Company Background – Chevron
Predictive Quality Stability & Remaining Useful Life (RUL) Optimization
SCG Chemicals Public Company Limited (SCGC) operates large-scale petrochemical assets, including steam cracker units that form the backbone of olefins production. Cracker run length directly determines plant throughput, maintenance timing, and overall profitability.
Unplanned or poorly timed shutdowns of a cracker unit can result in substantial production and margin losses.
Challenges
Variability and the Uncertainty of Equipment Lifespan
The TPE operation faced two challenges: product quality variability in Methanol levels and uncertainty about the Remaining Useful Life of critical equipment. Data fragmentation hindered visibility into process interactions, forcing operators to manually extract data. This lack of modeling led to reactive responses to lab deviations, causing process disturbances and yield losses.
The TPE operation faced two interconnected challenges:
- Product quality variability, particularly fluctuations in key chemical properties (e.g., Methanol levels in PY), leading to off-spec risk.
- Uncertainty in equipment Remaining Useful Life (RUL) due to gradual performance degradation.
Solution
A Predictive Quality and Asset Health Framework
SCG TPE implemented a predictive quality and asset health framework by integrating multiple plant historians into Seeq for centralized analytics and self-service model development.
01
Historian Integration
To create a unified, contextual analytics environment, real-time data streams—including reactor temperatures, pressures, feed compositions, flow rates, product quality lab measurements, and equipment condition indicators—were seamlessly integrated into a single platform. This completely unified the process, laboratory, and equipment datasets.
02
Self-Service Analytics
Rather than reacting to lab deviations after the fact, process engineers independently built multivariable models to identify dominant quality drivers. By leveraging good-run versus bad-run classifications, statistical boundary monitoring, and correlation mapping for Methanol variability, operations gained crucial predictive visibility into quality instability before it occurred.
03
Operational Optimization
Plant engineers utilized self-service analytics to analyze continuous degradation trends and dynamically estimate the RUL of critical equipment. By developing rate-of-change degradation indicators and threshold forecasting models, the team enabled proactive maintenance scheduling, ensuring interventions occurred precisely before performance or product quality was negatively impacted.
Benefits
Driving Yield, Stability, and Reliability
Operational Efficiency
- Reduced process variability
- Minimized reactive operating adjustments
- Improved stability of reactor conditions
Asset Reliablility
- Early detection of equipment degradation
- Improved maintenance planning
- Reduced unplanned interruptions
Sustainability
- Reduced material waste from off-spec batches
- Lower reprocessing and energy consumption
Organizational Impact
- Eliminated manual historian data consolidation
- Empowered engineers with self-service predictive modelling
- Established repeatable quality analytics methodology
Product Quality & Yield
- Reduced off-spec production risk
- Improved yield consistency
- Stronger compliance with customer specifications
Cost Saving
Quantifying the Value of Yield and Uptime
Reduced off-spec production, improved yield stability, and optimized maintenance timing generate substantial financial impact.
For a detailed breakdown of yield improvement valuation, avoided waste modelling, and maintenance optimization savings, connect with us to review the structured cost-saving assessment framework.
