt ≈ 90.1 % and ηts ≈ 80 %: a diffuser or the next stage must use the exit energy

Radial inflow turbine for a small turbocharger · Rotor tip diameter 60 mm at 120,000 rpm, total-to-static efficiency 0.82, rotor inlet whirl 0.88 U2, zero exit whirl, gas treated as air (cp 1,005, γ 1.4) entering at T01 = 1,000 K
  1. U2 = π × 0.060 × 120,000/60 = 377 m/s
  2. U2/c0 = √(ηts/(2 × 0.88)) = √(0.82/1.76) = 0.6826; c0 = 552.3 m/s
  3. Δh0s = c0²/2 = 152.5 kJ/kg; work = 0.88 U2² = 125.1 kJ/kg = ηts Δh0s
  4. Expansion ratio p01/p3 = (1 − Δh0s/(cp T01))^(−3.5) = 1.779; T0 falls by 124.4 K

Answer: U2/c0 ≈ 0.68, about 153 kJ/kg available and an expansion ratio near 1.78

Keep going

Recommended reading

  • Fluid Mechanics and Thermodynamics of Turbomachinery, S. L. Dixon and C. A. Hall
  • Turbines, Compressors and Fans, S. M. Yahya
  • Gas Turbines, V. Ganesan
  • Turbomachines, B. U. Pai
  • Fluid Mechanics and Hydraulic Machines, R. K. Bansal
  • Gas Turbine Theory, H. I. H. Saravanamuttoo, G. F. C. Rogers, H. Cohen and P. V. Straznicky

How this book was made

Chapters follow the official syllabi listed above. Every interactive's formulas and every worked-example number were checked by script. Explanations, animations and examples are original to Skill-Lync.

From the exam to a core job

Talk to a counsellor.

Ask how turbomachinery and CFD engineers use turbomachinery on real projects, and what the PG Program in CFD covers.

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.

Talk to a career counsellor

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.