Maintainability in maintenance: how to design better repairs

Maintainability is the ability of an asset to be restored safely, correctly and with controlled resources when maintenance is required. It influences repair time but extends beyond MTTR and beyond the maintenance team. Physical access, modularity, identification, test points, tooling, documentation, spares and design decisions determine how much effort an organization must mobilize to return a function to service.

A mature operation treats maintainability as an engineering attribute. The goal is to remove structural difficulty from the job, separate active repair from logistics delay and turn field evidence into a verifiable improvement. The complete workflow runs from an observed intervention to a change in the plan, work order and SAP PM history.

What maintainability measures

The central question is: under declared conditions, how predictable and executable is restoration of the asset function? The answer requires a boundary. The clock may cover diagnosis, preparation, isolation, removal, correction, assembly, testing and technical release. Waiting for material, authorization, an operating window or external support can be recorded separately. Combining everything in one average prevents the team from locating whether difficulty comes from design, process or support.

The RAM model separates the dimensions. Reliability addresses continuity of function and failure occurrence. Availability combines failure and restoration behavior within an operating definition. Maintainability addresses restoration, including physical characteristics, information and support means. Improving one dimension does not automatically improve the others.

MTTR summarizes a repair population, but it does not explain difficulty. Two pumps may have the same MTTR and require very different work. One may have simple failures followed by long material delays. The other may have every part available but poor access, ambiguous diagnosis and assembly prone to rework. Maintainability analysis opens those portions before comparing the number.

Assumptions to declare first

• Function and performance standard: state what the asset must deliver and in which context.
• Reference event: define which interventions enter the population and how recurrence, planned work and improvement are treated.
• Time milestones: record failure recognition, diagnosis start, active repair start, technical completion and operational return.
• Support condition: document team, shift, access, tools, parts, instructions and permits available.
• Quality criterion: repair is complete only after function, assembly, safety and evidence are verified.

Industrial equipment inspection provides field observations that help test these assumptions. These assumptions prevent convenient conclusions. A short repair followed by the same failure two days later does not demonstrate good maintainability. A long intervention caused by an external wait does not prove poor equipment design. History must distinguish execution, delay and recurrence.

An eight-step method

1. Select an intervention family. Compare similar jobs in the same context rather than every work order in the plant.
2. Reconcile time. Separate diagnosis, preparation, repair, testing and waiting with consistent rules.
3. Observe the work. Record movement, tool changes, access removals, ergonomic exposure, questions and returns.
4. Locate difficulty. Classify barriers in access, identification, diagnosis, handling, assembly, adjustment and verification.
5. Test the cause. Confirm whether difficulty lies in design, information, spares, procedure or coordination.
6. Develop alternatives. Consider a physical change, test point, modular set, kit, instruction, fixture or sequence change.
7. Assess risk and change. Engineering, operations and safety approve the solution through plant governance.
8. Verify in the field. Compare a new population with the baseline using the same definition.

Complete worked case: pump P-204

Every number in this case is didactic. It is not a customer result, benchmark or PM Run performance claim. Centrifugal pump P-204 transfers product to a packaging line. The selected event was corrective work on the seal, coupling and bearing assembly because this family concentrated downtime and showed high duration variability.

Function, population and window

Field	Case definition
Function	Transfer 42 m³/h within the process pressure range
Baseline window	July 1, 2025 through June 30, 2026
Population	12 comparable corrective interventions
Total active time	46.8 hours
Waiting recorded separately	19.5 hours for material, release and support
Didactic active MTTR	3.9 hours per intervention
Didactic 90th percentile	6.1 hours
Recurrence within 30 days	3 of 12 interventions

The 3.9-hour average concealed variation and recurrence. The team selected six work orders with complete evidence, interviewed technicians and observed a planned intervention that reproduced the same removal sequence. Individual speed was not a criterion. The review focused on sequence and barriers.

Observed evidence

• The coupling guard required removal of eight hard-to-reach fasteners and rested on the floor during the job.
• A pipe spool prevented frontal extraction and forced additional removal.
• No prepared point existed for the instrument used during initial diagnosis.
• Visually similar bolts had different specifications and were sorted only after disassembly.
• The assembly exceeded comfortable handling and lacked an approved dedicated lifting point.
• The instruction showed removal but omitted tightening sequence, alignment acceptance and return test.
• Two recurrences were associated with incomplete assembly and verification rather than a new failure mechanism.

Maintainability diagnosis

The team used an internal 1 to 5 scale only to compare alternatives. It was not presented as a universal standard. Access scored 2, diagnosis 2, handling 1, assembly 2 and verification 2. The dominant cause was not lack of skill. The design required access removals, did not support controlled handling and left critical criteria outside the instruction.

Three alternatives were compared. The first addressed only the instruction. The second combined a hinged guard, an approved removable spool, a lifting point and a test point. The third replaced the entire assembly with a different module. Engineering chose the second, conditional on management of change, safety review, lifting-point calculation and hygiene compatibility.

Approved decision package

• Hinged guard with captive fastening while preserving the protective function.
• Removable spool with a defined interface and approved assembly procedure.
• Engineering-calculated and identified lifting point.
• Test point that enables repeat measurement under the same condition.
• Spare kit by family without mixing specifications.
• Illustrated job plan with torque, alignment, inspection and functional test.
• Completion criterion requiring test evidence and condition confirmation.

