IIT Delhi Researchers Develop AI Models For Instantaneous Quality Check Of Cement
Concrete, comprising over 90% of the world’s-built environment, is the most used construction material globally. Cement, the key ingredient of concrete, is a critical industrial product that supports millions of jobs, drives urban development, and enables low-cost housing. Globally, the cement industry ranks among the largest industrial sectors, with India being the second-largest producer.
However, cement is one of the most carbon-intensive industries, contributing ~8% of the global carbon footprint. For context, producing just 1 tonne of cement—enough for a concrete staircase of a single floor—releases ~0.66 tonnes of CO₂, equivalent to driving a petrol car from Paris to Istanbul!
Read also: IIT Bombay To Conduct JAM 2026; Application Window Opens In September
With over 4.1 billion tonnes produced annually, and demand continuing to rise, advanced quality control is more urgent than ever.
Traditionally, cement quality is assessed by passing high-energy X-rays through clinker (partly- processed cement lumps). This process takes hours, causing delays that often lead to material waste and energy-intensive reprocessing when defective clinkers are found.
Addressing this challenge, Mr. Sheikh Junaid Fayaz, a PhD scholar under the supervision of Prof. N.M. Anoop Krishnan in the Civil Engineering Department at IIT Delhi, led the development of AI models that can predict clinker quality in fractions of a second. Not just ultra-fast, these models achieve ~88% lower errors compared to previously reported industrial-scale models, setting a new state-of-the-art benchmark.
“X-ray-based clinker checks can take up to 4 hours, while our AI models can predict clinker quality in just 1/100 of a second—making quality control a million times faster,” said Prof. N. M. Anoop Krishnan. “This real-time accuracy allows engineers to adjust plant parameters proactively, ensuring the target quality is met before production, rather than reacting to delayed X-ray analysis.”
“The work’s significance extends beyond cement manufacturing as it demonstrates how AI can transform traditional industrial practices and advance sustainability goals”, added Mr. Sheikh Junaid Fayaz. “Given the success of our models, several cement plants worldwide have expressed interest in adopting similar systems. We hope this paves the way for more AI applications in the industry.”
The research paper “Industrial-scale prediction of cement clinker phases using machine learning” was published in Communications Engineering (https://rdcu.be/enst8).
The research team also included Prof. Shashank Bishnoi from the Civil Engineering Department, IIT Delhi; Prof. Manuele Gatti, Prof. Matteo Romano and research scholar Mr. Néstor Montiel-Bohórquez from Politecnico di Milano, Italy.
