Guest Author : Margie Steele
What do cars, appliances and our homes have in common? They all require metal as a component. Steel production is labor- and equipment-intensive with high energy costs and expensive raw materials. A leader in the steel industry is adopting a better way to quality, profitability and increased production using Oracle’s IoT Asset Monitoring.
A major steel manufacturer wanted to focus on their hot strip mill process to understand issues that impact production and quality. The hot strip mill process takes a slab, which is the output of the casting process, and reduces the thickness to less than 1” that is used for cars and other consumer items. If you think back to when you were a child and played with playdoh, you would take a roller and roll it over the playdoh until it was a thickness you wanted. The same applies here as the slab is reheated to a malleable temperature then run through incrementally tighter rollers to reduce the thickness. At the end of reduction, it goes into coils which look like giant toilet paper rolls. The steel travels hundreds of feet per second and there are almost 200 sensor attributes such as temperature, side guides, and the tension of the rollers that can detect issues.
Common quality issues that can result are that the coils do not catch on the “toilet paper roll” and the metal will pile up because it does not roll. The coil could also telescope which is when it does not roll evenly and one side is tighter than another which telescopes the roll. They can also get staggered side edges if the side guides are not adjusted or ripples in the product if rollers do not have the proper tension. There is also the potential for worker safety issues depending on the manufacturing problem that occurs. All of these almost 200 sensor attributes are able to be detected electronically and evaluated as production progresses.
The steel industry utilizes a number of data collection systems for controlling processes which are highly sensored to capture the attributes so they can understand their processes better. When a pile up occurs, reviewing the values of sensors can point to the root cause. However, this is backward looking – what if you could detect that a sensor value was starting to vary from its required value? Even more important, what if there were groups of sensors that when combined together, gave significant meaning? This project for monitoring the hot strip mill did just that – found causal factors for issues and attributes that correlated. Now that the causal factors were determined, using IOT Asset Monitoring, we were able to predict 30-90 minutes before an issue occurred with 85% certainty regardless of whether a single attribute or combination of attributes were the cause. Now corrective actions can be put in place to stop the manufacturing issue from happening whether it be an adjustment to equipment, maintenance, or changing production specifications.
In the steel industry, maintenance costs can be as high as 10-15% of the cost of production. If you perform maintenance too early, you waste money because it wasn’t needed but if you wait too long, you end up with broken, expensive equipment. So maintenance is like the Goldilocks story - when is it just the right time? The above predicted issue that is causing production problems might be maintenance based. A future corrective action to the breakdown would be to have IoT Asset Monitoring create maintenance work orders for the equipment based on current equipment status from the sensors. As a side benefit, the maintenance would be performed at that “right time”. Stopping production for repairs is costly because of lower output. However, because there is a window of opportunity (30-90 minutes) before predicted breakdown, you have the ability to perform the corrective maintenance between production orders or a time that reduces impact to output.
Prediction models for failure based on built-in asset reliability algorithms are embedded in IoT Asset Monitoring but there is flexibility to add new or product-specific algorithms. The ability to take action to resolve the issue is a key factor to the success of this project. Since there are many potential causes to pile ups and problems in the hot strip mill, determining the precise corrective action to take is just as important so integration to maintenance, production, and quality systems is needed. If IoT Asset Monitoring has a prediction, the follow-up action is crucial whether it’s to alert someone to change a side guide, change the speed of the rollers in the control system or to create a maintenance work order in the maintenance system. This means that for all of the correlations that indicate a pile up, you must create a rule on how to handle that issue.
This IoT project was just in one area of a single facility. There are many other steps in production in which IoT can be leveraged as well as non-production areas such as inventory tracking, shipping, fleet to deliver, quality testing and facility maintenance. The benefits that real-time monitoring and predictive technology will bring to this steel manufacturer are more than just in production and maintenance cost savings but also in meeting performance and customer delivery expectations, reduced claim rates, and in worker safety. Maybe the Goldilocks saying that it’s not too early and not too late but it’s “just right” is this steel company’s fairy tale come true.