Read the major-accident bowtie
Original fictional UK teaching example, informed by fuel-storage guidance and Buncefield learning. It is not a reconstruction of Buncefield or a Crown Fuels design. The top event is tank overflow at atmospheric storage conditions; LPG, pipe rupture, chemical reaction and dust explosions need separate bowties. No setpoints, engineering design, SIL, LOPA credits, blast distances or operating authority are supplied. GB and NI requirements remain distinct. Older technical references are used for principles, not as current legal registers. Independent SHAREhub study support; it does not award the NEBOSH qualification or imply approval.
A large petrol inventory during tank filling
In everyday language: Petrol held and transferred in bulk can release flammable vapour and contaminate the environment. People both on and beyond the site may be exposed.
Why it belongs here: The hazardous material and inventory exist before anything goes wrong. The hazard is not the fire, the overfill or an injury.
What to check: Identify the actual product, inventory, transfer route and people or environmental receptors that could be affected. This example assumes an atmospheric storage tank, not LPG or a reacting vessel.
Petrol overflows the storage tank
In everyday language: Petrol crosses the tank’s containment boundary because filling has exceeded the level that the tank can safely contain. It can cascade down the tank and form a flammable vapour cloud.
Why it belongs here: This is the loss-of-primary-containment moment at the knot. Prevention acts before overflow; recovery limits what happens after it. Fire or explosion has not necessarily occurred.
What to check: Keep this one top event on every route. A pipe leak, tank collapse, thermal runaway and pressurised-vessel rupture require their own scenarios.
The planned receipt exceeds the tank’s spare capacity
In everyday language: The chosen tank has less usable space than assumed, or the incoming product is sent to the wrong receiving tank.
Why it belongs here: An incorrect starting inventory or destination can take the tank towards overflow even while equipment appears to work.
What to check: Reconcile the tank identity, starting level, usable capacity and receipt quantity before authorising the transfer.
The normal level indication is wrong or stops changing
In everyday language: The filling display or its associated alarm gives a misleading picture while petrol continues to enter the tank.
Why it belongs here: A stuck reading can hide the rising level. A screen and an alarm using the same failed measurement are not two independent protections.
What to check: Look for a reading inconsistent with flow or expected stock movement. Use the abnormal-operation procedure and stop the receipt when the safe condition cannot be established.
Receipt continues beyond the planned stopping point
In everyday language: The incoming flow, communication delay or actual stopping time leaves less margin than the receipt plan assumed.
Why it belongs here: A request to stop is not proof that inflow has stopped. The tank needs enough response margin for the real transfer arrangement.
What to check: Confirm responsibility at both ends, current flow, stopping authority and the verified response of the actual final stopping equipment.
Protection is bypassed or incorrectly restored after change
In everyday language: An inhibited trip, changed setting, wrong connection or incomplete maintenance hand-back leaves the overfill protection unavailable.
Why it belongs here: A protective device can exist on the drawing without providing its intended function during this receipt.
What to check: Review current impairments, change approval and the hand-back evidence before filling; do not treat an expired bypass note as proof of restoration.
Verify capacity, receipt quantity and stopping margin
In everyday language: Use trustworthy stock and tank data to plan a receipt that stays inside the approved normal operating envelope.
Why it belongs here: This reduces the chance of filling beyond the intended capacity. It is a planning control, not a replacement for independently justified overfill protection.
What to check: Verify the starting inventory, product, usable capacity, transfer rate and stopping margin against current controlled information. No generic setpoint is supplied here.
Owner: The receipt planner and responsible operations person.
What could undermine it: Wrong stock data, late rate changes or unrecorded simultaneous transfers.
Confirm the receiving tank and the actual flow path
In everyday language: Check that the physical route, valve line-up and receiving tank match the authorised receipt plan before flow starts.
Why it belongs here: The capacity calculation does not protect a different tank receiving the product by mistake.
What to check: Match tank and line identities to the real arrangement. Use the site’s specified verification and communication method; a copied checklist is not verification.
Owner: The operators controlling the sending and receiving interfaces.
What could undermine it: Ambiguous labels, a changed route, poor handover or checks made away from the actual plant.
Monitor filling and act on a reliable normal alarm
In everyday language: Operators follow the receipt and respond to a meaningful, functioning normal high-level alarm in time to stop filling.
Why it belongs here: This is the normal control-and-response path. A display, alarm and operator action that share one gauge must not be counted as independent layers.
