Human Factors and Usability Engineering Exercises

These exercises practise human factors and usability engineering as an engineering discipline: task demand, timing margin, error probability, alarm response, workload balance, validation evidence, risk control, ergonomic screening, training decay, and operational drift. The goal is not to make a screen look pleasant. The goal is to decide whether real users can perform safety-critical and production-critical work under realistic constraints.

Assume simplified deterministic or independent-probability inputs unless an exercise states otherwise. Real human factors work should also check user population, shift pattern, fatigue, language, lighting, noise, personal protective equipment, alarm load, training history, field workarounds, maintenance access, and the consequence of a delayed or wrong action.

How to Use These Exercises

For each usability calculation, define:

1. the task, user role, operating mode, and environment;
2. the performance criterion being tested;
3. whether the evidence comes from design analysis, observation, test, log data, or incident data;
4. the engineering action that follows if the criterion fails;
5. the residual risk after the interface, procedure, alarm, interlock, or training change.

The common mistake is treating human performance as a property of individual effort. In engineering review, repeated delay, confusion, workaround, omission, or recovery failure is evidence about the work system.

For each result, state whether it supports task redesign, alarm redesign, staffing change, interlock requirement, validation retest, training control, or post-change monitoring. A usability metric is engineering evidence only when it is linked to representative users, realistic conditions, and residual risk.

Exercise 1: Takt Pressure and Manual Task Margin

A manual setup station must support:

The net available production time after breaks, meetings, cleaning, and planned downtime is:

The observed task elements for one setup are:

Element	Time
Retrieve fixture	8 s
Scan work order	5 s
Install fixture	24 s
Confirm orientation	7 s
Start cycle	6 s
Record setup completion	6 s

Find the takt time, total manual task time, and margin. Decide whether the station is robust.

Solution

Takt time is available time per required unit:

The total observed manual task time is:

The time margin is:

As a percentage of takt:

Engineering Comment

The calculation barely passes. A 3.2 percent margin is weak for a human task because small disturbances can consume it: a misread work order, missing fixture, barcode retry, awkward reach, question from another operator, or glove-related handling delay.

This is not only a labor-efficiency issue. If the station is constantly near takt, the user may skip orientation confirmation or documentation to recover time. A robust design would add margin by reducing fixture retrieval time, improving part presentation, simplifying confirmation, splitting the work, or lowering demand during high-mix periods.

Exercise 2: Alarm Response Time and Process Safety Margin

A reactor alarm is triggered when a temperature trend predicts that the safety limit will be reached in:

The response chain has the following estimated times:

Response element	Time
Alarm processing delay	15 s
Operator recognition and diagnosis	35 s
Walk to local valve station	60 s
Manual valve action	12 s
Cooling effect becomes measurable	28 s

Find the nominal response time and margin. Then test a more conservative case where recognition and diagnosis are 45 seconds slower during an alarm flood.

Solution

Nominal response time:

Nominal margin:

Under alarm-flood conditions:

Flood-case margin:

Engineering Comment

The nominal case appears acceptable, but the abnormal condition fails. This is a human factors problem because the alarm asks the user to diagnose, travel, act locally, and wait for thermal response within a narrow process window.

Possible controls include earlier alarming, state-based alarm suppression, clearer alarm cause and response text, a remote cooling action, automatic permissive logic, better local access, or an interlock. Training alone is weak if the required response chain is physically longer than the available process margin during realistic workload.

Exercise 3: Expected Use-Error Escapes

A setup procedure has four critical confirmations per batch. The plant runs:

The observed omission probability for one confirmation is:

An independent downstream check detects 85 percent of omissions before release. Estimate:

1. expected confirmation omissions per month;
2. expected undetected omissions per month;
3. the expected interval between undetected omissions.

