Researchers from Heriot-Watt University and the Swiss Federal Institute of Technology Lausanne (EPFL) have generated a machine learning algorithm to more accurately forecast amine emissions in CO2 capture plants.
The algorithm aims to provide chemical engineers with more insights into the health hazards a plant may pose when capturing CO2 emissions.
Current CO2 capture processes have a success rate of 50-68% but scientists predict that number could reach more than 90% in the future. One of the downsides of the carbon capture technology currently is that it releases amines which can be harmful to the environment.
In an effort to provide an answer on why and when amine emissions happen during the CO2 capture process, the group of scientists designed the machine learning algorithm and tested it at Germany’s large coal-fired power plant Niederhauẞen.
Their findings were published in a research article for the academic journal Science Advances.
“We developed an experimental campaign to understand how and when amine emissions would be generated. But some of our experiments also caused interventions of the plant’s operators to ensure the plant was operating safely,” Professor Susana Garcia, associate director of carbon capture and storage at Heriot-Watt University, told the university press.
The scientists created a pattern-recognition algorithm to predict emissions caused by interventions and ones caused by stress tests. The model would allow operators and scientists to run scenarios to more accurately predict emissions, and work towards reducing them.
This research has the potential to have an enormous impact, said Garcia. The scientists were able to successfully forecast amine emissions from CO2 capture plants that use the most advanced amine-based technologies. “This type of forecasting proved to be very challenging with any of the conventional approaches, so it may change the way we operate industrial plants,” she also said.