AI and simulation: better together
Written by Andrea Macri’, Senior Product Manager, AVEVA
Data-driven artificial intelligence (AI) models and first-principles simulations were once considered distinct branches of engineering. But as technology has become more accessible, and industry has shifted focused toward higher efficiency, sustainability, and circular economies, these distinct disciplines are merging so industry can take advantage of their synergies.
When used together, AI expands the value of simulation by answering questions and solving problems that typically require extensive engineering hours to discover. While first-principles models are highly accurate across a wide range of operations, data-driven models make them more efficient and easier to deploy. AI used in concert with simulation software creates a powerful combination that proactively improves processes and reduces costs.
AI and digital twin technology
Industrial operators may feel pressured to design new facilities to justify their CAPEX and OPEX costs, when they could make existing plants more profitable and more sustainable to operate. AI increases the effectiveness of simulation, generating benefits at different stages of the plant life cycle, from design to operation. You can also use AI tools to develop a digital twin—a virtual replica of your plant and its assets that acts as a surrogate for simulation.
For instance, one of our customers in Asia wanted to minimize CAPEX by optimizing the design of their compressor anti-surge system. Using a first-principles model alone only permitted a few hundred dynamic scenarios and could identify an initial best run design, but lacked the capability to truly explore all possibilities. Instead, we helped them use the initial results from the first-principles simulation to train an AI surrogate to test hundreds of thousands of additional cases in just seconds. This produced a design that proved 12% cheaper than the initial best run.
Faster, more accurate operational intelligence
By combining AI and simulation in the same models, companies generate more value than by relying on high-fidelity models alone. A user in Europe was changing feed once per week, which took approximately eight hours to stabilize before production resumed, meaning the plant lost several days of production per year. We knew AI could improve their operation, but the simulation agent required more than 30,000 simulation runs. By concurrently running multiple instances of a dynamic simulation model on a digital twin, we were able to train the agent in just over a month—half the time it would’ve taken human operators.
The perfect blend for operational efficiency and cost savings
As my Italian grandma used to say, the secret to prepare the perfect dish is in the quality of the ingredients (love is one of them), and the chef’s ability to combine them. The same is true for AI and simulation. The ingredients in this case are:
- Access to big data
- Simulation capabilities
- AI technology
- Cloud maturity
AVEVA is an industry leader in the simulation space, and we are experts at mixing these ingredients. We have delivered more than 1,300 multipurpose dynamic simulators across all industrial verticals. We’ve invested in predictive analytics for over 13 years with patented AI technologies that include engineering, operations, and maintenance. Our cloud-native solutions allow you greater flexibility, ease of use, and cost savings that let you to use only what you need to improve your operational intelligence. To learn more about how AVEVA’s simulation solutions enable process innovation, read our white paper.