Types of maintenance: corrective, preventive, predictive and proactive

In one line

There is no such thing as "the best type of maintenance": there is the right mix of corrective, preventive, predictive and proactive, depending on how much it costs the plant when each piece of equipment fails.

What each one is

Corrective: you intervene after the equipment has already failed. It can be unplanned corrective (the failure stops the line and you have to react right away) or planned corrective (a symptom was detected, but the repair is scheduled for the next available window instead of stopping on the spot).

Preventive: you intervene BEFORE the failure, based on a fixed interval — hours of use, cycles, calendar time — defined by the manufacturer or by the plant's own experience. It does not matter whether the component still has service life left: it is replaced or inspected anyway, because the interval has been reached.

Predictive: you intervene when a measurement (vibration, temperature, oil analysis, electrical consumption) shows that the equipment is degrading, not when a calendar is met. It lets you intervene closer to the component's real end of life, instead of replacing it "just in case".

Proactive (or precision): instead of accepting the failure rate as a fixed fact, it attacks the root cause that generates it — alignment, balancing, correct lubrication, installation quality — so that the equipment simply fails less often, shifting the life curve to the right.

How to choose which one to use

The decision is not philosophical, it is economic: it depends on how expensive it is for that particular piece of equipment to fail without warning (line stoppage, safety risk, emergency repair cost) versus how much it costs to monitor it or intervene early. That is why the usual practice is to classify the equipment by criticality first, and only then define the strategy:

• Equipment with very low impact if it fails (there is redundancy, or it stops production for 10 minutes): planned corrective is enough; investing in predictive there is spending too much.
• Equipment with medium impact and a known failure pattern related to use/time: preventive, based on the interval dictated by the manufacturer or by the plant's own historical failure curve.
• Critical equipment, expensive to repair, or whose failure mode can be measured in advance (vibration in bearings, temperature in electrical panels): predictive.
• Recurring failures whose root cause can be identified (poor alignment, incorrect lubrication): proactive, rather than continuing to "treat" the symptom with preventive or predictive indefinitely.

Real example

PDVSA (Petróleos de Venezuela), in its Exploration and Production division in the Western zone, uses a standardized survey model to classify all the systems of a plant by criticality before deciding which strategy to maintain them with. The working team — people from operations, maintenance and the technical specialties involved — scores each system on six criteria: failure frequency, operational impact (percentage of production affected), average time to repair, repair cost, safety impact and environmental impact.

With those scores, a formula that combines failure frequency times consequence calculates a criticality value for each system. Sorting those values from highest to lowest in a bar chart reveals three clearly differentiated zones: high, medium and low criticality. Only once that classification is done does the plant define which mix of corrective, preventive, predictive or proactive corresponds to each zone — high-criticality systems concentrate the predictive and proactive effort, while low-criticality ones are covered with planned corrective, just as the previous section proposes.

The value of the method is not in the formula itself, but in the fact that it replaces a discussion of opinions ("this equipment is important because...") with a list that is ordered and agreed upon by all areas, and it is that list which ends up allocating the maintenance budget where it has the most impact.

Benefits and limits of each one

• Corrective: zero monitoring cost, but you do not choose the moment of the failure — it is the worst place to be for critical equipment.
• Preventive: predictable and easy to plan, but it wastes the service life of components that were still working fine, and it does not prevent failures that do not follow a time/use pattern.
• Predictive: makes better use of the component's real service life, but it requires instrumentation, training to interpret the data, and there is not always a measurable variable for the failure mode in question.
• Proactive: it is the only one that reduces the failure rate instead of merely managing it, but it requires root cause analysis — it is not a calendar task, it is an improvement project.

In summary

Most real plants combine all four types, distributed according to the criticality of each piece of equipment — they do not choose "the best type of maintenance" in the abstract. The starting point is always the same: classify the assets by failure impact before deciding which strategy covers them.