TOMSK, May 18. /TASS/. A model which allows to evaluate sites for geological storage of carbon dioxide with high accuracy has been created by scientists at Tomsk Polytechnic University (TPU). This will allow to select facilities faster and better, which will reduce the concentration of carbon dioxide in the atmosphere more effectively, the press service of the university said on Thursday.

Geological storage of carbon dioxide is noted to be a promising method of reducing the concentration of anthropogenic emissions in the atmosphere, which is widely implemented in international practices. During long-term storage, the share of free gas in the reservoir decreases as it transitions to a bound state due to interaction with rock particles. This is one of the important indicators for assessing the safety of this type of storage, but it is very difficult to fully model this due to the large number of parameters.

"Scientists from the Heriot-Watt Center at Tomsk Polytechnic University have developed a method for predicting the dynamics of carbon dioxide binding when injected into deep aquifers for long-term storage. The model proposed by the polytechnicians takes into account a large number of process parameters and has a high accuracy of prediction. This will simplify and accelerate the estimation of objects when selecting storage tanks," says the report.

The high accuracy of the forecast was achieved due to the large volume of the training sample and the detailed design of the experimentation plan. "Our model receives 5,450 sets of input data. It then determines the relationship between the variables and the outcome and learns to predict similar dependencies on the new data. Once trained, the model is able to perform target predictions with high accuracy," Shadfar Davoodi, a research engineer at the Heriot-Watt Center, is quoted as saying.

In the future, the scientists plan to further improve the quality of the forecasting model by optimizing the settings of the algorithm and applying a new methodology for preprocessing raw data. The research is supported by the Priority 2030 program.