These Pune engineers are optimising German steel plants

Steel plants are noisy, effluvia-emitting totems of our industrialised world. But the effluvium that is produced is less and less tolerated in our rapidly warming world.
German engineering giant Thyssenkrupp’s technologists in Pune were one day staring at the mountains of data coming from the company’s steel plant in Duisburg, Germany – one of the largest steel plants in Europe – when a thought struck them.The chemical engineers at the plants in Germany analysing this data have PhDs in their areas of specialisation, but they aren’t data engineers. Could the team of data specialists in Pune be able to derive insights from the data and improve the steel making process?
Rohit Gupta, head of technology for India at Thyssenkrupp, says the plant in Duisburg was superbly automated with sensors of all kinds dotted within its crevices. “The importance of data was realised very early on at the plant, and massive amounts of data were collected through these sensors and stored in data lakes," he says.
It is this data that Gupta and his team started looking at.
There is an anaerobic process in steelmaking where coal is converted to coke before it is used for making steel. The problem is the coal that is used is imported from a variety of sources globally, which means they usually have subtly different chemical properties – a consequence of different coal extraction processes. “Coal from different sources has different molecular structures as well different grindability indices,” says Gupta.
An important process of making coke suitable for steel production is related to its grindability index. “One of the sub-steps in the steel making process relates to adding moisture, like oil and water, to improve the grindability of the coal,” says Gupta. This moisture-laden coal is then fed through battery cars into charging ovens where the coal is converted into coke in a non-pyro anaerobic process.
In the beginning, Gupta says, he and his team just wanted to analyse the different process parameters and see what impact it has on the final output. They created charts gleaned from data analytics and presented it to the chemical experts in Germany. “We wanted them to confirm whether it aligns with the ground truths they were seeing,” says Gupta.
The experts in Germany were so impressed that Gupta and his team were encouraged to dig further into the data.
“Our focus then was to see if we could, irrespective of the input, optimise the moisture content that we add. We also wanted to see if we could, irrespective of the grindability index, ensure a stable output for the coke density that we get from a particular fixed input.”
Eventually, using data analytics, Gupta and his team were able to build models that could do just that. “Our analytics product provides the ability to syllogistically play with different coal compositions and see what kind of output one can get, but at the same time, it can feed on live data and provide insights and advisory.”
The product they developed is referred to as a solution for ‘bulk density optimisation from coal to coke.’ It aids in decreasing fuel consumption by approximately 5-6% required in the pre-processing stage and still delivers the best results while reducing the overall consumption of non-renewable resources, says Gupta.