The cost of an unplanned stop
Manufacturing runs on continuity. When a line stops without warning, the loss is larger than the repair. Material in process may be scrapped. Downstream stations stand idle. Deliveries slip. People wait. The failed component is often inexpensive. The stoppage it causes is not.
Most plants manage this risk with a combination of scheduled maintenance and experience. Parts are replaced at fixed intervals whether or not they are worn, and skilled technicians walk the floor listening for trouble. Both methods work, up to a point. Scheduled replacement discards useful life and still misses failures that arrive early. Experienced ears are scarce and cannot be everywhere.
Machines speak in physics
Rotating equipment has been understood through its vibration for generations. Imbalance, misalignment, looseness and bearing or gear wear each leave a recognizable pattern, tied to the speed and geometry of the machine. Reliability engineers read these patterns with handheld instruments and a great deal of training.
The same machines also speak in other ways. A bearing short of lubrication changes its sound before its vibration becomes severe. A motor under electrical stress shows it in its electromagnetic behavior and in its temperature. A compressed-air leak is silent to a person and loud in ultrasound.
AFFERENT is designed to turn physical behavior into machine-readable intelligence. The intent is to give every important asset on the floor the kind of attention that, until now, only the most critical machines have received.
The assets no one has time for
A plant typically protects its most critical machines well. Large turbines, presses and primary drives often have permanent monitoring. The gap is in the long list of ordinary assets: the pumps, fans, conveyors, small motors and gearboxes that number in the hundreds or thousands. Individually none of them justifies a dedicated monitoring system. Collectively they account for a great deal of lost production.
Covering those assets requires a device that is inexpensive to install, needs little configuration and can interpret what it sees without a specialist. That is the design brief for NODE.
How AFFERENT is designed to help
A NODE mounted on a motor or pump is designed to learn the normal behavior of that individual machine across its speeds and loads. Learning an individual baseline matters in manufacturing, where the same model of motor may be doing very different work at two stations.
Processing happens on the device. Plant networks are busy, segmented and protected for good reason, and a monitoring system should not add to their burden.
Findings are designed to arrive in a form a maintenance planner can use: which machine, what has changed, how confident the system is and what to check first. The intended destination is the maintenance system the plant already uses, reached through AFFERENT Cloud integrations.
The common fault families
A small number of fault types account for a large share of problems in rotating machinery. Each has its own physical character, which is why distinguishing them is possible.
- Imbalance. Mass is unevenly distributed around the axis of rotation, often from wear, deposits or a lost blade weight. It produces vibration once per revolution.
- Misalignment. Coupled shafts are not on a common line. It loads bearings and couplings and typically appears at multiples of running speed, with a strong axial component.
- Looseness. A mounting, base or bearing fit has slackened. The signature is rich in harmonics and often erratic.
- Bearing defects. Damage on a race or rolling element creates repeating impacts at frequencies set by the bearing's geometry, usually visible first at high frequency.
- Gear wear. Changes appear around the gear mesh frequency and in the pattern of sidebands beside it.
- Lubrication problems. Too little, too much or contaminated lubricant raises friction, with acoustic and thermal effects that precede mechanical damage.
- Electrical faults in motors. Problems in rotor bars, windings or supply quality show in the motor's electromagnetic behavior and in its current, and can produce vibration at frequencies linked to the electrical supply.
- Process-induced problems. Cavitation in pumps, surge in compressors and flow turbulence are caused by operating conditions and damage machines over time.
An alarm that says only that vibration is high leaves a technician to work out which of these is present. A finding that points toward a family of fault, and toward the part of the machine involved, lets the right person arrive with the right parts. That is the level of specificity AFFERENT is designed to reach, and it is the reason for combining several kinds of sensing. See multimodal sensing.
Starting small
Plants seldom adopt a new maintenance approach all at once, and they should not have to. A sensible first step is a single line or utility area with a known history of trouble: a group of pumps, a compressed-air system, a set of conveyor drives. Devices are installed, baselines are learned over a period of normal operation and findings are compared with what the maintenance team already knows about those machines.
That comparison is the real test. A system earns wider use when experienced technicians find that it agrees with them where they are confident and tells them something useful where they are not.
From calendar to condition
The practical goal is to let a plant maintain equipment according to its actual state. Components that are healthy stay in service. Components that are degrading are replaced during a planned stop. Technicians spend their time on machines that need them. Spare parts are ordered because a need has been identified, not because a date has arrived.
Over time, condition data also answers questions that are otherwise matters of opinion: which suppliers' components last, which operating practices shorten equipment life and where a redesign would pay for itself.
