Steam turbines: impulse and reaction principles; staging and compounding; components; condensing/bleeder/topping/extraction arrangements; governors; start-up/shut-down; condensers

Misjudge how a steam turbine works and the plant pays for it fast: overspeed it and it flies apart; let wet steam reach the low-pressure blades and they erode; pick the wrong arrangement and you starve the process header; lose condenser vacuum and you cook the last-stage buckets. A steam turbine converts the energy in high-pressure steam into rotating shaft work, and everything in this lesson grows from one fact.

The one fact everything hangs on. When steam expands through a properly shaped passage, its pressure drops and its velocity rises. How much of that drop happens in the fixed parts versus the moving parts, and how the drop is split into stages, defines the whole machine. Read every later beat against this single idea.

Terms defined first - meet them before they appear. A few words below are used early, so define them once now: a nozzle is a shaped passage (fixed) that turns pressure into velocity; buckets/blades are the vanes on the rotor that the steam pushes; a diaphragm is a stationary disc that holds a ring of nozzles between rotor wheels; a stage is one nozzle-set plus the blade row it feeds; a gland/seal is the packing where the shaft passes through the casing; critical speed is a shaft speed where the rotor naturally vibrates (you pass through it quickly, never dwell); soaking speed is a low speed you hold to let metal warm evenly; hogging/sagging is a stopped rotor bowing up (hot) or down (cold). Each is re-explained in context where it matters.

Variable cheat-sheet - keep this open for the two calculations. Only two quantitative relations live in this lesson; everything else is descriptive. Each symbol is also taught in its own beat.

Impulse stage - the pressure drops only in the nozzles. In an impulse stage the entire pressure drop happens in the stationary nozzles. The high-velocity jet strikes the moving buckets, changes direction, and hands over its momentum; across the moving blades the pressure stays essentially constant while velocity falls. That constant pressure across the moving blades is the distinguishing feature of an impulse stage.

Reaction stage - the pressure ALSO drops across the moving blades. In a reaction stage the moving blades are themselves shaped like nozzles, so the steam keeps expanding and losing pressure as it passes through them. The distinguishing feature is a pressure drop across the moving blades: velocity rises in the fixed blades and falls in the moving blades, while pressure falls across both. The contrast with impulse is a difference of WHERE the drop occurs, not an on/off switch - examiners test exactly that nuance.

Reaction blades leak and thrust - so they are built differently. Because there is a pressure difference across reaction blades, steam leaks around their tips, so clearances must be kept minimal. That same pressure difference produces a cumulative axial thrust in the direction of flow that must be balanced. Impulse blades, with no pressure drop across them, do not suffer this to the same degree - which is why reaction turbines use drum rotors, not discs.

Nozzles set up the energy conversion. A convergent nozzle (narrowing passage) handles small pressure drops; in a good design the steam reaches the critical pressure right at the exit. For larger pressure drops a convergent-divergent nozzle is used: the throat sits at the critical pressure, and the diverging (widening) section adds flow area so the expanding steam keeps accelerating without forming eddies.

The critical-pressure rule - 0.577 of inlet, and WHY. The critical pressure is the convergent-nozzle exit pressure for peak efficiency, and it equals $0.577 \times p_{inlet}$. Here is the intuition, not a magic number: at the critical pressure the steam reaches sonic velocity (the local speed of sound) right at the throat. A convergent shape cannot push steam past sonic; any further pressure drop in a convergent nozzle just churns into eddy currents (turbulence) instead of useful velocity - which is exactly why a bigger drop needs the diverging section. (The 0.577 figure is for steam and depends on the gas; treat it as the steam value to use.)

Worked example - critical pressure (forward)

A well-designed convergent nozzle has an inlet pressure of 1000 kPa. Find the exit pressure for peak efficiency.

• State the relation: $p_{crit} = 0.577 \times p_{inlet}$.
• Substitute: $0.577 \times 1000\ \text{kPa}$.
• Result: $p_{crit} = 577\ \text{kPa}$. Below 577 kPa the steam is already sonic at the throat, so extra drop in a convergent shape only makes eddies - that is when you reach for a convergent-divergent nozzle.

