The carbon conundrum: Digital technologies drive sustainability in LNG production and transmission

Shell’s Prelude FLNG (floating LNG) vessel is the largest floating production, storage and offtake vessel in the world, weighing as much as six aircraft carriers and stretching 1,601 ft. The offshore facility, which is currently anchored hundreds of miles off the coast of Western Australia, produces natural gas, then liquefies it at sea so that there is no need for long pipelines to deliver gas to processing plants onshore.

Of much concern to vessels like Prelude FLNG and to the larger LNG industry is a new initiative from the COP26 conference in Glasgow that is significantly increasing pressure on oil and gas to reduce gas emissions for cleaner energy production. Simultaneously, governments are also passing their own regulations to limit emissions, placing new caps and restrictions on production. On multiple fronts, LNG companies are facing rising demands to improve the efficiency and the sustainability of their operations.

The challenge is to accomplish these goals while avoiding massive new capital investments, which is why many LNG facilities have already turned to new digital tools to make the most of their existing assets and systems. Many of these use digital twin technology—a complete and integrated data model of all the engineering and operations information about the facility, which can be accessed using shared platforms by everyone working on, and remotely supporting, the site.

For example, Shell uses no-code operational digital twin technology to remotely monitor everything on the Prelude FLNG vessel, from the giant chain winches in the hull to the 50 MM liters (l) of ocean water used to cool the natural gas. From Perth, Shell engineers advise operators thousands of miles away, saving the company millions in transportation costs and enabling teams to spot potential issues before they become problems far out to sea. With its digital twin program, the company is not just building cleaner operations at sea, it is building more efficient, reliable and profitable operations, as well.

Site planning with the support of digital twins. Simply put, a digital twin is comprised of individual, focused twins—the process twin combined with the operational twin, and, for example, the financial twin, the asset registry twin, the human resources twin and so on—which, together, form an integrated system of systems.

Digital twin technology has been around for many years. In this time, industry users have built up a wealth of experience with these digital tools and have developed a much keener understanding of process stimulation. Cutting-edge users are adopting a layers-of-analytics approach, which moves beyond steady-state simulation models to more advanced, dynamic models fed with contextualized, high-quality, real-time data and associated intelligence. The results of the simulation models feed back to the operational twin to be operationalized in the form of targets and forecasts, which enables further layers of analytics, such as plan-vs.-actual analytics.

A process digital twin is a replica of a plant, with information about pumps, compressors, heat exchangers and any other relevant equipment modeled to mimic the workflow of the plant. This simulation can help improve plant efficiency throughout the entire plant lifecycle, from planning to process optimization to predictive analytics. Creating a process twin is a crucial step in optimizing the liquefaction process and detecting equipment problems before failures occur. The integration of the process twin and the operational twin, along with the other focused twins, produces a truly holistic digital twin of the business, which empowers remote and on-premises operator teams with the augmented intelligence they need to optimize operations and reduce emissions.

A digital twin also helps in the planning process for a new facility. The digital twin can evaluate hundreds of cases and scenarios within minutes to find a solution that will reduce emissions from the beginning, dramatically speeding up the engineering cycle. Modern steady-state simulation models can optimize design and the same model can be used later to monitor and optimize operations. It also enables engineers to model different scenarios and equipment locations in the front-end engineering and design (FEED) stage.

Steady-state simulation models can help optimize LNG plant design. Engineers can perform studies and look at different heat exchangers, distillation columns or flare designs to see which options will deliver maximum results, while increasing plant safety and minimizing both emissions from steady-state operations and fugitive emissions from operational failures.

For example, if an engineering team needs a certain throughput capacity for a reaction vessel, and the team is debating between using one large vessel or two smaller vessels, it can model the cooling and energy costs for the different scenarios, look at the pressure on specific lines and total throughput capacity to decide how the control design and operability of the two scenarios compare.

These types of major capital decisions create significant risks. A holistic digital twin can help companies manage the risk associated with huge capital projects by providing shared data and facilitating collaboration between project managers, vendors and suppliers. Simultaneously, a layer-of-analytics approach supports quicker and more accurate decision-making, while promoting better knowledge sharing between the different teams. By this approach, streaming analytics in the operational twin integrate bidirectionally with higher-level analytics that use modeling and simulation, machine learning and other advanced digital tools to provide an accurate, single source of truth.