Solution

Monthly confirmation opportunities:

Expected omissions:

Escape probability after downstream detection:

Expected undetected omissions:

Expected interval:

Engineering Comment

The downstream check reduces risk but does not make the process robust. An undetected omission about every two weeks is high if the confirmation protects quality, safety, traceability, or regulatory compliance.

The engineering response should focus on the design of the task: barcode-enforced confirmation, poka-yoke fixture orientation, electronic prerequisite checks, better step visibility, or an interlock that prevents release when a critical confirmation is missing. Additional reminders may help, but they do not remove the error opportunity.

Exercise 4: Workload Balance Across User Roles

A packaging cell uses three operators. The average task times per unit are:

Task	Assigned role	Time
Load product	Operator A	18 s
Apply label	Operator A	12 s
Visual inspection	Operator B	22 s
Reject handling	Operator B	6 s
Scan serial number	Operator C	9 s
Pack and seal	Operator C	21 s

The required takt time is:

Find each operator workload and identify the bottleneck.

Solution

Operator A:

Operator B:

Operator C:

All three are below takt:

Nominally, no role is over takt. The highest workload is shared by Operators A and C at 30 seconds per unit.

Engineering Comment

The arithmetic passes, but the design still needs observation. Operator B performs visual inspection, which can degrade under fatigue, glare, repetitive decisions, or ambiguous criteria. Operator C performs scanning and sealing, where serial-number mismatch or packaging rework may create interruptions not included in the average.

Human factors review should check peak workload, not only mean task time. If label misfeeds, barcode failures, rejects, or material shortages cluster, the bottleneck may move from nominal station time to recovery work.

Exercise 5: Usability Validation Completion Criterion

A usability validation test requires that representative users complete a safety-critical task without critical error. The acceptance criterion is:

In a formative test, 19 of 20 participants complete the task without critical error. Calculate the observed completion rate. Does the result prove that the design is acceptable?

Solution

Observed completion rate:

The point estimate meets the criterion.

Engineering Comment

Meeting the point estimate is not the same as strong validation evidence. With only 20 participants, one additional failure would reduce completion to:

The result is borderline. The engineering decision should examine the failed case, near misses, assistance requests, time pressure, user representativeness, and scenario realism. If the task is safety-critical, the team may need design correction and retest rather than accepting a fragile pass.

Human factors validation should not be reduced to a percentage. A single critical use error can reveal a design weakness even when the aggregate metric appears acceptable.

Exercise 6: Risk Priority Number Before and After a Design Control

A maintenance interface allows a technician to select one of two similar pump trains. A wrong selection can isolate the running train. The initial FMEA ratings are:

Rating	Value
Severity	8
Occurrence	4
Detection	5

The design team adds train-specific physical keying, clearer labeling, and a confirmation screen that shows live flow status. The revised ratings are:

Rating	Value
Severity	8
Occurrence	2
Detection	2

Calculate the initial and revised RPN values. What changed and what did not?

Solution

Initial RPN:

Revised RPN:

Risk-ranking reduction:

Engineering Comment

The controls reduce occurrence and improve detection, but severity remains 8. The hazard is still serious if the wrong train is isolated. That distinction matters: a lower RPN is not permission to ignore the failure mode.

The next review should verify that keying cannot be bypassed casually, labels remain legible in the field, live flow status is reliable, and the confirmation screen cannot become a routine click-through. A design control must change real task behavior, not only the spreadsheet rating.

Exercise 7: Queueing Effect of Interruptions

A control-room operator receives support calls during normal monitoring. Calls arrive at an average rate of:

Average handling time is:

Estimate utilization from calls alone. Interpret the result for alarm monitoring.

Solution

Convert service time to hours:

Utilization from calls:

The operator spends about:

of the hour handling calls.

Engineering Comment

A 70 percent call load leaves little attention margin for monitoring, alarm response, shift handover, logging, and abnormal events. Even if average call handling appears manageable, variability can create periods where alarms and calls compete directly.

