Norilsk Nickel built Data Platform "Data Lake" on Arenadata products

2024: Building a Technology Platform

Norilsk Nickel has built a technological platform for solving business problems using Big Data and Data Lake on the products of the domestic vendor Arenadata. The use of "Data Lake" allows Nornickel to reduce the Time to Market for the implementation of digital projects due to the lack of the need to design and implement integration tools, as well as significantly reduce the analytical burden on production management and dispatching systems. Arenadata announced this on August 19, 2024.

Nornickel works in the field of mining and production of non-ferrous metals, which is a technologically complex industry, which includes many E2E processes (end to end), around which terabytes of data are formed. Their processing allows you to increase the productivity of factories and workshops, make the working conditions more comfortable and safe. Therefore, the digitalization of production is one of the key strategic tasks of the company for the coming years. It includes: development and development of own digital innovation platforms, introduction of data-driven approaches to technological, production and business processes; AI operationalization.

Approximately 40% of cases in the overall business task register require unstructured or semi-structured data processing. They are implemented in the technological platform "Data Lake." This is the environment for the implementation of many initiatives aimed at digitalizing production based on AI advisers, AI auto advisers and analytical reports.

A single data access point and one of the main components of the platform is the analytical DBMS (ADB). In addition, the Data Lake uses the Arenadata Streaming (ADS) data streaming product and the Arenadata Hadoop (ADH) distributed scalable data distribution.

Building a Data platform landscape based on Arenadata products was significant for us, as it allows us to develop and launch real innovative cases. Arenadata's solutions meet our high requirements for systems that work with productive data, including security requirements, "said Mansur Gasratov, head of Norilsk Nickel's Data Lake.

In general, the Data Lake ecosystem consists of four large blocks:

• Data platform for processing and storing any data;
• Containerization for the launch and production of ML applications and IML instances
• ML-platform for corporate environment of ML-solutions development and sprinkling;
• ML clusters for a geodistributed ML infrastructure that allows you to run solutions at production sites.

The data platform allows Norilsk Nickel to process information from equipment sensors, packet data from relational systems, connect ML models, examine data and receive analytical reporting using BI tools connected to the Data Lake.

Data for enterprises is one of the key resources. And the experience of Norilsk Nickel clearly demonstrates how the company with their help increases operational efficiency and optimizes production processes. I am sure that, given the company's strategic plans for digitalization, including the use of the Arenadata product stack, new prospects await it, "said Maxim Vlasyuk, Director of the Arenadata Group Department for Work with the Industrial Sector.

As of August 2024, the total storage capacity in the Data Lake ecosystem is approximately 1.5 petabytes. Nine industrial integration data streams were launched, within the framework of which more than 2000 indicators go to the data platform in a mode close to real time, and more than 3000 indicators - offline. More than 10 business cases have been implemented, covering such areas as economics and finance, production and technological processes, voice assistants.

Examples of implemented business cases:

• Monitoring of the quality of production data at the Carbonyl Nickel Division of the Kola MMC and at the crushing and grinding line at the Bystrinskoye GC (Bystrinsky GOK). The technology helps to proactively identify incorrectly working measuring equipment and increase the speed of management decisions. Its implementation made it possible to reduce the time for detecting incorrect indicators by 30%.
• Advisory system for optimizing concentrate mixing (concentrate dewatering and warehousing workshop at Nadezhda Metallurgical Plant, Polar Branch). It helps operators choose the optimal process strategy.
• Advisor system for optimization of the suspended smelting furnace (at Nadezhdinsky Metallurgical Plant, Polar Branch).