PTQ Q3 2022 Issue

past parameters for current operations. There is no guaran- tee that the current working solution is the most efficient; it could just be the set of parameters that turned a profit. Without a way to simulate the assets being optimised, the experimentation required for production optimisation is sig- nificantly hindered. Using digital twins to model and predict how an asset will respond to various conditions can increase efficiency and help plan for extraordinary situations, espe - cially when deployed for operation training (OTS). A digital twin needs to be created with knowledge about the fundamental physical properties of the asset being modelled (see Figure 1 ). A few of the leading modelling suites, such as Aspen Performance Engineering, incorporate advanced properties systems, such as from the US NIST repository, that allow for the creation of accurate digital twins. Digital twins that combine this with AI, in hybrid mod- els, such as the patented Aspen Hybrid Models, supplement this knowledge with analytics from current and historical performance of the assets. An extensive suite of industrial sensors to collect data from the asset being digitally twinned is required to ensure the digital twin provides an accurate reflection of the asset. This data can be stored and input to the model afterwards or networked to stream the information to the digital twin in real time. In a production optimisation environment, the rigorous models can dynamically update the planning and schedul- ing models for higher accuracy and mapping to the current plant conditions. Schedulers can use them and the results they generate to ‘operationalise’ the plans the refinery plan - ner has delivered, taking into account real-world constraints. Shared models in data, never before accessible, unify the plan and schedule, leading to an easily adjusted and fully aligned schedule. At the operating level, teams work together to leverage multivariable process control centrally to run the plant closer to yield and energy limits. Adaptive self- healing technology maintains peak performance, increasing throughput by up to 5% in each unit. Beyond the improvement in each unit is the central jewel of unified production optimisation: dynamic optimisation that autonomously reconciles real-time data and applies it at the appropriate time to achieve new performance benchmarks across wide functional areas. This coordinates and adjusts the underlying advanced process control systems (APCs) to help refiners match and even exceed the plan. At the same time, it provides feedback to planners and schedulers to create a closed loop for continuous improve- ment. Digital models can simulate a variety of scenarios; adjusting this parameter or altering this arrangement may make sense in certain market conditions, but what happens if the price of the product drops? Operational teams can review many options quickly to identify the best path forward. The simulation can also help assuage any misgivings about data-based decision-making. Showing a living representa- tion of proposed changes may be more reassuring than sim- ply communicating, for instance, that the compressor should be running at 10% more than the normal pressure. Tightening up the integration between planning and sched- uling does help reduce operational margin leakage. However,

Process of efficiency Every industrial process will require a different approach, and each facility may have different requirements. There is no one-size-fits-all solution for production optimisation, at least in terms of the actual changes that need to be imple- mented. Companies should be prepared to undergo an itera- tive process; the changes and tweaks made at an early stage of optimising production may reveal further refinements and sources of waste. Some companies have changing feedstock while others function on long-term contracts, some are on an accelerated pace to incorporate biofeedstocks, some have started integrating distributed renewable power, and so forth. This is where process simulation software can be a huge boon for optimisation. By creating a digital model of the pro- cess being optimised, companies can explore the impact of changes in a simulation instead of risking a loss of produc- tion by actually making changes to an asset. With unified production optimisation, refinery planners can, for example, take into account selected crudes and committed product deliveries to develop an optimised operations plan. The digital twins can inform planning models through rap- idly updated unit models that account for fouling and other changes. They can calculate key sustainability metrics and inform the advanced control systems (APC). And when digi- tal twins are paired with APC, optimisation occurs. BPCL in India has won several sustainability awards for its combined digital twin and APC implementation, which help manage carbon emissions proactively while improving energy use and margins. Digital twins to the fore Digital twins are increasingly important to the planning and scheduling process within a modern refinery or olefins plant. The rationale behind their use is clear. Digital twins have a vital role to play in this context. Any company operating industrial processes that involve valuable assets and equipment must contend with the prob- lem of how to increase efficiency while keeping production online. Adjusting the operating conditions of an asset risks disrupting production, which can lead to an overreliance on Figure 1 Digital twin needs to be created with knowledge of the fundamental physical properties of unit-level assets

PTQ Q3 2022