Equipment Health Score
The composite condition index maintained for each critical asset — aggregating sensor readings, inspection results, failure history, age, operating hours, and maintenance compliance into a normalized health score that reliability engineers use to prioritize attention and predict degradation trajectories.
Why This Object Matters for AI
AI cannot provide actionable health dashboards or trigger condition-based maintenance without a formally managed health score model; without it, 'which assets are most at risk right now' requires an engineer to mentally synthesize dozens of data points per machine.
Maintenance & Reliability Capacity Profile
Typical CMC levels for maintenance & reliability in Manufacturing organizations.
CMC Dimension Scenarios
What each CMC level looks like specifically for Equipment Health Score. Baseline level is highlighted.
Equipment health is a gut feeling. The reliability engineer walks the floor, listens to machines, and says 'that one sounds rough.' When the plant manager asks 'which assets are most at risk right now?', the answer is the engineer's mental synthesis of a hundred variables they can't articulate. There's no formal health score — just experience and intuition that walks out the door at the end of the shift.
AI cannot assess equipment health because no health scoring model or condition data exists in any system.
Define a basic health scoring model — even a simple red/yellow/green rating assigned by the reliability engineer to each critical asset based on their assessment.
A basic health rating exists — the reliability engineer assigns red/yellow/green status to critical assets on a whiteboard or spreadsheet, updated when they feel like it. The criteria vary: one engineer rates based on vibration feel, another on age, another on recent failure frequency. 'Green' means different things depending on who scored it. Some assets have ratings; many don't. The spreadsheet represents one person's opinion, not a measured score.
AI can display the subjective health ratings but cannot analyze trends, predict degradation, or compare across equipment because the ratings lack objective criteria and consistent methodology.
Standardize the health scoring methodology — define the specific inputs (sensor readings, inspection results, failure history, age, operating hours), the weighting formula, and the scoring thresholds that produce each health level.
A standardized health scoring methodology exists with defined inputs: vibration readings, oil analysis results, inspection grades, failure count, maintenance compliance percentage, and age/operating hours. The reliability engineer calculates health scores monthly using a spreadsheet formula. All critical assets are scored consistently. But the score is a snapshot calculated manually — it reflects last month's data, not today's condition. Inputs come from separate systems and are manually assembled.
AI can trend monthly health scores and flag assets with declining trajectories. Cannot provide real-time health assessment because the score depends on manually assembled monthly inputs.
Link the health score model to live data sources — connect sensor readings, inspection results, and failure records so the health score calculation inputs update automatically rather than through monthly manual assembly.
Equipment health scores are calculated from connected data sources. Vibration readings, oil analysis results, inspection outcomes, failure history, and maintenance compliance feed into the scoring model automatically. The reliability engineer can query 'show me all assets with health scores below 60 and declining trajectory over the past 90 days' and get a current, reliable answer. Scores are current and findable — the health score dashboard is the reliability team's primary tool.
AI can prioritize maintenance based on health scores, predict which assets will cross critical thresholds, and correlate health score trajectories with operating conditions. Data-driven condition-based maintenance is feasible.
Formalize the health score as a machine-readable model with defined equations, weighting parameters, threshold logic, and confidence intervals — so AI can not only read the score but understand and optimize the scoring methodology.
Health scores are schema-driven with formal model definitions. The scoring model is documented as machine-readable equations with defined input weights, threshold rules, and confidence calculations. Each score carries a confidence interval reflecting sensor coverage and data freshness. An AI agent can ask 'what is Compressor 7's health score, what are the top contributing degradation factors, and what is the confidence level based on sensor coverage and data age?' and compute the complete answer from the formal model.
AI can optimize health score models themselves — adjusting weights based on actual failure outcomes, suggesting new inputs that improve prediction accuracy, and calibrating thresholds to operational risk tolerance. Full autonomous condition-based maintenance management is possible.
Implement real-time health score streaming — scores recalculate continuously from streaming sensor data and publish as events whenever they cross thresholds or change significantly.
Equipment health scores are continuous, self-calibrating, and real-time. Scores update continuously from streaming sensor data — vibration, temperature, current draw, acoustic signature, oil condition — recalculating the instant any input changes. The scoring model itself evolves: when a failure occurs that the score didn't predict, the model automatically adjusts its weights and thresholds to improve. Health scores are living, self-improving condition indices.
Fully autonomous equipment health management. AI maintains continuously accurate, self-calibrating health scores that drive condition-based maintenance decisions in real-time with zero manual assessment.
Ceiling of the CMC framework for this dimension.
Capabilities That Depend on Equipment Health Score
Other Objects in Maintenance & Reliability
Related business objects in the same function area.
Maintenance Work Order
EntityThe transactional record that authorizes and tracks a maintenance task — containing the target asset, problem description, work type (corrective, preventive, predictive), priority, assigned technician, parts consumed, labor hours, completion status, and root cause code upon closure.
Spare Parts Inventory
EntityThe managed stock of maintenance, repair, and operations (MRO) parts — including part numbers, criticality ratings, on-hand quantities, reorder points, lead times, interchangeability data, and the mapping of which parts serve which equipment assets.
Maintenance Procedure
EntityThe step-by-step instructions for performing a maintenance task on a specific asset type — including safety lockout/tagout requirements, tools needed, parts lists, torque specifications, inspection checkpoints, and expected completion time maintained by reliability engineers.
Equipment Failure History
EntityThe structured record of every equipment failure event — capturing failure date, asset identity, failure mode, root cause classification, affected components, time to repair, production impact, and the corrective action taken, linked to the associated work order and inspection findings.
Lubrication Schedule and Specification
EntityThe managed program defining lubrication requirements for each asset — specifying lubricant types, application points, quantities, frequencies, condition monitoring thresholds (viscosity, contamination), and the route maps that lubrication technicians follow on their rounds.
Repair-versus-Replace Decision
DecisionThe recurring judgment point where maintenance and engineering evaluate whether to repair a degraded asset or replace it — weighing remaining useful life estimates, cumulative repair costs, replacement lead time, production impact, and capital budget availability against defined thresholds.
Maintenance Priority Decision
DecisionThe recurring judgment point where maintenance planners determine which work orders to execute first given constrained labor, parts, and production windows — applying criteria such as asset criticality, safety risk, production impact, regulatory deadline, and health score degradation rate.
Preventive Maintenance Schedule Rule
RuleThe codified logic that determines when preventive maintenance tasks are triggered for each asset class — including time-based intervals, usage-based thresholds (run hours, cycle counts), condition-based triggers, and the escalation rules when PMs are deferred beyond acceptable windows.
Failure Mode Classification Rule
RuleThe taxonomy and classification logic that standardizes how equipment failures are categorized — defining failure mode codes, cause codes, effect codes, and the hierarchical structure (asset class → component → failure mode → root cause) that ensures consistent coding across technicians and shifts.
Work Order Lifecycle Process
ProcessThe end-to-end maintenance workflow from work request initiation through planning, scheduling, execution, quality check, and closure — defining approval gates, parts staging requirements, permit-to-work handoffs, technician sign-off steps, and the feedback loop that updates failure history and health scores upon completion.
What Can Your Organization Deploy?
Enter your context profile or request an assessment to see which capabilities your infrastructure supports.