Now you try (reverse - faded). A convergent nozzle is observed to choke (reach peak efficiency) at an exit pressure of 462 kPa. Work BACKWARD: since $p_{crit} = 0.577 \times p_{inlet}$, the inlet must be $p_{inlet} = p_{crit} / 0.577 = 462 / 0.577 \approx 800\ \text{kPa}$. Notice the reverse is just dividing by 0.577, not multiplying - the exam likes to see if you can run the relation both ways.

Why compound - one stage spins too fast. If all the expansion happened in one stage, steam could leave the nozzles at roughly 1100 m/s. Peak efficiency needs blade speed about half the steam speed (~550 m/s), which forces punishing rotational speed, huge centrifugal force, and high friction loss. Compounding spreads the drop over several stages so blade speed - and shaft speed - stay sensible. That single reason is behind all three compounding methods.

Pressure compounding - drop the pressure in stages (Rateau). Pressure compounding takes the total drop in two or more impulse stages in series inside one casing: nozzle set, then moving-blade row, repeated. Each stage takes only part of the drop, so velocity rises in each nozzle set and falls in each blade row. On its pressure-velocity diagram the pressure steps DOWN stage by stage while velocity saw-tooths up and down - the Rateau arrangement.

Velocity compounding - drop all the pressure once, then take the velocity off in steps (Curtis). Velocity compounding drops the WHOLE pressure in one inlet nozzle set, then bleeds the velocity off in two moving rows separated by a stationary redirecting row. On its diagram the pressure falls ONCE in the nozzles and then stays FLAT across every blade row; only velocity falls, in two steps. This is the Curtis arrangement, common as the first (HP) stage of larger turbines.

Pressure-velocity compounding - two Curtis sections in series. Pressure-velocity compounding puts two or more velocity-compounded sections in series on one shaft. The total drop is split between the nozzle sets (e.g. half and half), and within each section velocity rises in the nozzles and falls in steps through its blades. Its diagram is two velocity-compounding diagrams back to back. Its payoff: high inlet pressures usable at relatively LOW rotational speed. It is the combination of the first two, not a third unrelated idea.

Pin the three methods by their pressure-velocity DIAGRAM. This is the exam's favourite trap. Pressure compounding: pressure steps down in EVERY nozzle set. Velocity compounding: pressure drops ONCE, only in the first nozzle set, then flat. Pressure-velocity: pressure drops in STAGES, each stage followed by velocity steps. Name them by the diagram behaviour, not by their sound-alike titles - a learner who only memorised the words will mix them up the moment a question describes the curve instead of naming the method.

Arrangement comparison table - re-anchor the look-alikes. Eight arrangements follow; many sound alike. Keep this one-line-purpose table in view as you read each beat, so you never conflate a STAGE method with a whole-machine arrangement.

Condensing vs non-condensing - defined by the exhaust. A condensing turbine exhausts BELOW atmospheric into a condenser to extract maximum work - the classic generator driver where there is no process use for the exhaust. A non-condensing (back-pressure) turbine exhausts ABOVE atmospheric into a process header, trading some power for usable process steam, and runs at very high efficiency because there are no exhaust losses - but its power output is tied to the process steam demand.

Match the exhaust to the duty. If a plant needs process steam at a header pressure, fit a back-pressure turbine and let the process set the load. If the plant only needs shaft power from a limited steam supply, fit a condensing turbine and pull the deepest vacuum you can. Choosing condensing for a process header (or back-pressure for a pure generator) is the classic duty mismatch.

Bleeder vs extraction - uncontrolled vs controlled. A bleeder turbine taps UNCONTROLLED steam (up to about 20%) at intermediate points, usually for feedwater heating; the amount simply drifts with load. An extraction turbine taps a CONTROLLED amount at a set pressure for process use, and may be condensing or non-condensing. The one word that separates them is control: bleed is uncontrolled, extraction is held at a set pressure.

