Skip to content
Maintenance Management | Part 3 of 10

Choose the Maintenance Strategy per Failure Mode

Run to failure, time based, condition based, predictive or redesign: five questions to choose the right strategy for each failure mode of an A asset.

“We do preventive maintenance on this machine” says little. A machining center can fail in a dozen ways. Some give a warning weeks ahead, some are related to age, some happen without any sign. Each needs a different answer.

A maintenance strategy is therefore not chosen per asset. It is chosen per failure mode.

Five strategies

  • Run to failure: the failure is cheap, safe and fast to fix. Typical for C assets with a spare on the shelf.
  • Time based: wear is age related and the task is cheap. Filter changes, seals, inspections required by your local rules.
  • Condition based: a measurable sign appears before the failure, early enough to act. Vibration, temperature, oil condition, clamping force.
  • Predictive: enough data and a model forecast the remaining life of a component. Useful where sensor data is continuous and the failure is expensive.
  • Redesign: no task prevents the failure economically, so the design changes.

Five questions

This is the core logic of reliability centered maintenance, reduced to what a plant can apply in a two-hour session per A asset:

  1. What are the main functions of the asset, and how can each one fail?
  2. What happens when it fails: safety, output, quality, cost?
  3. Does a measurable sign appear before the failure, early enough to act? Then condition based.
  4. Is the failure age related, and is a time based task cheaper than the failure? Then time based.
  5. If neither: accept the failure or change the design.

Take the failure modes from the work orders of the last year. If the cause codes are missing, the session also shows how much the feedback on work orders needs to improve.

A worked example

The spindle of a bottleneck machining center:

Failure mode Sign before failure Strategy Task
Bearing wear Vibration rises 4 to 8 weeks ahead Condition based Reading every 2 weeks, limit 4.5 mm/s
Loss of clamping force Force drops gradually Condition based Monthly test, limit 18 kN
Coolant ingress Seal wear by operating hours Time based Seal check every 2,000 hours
Encoder cable break None Redesign New routing with larger bend radius

The three tasks cost about 17 hours per year, roughly EUR 1,000 at EUR 60 per hour. One spindle bearing failure as an emergency costs EUR 13,290. The cable redesign takes 18 hours once and removes the failure mode for good.

Limits and actions

A condition based task needs three things: a warning limit, an alarm limit and the job that follows. A vibration reading without a limit produces a growing spreadsheet and no fewer failures.

Review the strategy

Review the strategy of each A asset once a year and after every repeat failure. A failure between two tasks means that the interval, the task or the strategy is wrong. Tasks that never find anything are candidates for a longer interval or for deletion.

Simple checks, cleaning and lubrication points can move to the operators as part of autonomous maintenance, the topic of the TPM module in the Production System. That frees technician hours for the tasks that need their skills.

Newsletter

One Lean idea.
Every week.