Abstract
Background information
Natural gas liquid (NGL) production facilities, typically, utilize turbo-expander-brake compressor (TE) to generate cold for C2+ separation from the natural gas by isentropic expansion of feed stream and use energy absorbed by expansion to compress residue gas.
Experience shows that during operational phase TE can exposed to operation outside of design window that may lead to machine integrity loss and consequent impact on production. At the same time, there is a lack of performance indicators that help operator to monitor operating window of the machine and proactively identify performance deterioration.
For instance, TE brake compressor side is always equipped with anti-surge protection system, including surge deviation alarms and trip. However, there is often gap in monitoring deviation from stonewall region. At the same time, in some of the designs (2×50% machines) likelihood of running brake compressor in stonewall is high during one machine trip or train start-up, turndown operating modes.
Also, typical compressor performance monitoring systems does not have enough dynamic parameters that may indicate machine process process performance deterioration proactively (real-time calculation of actual polytrophic efficiency, absorbed power etc.) and help operator to take action before catastrophic failure occurs.
In addition, typical compressor monitoring systems are based on assumed composition and fixed compressibility factor and do not reflect actual compositions variations that may affect machine performance monitoring.
To overcome issues highlighted above, Hawiyah NGL (HNGL) team has developed computerized monitoring and advisory system to monitor the performance of turbo-expander-brake compressor, proactively, identify potentially unsafe conditions or performance deterioration and advice operators on taking necessary actions to avoid unscheduled deferment of production.
Computerized performance monitoring system has been implemented in HNGL DCS (Yokogawa) and utilized by control room operators on day-to-day basis.
Real-time calculation, analysis and outputs produced by performance monitoring system allow operator to understand how current operating condition are far from danger zone. Proactive deviation alarms and guide messages produce by the system in case of deviation help operators to control machine from entering unsafe region. Actual polytrophic efficiency, adsorbed power calculations provide machine condition status and allow identifying long-term performance deterioration trends.
Machine Learning Solutions for Bridge Scour Forecast Based on Monitoring Data