Trust IntegrityManagement Inc.

Home/Insights/Process safety

Insights · Process safety management · PSM tip

Audit the bypass register against the field, not the paperwork

A safeguard that has been quietly defeated looks exactly like one that works, until the day it is needed.

The tip

Pick five safety-critical trips at random. Go and look at each one. Then open the bypass register.

Most process safety management (PSM) audits check the bypass register for signatures, dates and risk assessments. That tells you the paperwork is in order. It does not tell you the plant matches the paperwork.

Turn the audit around. Start in the field and work back to the register. Any bypass you find that is not on the register, or is on it past its expiry, is a leading indicator. It is telling you about the program before an incident does.

What to look for

For each trip or interlock you picked, follow it from the sensor to the final element.

WhereWhat defeats the safeguard
Control systemForced inputs and outputs, inhibited alarms, trip setpoints moved, logic left in test or maintenance mode
Panels and junction boxesJumpers, lifted wires, pulled fuses, keyed bypass switches left on
InstrumentsClosed root valves, isolated transmitters, plugged or frozen sensing lines, a switch left in its test position
Final elementsTrip valves on handwheel or mechanically pinned, air supply isolated, solenoids disconnected
Relief pathsBlock valves closed or not locked open, gagged relief valves, blinded discharge lines

Then compare what you found with the bypass and inhibit log and the list of temporary changes under management of change (MOC).

What to count

Four numbers, trended monthly, say more about barrier health than most audit reports.

  1. Active bypasses on safety-critical elements. The count, and whether it is rising.
  2. Age of the oldest one. A temporary bypass that has seen two outages is a design change nobody approved.
  3. Share with a documented risk assessment and a compensating measure. Something has to stand in for the safeguard while it is out.
  4. Found in the field but not on the register. The target is zero. This is the number that matters most.

Why it works

Major incidents are rarely caused by a missing safeguard. They are caused by one that was there on the drawing and not there on the day.

Buncefield, United Kingdom, 11 December 2005. A gasoline storage tank overfilled and the vapour cloud exploded. The tank's level gauge had stuck, as it had repeatedly since August of that year. The independent high-level switch, the last automatic protection, could not operate because a padlock that held its test lever in the working position had not been fitted. Nobody on site knew the switch was out of service. No bypass register would have shown it. A field check of the device would have.

Texas City, United States, 23 March 2005. A distillation tower was overfilled during start-up and 15 people were killed. The United States Chemical Safety and Hazard Investigation Board (CSB) found that the tower's level transmitter gave false readings, the redundant high-level alarm did not sound, and the sight glass was unreadable. A work order written before the start-up recorded that the transmitter needed repair and deferred the work until the unit was running. The impaired safeguard was known, written down, and accepted.

One was a safeguard nobody knew was defeated. The other was a safeguard everybody knew was impaired. The same field check finds both.

The limits of this check

Five trips is a spot check, not a statistical sample. A clean result on five says little about the other five hundred, and it should not be reported as assurance.

It also finds only one kind of failure. A trip that is in service but has never been proof tested is no better than one that is bypassed. If proof tests and inspections of safety-critical equipment are overdue, fix that first. Use this check to find out whether the plant and the paperwork agree, then decide how deep the full audit needs to go.

Sources

  • Control of Major Accident Hazards (COMAH) Competent Authority, Buncefield: Why did it happen?, Health and Safety Executive, 2011. Root causes of the loss of containment: the automatic tank gauging system and the independent high-level switch.
  • United States Chemical Safety and Hazard Investigation Board, preliminary findings in the BP Texas City refinery accident, 27 October 2005, and the Board's final investigation report, March 2007.
  • United States Chemical Safety and Hazard Investigation Board, statement of Chairman Carolyn Merritt, August 2005, on the level transmitter work order and the high-level alarms.
  • CSA Z767:24, Process safety management, second edition, CSA Group, 2024. The Canadian national standard for PSM systems.

Are your safeguards actually in service?

Barrier health verification and PSM audits, done in the field by the inspector who writes the report.

Talk to the inspector