Path into SAP PM

The analysis does not end in a presentation. In the worked case, the approved modification creates an improvement order with operations, materials, owners and technical attachments. The corrective task list receives the new sequence. The equipment record retains the revision reference. Future notifications use a coherent catalog for symptom, failure mode and cause without replacing evidence with automatic selection.

SAP official documentation explains that maintenance notifications can record a technical object, malfunction, causes, tasks and activities, and that relevant information can be assigned to an order. The exact design depends on company configuration. The governing principle is traceability from abnormal condition through decision and executed work into history.

PM Run operates as a mobility, planning and execution layer over SAP PM. A technician can receive and execute the order and return confirmations and permitted evidence according to the configured process. Maintenance planning can organize capacity and scheduling. SAP remains the system of record. The platform does not define strategy, diagnose maintainability or replace engineering approval.

Indicators and verification window

Verification was set at 90 days or six comparable interventions, whichever occurred later. The minimum population prevents a conclusion from one order. The team retained the baseline and compared:

• median and 90th percentile of active restoration time;
• diagnosis, removal, assembly and test time as separate portions;
• 30-day recurrence for the same family;
• percentage of orders closed with functional-test evidence;
• safety and ergonomic deviations, which cannot worsen to reduce time;
• waiting hours, tracked separately so logistics are not assigned to design.

For the exercise, the internal criterion was to move the 90th percentile from 6.1 to no more than 4.2 hours, raise test recording from 42% to at least 95%, and have no assembly recurrence in six interventions. These are didactic case targets, not market references. A changed population or context requires a new baseline or stratification.

Why improvements fail

Standardizing a procedure does not correct impossible access. Buying a tool does not correct ambiguous identification. Creating a kit does not fix incorrect master data. Modularizing an assembly can reduce repair time and introduce another failure interface. Every alternative must be checked against function, safety, quality, cleanliness, energy, environment and available competence.

Another error is to chase only the average. Extreme durations carry risk and learning. Median shows the center, a high percentile shows residual difficulty, and recurrence checks repair quality. The MTTR and MTBF guide explains the calculations. Maintainability analysis explains why restoration follows that distribution.

Limits

• One intervention family does not represent every asset.
• Low event counts require qualitative review and a longer window.
• Observation changes behavior and must be combined with history.
• Maintenance time never replaces safety, quality or reliability.
• Physical modifications require engineering governance and change control.
• Missing data must remain a stated limitation rather than being inferred.

Specify maintainability for new assets

New projects allow structural difficulty to be addressed before it appears in a corrective order. A requirement should be verifiable during design review, installation, and commissioning. Wording such as "easy to maintain" provides no acceptance criterion. A specification can define removal envelopes, mass and center of gravity, lifting provisions, fastener access, identification, test connections, lighting, drainage, isolation, and replacement time under a declared support condition.

A maintenance demonstration strengthens acceptance. The team selects critical tasks, prepares the expected tools and documentation, performs the sequence under a safe condition, and records obstacles. For P-204, a demonstration before release would have exposed pipe interference, the missing lifting point, and the absent alignment criterion. Acceptance could then require correction before the asset entered the permanent maintenance plan.

Acceptance checks for the P-204 package

• remove the guard without placing components on the floor or using an unlisted special tool;
• handle the assembly with an approved device and defined gripping area;
• identify fasteners and spares without relying on technician memory;
• measure alignment at the specified point and condition;
• complete the functional test with result and responsible person recorded;
• repeat the demonstration after any change that alters access or sequence.

These checks do not impose a universal time. They convert physical and informational characteristics into observable acceptance conditions. Time can then be monitored as an outcome together with safety, quality, and variability.

Stratify results before attributing performance

Shift, experience, failure type, asset configuration, and support availability belong in the analysis. A sample involving three recently qualified technicians cannot be compared directly with a specialist-only sample without stratification. Work during a major outage also differs from intervention while a production line remains active.

The teaching case added context fields to the work order and reviewed the sample each month. A reduction visible with only one technician would not demonstrate a stable package. The result would be accepted when different trained professionals could follow the same sequence, return equivalent evidence, and meet the safety and functional-release criteria.

Technical references

• NASA Reliability-Centered Maintenance Guide, including first principles of maintainability design, accessed August 23, 2026.
• NASA, Reliability and Maintainability, accessed August 23, 2026.
• SAP Help Portal, Process Maintenance Notification, accessed August 23, 2026.

Frequently asked questions

Is maintainability the same as MTTR?

No. Maintainability is a characteristic of the asset and work system. MTTR is an indicator calculated over a population and a timing rule. Maintainability affects MTTR, but logistics, priority, operations and record quality also affect observed time.

Who owns maintainability?

Ownership is shared. Design defines access and modularity. Maintenance engineering structures method and evidence. Planning prepares resources and instructions. Operations release the condition. Execution returns actual difficulty. Safety and quality validate limits. One owner coordinates the improvement without replacing those interfaces.

How can a plant start without a large project?

Select a recurring family, observe one intervention, reconcile six to twelve orders and remove one proven barrier. Record the baseline, decision and verification window. A small verified improvement is more useful than a generic list of best practices.

To connect planning, field execution and feedback to SAP PM, learn about the PM Run operational layer. Maintainability decisions, asset changes and technical criteria remain the responsibility of the operation.