Functional Safety

SIL Verification

Verify safety integrity levels for safety instrumented functions using PFD calculation, architectural constraints, and systematic capability assessment per IEC 61508 and IEC 61511.

SIL 1

PFD: 0.01-0.1

Risk reduction factor 10-100

SIL 2

PFD: 0.001-0.01

Risk reduction factor 100-1,000

SIL 3

PFD: 0.0001-0.001

Risk reduction factor 1,000-10,000

Use this guide when:

  • Verifying SIF meets required SIL target
  • Calculating PFD for safety instrumented functions
  • Selecting voting architecture (1oo1, 1oo2, 2oo3)
  • Optimizing proof test intervals

1. Overview

SIL verification is the process of confirming that a safety instrumented function (SIF) achieves the safety integrity level (SIL) determined during the SIL determination phase (LOPA or risk graph). The verification must demonstrate three aspects: target PFD is met (random hardware), architectural constraints are satisfied, and systematic capability is adequate.

Random Hardware

PFD Calculation

Probability of failure on demand from component failure rates and test intervals.

Architecture

Hardware Fault Tolerance

Minimum redundancy based on SIL target and component SFF.

Systematic Capability

SC Rating

Software quality, design process, and management of systematic failures.

Common Cause

Beta Factor

Fraction of failures that affect all redundant channels simultaneously.

Three-legged stool: SIL verification requires ALL three requirements to be met: (1) PFD_avg within the SIL target range, (2) architectural constraints satisfied per IEC 61508 tables, and (3) all subsystem components have systematic capability rating equal to or greater than the SIL target. Failing any one requirement means the SIF does not meet the SIL target.

2. SIL Targets & PFD

SIL Levels for Low-Demand Mode

SILPFDavg RangeRisk Reduction FactorAvailability
SIL 10.01 to < 0.110 to 10090-99%
SIL 20.001 to < 0.01100 to 1,00099-99.9%
SIL 30.0001 to < 0.0011,000 to 10,00099.9-99.99%
SIL 40.00001 to < 0.000110,000 to 100,00099.99-99.999%
Low-Demand vs High-Demand Mode: Low-demand mode (most oil & gas SIF): Demand rate < 1 per year (or < 2x proof test rate) Performance metric: PFD_avg (probability per demand) SIL defined by PFD range (table above) High-demand / continuous mode: Demand rate > 1 per year Performance metric: PFH (per hour failure frequency) SIL 1: 10^-6 to 10^-5 per hour SIL 2: 10^-7 to 10^-6 per hour SIL 3: 10^-8 to 10^-7 per hour Most process industry SIF operate in low-demand mode.

3. PFD Calculation Methods

Simplified PFD Formulas

1oo1 Architecture (Single Channel): PFD_avg = (λ_DU × T_proof) / 2 Where: λ_DU = Dangerous undetected failure rate (per hour) T_proof = Proof test interval (hours) Example: Pressure transmitter: λ_DU = 2.5 × 10^-7 /hr Logic solver: λ_DU = 0.5 × 10^-7 /hr ESD valve: λ_DU = 5.0 × 10^-7 /hr Test interval: 1 year = 8,760 hours PFD_sensor = 2.5e-7 × 8760 / 2 = 0.00110 PFD_logic = 0.5e-7 × 8760 / 2 = 0.00022 PFD_valve = 5.0e-7 × 8760 / 2 = 0.00219 PFD_SIF = 0.00110 + 0.00022 + 0.00219 = 0.00351 This meets SIL 2 (0.001 to 0.01) ✓
1oo2 Architecture (Redundant): PFD_avg = [(λ_DU × T_proof)² / 3] + β_DU × λ_DU × T_proof / 2 Where: β_DU = Common cause failure fraction (typically 0.02-0.10) The first term represents independent double failures (both channels failing before the next test). The second term represents common cause failures that defeat both channels simultaneously. For small β and moderate failure rates: Common cause term often dominates Redundancy provides ~10x improvement (not 100x+)
2oo3 Architecture (Triple Modular Redundant): PFD_avg = [(λ_DU × T_proof)²] + β_DU × λ_DU × T_proof / 2 2oo3 provides: - Same dangerous failure PFD as 1oo2 - Better spurious trip rate than 1oo2 - Higher availability (fewer false trips) - More expensive (3 channels) Spurious trip rate for 2oo3 is much lower because two channels must fail safe simultaneously.

