Machine Design #46: Design for Maintenance — Replacement Is Not Enough; Recovery Must Be Correct
“The part can be replaced” is a weak definition of maintainability. A technician also needs a safe machine state, a clear access path, the correct tool, enough visibility, the correct replacement identity, a way to preserve datum and settings, a restoration method, functional verification, and an updated configuration record.
Maintenance is a real operating task, not an afterthought at the end of design. A good design asks what people must do, under which conditions, with which tools, how often, and what evidence proves that the machine returned to a known-good state.
1. Start from the maintenance task, not an empty gap
List the task: inspect, clean, lubricate, adjust, replace, calibrate, restore, troubleshoot, or remove a jam. Define the trigger, machine state, energy isolation, product condition, tools, skill, steps, expected result, and handover record.
An empty space around a component does not prove access. The technician may still need to reach through a guard, hold a heavy part, see a hidden fastener, or reset a datum after replacement.
2. Prioritize by frequency, duration, and consequence
A task performed every shift may justify more design effort than a rare repair. A rare task with a serious safety or contamination consequence may also be critical. Estimate frequency, active time, total downtime, exposure, skill, spare availability, and consequence of an error.
Use the result to prioritize access, tooling, diagnostics, spare strategy, and task trials. Do not optimize only for the fastest average replacement if the worst case is unsafe or difficult to recover.
3. Safe maintenance state is a design input
Define how the machine reaches a safe maintenance state: stop category, stored energy, gravity, pressure, temperature, robot pose, trapped material, access authorization, and restart control. The safe state must be reachable, visible, and documented.
A maintenance door that opens while a mechanism can move is not maintainability. Coordinate the task with the risk assessment and safety validation. A reset after maintenance should not silently re-enable motion before the technician confirms clearance.
4. Access paths have layers
Review the path from standing position to the task, not only the component envelope. Check guard opening, reach, posture, clearance, hand movement, line of sight, removal direction, fastener access, cable bend, and escape. Check both normal service and a failed component with damaged or awkward geometry.
If a tool must pass through an opening, include the tool and hand envelope. If a cover is removed, provide a safe place to set it down and a clear sequence for refitting.
5. Anthropometry is an input, not “the average person”
Maintenance teams vary in height, reach, strength, glove use, and physical capability. Select the user population and design range deliberately. Document whether the task is one-person or two-person, whether a lift aid is required, and what limits apply.
Do not use one average body measurement to justify every access decision. Validate representative tasks with the people who will perform them.
6. Tool envelope is often forgotten
Model the wrench, driver, puller, meter, lifting aid, and gloves. Check swing, approach angle, reaction force, fastener head, cable connector, and space for the technician’s hand. A component may fit in CAD while the required tool cannot turn.
Standardize fasteners and tool sizes where useful, but never make a standard that prevents safe access or correct torque.
7. Visibility and lighting belong to the task
Provide light at the inspection point, not only general room lighting. Make marks, connectors, datum references, labels, leak traces, and wear indicators visible. Consider glare, contamination, shadow, and the technician’s viewing angle.
Remote diagnostics can point to a region, but they do not replace physical visibility when the task requires confirmation.
8. Identify the correct part before removal
Part identity should be visible at the point of use or in a controlled maintenance record. Include item number, revision, orientation, connector, adjustment range, calibration state, and compatible machine or model.
Wrong-part installation is easier when similar sensors, cylinders, cables, or recipes share a shelf. Use keying, labels, error-proofing, and a verification step rather than relying on memory.
9. Remove and replace without destroying the reference
Define how to preserve datum, shims, home position, torque, cable routing, and calibration. If the item is adjustable, provide a controlled reference and a restoration method. If removal can disturb an adjacent item, include that impact in the task.
The task is not complete when the new part is tight. It is complete when the required function is restored and verified.
10. Modularization has trade-offs
A module can reduce access time, wiring errors, and alignment effort. It can also increase mass, cost, spare inventory, and the amount of configuration restored at once. Decide what should be modular based on the task and failure boundary, not on a generic preference for modules.
11. Restoration is often longer than active repair
After replacing a part, restore recipe, position, parameter, calibration, guards, lubrication, cleanliness, and documentation. Include controlled startup, homing, dry run, sample run, quality check, and release to production.
