By combining sensors, computer vision models, and artificial intelligence, CEAT Tire’s Chennai plant reduces defects, waste, energy usage and improves labor and machinery efficiency. Today’s factories run on algorithmsHumming machines and flashing sensor lights greet you as you enter a large seat tire factory in the dusty industrial area of Sriperumbadur on the outskirts of Chennai. The manufacturing process here involves more number crunching and computer screen monitoring than manual labor. Over the past few years, 70% of shop floor roles have been converted from heavy labor to skill-based, automated work. It also means more women participating.Manage blend complexityEach tire has multiple compounds and materials, and these vary depending on the purpose, such as driving a car on a normal day, driving the same car in the snow, or towing a truck. Durability and flexibility vary depending on each use case. Debashish Roy, chief digital transformation officer at CEAT, says the blending process is a critical first step in tire manufacturing. Think of this as the industrial equivalent of kneading dough in a bakery. Raw materials such as natural rubber, synthetic rubber, carbon black, and chemicals are put into a giant mixer to create the base material.The goal is to blend these ingredients until they are completely homogeneous.
.
Traditionally, factory machines followed strict, preset recipes. For example, it may be programmed to mix for 3 minutes. However, the machine was blindfolded and could not tell if the rubber was fully mixed after two and a half minutes or if it needed four minutes. To further complicate matters, natural rubber varies from lot to lot. It may be hard or soft depending on the season. If you keep the mixer running all day, the temperature of the mixer will also change. These parameters change the rate at which the rubber melts. Because the system could not adapt to these changes in real time, factories often ran mixers longer than necessary. This led to wasted time and reduced daily output.Roy says he built a gradient boosting regression model (a machine learning model) to optimize for these variations. The system continuously monitors key parameters such as temperature and energy consumption by comparing real-time data with past “golden” batches. Based on this comparison, intelligent adjustments are made to maintain optimal conditions and improve consistency of mixing operations.This reduced mix cycle time by 18%, reduced power consumption by 29%, and increased master mixer capacity by 32%. These are very important considering the size of the factory. The company now produces 350 types of tires on an average week, up from just 100 types six years ago.Prediction of mold dimensionsAs the company expands its presence geographically, there is an increasing need to develop new tires suitable for different terrains and environments. To develop a new tire, CEAT needs to design a new mold. This is a process that creates a metal template or profile cutter through which hot rubber is squeezed to form a specific tread, or the part that comes into contact with the road surface. Think of it like a toothpaste that squeezes paste into a specific shape depending on the nozzle attached. However, the elastic nature of rubber makes designing this metal plate difficult. As soon as the material is extruded from the mold, it relaxes and expands, a phenomenon known as “mold swelling.” Roy says if you want a 200mm wide rubber strip, you can’t cut a 200mm hole. The rubber will expand by 20mm, so you will need to drill a 180mm hole. The exact number depends on the rubber chemistry, temperature, and machine speed.Each tire variant has approximately 60 data points, ranging from compound viscosity to machine parameters. Previously, employees manually mapped these points into a spreadsheet to predict how the rubber would behave. This nonlinear calculation was often inaccurate and required up to three physical attempts to get it right. This iterative process increased product time to market and resulted in multiple scraps.To address this, the company introduced a machine learning model based on Gaussian process regression. The system accurately predicted mold dimensions, resulting in a 37% faster time to market and a 30% reduction in waste.Roy also introduced AI models and solutions for export container optimization and machine performance management. His team built an agent AI solution that allows junior engineers to quickly understand what’s wrong when a machine breaks down. Turn unstructured troubleshooting videos into a searchable knowledge base and retrieve past solutions through conversational chatbots. This improves troubleshooting efficiency.WEF lighthouseThis initiative has resulted in a cultural shift where employees are encouraged to think from a data-driven perspective. The team regularly attends external conferences and hackathons to learn about best-in-class technology. Before developing an AI use case, assess whether your production problem is a valid business problem and whether usable data exists. The Chennai plant’s technology has been validated by experts at IIT Madras and has received Lighthouse certification from the World Economic Forum. The Lighthouse Prize recognizes leaders in the field of technology-driven industrial transformation.Significantly increased productivityDigital efforts have improved yields and reduced energy usage, resulting in a 20-30% reduction in factory conversion costs (the total cost of converting raw materials into finished products). Order-to-shipment time has been cut by more than half, and export turnaround time has been reduced from 120 days to 55 days.decentralized digital teamCEAT built the technology solution from the ground up with an in-house team. This is thanks to digitalization efforts that began in 2021, including the installation of sensors, manufacturing execution systems, and dashboards. The company is currently investing 10% of its manufacturing capital expenditures in digital transformation efforts. A new role called Business Translator has been introduced to bridge the gap between technical teams and the reality on the ground. They have strong technical skills and work directly with shop floor operators to identify problems. Debashish Roy, chief digital transformation officer, says it’s a decentralized digital team. A centralized team may take years to learn the process, he says.
