Nippon Shokubai and NTT Communications (NTT Com) announced on February 18 that they have successfully automated a chemical manufacturing process, which has been considered difficult to achieve until now, using AI. This is expected to reduce the burden on operators and make it more efficient to transfer skills. In this initiative, AI was trained to learn the operation data of the continuous distillation process, the operation history of the operators, and operation know-how. By incorporating a trained AI model into NTT Com’s “AI Autopilot System,” automated operation of the continuous distillation process, whose conditions change in complex ways, was achieved. A wealth of knowledge and experience is essential to operate a chemical manufacturing plant. A wealth of knowledge and experience is required to operate a plant that manufactures chemicals, whose conditions such as temperature, pressure, and concentration are constantly changing, and automated operation has been considered difficult. In particular, the continuous distillation process, which increases the purity of chemicals, is easily affected by the composition of the reaction liquid, the amount of recycled distillate that flows out, and weather, and skilled operators had to operate it manually at all times with high concentration.
In addition, skill transfer is an important issue for maintaining stable operation, and training operators takes a lot of time. The accuracy of temperature control in the continuous distillation process is evaluated by the amount of effluent per unit time. The distillate not only flows out from the continuous distillation process, but is also intermittently reused, so the ideal distillate level (ideal liquid level) measured by the level gauge is constantly changing.
However, by incorporating a trained AI model, the error between the actual level gauge value and the ideal liquid level was reduced from an average of 2.38% under manual operation to an average of 2.06% under AI operation. This corresponds to an improvement of 13.5% compared to manual operation by an experienced operator, and it was confirmed that operation quality equal to or better than manual operation can be achieved.
Nippon Shokubai will expand this initiative to other plants in the future and promote transformation through DX. NTT Com will continue to expand the system’s functions to realize automated operation of plants where manual operation remains, such as batch plants and small-lot, multi-product production plants. In addition, the company plans to expand this technology beyond chemical plants to a wide range of industrial fields, contributing to solving issues such as labor shortages.
SOURCE: Yahoo
