Maintenance management
Knowing how to replace a contactor is the trade; knowing when, in what order, with which spare parts and at what cost is management. A plan that is not scheduled, not recorded and not measured ends up as a list of good intentions.
01Building the plan: inventory, criticality and frequencies
The three kinds of maintenance, thermography and the basic plan for an industrial switchboard are explained in Maintenance of electrical components. This page covers what comes next: how the work of a whole plant, building or workshop is organised so that the plan actually gets carried out.
To keep the function of each piece of equipment —not the equipment itself— with safety, with the availability that production needs and at the lowest total cost. The department has six functions: conserve, inspect, repair, improve, plan and record. The last two are what turn a group of technicians into a system.
Implementation stages. A preventive plan is not put together in one afternoon; it is built in order and reviewed every year:
| Stage | What is done | What comes out of it |
|---|---|---|
| 1 | Inventory and coding: survey every piece of equipment, give it a unique code (plant-area-type-number) and a legible nameplate. | List of equipment with nameplate data and location. |
| 2 | Criticality: classify each item by what happens if it fails. | Classes A, B and C. |
| 3 | Tasks and frequencies: define what is done on each item and how often, starting from the manufacturer’s manual. | Plan for each item. |
| 4 | Resources: staff by specialty, spare parts in stock, tools and calibrated instruments. | Organised tool crib and estimated labour hours. |
| 5 | Scheduling: spread the tasks across the year, coordinating shutdowns with production. | Annual calendar and weekly schedule. |
| 6 | Execution and records: issue work orders, carry them out and close them with data. | History of each item. |
| 7 | Evaluation: calculate indicators and adjust frequencies, tasks and resources. | Next year’s plan, improved. |
Criticality. Not all equipment deserves the same effort. For each item you ask what happens if it stops: whether it endangers people, whether it halts production or an essential service, how much it costs to repair, whether there is a standby unit, how long the spare part takes to arrive. The answers give three classes:
| Class | Criterion | Examples | Policy |
|---|---|---|---|
| A · critical | Its failure stops the plant or compromises safety; no standby. | Main switchboard, power transformer, fire pump, UPS of an essential service. | Strict preventive plus predictive; critical spares in stock. |
| B · important | Its failure affects production but there is a standby or a workaround. | Motors on lines with duplicate equipment, sub-switchboards. | Calendar-based preventive and periodic checks. |
| C · general | Its failure is a nuisance but stops nothing, and repair is cheap. | General lighting, socket outlets, auxiliary fans. | Visual inspection and corrective when it fails. |
Frequencies. They are set by calendar (weekly, monthly, yearly), by running hours, by number of operations (contactors, circuit breakers) or by measured condition (when a value crosses a limit). The starting point is always the manufacturer’s recommendation; it is then corrected with four pieces of data: the environment (dust, humidity, temperature and corrosive atmospheres shorten the intervals), the real load on the equipment, its criticality and its history. If a component finds nothing in three consecutive inspections, the interval is probably too short; if something always turns up, it is too long or the cause lies elsewhere.
Inspection route. This is the ordered round that the inspector makes with a checklist: it is built so that nobody walks more than necessary, no point is skipped and every point states what to look at and which value is acceptable. A weekly round of a plant’s electrical room might be:
| Order | Check point | What is looked at | Acceptable |
|---|---|---|---|
| 1 | Room access and signage | Free of objects, door, signs, fire extinguisher with seal and valid charge. | No obstacles; extinguisher in date. |
| 2 | Transformer | Oil temperature, level, leaks, noise, breather silica gel. | No leaks; noise unchanged. |
| 3 | Main switchboard | Voltage and current meters, indicator lamps, smell of burning, enclosure temperature. | Stable, usual readings; no smell. |
| 4 | Capacitor bank | Fuses, bulging of the units, current per stage. | Units not deformed; currents balanced. |
| 5 | Sub-switchboards | Closed, labels, accumulated trips, heat felt when bringing a hand close. | Closed and labelled; no abnormal heat. |
| 6 | Emergency lighting and earthing | Test of the self-contained unit, condition of the visible earth connections. | Lights up; connections tight and free of rust. |
02The work order and follow-up
The work order (WO) is the document that turns a need into a task with an owner, a date, resources and a record. It is born from an inspection, from an operator’s call, from a condition alarm or from the plan’s calendar; and it always travels the same path.
