Resolving a ‘crushing’ problem with IoT sensors

A leading steel company was looking for a solution to a problem it faced in its manufacturing units. The steel manufacturer wanted to detect and remove impurities in raw material before feeding it into blast furnaces. 

Heavy industries often face the dilemma of whether they should invest in state-of-the-art equipment or extend the useful life of existing machinery. While manufacturers realize the benefits of new-age technologies, prohibitive upfront investments pose a challenge. Retrofitting equipment such as internet of things (IoT) sensors to existing machinery could be a practical solution to this problem. TCS Research’s work on unobtrusive sensing enables organizations to adopt IoT with minimal effort and costs, and no disruption to existing manufacturing processes.

A steel manufacturer’s ‘crushing’ problem

Pulverized coal is used as a carbon source for reduction of iron in steel plants. Coal is pulverized through a crushing and grinding process at high temperatures. The presence of foreign particles, such as rocks, stones, rubber, concrete, and wood in the raw coal input stream causes outages in the grinding mills. The impurities are generally higher in density and larger in size than the coal input. These draw more electricity at variable speed drives, resulting in power trips. Outages typically last for about 6-7 hours and are more frequent during monsoons when raw material is stored in the open. 

Furthermore, these outages lead to increased consumption of raw input and production interferences. Higher costs are incurred due to damages to the blade and conveyer belts. Production is brought to a standstill for at least 6-8 hours, resulting in losses of over USD 80,000 per incident. 

TCS builds an unobtrusive, contactless alternative

A TCS Research team working on unobtrusive sensing proposed to customize their intellectual property to create an unobtrusive and contactless method to solve the problem of impurities. The system was built to detect and localize impurities before the inputs entered the grinding mills. A sensing and processing system was set up to observe the surface of the conveyor belt carrying the input.

Unobtrusive sensing 

TCS Research’s unobtrusive sensing work uses software-based fusion algorithms on sensing, off-the-shelf sensors, contactless sensing, and multimodal sensing to detect the health of machines.

TCS Research’s unobtrusive sensing is a non-contact sensing solution for asset, process, and material monitoring with appropriate hardware and software co-design that uses frugal off-the-shelf easy deployable transducers and sensors with special emphasis on fusion and multimodal sensing with augmented analytics enabling sensing from a distance with wider field of view.

Foreign particle detection using polarimetry, an unobtrusive sensing technique

The unobtrusive sensing system uses multimodal imaging and corresponding processing algorithms to detect the presence of foreign particles. The team installed cameras to continuously monitor the conveyor belt. By fusing conventional RGB (red-green-blue) imaging techniques with polarization imaging, the algorithms could differentiate between coal and foreign objects based on multiple parameters like shape, size, texture, and material type.

As a first step in the proof-of-concept (PoC), the team simulated the operation of the conveyor belt at a TCS Research lab. This acted as an experimental setup to monitor the pipeline in real time and replicate onsite operations. 

The actual PoC system deployed onsite is depicted in Figure 1.

Figure 1: Data Acquisition and Analysis Pipeline

A multimodal sensing camera with a light source was mounted on a stand to continuously monitor the belt. Data collected from the system was analyzed in real-time and alerts were generated when foreign particles were detected on the belt. The team also incorporated an additional facility to remotely monitor impurities.

Arming manufacturing operations with technology

.The retrofitted system made an impact on the customers manufacturing process with the following benefits:

• Ease of custom deployment: The unobtrusive system left the existing process flow unaltered. It required no additional skills from the operator. 

• Improvements in inferencing and accuracy of measurement: Early detection and removal of impurities improved the overall efficiency of the system.

• Justifiable returns on investment: Prevention of outages and reduced losses resulted in direct cost savings for the steel manufacturer. 

• Adaptation to dynamic business requirements: The imaging system detected impurities of multiple textures and generated alerts in real-time, helping operators to make timely decisions. 

Unobtrusive sensing—The scope and way ahead in manufacturing

The manufacturing industry is at a critical juncture to sustain growth while recovering from the disruptions caused by the COVID-19 pandemic. Given the increased availability of affordable sensors, unobtrusive sensing is a technology in manufacturing that is expected to grow over the next few years. Over 50 billion connected industrial IoT devices will be deployed globally by 2025 with retrofitted sensors as one of the major enablers, reports suggest. With unobtrusive sensing, TCS Research continues to work towards tapping into this growing market and enabling manufacturers to adopt the state-of-the-art with fewer disruptions.