Ormaven

8D problem solving: a worked example

A fictitious manufacturing case shows how to move from a recurring defect to a verified corrective action. The numbers are illustrative, not data from a real customer or an audit.

The example: 18 defects, but what changed?

A production line reports 18 assemblies with an incorrect fastener torque during a 500-unit shift. That is 3.6% of the batch. The team identifies 14 affected units at workstation A and four at workstation B. Those facts describe where to look, not yet why the defect happened. Define the defect precisely: part, torque specification, measurement method, time window and inspection evidence. Do not state that all 500 assemblies are defective, nor that the problem is solved merely because the line was stopped.

D1 — Put the right people in the room

Include the operator and supervisor, quality engineer, maintenance technician and process owner. Assign one leader and a deadline for each action. If a supplier component is suspected, invite the supplier contact rather than assuming responsibility. The record should show who can approve containment and who will verify effectiveness.

D2 — Write a testable problem statement

Use a statement such as: “On the morning shift, 18 of 500 assemblies failed a defined torque check; 14 were from station A and four from station B. The earliest known failure was found at 08:40.” Record which shift, line, product version and inspection method were involved. Include the exact requirement from the controlled specification. A vague statement such as “operators make mistakes” is not a root cause.

D3 — Contain risk without claiming the problem is fixed

Quarantine affected lots based on the known exposure window, apply a documented screening method to potentially impacted assemblies, and decide disposition under authorised quality controls. If products have shipped, evaluate customer notification and traceability obligations. Record what was inspected, what passed, what failed and which person released material. Containment protects customers; it is not the permanent corrective action.

D4 — Separate occurrence cause from escape cause

Suppose inspection shows a worn locating fixture can tilt the part under the tool. That is a hypothesis for the occurrence cause. A second hypothesis is that the torque-check sampling interval was too long to detect the condition promptly: an escape cause. Test both. Compare fixture alignment measurements, defect distribution and inspection records before and after a controlled fixture replacement. Reject hypotheses that cannot reproduce or explain the pattern. Do not conclude “human error” simply because an operator was present.

D5 — Choose actions linked to verified causes

If controlled tests confirm the locator is the occurrence cause, replace or redesign it, define wear limits and add a scheduled verification step. If the escape test finds a detection gap, update the monitoring plan or add a validated automated check. For every action, record owner, due date, evidence to collect and how the change is approved. A training slide alone is unlikely to correct a worn fixture.

D6 — Validate implementation and effectiveness

Document the fixture installation, measurement results, updated work instruction and any required operator briefing. Then verify performance over an agreed sample and time horizon under normal conditions. For example, inspecting a subsequent 1,000 assemblies with zero observed failures is promising, but it does not prove the true defect probability is zero. Monitor the process for recurrence and evaluate whether the new inspection can detect deliberately created test failures.

D7 — Prevent recurrence elsewhere

Review comparable stations and product families that use the same fixture design or sampling logic. Update preventive maintenance, process risk analysis, control plans and escalation rules only where evidence shows a related risk. Identify what triggered the original delay in detection and who owns each system change. A lesson learned has value only if it changes a relevant control.

D8 — Close with evidence, not a signature alone

An 8D can be closed when containment disposition is complete, causes and actions are supported by evidence, effectiveness criteria are met and outstanding risks have an owner. Record whether customer agreement is required. Recognise contributions, archive the evidence and schedule an audit of sustained use. This example is educational and does not replace customer-specific 8D rules, contractual requirements or professional engineering judgment.

Try a practical next step

Start with one real issue and write a two-sentence problem statement containing the product, defect, scope, when and where. List which facts are observed and which are only hypotheses. Then review the existing quality workflow and define the next measurement. Explore the Ormaven quality tools at /network/tools/quality-operations/ or describe a process problem at /request/.