Hydro Vibration and Shaft Monitoring
A gas turbine runs at 3,000 rpm. A hydro unit runs at a few hundred, and a large vertical machine at well under that. Almost everything that makes hydro vibration monitoring a separate discipline follows from that one fact.
Why the instruments are different
The first harmonic is near the bottom of the measuring range, not in the middle of it. A standard industrial accelerometer sold for rotating machinery is specified from around 10 Hz upward. On a machine turning at a few hundred rpm, once-per-revolution — the strongest and most diagnostic component there is — sits at or below that limit. Low-frequency transducers measuring absolute vibration displacement are what the duty needs, and choosing them by catalogue rather than by the machine's running speed is the commonest mistake on a hydro retrofit.
Displacement, not acceleration. At a few hertz, acceleration amplitudes are tiny even when the shaft is moving a great deal, because acceleration scales with the square of frequency. Displacement in micrometres is the quantity that carries the information, which is why the trip limits on a large vertical machine read as microns on the brackets rather than mm/s on the bearing housings.
Three different measurements, not one. On a steam turbine the question is usually how much the casing is shaking. On a hydro unit three separate things are measured, and they say different things:
- Absolute vibration of the bearing brackets — structural response, unbalance, hydraulic roughness.
- Shaft runout relative to the bearing — where the shaft actually is inside its bearing clearance.
- Rotor-to-stator air gap — rotor roundness, stator ovality, magnetic pull asymmetry.
The air gap measurement has no equivalent on a thermal machine at all. On a hydro generator the rotor rim is metres in circumference and the gap is millimetres; a rim that has gone slightly out of round, or a stator core that has settled, shows up there long before it shows up anywhere else.
Small machines are measured in velocity, large ones in displacement. On the 1.5 MW Pelton unit the protection trips on RMS velocity at 11.2 mm/s on the highest-reading transducer. On the vertical machine the protection reads bracket displacement, with alarm at 90 µm and trip at 100 µm, and shaft runout alarmed separately at 6 mm. Applying one machine's thresholds to the other is a common and expensive mistake.
Settings are useless without failure handling. On the vertical machine the vibration and runout channels were deliberately held out of the trip scheme until the vibration system itself had been commissioned and proved, and that was written into the setting schedule. Recording it there, rather than leaving the trips nominally in service and quietly disabled, is the difference between an honest protection scheme and a dangerous one.
What the signals are processed with
Raw thresholds are the floor, not the ceiling. Doing anything more needs the acquisition to support it: vibration and runout brought in on fast analogue input modules, in a controller that is not the one regulating the machine, with protection actions hard-wired across to the unit controller as well as passed over the network, so that a network problem cannot swallow a trip.
Shaft orbit. Two displacement transducers at 90° in the same plane give the path the shaft centre traces per revolution. The shape of that orbit says what is wrong: a circle is unbalance, an ellipse is stiffness asymmetry or misalignment, an inner loop is rub, and a drifting centre is a bearing clearance changing. A single probe and an RMS number cannot distinguish any of these.
Spectrum. A Fourier transform of the vibration record, with the harmonics referred to running speed, separates the mechanical causes from the hydraulic ones. On a hydro unit the components worth naming are the once-per-revolution component, the blade- or vane-passing frequency, the draft tube vortex rope at roughly a quarter to a third of running speed at part load, and the electrical components at twice line frequency that come from the generator rather than the turbine. Each has a different owner and a different fix, and on a slow machine they sit only a few hertz apart — which is why the transducer's low-frequency limit and the length of the record both matter.
Trend, not snapshot. Vibration data goes to the same archive as the process data, so that a change can be read against load, head, gate opening and temperature. A vibration level that is only a problem at a particular load is a hydraulic problem; one that tracks temperature is a bearing or alignment problem. The distinction is invisible in an isolated measurement.
What a retrofit is made of
A vibration and speed monitoring retrofit on an operating hydro unit is a small, dense project. The scope we quoted and built for the vertical machine:
Instruments. Six absolute vibration transducers on the machine structure, four air gap transducers around the generator, and an inductive speed sensor on the shaft with its own toothed wheel. Purpose-made mounting brackets for every one of them — on a machine of that age nothing has a transducer boss, and fabricating the mountings is a real part of the work.
Signal conditioning. A 200 V to 4–20 mA measuring transducer for the speed channel taken from the governor tachogenerator, and the existing speed monitor and high-speed trip module reused because they were sound, rather than replaced for the sake of it.
Controller. A fail-safe S7-1500 CPU per unit with analogue input and counter modules, local power supplies and a memory card. Around 160 signals per unit, six regulating loops, and roughly 600 m of screened instrument cable.
Where it lands. A vibration monitoring workstation in the main control room, an extension of the SCADA licence for the added signal count, and — the part most often forgotten in a retrofit — the new system wired into the plant's existing central alarm system, so that a vibration alarm reaches the operator the same way every other alarm does instead of living on a screen nobody watches.
Commissioning. Insulation resistance on every new line, software installation and configuration, functional configuration of around sixty functions, standalone commissioning, integrated commissioning and acceptance testing of the system as a whole.
Work we have done
On the vertical machine we added speed transducers with their trip relays set at 160 % of rated speed into the existing protection scheme, brought a fourth speed sensor that had been installed but never wired into service, added vibration transducers collected by a separate controller whose analogue outputs feed the unit control system, and changed the protection logic to use them.
On the Pelton unit we installed shaft runout measurement between turbine and generator, axial shaft displacement at the generator bearing, new mountings for the position, runout, displacement and speed transducers, and brought vibration into the trip scheme at an RMS velocity limit.
The settings those systems protect to, and the governing they protect, are described under Hydro Turbine Control and Protection.