REFERENCE

The maintenance glossary.

The floor's vocabulary, defined in plain language — from ladder logic to tribal knowledge.

AI maintenance agent: A tool that takes action on equipment problems in real time — reading the machine’s state, logic, and documentation, then guiding the person at the machine to the resolution. A chatbot answers questions; an agent acts. Jack is the AI maintenance agent built by Metis.
PLC (programmable logic controller): The industrial computer that runs a machine or production line. It reads inputs from sensors, executes control logic, and drives outputs like motors and valves. When a line faults, the explanation is almost always somewhere in the PLC’s logic.
Ladder logic: The dominant programming language for PLCs, drawn as rungs of contacts and coils that mimic relay wiring diagrams. Reading it fluently is a specialist skill — which is why a faulted rung can idle a line until the right person is available.
Rung: One line of ladder logic: a set of conditions (contacts) that must be true for an output (coil) to energize. Tracing a fault means following the failed rung back through the interlocks, timers, and inputs that gate it.
HMI (human-machine interface): The operator screen on or near a machine that displays its status and alarms. An HMI shows that a fault happened and its code; it usually does not explain which condition in the logic caused it.
CMMS (computerized maintenance management system): The system of record for maintenance: work orders, asset histories, PM schedules, and parts. A CMMS records and plans work; it is not built to diagnose a live fault at the machine.
Work order: The record that authorizes and tracks a piece of maintenance work — what needs doing, on which asset, by whom, and what was found. Work orders live in the CMMS and are the backbone of maintenance history.
PM (preventive maintenance): Scheduled maintenance performed to prevent failures rather than react to them — inspections, lubrication, part replacement on interval. PMs are planned in the CMMS and executed on the floor.
MTTR (mean time to repair): The average time from a failure to the equipment running again, across incidents. It captures diagnosis, waiting for the right person or part, and the repair itself — which is why expertise availability moves it as much as wrench time.
Unplanned downtime: Time a line is stopped by an unexpected failure rather than scheduled work. Its cost is the production not made while the line is idle, which makes diagnosis speed — not just repair speed — a direct economic lever.
Fault code: The identifier a controller or drive raises when something trips. A code names the symptom family; it rarely names the root cause, which still has to be traced through the logic to a physical component.
Interlock: A condition wired or programmed to prevent a machine from operating unless it is satisfied — a guard closed, a pressure reached, an upstream step complete. Unsatisfied interlocks and permissives are among the most common reasons a machine refuses to start.
VFD (variable frequency drive): The power-electronics unit that controls a motor’s speed and torque. Drives raise their own faults, which a technician has to read together with the PLC’s state to understand a stoppage.
SCADA (supervisory control and data acquisition): The plant-level system that supervises equipment, collects process data, and raises alarms across many machines. SCADA gives visibility over the operation; acting on a specific fault still happens at the machine.
OEE (overall equipment effectiveness): A composite score of availability, performance, and quality that measures how much of a line’s potential output is realized. OEE tells you that output was lost; it does not tell the technician what to fix.
Tag (PLC tag): A named data point in a control system — an input, output, or internal value like Line3_ClampPressure. A plant’s tag naming is local dialect: understanding it is a prerequisite for reading that plant’s logic.
Tribal knowledge: The unwritten, experience-based understanding of how specific equipment actually behaves, held by individuals rather than recorded in any system. It is the most valuable diagnostic asset in most plants — and it retires with the person holding it.
Predictive maintenance: Using condition data — vibration, temperature, current — to forecast when a component is likely to fail so work can be planned ahead of the breakdown. A prediction is a better alarm; it does not itself guide the repair.
Intelligent maintenance: The layer of a digital transformation that turns plant signals into guided action at the machine — closing the gap between knowing a problem exists and fixing it. It sits on top of IoT, OEE, and CMMS investments rather than replacing them.
Knowledge capture: Recording what was actually wrong and what fixed it, at the moment the work happens, so the answer is available to the next technician who hits the same problem. The alternative — binders and exit interviews — reliably goes stale.

See the vocabulary in action on your own line.

Book a demo