The technician may spend five minutes replacing a sensor and thirty minutes proving that the machine is safe and correct. Design and measure the whole recovery, not only the hands-on repair.
12. Verify after maintenance
Verification can include visual inspection, torque record, I/O state, calibration, home position, leak test, safety function, dry cycle, first-piece quality, alarm clear, and production release. Define who approves the return and what evidence is retained.
Do not let “it works” replace a known acceptance check. Different tasks need different evidence.
13. Diagnostics reduce search time; they do not replace access
Good diagnostics identify the affected function, state, timestamp, recipe, last transition, and likely next check. They should distinguish a missing signal from a mechanical misalignment, a communication fault, or an invalid parameter.
Diagnostics cannot remove a seized fastener, contaminated lens, hidden leak, or unsafe reach. Pair information with physical maintainability.
14. Design for inspection before failure
Provide wear indicators, inspection windows, test points, sample ports, trend data, access for measurement, and clear limits. An inspection task should be safer and shorter than a failure repair. Define what result triggers adjustment, replacement, escalation, or continued operation.
15. Cleaning and contamination
Cleaning may be frequent and may affect product quality. Define access, materials, drainage, trapped areas, compatible agents, dry time, and inspection. In clean or sensitive environments, avoid features that shed particles or retain fluid.
Cleaning instructions should state when energy is isolated, what may be removed, and how the machine is released after cleaning.
16. Lifting and handling
Heavy covers, motors, cylinders, and modules require a handling plan. Define mass, center of gravity, grip points, lift aid, path, two-person requirement, and storage. Do not force a technician to support a heavy component while reaching for hidden fasteners.
17. Maintainability metrics need a boundary
Choose a metric with a defined start and end: mean active repair time, total restoration time, inspection duration, first-time-fix rate, or task success without datum correction. Record assumptions, skill, tools, spare availability, and sample conditions.
An average number without a boundary hides waiting, search, troubleshooting, and validation. Use the metric to improve design, not to pressure technicians to skip verification.
18. Example: replacing a position sensor
Task definition
The machine is in a safe state. The technician identifies the sensor, isolates energy, removes the guard, replaces the sensor, sets the gap, restores cable routing, checks the input, homes the axis, runs a dry cycle, and records the result.
Design review
Check access, tool angle, connector, bracket datum, adjustment marks, label, spare identity, diagnostics, calibration, first-piece verification, and baseline update. Trial the task with representative gloves and lighting.
19. Validate maintainability with a task trial
Give the task to a representative technician without design-team coaching. Observe search, posture, tool use, mistakes, interruptions, safety checks, restoration, and evidence. Record actual times and questions. Fix the design or instruction, then repeat the trial.
20. Maintainability in design review
Review task list, safe state, access, anthropometry, tools, visibility, identification, datum preservation, module trade-offs, restoration, verification, diagnostics, inspection, cleaning, handling, metric boundary, training, spare, and baseline. Make maintainability an engineering gate, not a promise in the handover meeting.
21. Design-for-maintenance checklist
- [ ] Maintenance tasks, triggers, frequency, duration, and consequence are defined.
- [ ] Safe maintenance state and restart control are validated.
- [ ] Access includes guards, reach, posture, line of sight, and removal path.
- [ ] Tool and hand envelopes fit in the real task.
- [ ] Representative users and gloves were considered.
- [ ] Correct part identity, orientation, datum, and calibration are clear.
- [ ] Replacement preserves reference and has a restoration method.
- [ ] Cleaning, lifting, contamination, and storage are addressed.
- [ ] Diagnostics reduce search without replacing physical checks.
- [ ] Post-maintenance verification and release evidence are defined.
- [ ] Task trial measured real recovery, not only active repair.
- [ ] Metrics, training, spare data, and configuration baseline are updated.
Conclusion
Maintainability means returning the machine to a known-good, safe, and verified state with a task that real people can perform. Design the complete path from safe state to access, replacement, adjustment, restoration, verification, handover, and baseline update.
When MINATA treats maintenance as a design input, “replaceable” becomes a measurable capability instead of a claim made only after the machine is already difficult to service.
References
- ISO 12100:2010 — Risk assessment and risk reduction: https://www.iso.org/standard/51528.html
- ISO 13849-1:2023 — Safety-related control systems: https://www.iso.org/standard/73481.html
View all MINATA technical articles