1. Introduction: A Critical but Elusive Need
Everyone agrees: petrochemical enterprises must build a safety instrumentation failure database. But how exactly should it be done? Few have answers. Some industry experts stress the need for such a database during inspections, yet offer no actionable path forward. Meanwhile, frontline workers hold piles of failure records without a clue how to start, and managers are left puzzled over vague industry standards. This is not a single company’s dilemma—it is a shared pain point across the petrochemical sector.
When “data” becomes the lifeline of plant safety, the industry finds itself stuck in three key obstacles:
Tight Deadlines: Regulatory pressure with no clear roadmap.
Fragmented Records: Logs, reports, and systems scattered, with high integration costs.
Standard Ambiguity: Unclear technical criteria and lack of practical templates.
2. Root Causes of Failure: Analytical Methodologies
2.1 Correlation-Based Failure Analysis
Used for individual components or relatively simple systems, including:
Manufacturing process tracing.
Failure modes and patterns.
4M analysis (Man, Machine, Material, Method).
2.2 Systems Engineering Methods
Treats the system and human factors as one entity:
FTA (Fault Tree Analysis)
ETA (Event Tree Analysis)
Feature-Factor Mapping
Failure Rate Forecasting
FMEA (Failure Modes and Effects Analysis)
2.3 Statistical Analysis
Utilizes mathematical tools and software to identify trends and predict failures across large datasets.
3. Steps to Build a Failure Database
3.1 Data Collection & Sorting
Aggregate operational data from instrumentation, including:
Failure logs
Maintenance records
Inspection reports
3.2 Database Design
Develop a structured system that defines:
Input and output processes
Storage format
Query protocols
3.3 Data Updating & Maintenance
Ensure long-term validity through:
Regular updates
Integrity checks
Backup and recovery mechanisms
4. Regulatory Guidelines and References
GB/T 20438 (Functional Safety)
Section 7.4.9.4/7.4.9.5 outlines data needs for random hardware failures.
Encourages use of site-specific data when available (≥70% confidence level).
Notes the high dependency on operational environment and lifespan (e.g., temperature-sensitive capacitors).
GB/T 16855.1 (Mechanical Safety)
Section 6.2.11.7 calls for safety functions via programmable electronic systems.
Focuses on minimizing random hardware failure in control systems.
5. Case Study: Petrochemical Enterprise Implementation
5.1 Project Goals & Strategy
Core Focus: Database as foundation for SIL assessments.
Modeling Tools: Bayesian stats, FMEDA, benchmarking with EXIDA.
End Goal: Predictive maintenance and RCM-based management.
5.2 Instrumentation Scope
Covers departments in refining, ethylene, fertilizers, rubber, catalysts, utilities, etc.
5.3 Equipment Ledger Structure
Due to complexity, a four-tier categorization was used:
Level 1–4: Device type → Function → Control logic → Parameter type
5.4 Failure Data Architecture
Includes key attributes:
Device type, tag, manufacturer, model, unit
Failure time, mode, cause, frequency
Collected over 3,000 cleaned entries from 1,000+ raw failure logs.
5.5 Failure Dictionary & Classification
Failures are divided into four categories:
Main unit
Accessories
Wiring
Piping
Each with subtypes and symptoms (e.g., output drift, freeze, time delay).
5.6 Safety vs Dangerous Failures
Failures are categorized as:
Safe (λs): No impact on safety or function.
Detected Dangerous (λdd): High operational risk.
Undetected Dangerous (λdu): Minor impact due to current monitoring limitations.
6. Failure Rate Calculation Framework
For a certain device type with runtime T and quantity N:
Component failure counts: N1–N4
λ1–λ4: Failure rates by type (main, accessories, wiring, piping)
λs1–λs4, λd1–λd4: Safe/Dangerous breakdown per category
Calculations:
Safety Weighted Sum (E1): E1 = Σ(λi × Ni × λsi)
Dangerous Weighted Sum (E2): E2 = Σ(λi × Ni × λdi)
7. Final Thoughts: Not Just for Show
Creating a safety instrumentation failure database is not a checkbox exercise—it’s a vital safety revolution. It’s the bridge from reactive compliance to proactive risk control. Despite its complexity, it remains the only path forward for safer and smarter petrochemical operations.