A useful WO is short but complete. This is a model, with a preventive order already filled in and closed:
| Field | Filled-in example |
|---|---|
| No. and issue date | WO 0147 · issued on 04/03 |
| Equipment | P1-BOM-03 · discharge pump motor (15 kW, 380 V) · class A |
| Origin and type | Annual plan · preventive · normal priority |
| Tasks | Re-tighten terminals with a torque wrench, measure winding insulation, measure current per phase, grease the bearings, thermography under load. |
| Safety | Lock out breaker Q3 with padlock and tag; work permit No. 52; insulating gloves and face shield for the thermography. |
| Planned resources | 2 technicians · 3 h · torque wrench, insulation tester, clamp meter, thermal camera · 1 grease cartridge. |
| Planned date and execution | Planned for 12/03 · carried out on 12/03, from 8:00 to 10:40. |
| Measured values | Currents 28.1 – 28.4 – 28.2 A · insulation 640 MΩ · maximum ΔT at terminals 6 °C. |
| Actual hours and spares | 2 technicians × 2.7 h = 5.4 h · 0.3 grease cartridge. |
| Remarks and closing | L3 terminal with ΔT of 6 °C: check on the next round. Signed by the person who did the work and by the supervisor. |
Tracking sheet. The orders are entered in a sheet (today almost always a spreadsheet or a management program) that shows at a glance what was done, what is late and how much it cost:
| WO | Equipment | Type | Planned | Actual | Hours | Status |
|---|---|---|---|---|---|---|
| 0147 | P1-BOM-03 pump motor | Preventive | 12/03 | 12/03 | 5.4 | Closed |
| 0148 | P1-TGB-01 main switchboard | Preventive | 13/03 | 13/03 | 6.0 | Closed |
| 0151 | P1-CIN-02 conveyor belt | Corrective | — | 15/03 | 3.5 | Closed |
| 0152 | P1-CMP-01 compressor | Preventive | 20/03 | — | — | Overdue |
| 0155 | P1-TRF-01 transformer | Preventive | 27/03 | — | — | Open |
The first indicator that comes out of the sheet is plan compliance: preventive orders closed on time divided by those scheduled.
Alongside it, the proportion between preventive and corrective hours is tracked: when the plan works, corrective hours go down and preventive hours stop being the exception.
03Management indicators: MTBF, MTTR and availability
Three numbers sum up how a machine behaves over a period. All of them come from the corrective orders in the sheet: how many times it failed (n), how long it ran (Top) and how long it stood idle for repairs (Trep).
A compressor is scheduled for 2,000 h of service in the half-year. It suffers 4 failures and the repairs last 3, 5, 2 and 6 h. We calculate:
Trep = 3 + 5 + 2 + 6 = 16 h · Top = 2,000 − 16 = 1,984 h
Check by another route: . The compressor was available 99.2 % of the time.
After a year with the plan (re-tightening, insulation tests, oil analysis, filter changes), the next half-year brings only 2 failures of 4 h each: Trep = 8 h, MTBF = (2,000 − 8) / 2 = 996 h, MTTR = 4 h and A = 996 / 1,000 = 99.6 %. The MTBF almost doubled and the MTTR did not change: the plan improved reliability, not the speed of repair. Lowering the MTTR takes other measures: spares in stock, written procedures, trained staff.
99.2 % looks excellent, but it depends on when the 16 lost hours fall. If a single 16 h failure hits during harvest or over a full shift, the production damage is far greater than that of four failures spread out. That is why it is always read together with the MTBF, the MTTR and the cost of each hour of downtime. Also, scheduled stops for preventive maintenance are usually reported separately: the aim is to carry them out outside production hours.
04Costs, depreciation and service life
Maintaining costs money, and not maintaining costs more, but you have to be able to prove it with numbers. Maintenance costs are labour, spares and materials, contracted services and, the most underestimated, the cost of downtime: lost production, penalties, rework.