Topping turbine - power as a by-product of letting steam down. A topping turbine is a back-pressure turbine whose job is to take high-pressure header steam, drop it to a lower process pressure the plant still needs, and generate power on the way down. The classic case: old LP boilers are replaced by new HP boilers, but part of the process still needs the original lower pressure - the topping turbine bridges the two and drives a generator as a by-product.

Compounded turbines - two turbines in series (NOT compounding of stages). Compounded turbines run two separate turbines in series, the exhaust of the first feeding the second. Do not confuse this with compounding of STAGES inside one machine. A TANDEM arrangement couples the shafts together to one common load; a CROSS arrangement uses separate shafts, each driving its own load. Both are used where one turbine would be too large or too fast.

Double-flow - split the steam to cancel thrust. A double-flow turbine admits steam at the CENTRE of the casing and flows it toward both ends, so the blade thrust from each half cancels the other. It also lets many LP stages handle the huge steam volume at the condenser inlet without an excessively large disc diameter. It addresses axial thrust and steam volume - not pressure level or process supply.

Reheat turbine - dry the steam between sections. A reheat turbine extracts the whole flow partway through, returns it to the boiler reheater to raise its temperature, then admits it to the next (IP/LP) section. This keeps efficiency up and prevents excessive wetness in the LP stages, where steam near saturation would erode the blades. It exists to control wetness and recover efficiency - not to cut mass flow, raise back pressure, or remove the condenser.

Casings hold the clearances - through every temperature swing. The casing (shell) holds the nozzles, diaphragms, and bearing cases under high pressure and temperature; it must resist distortion through start-up and shutdown while keeping the tiny blade and labyrinth-gland clearances and the rotor true. The horizontally split casing is most common. Casing metal climbs with temperature: large LP casings are often welded plate, smaller LP casings cast iron to about 230 degC; cast carbon steel IP casings to about 425 degC; and cast alloy steel (3% chromium / 1% molybdenum) for HP/HT casings above about 550 degC.

The sentinel valve WARNS, it does not protect. A sentinel valve sits at the highest point of the casing and relieves a little steam to make a whistle when casing pressure is abnormally high - typically a start-up warning that the operator left the exhaust valve shut. It is an alarm, not a relief device; a true exhaust relief valve sized for full flow is fitted separately when the casing and exhaust piping cannot take full inlet pressure.

Shaft seals - keep steam in and air out. Where the shaft passes through the casing, shaft seals stop HP/IP steam leaking out and stop air leaking into the sub-atmospheric LP section of a condensing turbine. Small turbines (shafts under about 150 mm) use carbon-ring seals; large turbines use labyrinth seals - thin knife-edged rings with minute clearances that throttle the leakage. Blade tip strips do the same job at the reaction blade tips, where the pressure drop drives tip leakage.

Disc rotors for impulse, drum rotors for reaction. Impulse sections use disc (wheel) rotors - thin large-diameter discs on a small shaft - because with no pressure drop across the blades there is no disc thrust. Reaction sections use a drum rotor, which avoids the large disc faces that would otherwise turn the reaction pressure drop into a huge axial thrust. The rotor choice follows directly from where the pressure drops.

Balancing reaction thrust - dummy piston and thrust gear. Reaction thrust is severe and toward the exhaust end. Where a double-flow layout cannot cancel it, three tools handle it: the THRUST BEARING (often a Kingsbury tilting-pad bearing) holds the rotor axially and absorbs residual thrust; the DUMMY PISTON is an enlarged rotor section exposed to HP steam so its force opposes (nearly balances) the blade thrust, leaving a slight thrust toward exhaust; and THRUST-ADJUSTING GEAR lets the operator shift the rotor axially during start-up to protect clearances until temperatures settle.

Reheat intercept valves - the syllabus calls them interceptor valves. On a reheat turbine the path between HP exhaust and IP inlet runs through the reheater and reheat piping, which store a large volume of energetic steam. A reheat stop valve and a reheat intercept valve guard that path, closing on a trip so the stored steam cannot drive the machine into overspeed. The SOPEEC syllabus term interceptor valve is PanGlobal's reheat intercept valve - same component, two names.

