Build the mental model
Guided explanations, visible consequences and supported decisions for newer personnel.
HSG253 decision simulator
Isolation is a system, not a valve.
Learn to recognise hazardous energy, select and prove an isolation, control change, and reinstate plant without losing the thread.
Training demonstrator. This experience supports learning from HSG253. It does not replace company procedures, task risk assessment, permit-to-work controls, competent judgement or authorised personnel.
Live process74 bargOne decision away from containment loss
Apply and diagnose
Independent choices, P&ID reasoning and fault diagnosis for people planning or installing isolations.
Authorise and assure
ALARP, variations, interactions, change and reinstatement decisions for isolation authorities.
System view
The isolation lifecycle
The boundary remains controlled from task definition through reinstatement and learning. Select a stage to inspect its decisions and failure traps.
State laboratory
Build, de-inventory and prove
Manipulate a simplified isolation. The model tracks upstream pressure, valve condition, cavity pressure, downstream pressure and evidence.
Live/pressurisedIsolated/unprovenProved
Live process challenge
C-201 compressor seal replacement
Gas compressor C-201 requires intrusive work while the adjacent train remains live at 74 barg. Your decisions change the isolation state and later options.
74 bargIsolation envelope
Not provedIsolation evidence
2 linkedDependent permits
HIGHCurrent exposure
Competence evidence
Decision assessment
Your first answer is recorded. Correcting an answer supports learning but does not erase the original decision.
Learning record
Competence profile
Evidence and integrity
Assurance dossier
Design rationale, HSG253 traceability and executed checks for this prototype.
Learning architecture
The product follows a decision-first architecture. The lifecycle establishes the system model; the isolation laboratory exposes causal plant behaviour; the compressor scenario integrates hardware, documentation, people and time; the assessment samples transfer. Three routes alter scaffolding and feedback tone. The experience deliberately returns to proving, change and reinstatement because these are points where a superficially complete isolation can fail.
Design decisions
Plant state is held as data: valve positions, pressure, proof, bleed condition, motive power, permits, scope and reinstatement dependencies. Visual status derives from that state. Unsafe choices identify the remaining hazard and the evidence needed before work proceeds. The model avoids presenting positive isolation as infallible or using a universal hierarchy without task risk context.
Competence evidence
Assessment retains first-attempt performance, domain scores and critical errors. Repeated guessing cannot create a competent result. “Competent in this demonstrator” is distinct from site authorisation, which remains dependent on workplace assessment and the organisation's competence system.
Most ambitious feature
The isolation laboratory couples valve, bleed, pressure, proof and hidden-fault states. A blocked bleed can create a false impression from the gauge; proof logic then rejects the evidence. The scenario carries earlier choices into change-control and reinstatement decisions.
| Component | HSG253 basis | Application |
|---|---|---|
| System principles | paras 12–16, 22–25 | ALARP, tolerability, controlled lifecycle |
| Human factors and competence | paras 28, 42–57 | Error traps, route depth, competence caveat |
| Work control and P&IDs | paras 69–84 | Certificates, cross-references, common isolation and change |
| Hazard/risk assessment | paras 87–108 | Scope, substances, task hazards, live-plant decision |
| Method selection | paras 109–124; Figure 4 | Positive/proved/non-proved, DBB/SBB, risk basis |
| Variations | paras 128–141 | Authority challenge and approval controls |
| Planning and installation | paras 142–155 | Walkdown, initial/final isolation, security and tags |
| DVPF | paras 156–164 | Pressure and inventory removal |
| Proving and monitoring | paras 165–183 | Proof, false indications, monitoring and failed proof |
| Reinstatement | paras 184–188 | Permit dependencies, overrides, blinds and post-start checks |
Executed programmatic checks
- Navigation targets and all queried element IDs checked against the document.
- Scenario data contains eight reachable decisions and feedback/reference content for every option.
- Assessment scoring preserves first attempt and flags critical choices.
- Reset clears stored state and rebuilds scenario, laboratory, assessment and report.
- No external resources or network calls are present.
Runtime inspection required
- Desktop and tablet layout, keyboard sequence, visible focus and browser print should be inspected in the delivery browser.
- Automated screen-reader testing was not executed. Controls use native elements, accessible names and polite live regions.
- Colour contrast was designed for dark presentation but was not certified with an external WCAG analyser.