Hourly cost. To make decisions about a machine, what it costs for each hour it works is expressed:
Depreciation. This is the way of spreading the value of a machine over its service life, that is, the period in which it performs at a reasonable cost. With the straight-line method:
The compressor cost 18,000 USD, a residual value of 2,000 USD is estimated after 10 years and it works 2,000 h/year. Its maintenance (labour and spares) runs about 1,200 USD/year.
Mh = 1,200 / 2,000 = 0.60 USD/h, so, not counting energy, the machine costs 0.80 + 0.60 = 1.40 USD per hour worked. At 4 years its book value is 18,000 − 4 × 1,600 = 11,600 USD. Check: 16,000 / 20,000 h of life = 0.80 USD/h.
Replace or repair. When an important machine fails, the decision is not made by intuition or by looking only at the repair invoice. Many workshops use as a rule of thumb not to repair if the cost exceeds half the price of a new unit, but that rule ignores two things that matter: the energy the old machine consumes and the years of life each option buys. The right way is to compare the equivalent annual cost of the alternatives:
A 15 kW motor running at 75 % load for 4,000 h/year burns out; energy costs 0.10 USD/kWh. Teaching assumptions:
- Rewind: 1,150 USD, efficiency afterwards 90.0 %, 4 more years of life.
- New high-efficiency motor: 2,400 USD, efficiency 93.0 %, 12 years of life.
The useful power is 15 × 0.75 = 11.25 kW. The electrical power each one draws:
| Item (USD/year) | Rewind | New motor |
|---|---|---|
| Investment spread over its life: 1,150 / 4 and 2,400 / 12 | 287.5 | 200.0 |
| Energy: 12.50 × 4,000 × 0.10 and 12.097 × 4,000 × 0.10 | 5,000.0 | 4,838.7 |
| Equivalent annual cost | 5,287.5 | 5,038.7 |
The new motor costs 248.8 USD/year less (4.7 %), even though rewinding is only 47.9 % of the price of a new one. Check over 12 years: new, 2,400 + 12 × 4,838.7 = 60,464.5 USD; rewound (three times), 3 × 1,150 + 12 × 5,000 = 63,450 USD; difference 2,985.5 USD, or 248.8 USD/year. And downtime has not been counted yet: rewinding takes days; swapping, if there is stock, hours. With a high cost of downtime the conclusion is reinforced.
Service life and the bathtub curve. The failure rate of an electrical machine is not constant over its life: it has three very different stretches, and each calls for a different action.
In the middle stretch the failure rate λ is approximately constant and equals the inverse of the MTBF; with it we estimate the probability that the machine keeps working after a time t:
05From preventive to predictive: periodic checks and control charts
Preventive maintenance changes a part “just in case”; predictive maintenance measures a variable that follows the deterioration and intervenes only when the trend justifies it. For that you need to take the same measurement, at the same point and under comparable conditions, at regular intervals. This is called a periodic check.
| Technique | What it detects | Measured with |
|---|---|---|
| Vibration | Unbalance (vibration concentrated at 1× the rotation speed), misalignment (often shows up at 2×), looseness and bearing wear, in motors, pumps and fans. | Accelerometer with a data collector; vibration velocity is recorded in mm/s RMS. Severity is assessed with the ISO 20816 series. |
| Thermography | Contacts and terminals with abnormal resistance, overloads, imbalance. See the criterion based on temperature difference. | Infrared camera, with the installation under load. |
| Oil analysis | In transformers: moisture, acidity, dielectric strength and dissolved gases that reveal overheating or internal discharges. In gearboxes: wear particles, viscosity, contamination. | Sample taken following a procedure and sent to a laboratory; compared with the previous analysis. |
| Insulation | Degradation of the winding insulation from moisture, dirt or temperature. The trend of the values matters more than a single number. | Insulation tester; always measured at a similar temperature, which is recorded. |
| Current and power quality | Imbalance, growing overload, harmonics, falling power factor. | Clamp meter, power analyser (see digital instruments). |
On the electrical side of an installation, the AEA 90364 regulation also requires periodic verifications (its Part 6 deals with the verification of installations), and Res. SRT 900/2015 sets the protocol for measuring earthing and the continuity of exposed conductive parts that is required in the workplace. Those measurements are a natural part of the plan’s calendar. Motors and transformers have their own particularities in AC motors and transformers.
