Cross-cutting¶
Professional ethics¶
Rules found in this domain: P1, P2, P3, P6.
Illustrative, not evidence.
Scenario. A consulting engineer is asked to approve a bridge design. Its bearings come from a firm in which she holds shares, and bearings are outside her field. She declares the share holding on her firm's register of interests and steps out of the bearing review. A named colleague with bearing expertise checks that part instead. The project is behind schedule, and the contractor offers cheaper, untested bearings to save time. The release gate has a safety sign-off item, so the swap waits for the same check.
- P6: She would review work in which she has a personal interest, so she discloses it and steps out of that review. The register is checked before the approval.
- P2: Bearings are outside her competence, so the bearing part goes to a named checker with expertise in that domain.
- P1: Public safety goes above cost and schedule. The release gate contains a safety sign-off item, so the cheaper, late substitute does not skip it.
Systems engineering¶
Rules found in this domain: P9, P54, P59.
Illustrative, not evidence.
Scenario. A team builds a train door system that is safety-critical and regulated. The requirement "doors must close quickly" is rewritten as "doors close in under 4 seconds", with a link to the passenger need for short stops. Each requirement is linked to design and test. Before handover, the team tests the whole system against the operator's requirements, and the operator signs an acceptance.
- P9: Each requirement is written as measurable, with an acceptance criterion and a link to a user need.
- P59: The system is safety-critical and regulated, so every requirement is linked in both directions to design and test, with no requirement lacking a test.
- P54: The team does verification and validation. Before handover, all parts are tested together against the owner's requirements, and both a test result and a stakeholder acceptance exist.
Configuration management¶
Rules found in this domain: P64.
Illustrative, not evidence.
Scenario. A plant changes the set point of a pressure controller. The change is recorded as a new version of the settings file. Before it goes live, a reviewer checks its technical basis, the safety impact and the operating procedures it affects. A manager approves it, and operators are told before go-live. Later, an audit finds that the deployed settings match the approved version.
- P64: Every artifact is versioned. A change is reviewed for basis, safety impact and affected procedures, then approved, and the affected people are informed before go-live. The deployed state matches an approved version.
Reliability and safety engineering¶
Rules found in this domain: P22.
Illustrative, not evidence.
Scenario. Before building a robot arm for a packing cell, the team lists how each part can fail, such as a cable, a motor or a sensor. It also lists how parts that have not failed can interact unsafely, such as a sensor and a controller that disagree. The team adds how the arm can cause harm while working as designed, for example when a worker reaches in. The list is reviewed.
- P22: Before building, the team lists how each part can fail, how parts that have not failed can interact unsafely, and how the system can cause harm while working as designed. The list exists per component and is reviewed.
Accident research¶
Rules found in this domain: P82.
Illustrative, not evidence.
Scenario. A hydraulic lift in a workshop sometimes drifts down a few centimetres when it is loaded. Nobody has been hurt and the crew says it always did that. A new technician records it as an anomaly in a ticket. The team investigates it and finds a worn seal. They fix the seal and the ticket holds the decision.
- P82: The system behaves differently from what was expected, even though nobody was harmed. The team records the anomaly and investigates it before accepting it as normal. The ticket holds a decision: fix it, or accept it with a written reason.
Metrology¶
Rules found in this domain: P58.
Illustrative, not evidence.
Scenario. Two labs compare the stiffness of the same polymer. Their numbers differ by 8 percent. They find that one lab measured at 20 degrees Celsius and the other at 30. They agree on a reference temperature and report it. Each report also states the uncertainty. Both labs calibrate their load cells against a traceable reference on a schedule.
- P58: Measurements are used to compare, so they are made under stated, controlled reference conditions, with uncertainty reported with each result. Instruments are calibrated regularly against traceable references.
Risk regulation (UK HSE)¶
Rules found in this domain: Master.
Illustrative, not evidence.
Scenario. A site owner reviews a risk from a storage tank. The risk is first sorted into a region: unacceptable, tolerable or negligible. An unacceptable risk must be reduced whatever the cost. A tolerable risk is reduced while it is reasonably practicable. A negligible risk needs no further reduction. The tank's risk falls in the tolerable region, so the owner keeps adding controls until the next one costs grossly more than it gains.
- Master: The master rule says to keep reducing risk until the cost of the next reduction is grossly disproportionate to the benefit (ALARP). The three risk regions set how far to go. The tank's risk is tolerable, so the owner reduces it while that is reasonably practicable and stops at the grossly disproportionate point.
Design engineering¶
Rules found in this domain: P16.
Illustrative, not evidence.
Scenario. Early in the design of a new handheld power tool, the team is unsure whether a brushless or a brushed motor will meet cost and noise targets. It develops both in parallel. As test results arrive, it drops the weaker option. It commits only once feasibility is shown. The design records list the alternatives kept and the evidence that removed each rejected one.
- P16: Uncertainty is high early, so several alternatives are developed in parallel and weak ones are eliminated as evidence arrives. The team commits only after feasibility is shown, and the records list the evidence.
Engineering physics¶
Rules found in this domain: P12.
Illustrative, not evidence.
Scenario. A team uses a heat flow simulation to decide how thick an insulation layer must be for a furnace wall. The model was validated for temperatures up to 600 degrees Celsius, but the furnace runs at 900. The team does not rely on the result. It extends the validation to the higher range and checks the output against a rough hand estimate of the right order of magnitude.
- P12: A decision depends on a model, so it is validated for its intended use and used only inside its validated range. Each result is checked against an independent rough estimate. Once the validation is extended to 900 degrees, the decision lies inside the stated range.