Chemicals and materials / Basic chemicals
Evonik Oxeno: Recommending chemical plant operating conditions using precomputed simulations
- Company
- Evonik Oxeno
- Country
- Germany
- Adoption stage
- In operation
- Source published
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
Not every process inside a chemical plant can be observed directly. When feedstock streams change, the most efficient operating conditions change too, but conventional process simulations are complex and slow to compute, making them difficult to use immediately in operational decisions. Evonik Oxeno wanted to connect process knowledge to actual operating decisions and reduce energy use.
Technology and data
The COMPAS project precomputed simulations of plant operating conditions and used the results to train AI surrogate models. Evonik said the simulation library contains more than one million runs. A gray-box approach combines process knowledge of thermodynamics and kinetics with trained models to produce realistic operating recommendations within seconds from previously calculated scenarios. The company said it uses the technology in live operations in Marl, Germany, and Antwerp, Belgium. Automatic transfer of model results to the control system is included among additional implementations, so this does not mean fully autonomous control has been completed.
Results
Evonik Oxeno said close agreement between models and actual operating data enabled further process optimization, and initial applications reduced the energy needed to manufacture products and the associated carbon dioxide emissions. It also reported faster decisions when feedstock streams change. The company is gradually extending the application scope while reusing knowledge built by operations, engineering and research teams.
Limits and open questions
The source gives no percentage reduction in energy or emissions and no comparison period by process. Automatic control-system integration and an entirely self-optimizing plant remain follow-on work being expanded.
Sources
Compiled from public sources. These are not results from ATF Works customers.