The Smart Water Leakage Advisor is an AI-powered leakage detection application for safe recovery boiler operation, combining traditional mass balance calculations with machine learning algorithms to predict boiler conditions and detect abnormal process conditions associated with water or steam leaks. The analysis is based on standard recovery boiler instrumentation, with optional additional sensors to improve detection accuracy.
The Challenge
Recovery boilers are among the most critical and hazardous units in a kraft pulp mill. A water or steam leak in a recovery boiler — if undetected — can lead to a smelt-water explosion, posing severe risk to personnel, equipment, and production continuity. Recovery boiler shutdowns triggered by suspected leaks are costly in lost production time and chemical recovery capacity. Traditional mass balance-based leak detection approaches can generate false alarms or miss developing leaks that do not yet show in conventional measurements, making reliable early detection a persistent challenge that directly affects both safety and operational efficiency.
The ANDRITZ Solution
The Smart Water Leakage Advisor (WLA) is a Metris platform application that provides AI-powered leakage detection for recovery boiler operation. It combines traditional mass balance calculations — the standard method for detecting water imbalances in the boiler — with machine learning algorithms that predict normal boiler operating conditions and identify deviations associated with water or steam leaks. The result is a detection system that is both more sensitive and more specific than mass balance alone, reducing both false alarms and missed detections.
How It Works
The Smart Water Leakage Advisor continuously monitors recovery boiler instrumentation data, building a predictive model of normal process conditions across the boiler's operating range. The mass balance calculation provides a quantitative measure of water imbalance, while the machine learning layer analyzes the full combination of process signals to detect abnormal patterns that may indicate a developing leak — including conditions that do not yet register in the mass balance. The system generates alerts when it identifies conditions consistent with water or steam leakage, enabling the operations team to take preventive action before a developing leak escalates. The analysis is based on standard recovery boiler instrumentation already present in the plant DCS. Optional additional instrumentation — where available — can be integrated to further extend detection coverage and improve accuracy.
Key Benefits
- Combined mass balance and machine learning approach reduces both false alarms and missed leak detections compared to mass balance alone
- Earlier identification of developing water or steam leaks supports safer recovery boiler operation
- Based on existing recovery boiler instrumentation — no mandatory additional hardware required for deployment
- Optional additional instrumentation can be integrated to improve detection coverage and sensitivity
Integration & Compatibility
The Smart Water Leakage Advisor integrates within the Metris platform and connects to standard recovery boiler instrumentation through the plant's DCS. It is designed for kraft pulp mill recovery boiler applications. Optional additional sensor interfaces, alert configuration, and model training requirements are confirmed during implementation.
