Access gap
“
The conveyors at our hot strip mill are a critical part of the production process, but it’s virtually impossible to use manual or vibration-based techniques to assess their condition.
The Monitoring Blind Spot
The hardest assets to monitor are often the ones most painful to lose.
Most condition monitoring depends on getting close to the machine. That works for accessible bearings, motors, and gearboxes. It breaks down when critical motor-driven assets are submerged, enclosed, hazardous, remote, or too numerous to instrument one by one. The result is a structural blind spot: assets that matter, but do not produce enough continuous condition data. SAM4 closes that gap from the motor control cabinet. It reads current and voltage, detects developing fault patterns, validates customer-facing findings, and delivers maintenance actions your team can use before failure.
Monitoring coverage often tracks access. Risk does not.
Easy to reach
Hard to reach
High risk
Where vibration earns its keep.
Accessible critical assets get deep diagnostic coverage. Direct measurement, frequency-rich data, mature tools, trained analysts.
The blind spot.
High-consequence assets where mounting and maintaining sensors is unsafe, expensive, intermittent, or impractical. Coverage is sparse precisely where consequence is high.
Low risk
Often over-monitored.
Easy access invites instrumentation. Useful, but not always where the next consequential outage is hiding.
Reasonably ignored.
Low-consequence assets where access is difficult and the risk-adjusted value of monitoring is low.
The problem is not that these assets are unimportant. The problem is that the monitoring model does not scale to them.
Condition monitoring was built around proximity. That is the constraint.
Vibration, ultrasound, thermography, visual inspection, and oil analysis all depend on getting close to the machine. That works for accessible assets. It breaks down when assets are submerged, sealed, hazardous, remote, or distributed across hundreds of motors. These are not edge cases. They are the structural blind spot left by proximity-based monitoring. For years, the result was mainly a reliability problem: faults developed where teams had little or no continuous condition data.
That reliability blind spot is becoming an asset-level data gap.
Maintenance workflows, energy programmes, operational dashboards, digital twins, and AI optimisation systems all need trusted asset-level evidence. But many motor-driven assets still do not produce continuous data on condition, load, runtime, energy use, or operating behaviour. The CMMS knows the work orders. SCADA knows the process. Energy systems know meter-level consumption. Digital twins know the model. What is often missing is the asset's own operating evidence. Without it, twins run on assumptions. Energy programmes rely on estimates. Predictive maintenance defaults to schedules. AI models learn from inferred state, not observed state. SAM4 closes that gap from the motor control cabinet: current and voltage become a continuous asset data layer for motor-driven equipment.
Three forces pushing the old model past its limits.
Proximity-based monitoring has worked for decades on assets teams could reach. The assets outside that model are becoming harder to ignore. Three pressures are pushing the old model past its limits: people, risk, and data.
01
Human capacity is shrinking
Industrial diagnostic judgement still depends heavily on experienced engineers. Many are retiring faster than organisations can replace them.
That judgement has to move into systems: validated field data, repeatable diagnostics, and clear recommendations that help less experienced teams act with confidence.
02
Operational and energy exposure is rising
Under-monitored assets now carry more consequence. Downtime, safety risk, pollution events, energy waste, carbon reporting, and insurance pressure make periodic snapshots and reactive maintenance harder to defend.
Motor-driven systems are also a large part of industrial electricity use. Without asset-level data on load, runtime, energy use, and abnormal operation, teams struggle to identify which assets create waste and which interventions are worth prioritising.
03
Data demand is increasing
CMMS automation, digital twins, energy platforms, carbon reporting, and AI optimisation systems all need trusted asset-level evidence.
Models and estimates help, but they cannot replace continuous operating data from real assets: condition, load, runtime, energy use, and abnormal behaviour. Without that data, digital systems become better interfaces for incomplete assumptions.
The cabinet becomes the asset data point.
The motor control cabinet already carries the electrical signal that powers the asset. SAM4 turns that signal into continuous asset evidence. Some issues appear directly in current and voltage. Others appear because faults or process changes alter torque, speed, load, or electromagnetic balance. That gives SAM4 a way to monitor hard-to-reach motor-driven assets without mounting sensors on the machine. It also closes the deeper data gap: condition, load, runtime, energy use, and operating behaviour can become available to the systems that need them.
Signal
High-resolution current and voltage captured at the motor control cabinet.
Models
Physics-based signatures and asset-specific baselines separate developing faults from normal operating variation.
Validation
Reliability engineers review ambiguous, urgent, and edge-case findings before notification.
Workflow
Validated findings and asset-level metrics route into CMMS, dashboards, APIs, reports, or team notifications.
See how ESA works →·Explore what SAM4 monitors →·View the installation model →
One signal. Three levels of decision support.
The same cabinet signal can support more than fault detection. Current and voltage can help teams understand condition, operating behaviour, and energy performance. The boundary matters. These outputs are not equally direct. Condition findings are closest to maintenance action. Operating performance and energy insights create decision space, but customer process context determines the right intervention.
01 — Direct action
Condition findings are the most directly actionable. SAM4 detects developing fault patterns, validates customer-facing findings, and sends maintenance teams the likely fault type, severity, supporting evidence, and recommended next action.
Best for: inspections, planned maintenance, fault escalation, CMMS work orders.
