Chemical¶
Chemical process safety¶
Rules found in this domain: P64, P38.
Illustrative, not evidence.
Scenario. A plant plans to change the solvent in a reactor to a more volatile one. A hazard is found: the new solvent can form a flammable vapour in the vessel headspace.
- P64: The change is a versioned record. Before go-live, the team reviews its technical basis, its safety impact and the procedures it affects, approves it, and informs the operators. The deployed state matches the approved change.
- P38: The team first asks whether the change is needed at all, or whether a less volatile solvent would do. The process needs the new solvent, and the record states why. The team then adds an engineering control at the source, a nitrogen blanket that keeps the headspace too low in oxygen to burn, before any procedure or protective equipment.
Oil and gas¶
Rules found in this domain: P40.
Illustrative, not evidence.
Scenario. An operator runs a gas processing facility. A relief valve protects a vessel from overpressure. Gas detectors and an emergency shutdown system also act on the same hazard.
- P40: The vessel has several independent and verified control layers: the process control loop, the relief valve and the shutdown system. Each layer is tested on its own, so the loss of one does not leave the vessel unprotected.
Food safety¶
Rules found in this domain: P72.
Illustrative, not evidence.
Scenario. A dairy pasteurizes milk. The team lists the process steps and finds that heating is the step where control is essential.
- P72: The dairy defines pasteurization as a critical control point, with a temperature and hold-time limit. It monitors the temperature continuously, has a predefined corrective action for a low reading, keeps records, and regularly verifies in practice that the control works.
Pharma and clinical trials¶
Rules found in this domain: P11.
Illustrative, not evidence.
Scenario. A clinical trial records patient blood pressure readings in an electronic system. A reading is later found to be wrong and is corrected.
- P11: Each record shows who entered it and when, and it was recorded at the time of measurement. The correction keeps the original entry visible and records who changed it, when and why, so a sample of records shows author, time and full change history.
Textile¶
Rules found in this domain: P58.
Illustrative, not evidence.
Scenario. A lab tests the colour fastness of two dyed fabrics from different suppliers. The two reports used different light sources and different humidity.
- P58: The lab repeats both tests under stated, controlled reference conditions and reports the uncertainty with each result. It calibrates the instrument against a traceable reference, so the two fabrics can be compared fairly.
Electrochemical (batteries)¶
Rules found in this domain: P41.
Illustrative, not evidence.
Scenario. A battery pack for a tool is built from many cells. One cell can fail and heat up. The design keeps heat and failure from spreading to the neighbouring cells.
- P41: The pack is divided into isolated compartments, each with its own barrier and fuse. In a failure test, one cell is forced to fail, and the other compartments keep working.
Molecular and biomolecular engineering¶
Rules found in this domain: P31, P81.
Illustrative, not evidence.
Scenario. A lab wants an enzyme that works at a higher temperature. No model predicts which changes help. The team makes thousands of variants and tests them in rounds.
- P31: No model is good enough to design the enzyme analytically, so the lab generates many variants, tests them, keeps the best and repeats. Each round has a defined activity test and records which variants were kept.
- P81: The lab repeats the cycle plan, do, check, act. After each round it records the follow-up actions, such as changing the screening temperature, and tracks them until they are completed.