JSW Steel using AI, machine learning to deliver quality and reduce costs

Using Internet of Things, JSW Energy can accurately predict how much power it has to generate and supply to the plant in every 15 minute cycle

In ancient Greece, Athena was the goddess of wisdom, handicraft and warfare. Interestingly, when JSW Steel embarked on its digital transformation journey back in 2017, it codenamed the initiative as ‘Athena’. Two years on,  the company has made significant progress in this effort by implementing Industry 4.0 technologies in over 100 projects. The Sajjan Jindal-run steel major claims that the initiatives have so far resulted in a cost saving of over Rs 180 crore. This year, it expects to make an additional saving of Rs 300 crore.

Last year, the company adopted several Industry 4.0 technologies such as machine learning, Internet of Things, artificial intelligence, big data and advanced analytics and computer vision, to name a few. Taking the initiative further, earlier this year, it roped in Vineet Jaiswal, a top executive from GE, as its chief digital officer (CDO), a position which is becoming common in the technology industry globally, though not so much in traditional sectors like manufacturing. 

One of the first things that Jaiswal did was to create a Digital Playbook to focus on several digital initiatives with clear value creation goals. The purpose was to create a network of “connected factories” built over multiple cyber physical systems, the groundwork for which had already been laid out. Digital and Industry 4.0, the company says, are strategic levers to increase throughput, optimise cost performance, deliver best quality, establish predictive maintenance and enhance safety. 

“The entire world is going digital. We know we will not be competitive unless we are digitally equipped,” says Seshagiri Rao, joint managing director and group CFO. Rao says that JSW Steel is leveraging multiple digital technologies to optimise operations. These include an Industrial Internet of Things (IIoT)-driven power prediction model to optimise power consumption in steel-making, custom optimisation models enabled through advanced analytics to ensure efficient use of raw materials and energy, real-time scheduling and tracking, as well as an analytics-enabled system to optimise logistics. A multi-location team comprising over 100 technology and operations professionals is implementing the company’s digital initiatives.

Explaining how the new technologies are increasing efficiencies, Rao says that at the company’s Dolvi plant (Dolvi Works) in Maharashtra, the power is sourced from JSW Energy in Ratnagiri. Typically, if the power is not consumed within 15 minutes of being procured, it goes back to the grid, resulting in a loss to the tune of Rs 80-Rs 100 crore per annum. So the company needed to accurately predict how much power JSW Energy had to generate and supply to the plant in every 15 minute cycle. 

To do this, the company implemented an IIoT-based power prediction model at its steel melting shop (SMS). “By measuring parameters such as arcing schedule, hot metal quantity and oxygen blowing characteristics and running these through a complex algorithm, we can predict the exact quantity of power required by the SMS,” says Rao. “This not only prevents any over or under injection of power from the grid, but also helps in efficient utilisation of critical resources,” he adds. 

That’s not all. JSW Dolvi Works has deployed a number of digital tools to track and predict the schedule of its ongoing expansion projects and drive on-time completion of these within the budgeted cost. Computer vision-aided video analytics are being used to reduce the time for manual processes, enhance throughput and improve SMS productivity. 

The company is also employing digital technologies to customise products according to the client’s requirements while ensuring minimal wastage. For instance, the quantity of ferroalloys and other materials it needs to get a particular quality of product may vary. Given that ferroalloys are quite expensive, JSW has automated the raw material dispensing mechanism using an algorithm. To get a desired quality of steel, the machine releases the correct quality and quantity of ferroalloys and other raw material to the bath. Thanks to their machine learning and AI capabilities, the systems keep learning from previous data — not just to develop the best customisable product, but also to optimise the use of raw materials and power. This solution has now been deployed at Dolvi as well as JSW’s Vijaynagar plant in Karnataka.  

In addition, a Robotic Process Automation (RPA) technology is being used to automate repetitive tasks in areas like account payables (supply and services), vendor support cell, payroll processing and logistics bills processing. After a successful pilot of contract workforce management at its Vijayanagar plant, company is now ready to adopt the system in its other plants this year.