02 — Operating evidence
Performance insights show how assets run: runtime, starts, stops, load, speed, duty cycle, abnormal operating patterns, and supported pump-performance estimates.
The data shows the operating pattern. Your team applies process context to decide whether to change sequencing, control logic, duty allocation, or maintenance planning.
Best for: operational review, control improvements, duty-cycle analysis, digital twins, and asset-performance models.
03 — Decision support
Energy insights show consumption, load, losses, operating efficiency, and saving potential. SAM4 can rank opportunities and show where waste is likely occurring.
It does not replace process engineering. Setpoint changes, control strategy, motor rightsizing, retrofit choices, and capital decisions remain customer decisions.
Best for: energy reviews, ISO 50001 programmes, sustainability reporting, asset replacement planning, and energy-saving project prioritisation.
Together, these outputs turn hard-to-reach motors into an asset-level data source: condition actions for maintenance, operating evidence for reliability and operations, and energy decision support for engineering and sustainability teams.
Where SAM4 fits, and where it stops.
SAM4 is not a process optimiser, SCADA replacement, vibration replacement, or generic dashboard layer. It produces trusted asset-level evidence: validated fault findings, condition trends, energy use, runtime, load, operating behaviour, and performance indicators. Your CMMS, data platform, digital twin, energy programme, process team, or reliability team uses that evidence to decide what happens next. ESA and vibration are complementary. Vibration remains strong for accessible, high-criticality machines where sensors can sit close to the fault path. ESA adds coverage where access, signal transmission, or deployment economics make vibration hard to scale: submerged assets, ATEX-zone motors, enclosed machines, remote stations, distributed motor fleets, electrical issues, and process or load-path changes. The principle: SAM4 sends better asset data into the systems and teams that already make industrial decisions. It does not replace those systems or teams.
Why fleet-scale ESA is hard to build.
Electrical Signature Analysis has existed for decades. But fleet-scale ESA is not a formula library. It requires a full production stack: reliable measurement, physical interpretation, asset context, field validation, and workflow delivery.
01
Measurement
Capture usable signals in real cabinets.
Industrial cabinets are noisy. VFDs, variable loads, power-quality effects, switching harmonics, and incomplete asset data all affect the signal. Production ESA needs acquisition hardware, edge sampling, and signal conditioning built for real cabinets, not lab waveforms.
02
Interpretation
Turn signal patterns into fault evidence.
Fault evidence appears through different paths: direct electrical effects, electromechanical response, and process or load-path changes. Signatures depend on motor type, drive setup, speed, load, asset configuration, and failure mode. Engineers need to understand why a fault was flagged, not just see a model score.
03
Fit and context
Apply it to this asset and operating regime.
ESA strength varies by asset type, operating regime, drive configuration, load stability, and target failure mode. Production ESA needs asset-fit scoping, nameplate data, configuration context, and asset-specific baselines before confidence can be judged.
04
Field validation
Prove and improve from real outcomes.
Models need confirmed field outcomes, not lab-only examples. SAM4 learns from confirmed faults, false alerts, missed findings, inspection feedback, resolved work orders, and operating context.
05
Workflow delivery
Convert findings into actions.
A spectrum is not a work order. Customers need fault type, severity, evidence, confidence, recommended action, and delivery into the systems they already use.
Field learning compounds
Production ESA improves when real outcomes close the loop. New asset types extend the signature library. Confirmed faults refine models. False alerts tighten validation. Missed findings expose blind spots. Resolved work orders connect signal evidence to physical reality.
The advantage is not only the algorithm. It is the combination of measurement infrastructure, physics, asset context, field-labelled outcomes, and operational feedback.
The monitoring gap, in customers' words.
Three customers. Three views of the same problem: assets that are critical, hard to monitor, and easier to manage once the signal becomes visible.
Visibility gap
“
We invested heavily in condition-based maintenance — vibration, oil analysis, airflow, the lot. But those approaches can still be snapshots. SAM4 added continuous visibility.
Planning value
“
SAM4 brought to my attention issues with one of our submersible pumps located in a busy office carpark. This insight helped me to plan crews, issue notifications and permits, as well as organize a 25-ton crane.
Field evidence, not lab claims.
SAM4 performance is measured on resolved customer-facing incidents from live industrial deployments. Performance varies by asset type, operating regime, fault mechanism, and signal path. That is why headline metrics should be read with the methodology, not as universal guarantees for every asset and every failure mode.
Recall
95.5%
Share of confirmed real conditions detected by SAM4.
False-alert rate
2.1%
Post-review false-alert rate.
Scored incidents
1,467
Scored customer incidents in the reporting set.
Assets monitored
7,000+
Live industrial assets across 80+ customers and five continents.
Methodology
Recall = fault detected / (fault detected + fault missed). False-alert rate = false communication / (fault detected + false communication).
Reporting window: 1 May 2025 to 1 May 2026. Source: 2,087 customer incidents from live industrial deployments. Inconclusive, not applicable, and unresolved outcomes are excluded from the metric denominators.
Metrics are reviewed quarterly. Performance is published by asset type and failure mode where the evidence base supports it.
Find the assets your monitoring programme cannot reach.
In 30 minutes, we review your asset types, access constraints, failure history, and maintenance priorities to identify where cabinet-based monitoring is likely to fit. Engineer-to-engineer. No SDR layer.