Maintenance Task Analysis · Chapter 5

Reading the Results

How to read the outputs and what they let you decide.

The deliverable is a set of records, one per task, and almost nobody reads them as a document. They are read as inputs by six other activities, each taking a different column.

1. The task record itself

What one task carries. The identity and effort fields are what get filled in; the resource and condition fields are what the rest of the programme is built from.
What one task carries. The identity and effort fields are what get filled in; the resource and condition fields are what the rest of the programme is built from.

Identity, effort, resources and conditions, structured rather than written as prose. The structure is what allows a provisioning demand, a support equipment list and a manpower estimate to be derived rather than compiled, and it is why the data standards specify tables and named elements instead of a report format.

2. Two times and a frequency, per task

OutputConsumer
Elapsed timeAvailability, turnaround, maintenance planning
Man-hours, by trade and skillManpower establishment, shift planning, support contracts
FrequencyEverything that needs a workload rather than a duration

Reported together, always. The three are the analysis's signature, and any two of them without the third support a decision somebody will get wrong.

3. A workload, at system and fleet level

annual man-hours by task, by trade, by level, and MMH per operating hour for the system. On the worked example this comes out at 238.5 man-hours a year and 0.0110 MMH per flight hour, of which 69 per cent is scheduled work, which is the sort of split that changes what a programme argues about.

4. Resource requirements, quantified and located

OutputForm
Support and test equipmentA list with quantities per site, and the tasks that justify each item
Special toolsThe same, and usually the items nobody else would have asked for
Spares and consumablesDemand per task, which becomes demand per year through the frequency
FacilitiesBays, benches, power, lifting and clean areas, each traced to its tasks
SkillsTrades and levels, including any that do not exist yet and have to be created

Each of these is a justification as much as a list. The data standards ask for facilities and new skills to be recorded as consequences of the task analysis, which means the answer to "why are we buying this" is a task identifier rather than an opinion.

5. Inputs to the analyses downstream

ConsumerWhat it takes
Maintainability predictionTask times, weighted by frequency, to produce MTTR and its percentiles
Level of repair analysisTask times and resources at each candidate level, and the fixed costs each level implies
Availability modellingElapsed times, and the downtime the fleet actually sees
Technical publicationsThe validated step sequence, with cautions and prerequisites
TrainingThe tasks, the skills, and the hours each trade will spend
ProvisioningDemands per task, and the range and depth those imply

What the results do not support

  • They are not a measured MTTR. Predicted times weighted by predicted rates give a predicted mean. The demonstration measures it, and the field corrects it.
  • They are not a maintenance schedule. The analysis says what a task costs; when tasks are grouped into packages, and how they share access, is maintenance planning and it uses these numbers as inputs.
  • They do not prove a task is worth doing. A scheduled task with a well-analysed 4.5 man-hours is still a task somebody chose. The interval and the justification belong to the preventive maintenance analysis.
  • They are not valid across configurations. A task analysed on one build standard is an estimate on another, and the difference is usually access rather than the work itself.
  • They are not true until walked. Before validation the numbers are a structured argument. After validation they are a record, and the difference between the two is the point of the event.

Want to see this on a live system model? Request a walkthrough.