The design response may be a triage role, call routing, protected alarm-monitoring periods, better self-service diagnostics, or a redesigned support workflow. Asking the operator to simply “pay more attention” does not solve the workload conflict.

Exercise 8: Ergonomic Screening of a Repeated Manual Lift

A technician lifts a 12 kg calibration kit from a lower cabinet to a bench:

Each lift requires a forward reach that adds an estimated ergonomic multiplier of:

Use a simple screening load:

where L is the physical load. Compare the screened load with a local review threshold of 15 kg.

Solution

Screened load:

Comparison:

The task exceeds the screening threshold.

Engineering Comment

This simplified screen does not replace a formal ergonomic assessment, but it is enough to justify review. The issue is not only the 12 kg mass. The forward reach, repetition, cabinet height, grip quality, floor condition, fatigue, and awkward posture change the usability of the maintenance task.

Engineering controls include storing the kit at bench height, splitting the kit, adding a slide-out shelf, using a cart, changing the calibration workflow, or relocating test points. Training in lifting technique is weaker than reducing the reach and handling demand.

Exercise 9: Training Decay and Checklist Control

After initial training, a rarely used emergency procedure has an observed correct-step completion rate of:

Six months later, the observed completion rate is:

After adding a simplified checklist and simulation refresher, the completion rate becomes:

Calculate the relative loss after six months and the recovery achieved by the checklist and refresher.

Solution

Relative loss from initial training:

Absolute recovery after the checklist and refresher:

Recovery as a fraction of the lost performance:

Engineering Comment

The intervention recovers most of the observed loss, but the final rate remains below the original 92 percent. For an emergency procedure, the remaining gap may still be unacceptable depending on severity and time pressure.

The engineering lesson is that rare tasks should be designed for recognition and guided execution, not memory. Checklists, simulation, labels, pre-positioned tools, and decision aids are system controls. Refresher training is useful, but it should support a task design that remains usable after long intervals.

Exercise 10: Post-Change Drift from a Procedure

A new electronic maintenance workflow is deployed. During the first month, audit data show:

Signal	Count
Completed work orders	900
Work orders with manual side notes	126
Work orders reopened for missing evidence	54
Supervisor override approvals	36

Calculate the rate of each signal per completed work order. Which signal should be investigated first?

Solution

Manual side-note rate:

Reopened work-order rate:

Override approval rate:

Engineering Comment

The highest rate is manual side notes at 14 percent. That should be investigated first because it may reveal missing fields, poor terminology, slow screens, inadequate mobile usability, or information that the formal workflow does not capture.

The reopened work orders are also important because missing evidence affects traceability and validation. Overrides may be justified in unusual cases, but a sustained 4 percent override rate can indicate poor configuration or excessive approval friction.

Post-change drift is not automatically noncompliance. It is evidence that the engineered workflow may not fit the real maintenance task. The review should observe users, inspect examples, classify the reasons for side notes and reopenings, and decide whether the fix is interface design, procedure simplification, data model change, training, staffing, or governance.

Review Checklist

When reviewing a human factors or usability result, ask:

• Does the calculation describe real task demand or only an idealized average?
• Are abnormal, degraded, night-shift, high-mix, alarm-flood, maintenance, and recovery modes represented?
• Does the metric connect to an engineering decision?
• Could the proposed control be bypassed, ignored, misunderstood, or overloaded?
• Is the residual risk still high because severity remains high?
• Does validation evidence include representative users, realistic context, and observed behavior?
• Do field data show workarounds, delays, manual notes, repeated questions, or repeated recovery actions?
• Are fatigue, PPE, language, lighting, noise, shift pattern, alarm load, and maintenance access represented where they affect the task?
• Are training, checklist, interface, interlock, staffing, and procedure changes verified by observed behavior rather than intent?

Human factors engineering is strongest when numbers and observations are used together. Timing margins, rates, probabilities, RPN values, and completion percentages are useful only when they change the design of the work system.