Diagnostic Coverage Effect

ParameterDefinitionEffect on PFD
λ_D (dangerous)Total dangerous failure rateDetermines base failure probability
DC (diagnostic coverage)Fraction of dangerous failures detectedSplits λ_D into DD and DU
λ_DD (detected)λ_D × DCDetected quickly, repaired before demand
λ_DU (undetected)λ_D × (1 - DC)Only found by proof test, drives PFD

4. SIF Architecture Selection

Architectural Constraints (IEC 61508 Route 1H)

SIL TargetSFF < 60%SFF 60-90%SFF 90-99%
SIL 1HFT = 1HFT = 0HFT = 0
SIL 2HFT = 2HFT = 1HFT = 0
SIL 3HFT = 3HFT = 2HFT = 1
SIL 4Not allowedHFT = 3HFT = 2
Key Definitions: SFF = Safe Failure Fraction = (λ_S + λ_DD) / (λ_S + λ_D) = (Safe failures + Detected dangerous) / Total HFT = Hardware Fault Tolerance = Number of faults that can be tolerated 1oo1: HFT = 0 (no fault tolerance) 1oo2: HFT = 1 (tolerates one fault) 2oo3: HFT = 1 (tolerates one fault) 1oo3: HFT = 2 (tolerates two faults) IEC 61511 Route 2H (Simplified): SIL 1: HFT = 0 (no redundancy required) SIL 2: HFT = 1 (redundancy in one subsystem) SIL 3: HFT = 1 (all subsystems) or HFT = 2 Route 2H does not require SFF data and is widely used in the process industry per IEC 61511.

5. Proof Test Optimization

Proof Test Interval vs PFD: PFD_avg = λ_DU × T / 2 (for 1oo1) This is linear: doubling T doubles PFD. To achieve a target PFD: T_max = 2 × PFD_target / λ_DU Example (SIL 2, PFD < 0.01): λ_DU = 5 × 10^-7 /hr T_max = 2 × 0.01 / 5e-7 = 40,000 hours = 4.6 years With partial stroke test (PST) credit: λ_DU splits into: λ_DU_PST (detectable by PST): 60-70% λ_DU_FT (requires full test): 30-40% PFD = λ_DU_PST × T_PST / 2 + λ_DU_FT × T_FT / 2 Monthly PST can extend full test to 3-5 years while maintaining the same PFD.

Proof Test Coverage

Test TypeCoverageWhat It Tests
Partial valve stroke60-70%Valve movement, actuator, solenoid
Full valve stroke90-95%Full closure, seat integrity
Transmitter simulation70-80%Electronics, output signal
Transmitter process test90-95%Sensing element, process connection
Logic solver validation95-99%Input/output, logic execution
Less-than-perfect testing: If proof test coverage is less than 100%, a residual undetected failure probability accumulates over the life of the SIF. This must be included in the PFD calculation as: PFD_residual = λ_DU × (1 - coverage) × T_life / 2, where T_life is the expected SIF lifetime (typically 15-25 years).

6. Systematic Capability

Systematic capability (SC) addresses failures caused by design errors, specification mistakes, software bugs, and manufacturing defects. Unlike random hardware failures, systematic failures cannot be quantified probabilistically. Instead, IEC 61508 requires that the development process meets defined quality requirements.

Systematic Capability Assessment: Each SIF subsystem component must have an SC rating ≥ the SIL target. Sources of SC rating: 1. IEC 61508 certification (TUV, exida, FM): Component certified to SC 1, SC 2, or SC 3 This is the most reliable evidence 2. Prior use (IEC 61511 Clause 11.5): Documented operating history in similar service Minimum 10 device-years of experience No dangerous systematic failures observed Configuration management maintained 3. Proven-in-use (conservative approach): Extensive field history with positive results Supported by manufacturer data Requirement: All components in the SIF must have SC ≥ SIL target The weakest link determines the maximum achievable SIL

7. Practical Application

SIL Verification Workflow

Step-by-Step Verification Process: 1. Define SIF: sensors, logic solver, final elements 2. Obtain failure rate data (λ_D, λ_S, DC, β) Sources: exida SERH, OREDA, manufacturer data 3. Select architecture (1oo1, 1oo2, 2oo3) 4. Set proof test interval and coverage 5. Calculate PFD_avg for each subsystem 6. Sum subsystem PFDs for total SIF PFD 7. Check against SIL target PFD range 8. Verify architectural constraints (HFT) 9. Verify systematic capability of all components 10. Document in SIL verification report

Common Pitfalls

PitfallImpactPrevention
Ignoring common causeOverestimates redundancy benefitAlways include β factor (0.02-0.10)
Assuming 100% test coverageUnderestimates residual PFDUse realistic coverage (70-95%)
Missing systematic capabilitySIL not truly achievedVerify SC for all components
Ignoring process demand rateMay need high-demand analysisVerify demand rate < 1/yr for low-demand
Not accounting for MTTRDD failures add to PFD if repair is slowInclude mean time to repair in PFD model

Failure Rate Data Sources

SourceCoverageUse
exida SERHSensors, logic, valves, solenoidsPrimary source for SIL calculations
OREDA HandbookOffshore equipment focusSupplementary, subsea equipment
IEC 61508 SN 29500Electronic componentsLogic solver component-level data
Manufacturer FMEDASpecific product dataMost accurate for selected hardware
SIL is not just a number: Achieving a SIL target requires more than passing the PFD calculation. The complete SIS lifecycle from hazard analysis through design, installation, commissioning, operation, maintenance, and decommissioning must follow the systematic requirements of IEC 61511. A properly verified SIF with inadequate maintenance procedures will not maintain its SIL rating over time.