Industrial AI Can Help Realize the Vision of the Self-Optimizing Plant

We wish to enable our customers’ digital transformation journey by assisting them to run their assets safer, greener, longer and faster, and to achieve these objectives simultaneously. That was the message from Ching Chye Hou, Senior Principal Solution Consultant, AspenTech, as he kicked off the company’s English Track presentation, Entering a New Era of Autonomy for Production & Value Chain Optimization, at the ARC Industry Forum Asia.

In his session, Mr Ching introduced and detailed key concepts and solutions around Industrial AI, which AspenTech believes will generate new levels of value in the capital-intensive industry it has served for many years with a suite of software solutions focused on asset optimization. In particular, Industrial AI can help realize the vision of the self-optimizing plant, a facility that provides autonomous and rapid response to changing conditions, thus enabling higher margins, safer operations, reduced emissions, and improved reliability.

For AspenTech, as Mr Ching explained, asset optimization is built around a combination of Performance Engineering, which determines what the plant can produce, Asset Performance Management, which determines the condition of the plant, and Production Optimization, which uses engineering and asset information to optimize production and fully realize asset potential. With artificial intelligence, Performance Engineering, Asset Performance Management, and Production Optimization solutions develop to become more autonomous, a necessary precursor for the self-optimizing plant explained Ching Chye Hou.

For production optimization, AspenTech has solutions from planning and scheduling down to dynamic optimization and advanced process control (APC). The latest aspenONE V12.1 release brings these layers even closer together. What this means is that while all the constituent applications have delivered differentiated value in their own way in the past, they deliver even more value now together through a range of synergies, notably sharing of model components in a “model alliance” as well as information about constraints.

Additionally, AspenTech’s dynamic optimization and APC technologies are adaptive, such that they adapt to changing process conditions over time and send estimates of changing model parameters up into the planning and scheduling layers. Bringing planning and operations into closer alignment in this way has a major impact and significantly reduces the gap between plan and actual performance. Furthermore, integrated business processes operating together in shorter timeframes enables plants to respond with greater agility as nodes in a smart enterprise’s value chain.  

Integrating AI

In August 2020, to integrate its recent investments in Industry 4.0 capabilities, AspenTech announced AIoT Hub, which enables seamless and flexible data mobility and integration across the enterprise from sensors to the edge and cloud and, importantly, accelerates the delivery of visualization and insights for capital-intensive industries.

In a layered representation, AIoT Hub has Aspen Connect and Aspen Intelligent Edge at its base, enabling connection to a broad variety of third-party data sources as well as to AspenTech solutions. In the middle is a highly scalable cloud-based system for ingesting high volumes of data, cleansing the data, and performing data processing. At the top, Industrial AI Apps are tools to build production-grade artificial intelligence applications, and the enterprise insights layer delivers cloud-based visualization and workflow.  

One of those Industrial AI Apps is Aspen AI Model Builder. While rigorous first principle models are traditionally used for modeling of process industry assets, there is now the ability to deploy machine learning-based models. However, as Mr Ching cautioned, as well as being data intensive, such models can easily disobey the law of physics and chemistry.

Accordingly, taking a hybrid modeling approach, AspenTech judiciously integrates first principles-based process simulation models and domain expertise together with AI and analytics algorithms. The result is a hybrid modeling system that delivers a comprehensive, accurate model more quickly and without requiring significant expertise. Developed models can be incorporated into AspenTech production optimization solutions, like PIMS and GDOT.

Industrial AI can also be applied in the advanced process control area. As Mr Ching explained, the traditional approach to building APC models involves multiple and time-consuming steps such that it often requires up to two weeks to build a preliminary model. Aspen Maestro for DMC3 reduces the time down to a matter of hours by applying machine learning algorithms to mine historical process data and build seed models, which are then deployed as self-learning DMC3 applications.

Sustainability and the Smart Enterprise

Sustainability is one of the key objectives for the self-optimizing plant, emphasized Ching Chye Hou, and AspenTech can help customers make a big impact in this area by bringing together sustainability information and metrics across the value chain and powering measurements of relevant KPIs. For example, the company is working with several customers on developing model-based plantwide emission monitoring systems to make emissions more visible. And through advanced control and optimization technologies, the company is helping customers minimize energy use as well as CO2 emissions. Crucially, taking CO2 into account in optimization objectives enables plants to optimize sustainability performance as well as financial performance.

To help move the self-optimizing plant from vision to reality, AspenTech is intent on building an increasingly powerful range of model types within its model alliance and unified infrastructure. These available today include a Feedstock Selection Model, a Crude Scheduling Model, and a Full Refinery Scheduling Model. Against a backdrop of owner-operators increasingly building integrated refinery and petrochemical complexes, an important one for the future is the Oil to Chemicals Model.

That interest extends outside of the plant and into managing and optimizing the oil to chemicals value chain, said Mr. Ching. Companies want to understand how to bring the petroleum and chemicals value chains together and how to develop integrated value chain optimization strategies. For example, ensure that feedstock optimization on the refinery side, takes into account effects on the downstream chemicals value chain.

Oil to chemicals value chain optimization is an example of AspenTech’s efforts to go outside the four walls of the plant and get visibility across the entire value chain to support the vision of a smart enterprise. This is built around the concept of a Value Chain Control Tower bringing the important information together in order to easily see impacts and to make decisions. When it comes to performance and sustainability, the control tower provides governance and strategies to enable operators to act effectively and manage the whole value chain and get to the sweet spot of optimizing performance and sustainability at the same time.

Ching Chye Hou concluded his session at the ARC Asia Forum 2021 with the following words: “As you can see from this presentation, the vision of the self-optimizing plant and smart enterprise is highly exciting and relevant to our industry, especially as we navigate the energy transition. AspenTech is fully committed to this vision.”