What to check: Check the measurement, alarm presentation, workload, response procedure and demonstrated stopping time. Do not use a high-level alarm as the routine target fill level.
Owner: The receiving operator and control-room supervisor.
What could undermine it: Stuck instruments, alarm floods, distraction, unclear ownership or insufficient response time.
An independent high-high function stops receipt
In everyday language: In this example, a separately engineered overfill function detects an unsafe high level and automatically stops incoming petrol before overflow.
Why it belongs here: The whole chain matters: sensor, logic, utilities and final stopping equipment. Its required performance and independence need an engineering assessment; the diagram assigns no SIL or risk-reduction credit.
What to check: Trace the tested cause-and-effect through the actual shutdown equipment. Check action level, response time, proof-test coverage, failure response and restoration from bypass against the approved design.
Owner: The accountable operations owner and competent instrument/control engineers.
What could undermine it: A shared sensor or failure path, an ineffective final valve, a hidden inhibit or a test that omits the stopping action.
Resolve changes and protection impairments before receipt
In everyday language: Changes, temporary bypasses and maintenance hand-back are reviewed against the real receipt scenario before the tank is returned to filling service.
Why it belongs here: This supporting arrangement helps keep the physical safeguards available. It is not an independent instrumented protective layer, and a permit alone does not stop an overfill.
What to check: Confirm who authorised the change, the current operating restriction, completed checks and evidence of restored protection. Do not continue without a defensible safe operating basis.
Owner: The change authority, maintenance hand-back owner and operations manager.
What could undermine it: Production pressure, informal overrides or a bypass that survives a shift change.
Retain released liquid and protect drainage routes
In everyday language: A suitable liquid-retaining bund and controlled drainage limit the spread of released petrol and contaminated water.
Why it belongs here: This acts after primary containment is lost. A bund is not a vapour-tight enclosure and does not guarantee that a flammable cloud stays within its walls.
What to check: Check usable capacity, wall and floor integrity, joints, penetrations, drain status and the assessed firewater/run-off arrangement. A generic percentage or visual check alone does not establish adequacy.
Owner: The site containment owner and operations team.
What could undermine it: Open drains, damaged joints, accumulated rainwater or fire/explosion damage.
Keep ignition sources away from the possible release
In everyday language: Suitable equipment, controlled hot work and managed vehicles reduce opportunities for released petrol vapour to ignite.
Why it belongs here: For this overfill top event, this control limits a subsequent consequence. It must already be established; do not enter a vapour cloud to switch equipment off. Zoning is not a spill or explosion boundary.
What to check: Use the assessed release conditions, equipment suitability and simultaneous-work controls. A normal hazardous-area drawing or one gas reading cannot authorise work during an abnormal release.
Owner: Operations and the people controlling equipment and work permits.
What could undermine it: Uncontrolled vehicles, unsuitable temporary equipment or changed work close to the tank.
Detect the release, alarm and protect people
In everyday language: The scenario-based response warns people, stops further receipt from a safe location where feasible, and directs evacuation or shelter under the site plan.
Why it belongs here: Isolation can limit additional material but cannot recover petrol already released. Detection and response take time; escalation protection and specialist firefighting must be separately assured.
What to check: Check detection coverage, alarm and communication paths, safe stopping locations, escape arrangements, accountability, responder access and contingencies for failed shared services. Recovery and restart need a separate decision.
Owner: The incident controller, operations and the competent emergency-response organisations.
What could undermine it: Delayed detection, unavailable communications, blocked access or an untested response assumption.
People are exposed and fuel reaches soil or water
In everyday language: Released petrol can expose people and reach drains, soil or water even if it does not ignite.
Why it belongs here: Loss of containment can cause harm without a fire. Failed secondary containment can extend the environmental consequences.
What to check: Include off-site routes, drainage destinations and vulnerable receptors in the scenario. Response work must not put people into an uncontrolled atmosphere.
A vapour cloud ignites and harms people
In everyday language: Released petrol can form a flammable cloud. Ignition may cause a flash fire; damaging explosion overpressure depends on the release and combustion conditions, including congestion or confinement.
Why it belongs here: An explosion is a possible consequence, not an automatic result of every overfill. The cloud may extend beyond a bund or the usual classified area.
What to check: Assess actual dispersion, ignition, occupancy and layout. This bowtie predicts no cloud distance, blast pressure, casualty count or accident frequency.
