Skip to main content

Giving value a whole new meaning

Published by , Editor
LNG Industry,

Jason Chadee, Director at SparkCognition, USA, details how artificial intelligence can help bring a new meaning to terminal value.

The volume of LNG traded globally has quadrupled in the last two decades, and is set to double in the next two. The development and operating costs associated with the value chain have shifted due to an abundance of low-cost gas supply from the shale revolution, which unlocked vast natural gas reserves previously inaccessible or economically unviable to extract. This significant increase in natural gas supply has led to lower domestic natural gas prices and allowed for more extensive LNG production to meet global demand.

There are currently seven operating LNG terminals that can collectively export 12% of all US gas production. Japan purchased 98.3 billion m3 of LNG in 2022, making it the world’s most prominent LNG importer. China fell to second place with 63.44 million t of LNG imported in 2022. Europe is also a net importer of LNG, with 28 large scale LNG import terminals, including non-EU Türkiye.

The continued viability of the LNG market in the US is based upon a complex supply chain involving the most extensive natural gas production, pipeline infrastructure, gathering, processing, and storage systems in the world.

The industry is under immense pressure to maximise production and energy efficiency, productivity, safety, and sustainability at existing terminals, requiring significant agility and tight controls between supply chain activities. There is immense pressure on infrastructure and operational procedures for the safe arrival and departure from terminals and, when necessary, safe aborting entry or egress manoeuvres during an on-board or onshore emergency. Risks related to collisions, groundings, contacts, fire, and explosion on board – and, if necessary, the release of gas and any other deleterious consequences – all need measurements and evaluations.

Terminal and port authorities require explicit details on the total safety level of LNG shipping operations as it relates to existing infrastructural and surrounding ship-ping movements.

Digital transformation has become a critical strategic priority for LNG infrastructure companies. They have adopted and implemented sensors, analysers, and control and information systems – a network of technologies designed to collect real-time data on various parameters such as temperature, pressure, flow rates, equipment health, and safety conditions – generating massive amounts of data.

The challenge: Too much data

Digital technologies present unique opportunities for the LNG industry. However, challenges still need to be solved in the massive amounts of data being produced that need to be analysed, leaving massive gaps in optimisation. Most companies engaged in digitisation have lost control of their data – or never had control of it in the first place. They are inadvertently stockpiling massive amounts of data in unstructured and structured repositories, keeping it indefinitely, and bleeding it out through accidental loss, careless but well-intentioned sharing, unfettered collaboration, and insider theft. Without harnessing it, companies are oblivious to what they have, who is using it, how it is being used, or why. Engineers are estimated to spend around 50% of their time looking for data they need and end up using only about 10% of the data being gathered.

LNG terminals are more complicated than ever before, yet time for calculating and estimating terminal behaviour is shorter. The overwhelming abundance of data and the persistence of elusive physical laws to explain the complexity of assets and operations promote a renowned interest in more powerful technologies to extend current model capabilities and decision workflow practices.

The solution to the data problem: Artificial intelligence

To fully capitalise on the digital technologies being adopted by LNG companies and leverage the data that will truly drive the digital transformation, the industry has turned to artificial intelligence (AI).

AI can aggregate multi-source data and visualise it all in one area, making it a technological game changer. AI can predict equipment failures, optimise processes, and identify opportunities to improve terminal operations. Companies can then move away from time and tactic-based activities to proactive and predictive management of terminal assets, thus improving safety and reliability.

AI removes what was previously limited to the judgment and limits of human cognition and evaluation. It is not to say humans can be removed from the equation. Only subject matter experts (SME) can evaluate whether the definitive correlation found by the data analysis is a possible phenomenon.

The technology

SparkCognition’s Industrial AI Suite (IAS) uses advanced model-building techniques to reduce AI predictive maintenance deployment times to just weeks or days. With a wide range of its capacity to ingest and analyse new data, IAS scales to suit increasing workloads and changing business requirements. The continual learning and adaptive algorithms capture SME knowledge with natural language processing (NLP) technology that extracts insights from unstructured data – even sparse, unlabelled, and dirty data. AI can discover patterns, improve models, and reduce time to resolution with faster root-cause analysis, avoiding model drift with normal behaviour modelling (NBM) techniques.

Normal behaviour modelling is an automated AI/machine learning (ML)-enabled anomaly detection methodology for evaluating and describing the behaviour of a system or piece of equipment under normal operational and environmental conditions. NBM models ingest large volumes of quantitative time-series data (temperature, pressure, flow rate, etc.) from multiple sensors, both initially for training purposes and continually thereafter for ongoing monitoring and periodic system retraining. Once trained to understand the quantitative characteristics that define ‘normal’ for the system being monitored, the model continues to evaluate the incoming sensor-provided data stream and generates alerts whenever an out-of-normal condition is detected. Managers and technicians can then use these alerts to undertake maintenance and repairs of the system more proactively than doing so only upon system failure. As a result, an organisation saves time and money and improves the overall productivity and safety of the system, the facility in which it operates, and the workers who interact with it.

Enjoyed what you've read so far? Read the full article and the rest of the December issue of LNG Industry by registering today for free!

Read the article online at:

You might also like


Embed article link: (copy the HTML code below):


This article has been tagged under the following:

Natural gas news UK LNG news