Severstal spent 2 billion rubles on the development and implementation of a digital quality management system

2026: Severstal spent RUB 2 billion to develop and implement a digital quality management system

On February 17, 2026, the vertically integrated mining and metallurgical company Severstal announced the completion of a key stage in the implementation of a digital product quality management system. About 2 billion rubles were spent on the project.

The developed digital complex helps employees make decisions about the suitability of rolled metal based on a previously inaccessible large amount of information. Data is collected at all stages of production. For this, industrial units are equipped with special meters based on computer vision and artificial intelligence technologies, which are responsible for fixing certain deviations. The generated predictive models perform calculations of the probability of occurrence of defects invisible to the human eye based on technological parameters. Metal tracking tools provide data binding to each meter of a particular product, and incident detection tools monitor critical areas affecting quality.

𝓲

Фото: Freepik

The new complex helps employees to solve certification tasks, and also ensures a timely response to deviations in technological processes. In total, more than 17.5 thousand parameters are estimated, of which 500 are certification. Based on the data obtained, the Sherlock auto-certification system recommends a decision on the suitability of rolled metal.

The platform is deployed at the Cherepovets Steel Mill, where 35 key units are included in its contour. Every four seconds, the company certifies a product, every day - about 20 thousand certifications. The system covers the entire chain of production of polymer rolled products, galvanized rolled products, hot-rolled rolled products, etched rolled products, commercial slabs. In general, digital quality control is implemented for 60% of marketable products.^[1]

2024: Spending ₽1,5 billion on digitalization of product quality control

Severstal has completed a large-scale stage of digitalization of product quality control, investing ₽1,5 billion in this program over the past seven years. The project includes the introduction of unique computer vision and artificial intelligence systems, as well as the creation of an automatic certification platform for rolled metal products. This was announced in mid-November 2024 by the director for technical development and quality of the company Pyotr Mishnev.

According to Metalinfo, the VERA computer vision model developed by Severstal has become a key tool in the project. It was introduced at the Cherepovets Steel Mill and controls the quality of metal at the etching unit No. 4. The neural network analyzes images obtained from cameras, detects and classifies coil defects, which allows experts to quickly make decisions on product compliance with customer requirements.

𝓲

Фото: Severstal Media Photo Bank

The effectiveness of the VERA model is measured by two measures: the proportion of errors in the detection of defects ("malfunction") and the incorrect classification of defects ("overkill"). These figures are 14% and 12%, respectively, which is significantly better than the results of similar market decisions. The use of the model made it possible to reduce the number of notifications from customers about defects in finished products by 1.8 times in the first half of 2023, which brought an economic effect of ₽2 million.

In addition, the project introduced the Sherlock automatic certification system, as well as a set of digital solutions for predictive analysis and recording of incidents at all stages of rolled metal production. More than half of the company's products now undergo digital quality verification.

As noted by Petr Mishnev, the development and implementation of its own solutions allowed the company to reduce dependence on foreign equipment and adapt systems to unique production tasks.

The company said that the VERA model and other digital tools are being developed jointly with Severstal Digital and the Digital Quality Development Center. These technologies can be scaled to other parts of the production chain, which will increase their coverage and efficiency.^[2]

2023: Digitalization of product quality control

On September 28, 2023, Severstal announced that it was implementing a large-scale project to digitalize product quality control. The company's experts have developed a system for end-to-end automatic certification of rolled metal products. The volume of investments amounted to about 1.3 billion rubles.

The development of a comprehensive quality control system was a response to a number of challenges - both in terms of creating and mastering technologically complex new types of products, and meeting the growing requirements of customers, including in new market segments for us. The use of traditional methods alone did not solve these problems. The key goal of the project was the creation of a single system in which every deviation ever detected will be monitored. The set of digital tools works in order to help our employee decide on the suitability of rolled metal based on the previously inaccessible large amount of information received at all stages of manufacture. We have achieved digital control of the production of about half of the volume of metal products produced per year, - said Pyotr Mishnev, Director of Technical Development and Quality of Severstal.

The implementation of the project makes it possible to provide a number of significant advantages for the company's customers. These include quality assurance for 100% of the product length due to continuous measuring systems and minimization of the human factor, reduction of the customer's sorting losses due to improvement of the basic product quality, quick response to deviations and delivery on time. For the period from 2018 to 2022, the share of received notifications from customers for four types of products covered by the system decreased by 55%.

The quality control system includes dozens of digital tools on the life path of product creation. These include equipping industrial units with meters using technologies computer vision and, artificial intelligence which are responsible for fixing visible deviations. The developed predictive models perform calculations of the probability of occurrence of defects invisible to the human eye based on technological parameters. Metal tracking systems provide binding data to each meter of a specific product, and incident recording systems monitor critical areas that affect its quality. Their use makes it possible to eliminate the human factor in matters related to the correct operation of equipment and certification, as well as to ensure a timely response to deviations in technological processes.

Based on the data obtained, the decision on the suitability of rolled metal is recommended by the Sherlock auto-certification system. The digital assistant has tools for finding the root causes of quality deviations, analyzing trends and building models, ensures the availability of data for any user and the ability to quickly select orders for manufactured and on-hand products for incoming orders. The results of the system application are not only a decrease in the number of notifications of deviations and internal losses, but also an increase in the accuracy of delivery times.

The project for the introduction of a digital auto-certification system has been going on for seven years. During this time, it covered the through chain of production of galvanized rolled products, hot-rolled rolled products, etched rolled products, commercial slabs. At the same time, 43% of ever detected defects are monitored. The system is planned to be fully completed by 2028.

Severstal's team of experts with the necessary competencies from the Technical Development and Quality Directorate, the production divisions of Cherepovets Steel Mill, as well as the specialized companies Severstal Digital and Severstal-Infocom are implementing initiatives.