Reduced process variability for increased energy efficiency. Once the process twin is created for the engineering planning stage, construction is completed, and startup commences. Then, the engineering teams can hand that process twin over to the operations team to integrate with the operational twin to continue refining processes and proactively manage equipment to prevent unexpected shutdowns and equipment failures. The digital twin does this by reducing process variability, using predictive analytics to monitor equipment health and conduct overall process performance monitoring. For example, the operational twin can access pump curve data from the process twin and compare it against real-time pump efficiency. When the system detects a deviation from the best efficiency point, it issues an alert to operators so that they can address the issue immediately.

Advanced process control (APC) is used in multiple cryogenics operations to stabilize throughput and reduce variability from different feedstock compositions, ambient temperatures or equipment conditions. APC adjusts conditions and processes to manage changes in feedstock rate and quality, while also keeping plant operations within safe operating conditions and horsepower constraints.

At most LNG plants, implementing APC can deliver a 3%–4% increase in throughput and yield and a 2%–5% drop in energy consumption as measured by MMBtu/t. APC also reduces the need for operator interventions and improves plant stability.

A more proactive approach to maintenance with layers of analytics. Along with the increasing sophistication of digital twin capabilities, improvements in predictive analytics are being made, as well. This enables more precise recommendations earlier and requires less historical data to accurately anticipate failures. If a piece of equipment fails in production, it can cost tens to hundreds of thousands of dollars to replace the asset. Unplanned plant shutdowns can cost upwards of $1 MM/d in lost production revenue. However, plant overruns can quickly add unexpected costs at an even faster rate, to say nothing of possible safety implications.

Unplanned shutdowns can hit the natural gas industry especially hard because LNG trains are extremely cold during operation. If a shutdown occurs, it will require more time to restore full production than it would for operations near ambient temperatures. A layers-of-analytics approach enables the modeling of rotating equipment using machine learning and advanced pattern recognition with streaming and predictive analytics integration.

Advanced predictive analytics compares real-time operations data to the digital twin models and flags any subtle deviations from normal equipment behavior. This early warning system means that reliability and maintenance teams can identify, evaluate and address problems before a major breakdown takes place. It also means teams can make an informed assessment when a particular piece of equipment starts showing signs of wear.

TransCanada—one of the largest LNG pipeline operators in the world—calls this ‘fixing little things;” things that can have million-dollar impacts if they are neglected. TransCanada manages 31,000 mi of pipeline across North America with a very diverse fleet, ranging from small reciprocating 300-hp units to 35,000-hp turbines, while tracking more than 16,000 streams of data.

Like Shell’s Prelude FLNG, TransCanada has also adopted a layers-of-analytics approach. The company’s analytics journey began with relatively modest objectives: digital asset construction and basic anomaly detection. Once TransCanada had built a solid foundation of operational data management and streaming analytics, it began to evolve its analytics overlay with more advanced algorithms and event frame generation to monitor key performance indicators (KPIs) and performance metrics. By 2017, these layers of analytics had detected 129 anomalies and netted more than $10.65 MM in prevented expenses.

Maximizing throughput with real-time performance monitoring. There is one other key area where digital twins can reduce the carbon footprint for LNG operations: process performance monitoring. From the beginning, an LNG plant is designed using APC to operate at maximum safe efficiency based on the characteristics of the assets, plant layout and production pipeline. However, as soon as operations start, variations in feedstock and fuel gas composition or changes in ambient conditions inevitably impact throughput, which can then affect operating profits.

Performance monitoring studies live operations to resolve non-linear relationships that may impact plant throughput, including process constraints, feedstock and mixed-refrigerant composition, to reconcile material use and look at energy and equilibrium balances around a piece of equipment. With better tracking of real-time performance, pipeline companies can also identify inaccurate instrument readings. This acts as a data validation layer, while also pinpointing material and energy loss locations so operators can quickly find the source of any fugitive emissions that occur.

More-efficient LNG, a more sustainable future. LNG producers and pipeline operators are facing demands to make their operations greener and more sustainable. Simultaneously, they must maximize performance, minimize costs and delays, and ensure efficient organizational operations. To achieve all these aims at once, natural gas companies need to synchronize every facet of their business, from planning and engineering to operations and onward. By reducing emergency flare-ups and unplanned shutdowns, a holistic digital twin enables producers to thread that needle—reducing energy usage and emissions while increasing performance and efficiency all at the same time. By embracing a layers-of-analytics approach, LNG companies can help ensure efficient use of the world’s precious resources. GP