Machinery vibration monitoring is the continuous measurement of how a rotating machine shakes, used both to protect the machine from a fast-developing fault and to reveal wear long before it becomes a failure. It is the core of a machinery protection system on critical rotating equipment: sensors watch the shaft and the casing, electronics turn their signals into meaningful values, and alarm and trip logic acts when those values cross safe limits. This page gives a rotating-equipment focus, covering the sensors, the difference between overall and spectral values, and the protection logic that sits behind them.
Vibration Monitoring in one line: Machinery vibration monitoring uses shaft proximity probes and casing accelerometers to measure a rotating machine's vibration continuously, converting it into overall levels and frequency spectra that reveal imbalance, misalignment, bearing wear, and instability. A machinery protection system compares those values against alarm and trip setpoints and shuts the machine down before damage occurs. Built to standards like API 670, it is the primary defense for critical turbomachinery and the richest source of early-warning condition data.
The right sensor depends on how the machine is built. On machines with fluid-film bearings, where the shaft floats in an oil film, eddy-current proximity probes watch the shaft directly and measure its displacement and position, because the shaft moves relative to a casing that stays comparatively still. Two probes at each bearing describe the shaft's orbit, and a shaft-speed reference probe ties the vibration to rotor angle for detailed analysis.
On machines with rolling-element bearings, casing vibration tracks bearing condition well, so seismic accelerometers bolted to the housing are the sensor of choice; they measure the acceleration of the case and are integrated to velocity for most fault work. Many critical machines carry both kinds, proximity probes watching the shaft and accelerometers watching the case, so that a fault which shows in one but not the other is not missed. Choosing and placing sensors is not arbitrary, and the reference for doing it consistently is API 670, the machinery protection systems standard.
Whatever the sensor, the goal is the same: get a faithful, continuous electrical picture of the machine's motion at each bearing. That signal chain, from transducer through signal conditioner to monitor, has to be reliable enough to trust with a shutdown decision, which is why it is engineered and documented to a recognized standard rather than assembled ad hoc.
Vibration data is used at two levels of detail. The overall value is a single number, the total vibration energy at a point, and it is the value the protection system uses for alarm and trip decisions because it is simple, robust, and fast. A rising overall level says something is getting worse, and that is enough to protect the machine. What the overall value does not tell you is why the vibration is rising.
Spectrum analysis answers the why. By breaking the vibration into its component frequencies, an analyst reads the fault directly from where the energy sits: a peak at running speed points to imbalance, a strong peak at twice running speed points to misalignment or a coupling problem, sidebands and high-frequency energy point to bearing or gear defects, and a peak below running speed points to an oil-film instability. Overall values catch that a problem exists; the spectrum names it.
The protection layer turns overall values into action through alarm-and-trip logic. When vibration crosses the alarm setpoint, operators are warned so they can investigate and plan; when it crosses the higher trip setpoint, the system shuts the machine down to prevent damage. To avoid tripping on a single bad reading or a wiring fault, critical machines use voting logic, requiring more than one channel to confirm before a shutdown, so the system is both protective and dependable.
The trip logic that protects a machine has to be fast and self-contained, so it lives in a local machinery protection system at the unit, independent of any network. That independence is deliberate: a critical shutdown must never wait on a link to a supervisory system or fail because that link is down. The protection system stands alone and does its job whether or not anything upstream is listening.
But the values that protection system produces, overall vibration at each bearing, axial position, and alarm and trip status, are exactly the data a reliability team wants historized and trended across the whole fleet. That is the supervisory role, complementary to and separate from protection. Where protection asks is this machine safe right now, supervisory monitoring asks how is this machine, and every sister machine, trending over weeks and months.
A cloud SCADA platform reads those vibration values over the same protocols as the process instruments and trends them against each machine's own baseline. A bearing whose overall vibration creeps up a little each week, or a fleet of pumps where one stands out from its siblings, is visible long before any trip point. For unattended stations Merobix alarms on-call staff the moment a machine trips or a trend crosses a limit, and preserves the vibration history that led up to it, turning point-of-machine protection into fleet-wide predictive insight in one web-native view.
Overall vibration is a single number representing the total vibration energy at a measurement point, and it is what the protection system uses to alarm and trip because it is simple and reliable. A spectrum breaks that vibration into its component frequencies so an analyst can see where the energy sits and identify the specific fault, such as imbalance at running speed or misalignment at twice running speed. Overall values tell you a problem exists; the spectrum tells you what it is.
API 670 is the industry standard for machinery protection systems, covering how vibration, axial-position, and related sensors are selected, placed, and wired, how the monitors and trip logic are configured, and the whole chain from transducer to shutdown. Following it gives critical machines consistent, dependable protection regardless of the equipment or instrument vendor. It is the common reference operators and manufacturers use when specifying vibration monitoring for turbomachinery.
A machinery protection system must be able to trip a machine fast and reliably, so it is engineered as an independent, self-contained system that does not depend on a network or on the process control system to do its job. That independence ensures a critical shutdown never waits on a communication link or fails because that link is down. Supervisory platforms then read the protection system's values for trending and alarming, but they sit alongside protection rather than being part of the trip path.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.