Operation, maintenance and proof testing
Keeping the SIL you designed
A safety instrumented system (SIS) only provides the risk reduction it was designed for if it is operated, inspected and maintained as the design assumed. The SIL verification made assumptions about proof test intervals, test coverage, repair times and bypasses. In operation, those assumptions become obligations.
The Buncefield explosion in the UK in 2005 showed what happens when protection layers are not maintained. An independent high-level switch had been left inoperable, and a tank overfilled. Well-designed safety layers are only effective if they are also maintained and tested at regular intervals.
Why proof testing matters
Most process industry safety instrumented functions (SIFs) operate in low demand mode. They sit dormant, sometimes for years, waiting for a real demand. Some equipment, valves in particular, has no self-diagnostics. A dangerous failure can stay hidden until the next proof test, or until a real demand finds it.
That is why low demand SIL verification credits proof testing, and why the test interval directly affects the PFDavg:
- If the interval is extended (a test is deferred, for example), the PFDavg rises and the risk reduction falls. A SIF that met SIL 2 with annual tests may only reach SIL 1 if it is tested every three years.
- If the test is not fully effective, some dangerous failures are never found. The PFD then creeps up year after year, and the risk reduction actually achieved falls during the system’s life. Extending the operating life has the same effect.
In high demand or continuous mode, proof testing is not credited in the same way, and PFH is the measure.
Inspection and test methods
Proof test procedures should be written for each SIF, usually based on the validation procedures. They should test the whole function, from sensor to final element, as close to real process conditions as is safe:
- Sensors. Apply the process variable where possible, or inject a signal as near to the sensing element as possible. Check the trip point and the diagnostics, and look for impulse line blockages.
- Logic solvers. These have high diagnostic coverage, so proof testing is relatively limited. It should check the application program version, voting and outputs.
- Final elements. Fully stroke the valve, measure the closure time against the SRS, and check tight shut-off where the SRS requires it. Partial valve stroke testing between proof tests can find some failures earlier, but it does not replace the full test.
Each test should record the as-found and as-left condition. A failure found during a proof test is valuable information, not just something to fix and forget.
Bypasses and overrides
Bypasses and overrides are sometimes needed: for testing, for maintenance, or for start-up, when a trip would otherwise act on a condition that is normal at that time. But every bypass removes a protection layer while it is in place. Uncontrolled bypassing is a well-known feature of serious incidents.
Good bypass management includes:
- a written procedure, and authorisation by a competent person;
- compensating measures that manage the risk while the SIF is bypassed, written down before the bypass is applied;
- clear indication in the control room of every active bypass, and an alarm or limit on how long it can stay on;
- a log of every bypass, with the time it was applied and removed;
- start-up overrides that remove themselves automatically when the process reaches its normal state.
The SRS should specify the bypass and override requirements for each SIF, so that they are designed in, not improvised.
Track demands and failures
The SIL verification assumed a demand rate and failure rates. Operation is where those assumptions are tested in real life. The owner should record and review:
- demands on each SIF: real trips caused by the process;
- failures found by proof tests, diagnostics or real demands, with their causes;
- spurious trips, which damage confidence in the SIS and encourage bypassing;
- bypass history.
If demands happen more often than the LOPA assumed, the hazard analysis may need revisiting, because the risk is not what was assessed. If failures happen more often than the data predicted, the SIL may not be met, and equipment, test intervals or design may need to change. Either way, the change follows the modification procedure in the next chapter.
IEC 61511 also expects periodic functional safety assessment in operation (FSA stage 4). It checks that operation and maintenance follow the assumptions made in design, and that the hazard analysis is still valid.
Competence in operation
Operators need to understand what each SIF does, what its alarms and bypasses mean, and how to respond. Maintenance staff need to be competent to test SIS equipment safely and record results accurately. Both should be covered by the competence management arrangements described in chapter 11: Functional safety management.
Summary
Operation and maintenance are where the SIL is either kept or lost. Proof test at the intervals the calculation assumed, with tests that find dangerous failures. Control every bypass. Track demands and failures so that real performance can be compared with the design, and act through management of change when it differs.
This chapter summarises the SIS operations, bypass and override, and inspection and proof testing lessons in our IChemE-accredited course SIS and Functional Safety for IEC 61511 (eLP003). See also What is proof testing?