Control chart. This is the tool for knowing whether a measurement “has left the normal range”. With the first measurements, taken while the machine is in good condition, the centre line (the average) and the control limits at three standard deviations are calculated. Each new measurement is then plotted:
Ten vibration velocity measurements (mm/s RMS), taken at the same point: 2.1 · 2.3 · 2.0 · 2.2 · 2.4 · 2.1 · 2.2 · 2.3 · 2.0 · 2.4.
Average: 22.0 / 10 = 2.20 mm/s. Sum of squared deviations: 0.01 + 0.01 + 0.04 + 0 + 0.04 + 0.01 + 0 + 0.01 + 0.04 + 0.04 = 0.20, so σ = √(0.20 / 9) = 0.149 mm/s and 3σ = 0.447 mm/s.
UCL = 2.20 + 0.447 = 2.65 mm/s · LCL = 2.20 − 0.447 = 1.75 mm/s. (Check by another route: ∑x² = 48.6 and 48.6 − 10 × 2.20² = 0.20, the same numerator.)
The following measurements give 2.4 – 2.5 – 2.6: all inside the limits, but rising. That is the signal to shorten the interval between checks. The next one gives 2.9 mm/s, above the UCL: a scheduled corrective order is issued; a worn bearing is found and replaced. The following measurement (2.1 mm/s) is back to the average.
The most expensive instrument is of little use if the measurements are not coded, dated and comparable. Predictive maintenance rests on the same things as preventive: inventory, criticality (it is justified on class A and some class B equipment), work order and record. The only thing that changes is what triggers the intervention: no longer the calendar, but the condition.
06Safety in maintenance tasks
Most serious electrical accidents happen during maintenance tasks, not in normal use. The basic procedure —disconnect, lock out, verify absence of voltage, ground and short-circuit, and delimit the area— is the five golden rules, explained in Occupational health and safety. What follows is what management adds so that those rules are always followed and do not depend on each technician’s memory.
Lockout prevents someone from re-energising the equipment while another person is working on it. An orderly system requires:
- a written procedure for each machine listing all its energy sources —electrical, pneumatic, hydraulic, capacitors or loaded springs— and where each one is isolated;
- a personal padlock and tag with each worker’s name, date and phone number; if three people work, there are three padlocks, and a group lockout box is used for the keys of several people;
- checking that the lockout works: trying to start the machine from the local control, after verifying that there is no voltage;
- that the padlock is removed only by the person who placed it, and that before re-energising it is confirmed that everyone has left the area.
It is a document that authorises a high-risk task and sets its conditions. It does not replace the work order: it complements it. It is required when the risk goes beyond the routine: work on medium voltage, live work, work at height, in confined spaces or with fire. It contains:
- description of the task, place, date and period of validity;
- who carries it out and who authorises it, with both signatures;
- identified risks and measures: isolations and lockouts made, grounding, personal protective equipment, delimited area;
- a closing: signed when the task is finished, people and tools have been withdrawn and the lockouts lifted.
Live work is also subject to what Res. SRT 592/2004 establishes and requires trained, authorised staff.
The general legal framework is Law 19.587 on Occupational Health and Safety and its Decree 351/79: the employer must train staff, provide protective equipment and ensure the procedures. If maintenance is outsourced, the work permit and the lockout are also the moment in which who is responsible for what is coordinated with the contractor.
07In the workshop
Materials: survey sheet, tape or labels for coding, flashlight. Steps: survey ten pieces of equipment in the workshop or school (switchboards, motors, test benches), assign each a code and copy the nameplate data; rate each one by the consequences of its failure and classify it as A, B or C; for class A, propose tasks and frequencies. Acceptance criterion: all equipment with a unique code and a class justified in writing; each class A item with at least four tasks and a reasoned frequency.
