Condition-Based or Calendar-Based Electrical Maintenance? A Factory Scope Decision Guide

Executive answer: an FDI factory should not choose between calendar-based and condition-based electrical maintenance as if one replaces the other. Use a hybrid programme. Keep time- or usage-based work where the OEM, approved maintenance plan, safety case, insurer, authority or test method requires it. Use measured condition to prioritize inspections, plan outages and justify earlier or later intervention only when the signal, baseline, alarm logic and decision authority are controlled.

This guide gives an owner-side method for comparing maintenance proposals for 22–35 kV switchgear, transformers, LV assemblies, protection systems, cables and critical motors in Vietnam. The examples are hypothetical. They do not prescribe universal intervals, alarm limits or legal duties. Final tasks and limits must come from the actual equipment instructions, approved design, risk assessment, applicable Vietnamese requirements and competent engineering review.

1. Define the two strategies precisely

Calendar-based maintenance triggers work after a stated time or usage interval: for example, months in service, operating hours or breaker operations. It is repeatable and easy to budget, but it can open healthy equipment unnecessarily or miss a rapidly developing defect between intervals.

Condition-based maintenance uses evidence such as temperature, insulation condition, oil results, operating time, vibration, discharge activity, mechanism signatures or protection diagnostics to trigger a decision. It can focus resources, but it is only as reliable as the sensors, measurement method, baseline, data quality and response process.

Predictive maintenance estimates future condition or failure risk from trends or models. A dashboard forecast is not proof. The owner should know the model inputs, confidence, failure modes covered, false-alarm treatment and action that follows each output.

2. Start with failure modes, not available instruments

Build an asset–function–failure-mode register before buying monitoring. For each asset, record:

• the function and loads affected;
• credible failure modes and warning time;
• safety, production, quality and environmental consequences;
• redundancy and restoration path;
• detectable condition indicators;
• required outage, tools, spares and competent persons;
• the approved maintenance source: OEM instruction, project specification, site procedure or applicable requirement.

A thermal camera cannot detect every internal insulation defect. Dissolved-gas analysis does not cover a dry-type transformer. A breaker operation counter does not prove contact condition. Match the method to the failure mode and state what remains undetected.

3. Use a hybrid decision matrix

Asset or function	Calendar/usage baseline	Condition evidence	Owner decision
MV switchgear	OEM inspections, mechanism service, interlock and protection tests	Operation count, timing, trip/close-coil signature, contact resistance where applicable, gas or pressure indication, thermal and discharge observations	Keep mandatory tasks; use trends to prioritize panels and outage scope
Oil-filled transformer	Routine inspection, cooling-system checks, protection tests and prescribed oil work	Oil quality, dissolved gases where applicable, temperature/load history, leaks, bushings and frequency-response evidence after defined events	Investigate abnormal trends before changing interval
LV assembly	Cleaning, connection checks where approved, protective-device testing and functional checks	Thermography under recorded load, trip-unit diagnostics, breaker operations and environmental condition	Target hotspots and deteriorating devices without omitting protective-function tests
Cable system	Route inspection, terminations and project-defined electrical tests	Thermal condition, sheath/earth evidence, partial-discharge or diagnostic tests when technically justified	Select a method suitable for cable type, voltage and failure mode
Critical motor	Lubrication and OEM service by time or running hours	Vibration, temperature, current signature, insulation evidence and starts	Plan intervention from corroborated trends, not one reading
Protection/DC supply	Functional tests, battery and charger checks, settings control	Self-diagnostics, event logs, battery impedance/voltage trends and alarm history	Do not let online alarms replace end-to-end trip-path proof

4. Score consequence separately from condition

Do not let a “green” condition score hide a high-consequence single point of failure. A useful owner register keeps at least three fields separate:

• Consequence: safety, production, quality, environment and restart impact;
• Likelihood evidence: age, duty, environment, defects, history and measured condition;
• Detectability and warning time: whether the selected method can provide usable notice.

A simple screening score may multiply agreed ordinal ratings, but it is a prioritization tool, not a failure probability. For illustration, consequence 5 × likelihood 3 × poor detectability 4 gives 60. If inspection improves detectability from 4 to 2, the score becomes 30; the consequence remains 5. The team must define rating meanings and escalation rules before scoring assets.

