SK gas, a natural gas provider affiliated with South Korea's SK Group, claimed to have developed the world's first artificial intelligence-based chemical process yield prediction and simulation system that can replace the conventional method of detecting failure factors through numerous hypotheses.
The system can optimize propane dehydrogenation (PDH), a process to make propylene, by removing hydrogen from propane. Traditionally, chemical processes have used Excel and other empirical methods. SK gas said its AI analysis system is capable of analyzing the correlation of various factors to provide simulations for future prediction. It allows engineers to quickly carry out their data analysis task, which used to take more than 5 hours with Excel.
When a problem is detected in the PDH process using explainable AI (XAI), operators can find the most relevant factor in real time instead of tedious hypotheses and verifications. The technique was applied to SK advanced which produces propylene.
By Kim Joo-heon