ABSTRACT

The Woodside-operated Karratha Gas Plant (KGP) is a large-scale integrated gas production system located in Karratha, Western Australia. Producing liquefied natural gas (LNG), domestic gas, condensate and liquefied petroleum gas (LPG) through five LNG processing trains; two domestic gas trains; six condensate stabilisation units and three LPG fractionation units. Woodside actively pursues opportunities to reduce greenhouse gas (GHG) emissions in operations, including the use of data analytics techniques to inform our operations teams on plant energy efficiency optimisation. Two examples of data analytics application in LNG plant energy efficiency optimisation are presented: (1) power generation config explorer – an analytical and logic solver model with ‘now-casting’ capability and (2) a live plant-wide energy efficiency metric with built-in thermodynamic calculation functionality. The power generation config explorer tool is an advisory application which provides a recommended operation config (number of genera-tors and type) to meet operational constraints, maximise energy efficiency and reduce GHG emissions. The tool uses machine learning techniques to overcome the challenge of predicting reactive power in a complex alternating current (AC) power network and a logic solver to mimic advanced process control behaviour. Core to energy management is accurate measurement of energy consumption and energy production. The conversion of LNG product ‘in-tank’ to an energy equivalent basis is a common challenge due to the need to correct for boil-off gas losses. A data analytics approach has been applied using live plant data integrated with thermodynamic equation of state calculation and numerical optimisation methods to account for heat losses and other uncertainties.