5. Compare outage exposure in hours, not slogans

A calendar task can create planned outage exposure; a missed defect can create unplanned outage exposure. Compare both with declared assumptions.

Hypothetical example: one transformer inspection requires a 6-hour planned outage. A condition-monitoring package is expected to reduce, not eliminate, the chance of a 30-hour unplanned outage from 8% to 3% in the planning year. The screening exposure changes from 0.08 × 30 = 2.4 h/year to 0.03 × 30 = 0.9 h/year, a difference of 1.5 expected hours. This does not automatically justify the package: the assumptions need evidence, the outage consequence is not purely linear, and safety or mandatory work cannot be traded away.

Show planned production loss, temporary-supply risk, switching risk, monitoring cost, sensor maintenance, false alarms and spare strategy separately. Do not invent a monetary value without owner-approved production and recovery data.

6. Establish a trustworthy baseline

Condition data becomes useful only when it is comparable. Record asset ID, equipment state, load, ambient conditions, instrument and calibration, sensor position, method, units, operator, timestamp and raw file. Preserve the first accepted baseline and any post-maintenance baseline.

A 15 °C temperature rise cannot be compared across surveys if loading, emissivity, viewing angle or enclosure state changed. A vibration trend is weak if the sensor point moves. Oil results need traceable sampling, container, laboratory method and equipment identity. Define invalid-data rules instead of forcing every point into a trend.

7. Set action bands without inventing universal limits

Each condition indicator needs four elements:

1. the source of the limit or rule;
2. the confirmation method;
3. the authorized decision owner;
4. the response time and safe operating restriction.

Use bands such as normal, watch, investigate and remove from service only after project-specific criteria are approved. A dashboard supplier should not unilaterally define an electrical isolation decision. Conversely, the facility team should not ignore a protection or OEM limit because production is busy.

8. Write the maintenance scope as measurable work

The electrical maintenance tender-scope guide explains the asset, access and evidence boundaries. For the strategy comparison, require bidders to return:

• asset list and included failure modes;
• calendar/usage tasks with source and interval basis;
• condition measurements, locations, units and instruments;
• alarm and escalation matrix;
• planned outage, isolation and temporary-supply needs;
• data ownership, raw files, dashboards, licences and export format;
• spares, consumables, labour and specialist subcontractors;
• deliverable templates, acceptance criteria and exclusions.

9. Build a BOQ that exposes hidden dependencies

BOQ group	Measurement basis	Minimum scope definition
Asset verification	Asset / location	Identity, drawings, nameplate, condition baseline and criticality
Planned tasks	Task / asset / interval	Method, access, outage, labour, tools and record
Condition measurements	Point / channel / survey	Sensor or instrument, units, location, state and raw data
Online monitoring	System / channel	Hardware, power, communications, historian, cybersecurity and licences
Testing	Test / asset	Procedure, equipment, calibration, acceptance source and witness point
Corrective allowance	Rate schedule	Labour, mobilization, common spares and approval threshold
Reporting/CMMS	Package / asset	Raw files, trends, recommendations, work orders and closure evidence
Outage/HSE	Window / permit	Switching, LOTO, barriers, PPE, rescue and restoration coordination

10. Keep HSE controls independent of maintenance philosophy

Condition-based does not mean work may be performed energized. The approved method statement must identify operating authority, switching authority, permit-to-work boundary, lockout/tagout, test-for-dead, earthing where required, stored energy, backfeed and induced-voltage hazards, arc-flash controls, access barriers, rescue arrangements and restoration authority.

Online sensors can introduce their own hazards and failure paths: CT secondary circuits, auxiliary power, cable penetrations, communications, battery supplies and cybersecurity access. Review modifications through engineering change control. Never open an energized enclosure solely to collect a reading unless the approved risk assessment and authorized procedure explicitly permit it.

11. Use an ITP for maintenance and monitoring deployment

Stage	Hold/witness evidence	Acceptance question
Survey	Asset register, SLD, criticality and access review	Are identities and boundaries correct?
Method approval	Task procedure, instruments, calibration and HSE plan	Is the method suitable and safe?
Baseline	Raw data with operating state and measurement metadata	Can later readings be compared?
FAT for monitoring panel	I/O, alarms, time sync, fail-safe and data export	Does the supplied system match the approved design?
SAT/T&C	Sensor mapping, polarity, scaling, communications and alarm challenge	Does the installed system report the correct asset and value?
Maintenance execution	Before/after results, defects, work orders and restoration checklist	Was the task completed without losing configuration control?
Review	Trend, exceptions, deferred work and approval	Is the next action justified and owned?