A pool fire spreads to nearby plant and people
In everyday language: Ignited released liquid may form a pool fire. Heat, smoke and damage to nearby equipment can escalate the incident, while contaminated firewater creates further pollution.
Why it belongs here: One tank’s event can affect neighbouring equipment and emergency resources. Separation, passive protection and any assessed active fire protection need their own performance evidence.
What to check: Review the credible multi-equipment scenario, protection duty, water/media availability and run-off retention. Do not assume a pump start or the arrival of responders proves successful control.
Tank, product and route information is correct
In everyday language: The people planning and checking the receipt use current, consistent information about the tank, petrol, capacity and connections.
Why it belongs here: The plan and line-up can both be wrong if they share the same incorrect drawing or inventory figure.
What to check: Reconcile controlled records with the real plant and resolve discrepancies before using the information.
Owner: The process-information owner and operations.
What could undermine it: Unrecorded modifications or unreliable stock reconciliation.
People have time, competence and authority to stop
In everyday language: The people making receipt and abnormal-operation decisions understand the task and can stop it without production pressure overriding safety.
Why it belongs here: Instructions cannot compensate for an unworkable workload, missing task skill or an alarm that no one owns.
What to check: Check task-specific competence, staffing, workload, handover and the escalation route for an uncertain safe condition.
Owner: Operations management and task supervisors.
What could undermine it: Fatigue, rushed handovers or normalisation of repeated faults.
Inspection and tests cover the required function
In everyday language: Checks show what the equipment actually does, with defects, test limits and restoration recorded.
Why it belongs here: A sensor-only test cannot establish that the final stopping equipment closes. An inspection certificate does not demonstrate every containment or protection function.
What to check: Match the test to the approved performance requirement and full path, including demand conditions, intervals and unresolved defects.
Owner: The competent integrity, instrument and maintenance owners.
What could undermine it: Overdue tests, incomplete coverage or a test override left active.
Power, actuation and communications behave as designed
In everyday language: The controls have the services they require, or move to their engineered safe condition when a service is lost.
Why it belongs here: A shared utility failure can affect several functions. Some failures create a safe stop; others defeat protection. The actual design determines the result.
What to check: Trace utility dependencies, loss-of-service response, backup capacity and survivability. The what-if highlights review needs; it does not assert that every affected function fails.
Owner: The utilities and protective-systems owners.
What could undermine it: Common supply loss or backup equipment that cannot deliver its duty.
Claimed independent paths have no hidden common failure
In everyday language: The normal level-control path and the independent protective function are assessed for shared sensors, logic, final elements, services and maintenance errors.
Why it belongs here: Two boxes on a diagram are not automatically two independent layers. Common failures can defeat both at once.
What to check: Review the actual architecture and evidence before crediting independence. The qualitative model neither proves independence nor performs a LOPA calculation.
Owner: The competent functional-safety and process-safety reviewers.
What could undermine it: A common measurement, valve or systematic configuration error.
Bund and drainage boundaries stay intact
In everyday language: The containment system remains liquid-retaining, and drainage follows the assessed controlled arrangement.
Why it belongs here: An open outlet or failed joint can bypass the bund even when its wall is visibly present.
What to check: Verify the real drain destinations, valve positions, integrity and available retention capacity against the current scenario.
Owner: The containment and drainage owners.
What could undermine it: Uncontrolled rainwater discharge or an unrecorded drain connection.
Changes, bypasses and nearby work are controlled
In everyday language: Receipt, maintenance, hot work and temporary arrangements are coordinated so that one activity does not defeat another’s safeguard.
Why it belongs here: A maintenance action or vehicle route can affect overfill protection and the ignition conditions at the same time.
What to check: Check the live impairment register, authorisation, shift handover, work interfaces and verified restoration before release.
Owner: Operations, maintenance and the work-control authority.
What could undermine it: Informal bypasses, incompatible simultaneous work or incomplete hand-back.
The response resources can perform their actual duty
In everyday language: People, alarms, access, communication and any required specialist protection resources are available for the credible incident.
Why it belongs here: A written plan alone cannot warn, evacuate or protect people. Response may itself depend on damaged or shared services.
What to check: Exercise the real interfaces and contingencies; verify responder capability and protection duty without assuming immediate success.
Owner: The emergency planning owner and participating response organisations.
What could undermine it: Unavailable resources, blocked access or unresolved exercise findings.