Barring (turning) gear - turn it slowly so it does not bow. A cold rotor at standstill SAGS between its bearings; a hot rotor at standstill HOGS (bows upward) as its lower half cools faster. Either way it will not start smoothly. Barring (turning) gear is a small motor-and-reduction-gear that rotates the shaft slowly (about 20-40 rpm, as low as 1 rpm on the largest units) before start-up (often about 3 hours) and after shutdown (often up to 24 hours) so the rotor warms and cools evenly. Jacking oil lifts the shaft off the bearings while it barrs.

Governors - sort them by what they HOLD. Speed-sensitive governors hold turbine speed against load changes; PanGlobal lists three speed-sensitive methods: nozzle governing (impulse turbines only - opens nozzle valves in sequence), throttle governing (always used on reaction turbines - throttles all the inlet steam through one or two valves), and bypass/overload governing (a second admission point for overload power). Pressure-sensitive governors instead hold a pressure: back-pressure governing holds a steady exhaust pressure, and extraction governing holds a set extraction pressure while still controlling speed.

Governor families - mechanical, mechanical-hydraulic, electro-hydraulic. A mechanical governor links the flyweights directly to the steam valve, so the speed must change before it can move - that is why it has a HIGH droop (around 10%) and suits pumps and fans, not generators. A mechanical-hydraulic governor inserts a pilot valve and oil-powered servo between the flyweights and the valve, cutting the force needed and driving droop toward zero. An electro-hydraulic governor measures speed electronically (a magnetic pickup), compares it to a reference, and positions a servo-valve - the most precise, and standard on large modern units.

Droop - the speed change from no-load to full-load. Droop is the speed change from no-load to full-load as a percentage of SET speed: droop = (no-load rpm - full-load rpm) / set rpm x 100. Low droop means tighter control. The divisor is the SET (no-load) speed, not the full-load speed - getting that backwards is the classic slip.

Worked example - governor droop (forward)

A turbine has a set (no-load) speed of 3000 rpm and the governor holds 2880 rpm at full load. Find the droop.

• State the formula: droop = (no-load rpm - full-load rpm) / set rpm x 100.
• Speed change: $3000 - 2880 = 120$ rpm.
• Substitute and solve: $\frac{120}{3000} \times 100 = 4.0\%$. Dividing by 2880 instead gives 4.17% - the classic trap (wrong divisor).

Now you try (reverse - faded). A governor has a droop of 5% and a set (no-load) speed of 1800 rpm. Work BACKWARD to the full-load speed: the speed change is $5\% \times 1800 = 90$ rpm, so full-load speed = $1800 - 90 = 1710$ rpm. The reverse uses the SAME divisor (set speed) - multiply set speed by the droop fraction to get the drop, then subtract.

Isochronous governing and hunting. An isochronous governor holds CONSTANT speed (zero droop) and is used only when a unit runs ALONE. Paralleling two isochronous units makes them fight for control, causing hunting (load and speed cycling) that can leave one unit fully loaded and the other unloaded. Units sharing a grid must therefore run with some droop, never isochronous.

Start-up logic - oil first, warm slow, race through criticals. The cold start-up sequence has a logic worth memorising even if exact steps vary by machine: establish lube oil (and, on a large unit, jacking oil and barring gear for hours) BEFORE steam; drain condensate from steam lines so no water slug hits the blading; on a condensing unit establish condenser level, cooling water, gland sealing steam, and vacuum before admitting steam; then warm at low speed and advance QUICKLY through the critical speeds to the soaking speed per the manufacturer. The recurring theme: protect bearings and clearances, and never admit steam to a cold, un-drained, un-sealed machine.

Shutdown logic - cool it evenly, keep it sealed. Reduce load to zero, trip or close the steam supply, let the machine coast down, then put a large unit on jacking oil and barring gear for its cooldown period (often up to 24 hours) so it cools without hogging. Keep gland sealing steam and the vacuum logic correct until the machine is cold; pulling vacuum or seals too early lets cold air thermally shock the casing.