The owner’s electrical ITP guide provides a broader hold-point structure. The switchboard FAT evidence guide helps distinguish factory evidence from installed-condition SAT.

12. Test the decision process, not only the sensors

Commissioning should inject or simulate normal, warning, invalid and lost-signal states. Verify timestamp, alarm route, acknowledgement, work-order creation, escalation and safe fallback. Confirm that a communications failure cannot block protective functions or create an unsafe automatic command.

Run a tabletop exercise: one transformer oil result moves to “investigate”, an MV breaker shows slower operation, and production requests a four-week deferral. The team should demonstrate who validates the evidence, who sets restrictions, who approves outage timing and who can require immediate shutdown.

13. Define monthly and annual KPIs carefully

Useful KPIs include overdue safety-critical tasks, measurement completeness, invalid-data rate, alarm-to-review time, defect-to-work-order time, repeat defects, planned versus unplanned outage hours, and percentage of recommendations closed with evidence. Avoid rewarding the contractor for a low alarm count or high number of tests alone.

Review whether the programme found actionable defects without excessive false alarms, whether intervals remain supported, and whether production or equipment changes altered criticality. The operational-readiness checklist helps connect maintenance to operating authority and configuration control.

14. Standards: use the correct scope

IEC 60300-3-10:2025 gives general guidance on maintainability and maintenance programmes, information and life-cycle application; it does not prescribe an interval for a specific transformer or breaker. IEC 60300-1:2024 frames dependability across technical, financial and organizational considerations.

IEC TR 63482:2024 addresses maintenance approaches for LV switchgear and controlgear assemblies within its voltage scope. IEC 60422:2024 gives supervision and maintenance guidance for mineral insulating oils; it explicitly does not make every condition-monitoring method part of its scope. IEC 60076-18:2012 covers frequency-response measurement when that measurement is required; it is not a universal periodic-test rule.

ISO 55001:2024 sets requirements for an asset-management system and emphasizes decision-making, data and life-cycle management. Certification to a management-system standard would not prove the condition of a specific electrical asset.

15. Owner decision checklist

• Asset, function, failure mode and consequence are defined.
• Calendar/usage tasks retain their approved source and interval basis.
• Condition methods match detectable failure modes.
• Baselines, units, instruments and invalid-data rules are controlled.
• Action bands name confirmation, authority, response and restriction.
• BOQ includes sensors, data, licences, outage, HSE and corrective workflow.
• FAT, SAT/T&C and ITP evidence are measurable.
• Raw data and configuration remain available to the owner.
• Every deferred action has risk acceptance, expiry and review.
• The programme is revalidated after material equipment or duty changes.

Frequently asked questions

Can condition monitoring replace annual maintenance?

Not automatically. It may support risk-based adjustment, but OEM instructions, approved plans, safety duties and tasks that verify protection or mechanical function may still require periodic work.

Is thermography enough for switchboards?

No. It can identify some thermal anomalies under suitable load and access conditions. It does not prove insulation, protection, mechanical operation, internal joints outside view or every failure mode.

Should old equipment always have shorter intervals?

No universal rule applies. Age is one input alongside design, duty, environment, history, condition, supportability, spares and consequence.

Who owns a condition-based maintenance decision?

The contract should name data reviewers, competent technical authority, operating authority and risk approver. A monitoring vendor should not silently become the plant’s switching or safety authority.

What should be handed over?

Asset hierarchy, baselines, raw data, methods, instruments, alarm rules, configurations, licences, trend reports, work orders, defect closure and updated maintenance plans. See the electrical handover dossier guide.

Next step

Ask bidders to price one hybrid maintenance matrix: asset, failure mode, calendar task, condition indicator, trigger source, outage, HSE control, evidence and decision owner. Quang Anh can help an FDI factory structure the electrical asset register, maintenance RFQ, BOQ, ITP and commissioning evidence for technical comparison. Final maintenance intervals, limits and operating decisions remain subject to approved equipment data, applicable requirements and competent site authorization.

[contact-form-7 id="7239967" title="Contact form 1"]