APM, AI, SOP, predictive, prescriptive maintenance

Cake, Pie and Process Simulation

Revolutionizing Process Simulation With AI

May 27, 2021

What do cake and pie have to do with process simulation?

You probably never thought that you would be seeing these words in the same sentence. But let’s try this: “process simulation is a piece of cake,” or “process simulation is easy as pie.” Have these combinations of words crossed your mind before? 

Maybe. But can you say the same thing about tuning your simulations to existing operations? While some might consider it easy, it can also involve multiple steps to calibrate reactors, many iterations to tune equipment efficiencies, or numerous regressions to obtain kinetic parameters of complex reactions. And while this is not typically considered easy, it can be a piece of cake. 

As process engineers, we want our models to represent what is going on at the plant so we can make good decisions. But catalysts in reactors degrade, heat exchangers get dirty and foul, equipment ages or its performance can be affected by mechanical interventions. And in a VUCA (volatile, uncertain, complex and ambiguous) world, engineers need to take this into account and to have real behavioral information readily available to support high value decisions in the shortest time possible.

AspenTech is uniquely positioned to help with these challenges. With Aspen Hybrid Models™ and the recent addition of First Principles Driven Hybrid Models, process engineers can use AI to easily tune simulations to reality and make quick and accurate decisions that will make the difference in this dynamic environment. And making model calibration easier than ever before. 


AI is helping process engineers 

At OPTIMIZE 2021 we listened to global industry leaders from Dow, Process Ecology, Petrorabigh and others talk about how they are using Aspen Hybrid Models to solve complex and difficult problems that could not easily be solved before. You can still watch the recordings until June 30, where you will hear from many others who also talked about incorporating Industrial AI as the next step to improve their plant Digital Twins. 

Aspen Hybrid Models is a unique solution combining AI, first principles and domain expertise. As part of the solution, we introduced AI driven hybrid models to quickly help engineers create new models using data from operations, and Reduced Order Hybrid Models to help transform complex models into robust simplified simulations that can easily be used in multiple applications. This demonstrates how AI can easily help process engineers address many important challenges. 

But can AI also help with tuning models to reality? The answer can be found through First Principles Driven Hybrid Models.  


Revolutionizing process simulation with AI-enabled Model Calibration 

With our most recent release of aspenONE V12.1, we are expanding our Aspen Hybrid Models with First Principles Driven Hybrid Models. This new technology, that is changing how we simulate processes, embeds machine learning inside Aspen Plus® and Aspen HYSYS® to automate model calibration. 

With First Principles Driven Hybrid Models, machine learning algorithms use data to learn to easily calibrate Aspen Plus and Aspen HYSYS models, providing a new solution to quick and accurate modeling without manual and time-consuming iterative tuning. 

First Principles Driven Hybrid Models provide AI-enabled automatic model calibration so that engineers can get higher value from their simulations and make more agile and informed decisions in less time. And with this revolutionizing technology, they can quickly expand these benefits to a wider scope of operations that could not be easily and accurately modeled before. 



First Principles Driven Hybrid Models in Action

Less than two weeks after V12.1 was released, Nissan Chemical Corporation was presenting at OPTIMIZE 2021 how to use First Principles Driven Hybrid Models to create a rapid an accurate steam reformer model for an ammonia process.  

While conventional rigorous reactor modeling for steam reforming requires process fluid temperature profiles, it is difficult to estimate or measure temperature in the furnace. They used First principles Driven Hybrid Modeling combining Aspen Plus First Principles model knowledge and machine learning from data for unknown phenomena to tune reaction rates. Using neural networks, they produced a more accurate model in less than 50% of the time. 


“Using the hybrid model, we were able to create a model that can reproduce real plant data more accurately than the conventional reformer model. We were able to create a highly accurate model in a short period of time.” 

– Mr. Takuto Nakai, Production Department, Nissan Chemical Corporation


Nissan Chemical Corporation estimates that with this model they could save 1% steam use in the reformer, a significant amount for an energy intensive process. And they are looking to apply this modeling technique to other units in the process to expand the benefits. 

Just as with Nissan Chemicals, AI-enabled model calibration using First Principles Driven hybrid models provide great benefits and is a necessary tool for engineering leaders during current market dynamics. 

With AI- embedded capabilities in Aspen Plus and Aspen HYSYS, we are revolutionizing process modeling and creating a new world of possibilities for process engineers. Are you ready to embrace Aspen Hybrid Models as part of your digitalization journey? It should be a piece of cake.


To learn more, visit our solutions page here




There was a problem storing your subscription

Leave A Comment