Materials: a decommissioned three-phase motor with its control switchboard, torque wrench, insulation tester, clamp meter, padlocks and tags, WO and work-permit templates. Steps: (1) issue the WO with tasks, resources and risks; (2) disconnect, lock out with padlock and tag, verify absence of voltage testing the instrument before and after; (3) re-tighten terminals to the specified torque and measure the winding insulation; (4) remove the lockout and measure current per phase under load; (5) close the WO with measured values, hours and remarks, and enter it in the sheet. Acceptance criterion: WO complete and signed; insulation recorded with the approximate temperature; current imbalance between phases below 5 %; no empty field.
Materials: a laboratory motor running, infrared thermometer or contact sensor, spreadsheet. Steps: measure ten times, always at the same point of the casing and at the same time of day, the temperature with the motor at steady state; calculate the average, σ and the ±3σ limits; plot. Then cause a change (partly block the ventilation) and record three more measurements. Acceptance criterion: average and limits calculated by hand and checked with the spreadsheet; the change you caused is identified on the chart, either by leaving the limit or by an upward trend.
08Frequent mistakes
| Mistake | Consequence |
|---|---|
| Applying the same plan to all equipment | Effort is spent on equipment that does not need it and the critical items are neglected. Criticality exists to allocate resources. |
| Issuing orders without closing them with data | There is no history, no indicators can be calculated and the plan does not improve. |
| Calculating the MTBF by dividing the whole period by the failures | Downtime is included as if it were running time. The repair time must be subtracted. |
| Believing the MTBF is a guaranteed time | It is an average: with random failures, only 36.8 % of machines reach the MTBF without failing. |
| Deciding to repair by looking only at the cost of the repair | Energy and years of life are ignored: an old machine that is cheap to repair can cost more per year. |
| Taking predictive measurements under different conditions each time | The measurements cannot be compared and there is no trend. Same point, same load, same instrument. |
| Waiting to leave the limit before acting | A rising trend inside the limits is already a warning; waiting turns predictive into corrective. |
| Doing the lockout “by word of mouth” or with one padlock for the whole machine | If someone re-energises while another is working, the accident happens. One padlock per person, always. |
09Self-assessment
What is the point of classifying equipment by criticality?
To allocate maintenance resources: class A equipment, whose failure stops the plant or endangers people, gets strict preventive and predictive maintenance; class C can wait for corrective.
In one month 24 preventive orders were scheduled and 21 were closed. What is the compliance?
21 / 24 = 0.875, that is 87.5 %. The remaining 3 are recorded as overdue, with the reason.
A machine is scheduled for 3,000 h in the period and has 5 failures adding up to 30 h of repair. Calculate MTBF, MTTR and availability.
If 3,000 h is the scheduled period, Top = 3,000 − 30 = 2,970 h. MTBF = 2,970 / 5 = 594 h; MTTR = 30 / 5 = 6 h; A = 594 / (594 + 6) = 0.99, or 99 %. Check: (3,000 − 30) / 3,000 = 0.99.
A machine cost 50,000 USD, is worth 5,000 USD after 15 years and works 3,000 h/year. What is the depreciation per hour?
D = (50,000 − 5,000) / 15 = 3,000 USD/year; Dh = 3,000 / 3,000 = 1.00 USD/h. Check: 45,000 / (15 × 3,000) = 1.00.
A control chart has an average of 2.20 mm/s and σ = 0.15 mm/s. What are the limits and what should be done if a reading of 2.8 appears?
3σ = 0.45; UCL = 2.65 mm/s and LCL = 1.75 mm/s. A reading of 2.8 is outside the UCL: a scheduled corrective order is issued to look for the cause (for example a bearing or misalignment), first confirming that the measurement was taken under the same conditions.
What is the difference between a work order and a work permit?
The work order organises the task (what, when, with which resources); the work permit authorises carrying it out under risk conditions and sets the safety measures, with who authorises and who executes. A high-risk task usually needs both.
Why is one padlock per person used instead of one per machine?
Because every worker must control their own safety: the machine can only be re-energised when all the padlocks have been removed, each by the person who placed it.