Data Analytics Leveraged in Designing Flaring-Reduction Strategy

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 226368, “Leveraging Data Analytics To Design an Overarching Flaring-Reduction Strategy,” by Farras Sailendra and Bugi Setiadi, SPE, BP. The paper has not been peer-reviewed. _ This paper details a data-driven methodology applied at the Tangguh liquefied natural gas (LNG) facility in Indonesia to enhance flare-emission visibility and enable targeted reduction strategies. By integrating real-time process data with engineering models, flare contributions were quantified at the valve level. A phased implementation roadmap was developed to reduce flaring through three strategic focus areas: optimizing startup sequences, rationalizing purge-gas consumption, and adjusting operating parameters during LNG loading and dryer-bed changeover. Introduction Tangguh LNG, in Bintuni Bay, Indonesia, is a major facility with a total production capacity of 11.4 million tons per year across three processing trains. Traditionally, flare monitoring at Tangguh relied on flowmeters installed at the knockout drums of each major flare system (dry, wet, and tankage headers). While this provided a high-level summary of flare volumes, it offered little insight into the specific equipment or events contributing to flare load. As a result, opportunities for flare reduction remained largely reactive and untargeted. A 2022 internal audit revealed that over 60% of flare volume originated from preventable sources. Because of the lack of granular, real-time data, however, plant teams were often forced to conduct manual retrospective investigations. Methodology and Analytical Framework This section of the complete paper outlines the structured analytical approach used to transform raw process data into actionable flare-reduction strategies. The methodology follows the data-information-knowledge-insight-wisdom framework. This hierarchy ensures a systematic progression from unstructured information to strategic action. From Data to Information: Tagging and Labeling Flare Contributors. Initially, these valves existed only as tag numbers, unstructured and without clear contextual meaning. The first transformation occurred when each valve was systematically labeled based on its function and flaring behavior. This was achieved through process knowledge, operator input, and pattern analysis from historian trends. The result was the creation of an information layer. Each valve now carried a functional identity. The labels were categorized into the following operational flare types: - Safety-related flaring - Routine flaring - Nonroutine flaring - Startup and shutdown flaring To support this transformation, the flare network was visualized as a node-link diagram in which each was represented as a node and color-coded based on its assigned flare category. From Information to Knowledge: Grouping Contributors into Functional Categories. Once each valve in the flare network had been labeled based on its flaring behavior, the next step was to organize this information into structured, functional categories. This involved grouping the identified valves into “rooms” that represented distinct flare contributors correlating with the four previously described operational flare types. By visualizing each group of flare contributors as a contained cluster, determination of which valves collectively contributed to a specific type of flaring event was possible.