The Uralchem case: minimizing the influence of the human factor on the production process
The crux of the problem
In 2018, Uralchem, one of the largest Russian manufacturers of mineral fertilizers, was tasked with increasing the efficiency of production processes and reducing the influence of the human factor on production. To solve this task, the company created the Department of Digitalization and Technological Development, whose tasks include introducing innovations to increase productivity.
One of the factories faced challenges controlling the DGD apparatus (drum-granulator-dryer). Only one operator out of ten was experienced enough to efficiently operate it. Not all operators can see how their actions affect the product due to the inertia of the process. As a result, the product was usually returned multiple times for improvement. The company concluded that the human factor was the main problem: performance could be improved by reducing its influence.
The solution
After analyzing production processes, it was decided that the most effective solution was the introduction of an artificial control system — an online advisor that helps each operator to operate at the level of their most experienced colleagues.
The online advisor operates based on Big Data analytics and Artificial Intelligence. It collects and analyzes all technological process parameters in real-time, including temperature and air pressure on equipment components, to determine optimal parameters and give the operator recommendations.
Unexpectedly, the main challenge in the fight against the human factor was the human factor itself. At first, specialists were skeptical about the new service and ignored its recommendations. Therefore, initially, they were allowed not to follow recommendations that they considered incorrect. The situation began to improve when the operators realized that they were making the same decisions the system had already recommended, but with a delay of 5 to 10 minutes.
The results
Uralchem says the testing of the new system has shown positive results. The company managed to improve the stability of the process and increase output by 2–6 %. The experiment was deemed successful; now, it is planned to increase capacity, introduce automatic control, and scale the solution to other factories of the company.