Condenser - vacuum buys you more work. A condenser improves cycle efficiency by holding a vacuum (around 6.9 kPa abs / 710 mm Hg) at the turbine exhaust so steam expands further and does more work, returns condensate to the boiler, and removes non-condensable gases. PanGlobal puts the gain at about +50% turbine efficiency: roughly 20% exhausting to atmosphere, 30% condensing, and 80% or better when the exhaust is also used for process heat. Raising exhaust pressure is the exact opposite of its function.

Two condenser families - contact and surface. A contact (jet) condenser sprays the cooling water straight into the exhaust steam, so condensate and coolant mix and leave together - cheap but it contaminates the condensate, so it is rare; the barometric and ejector condensers are jet types. A surface condenser keeps steam and cooling water apart: cooling water runs through tubes, steam condenses on the outside, and the condensate is clean enough to return to the boiler - which is why surface condensers dominate power plants. (The SOPEEC syllabus also lists a 'Panier style' condenser; that name is not used in the PanGlobal text - see the lesson notes/flag.)

Surface-condenser sub-types - by coolant and by steam path. A water-cooled surface condenser rejects heat to circulating water (river, or a cooling-tower loop); an air-cooled condenser uses finned tubes and fans where cooling water is scarce. By steam PATH, a downflow condenser takes steam in at the top and drops condensate to the bottom hot well, while a central (radial) flow condenser draws steam radially inward toward central air extraction. All share the same job: condense at the lowest practical pressure to hold vacuum.

Air removal - eject it or pump it out. Because the shell sits below atmospheric, air leaks in and blankets the tubes, weakening the vacuum and heat transfer, so air must be removed continuously. A steam-jet air ejector expands HP steam through a nozzle and venturi to entrain and carry off the air (single-, two-, or three-stage for deeper vacuum; a start-up 'hogging' ejector pulls the initial vacuum). A vacuum pump - typically a liquid-ring type - does the same job mechanically. Both raise the air back to atmospheric pressure to vent it.

Feedwater heaters - spend the bleed steam, not the fuel. Bleed/extraction steam piped to feedwater heaters preheats the boiler feedwater, so less fuel is burned and less latent heat is dumped to the condenser - a direct efficiency gain that ties the turbine arrangement to the condenser circuit. Low-pressure heaters sit on the condensate-extraction-pump side; high-pressure heaters sit on the boiler-feed-pump side, usually with a deaerator between them. This is why bleeder and extraction arrangements exist in the first place.

Common misconceptions and exam traps. (1) Impulse = constant pressure across the moving blades; reaction = pressure drop across the moving blades - a difference of WHERE the drop is, not on/off. (2) Compounding of STAGES (pressure / velocity / pressure-velocity) is NOT compounded TURBINES (tandem / cross) - different ideas that sound alike. (3) Tell the three compounding methods apart by their pressure-velocity diagram, not their names: pressure compounding drops pressure in EVERY nozzle set; velocity compounding drops it ONCE then flat; pressure-velocity drops it in stages with velocity steps inside each. (4) Bleed is uncontrolled; extraction is controlled at a set pressure. (5) Back-pressure suits a process header; condensing suits a pure power duty - matching them backward is the duty trap. (6) Double-flow cancels thrust and handles volume; reheat dries the steam - do not swap their purposes. (7) Disc rotor = impulse; drum rotor = reaction (because of where the pressure drops). (8) The sentinel valve WARNS, it does not relieve. (9) In the droop formula the divisor is the SET (no-load) speed, not full-load; the reverse calculation uses that same divisor. (10) Below critical pressure a CONVERGENT nozzle just makes eddies (steam is already sonic) - you need a divergent section to accelerate further. (11) The syllabus 'interceptor valve' is PanGlobal's reheat intercept valve. (12) Contact/jet condensers contaminate condensate; surface condensers keep it clean - which is why surface types dominate.

Source: PanGlobal Power Engineering Third Class, Part B2, Book 2 (E30), Chapters 1-3 (Steam Turbine Principles and Design; Auxiliaries and Operation; Turbine Condenser Systems). SOPEEC 3rd Class Paper 3B2. The 'Panier style' condenser named in the SOPEEC syllabus is not a PanGlobal term and is flagged for verification.