Metal casting

Improve Your Continuous Casting Process with Prescriptive Analytics

January 22, 2019

Based on a S&P Global Q3’18 report, more than half of top mining companies beat Q3’18 earnings estimates by decreasing costs and increasing production. One steel producer beat estimates by more than 55 percent and saw its net income double year over year. However, with steel prices forecasted to decline in 2019, similar margins will require some sharpening of the pencil. 

The evolution of metal refining and steelmaking is focused on decreasing operating cost per ton to better weather lower realized prices. Multiple factors can drive up operating costs and erode profits. Steelmaking not only involves extreme temperatures to melt new iron ore or scrap steel; it also has a complex system of hydraulics, rotating equipment and cooling machinery. Problems with equipment can severely impact the refining and casting processes, driving up costs.   

The numerous moving parts within the continuous casting process can be prone to expensive equipment failures. The turret, oscillating mold, cooling chamber, induction stirrer and numerous rollers must function within specified parameters or they can cause serious problems. Issues such as breakouts, oxygen contamination and carbon boil can spawn time-consuming clean-ups and turnarounds, potentially harming workers (and earnings). 

What if instead of responding to incidents by breaking apart spilled steel with torches and heavy equipment or replacing damaged machinery, mining companies could prevent problems before they occurred? They can, thanks to early detection and detailed prescriptive analytics provided by machine learning. 

Damage can be prevented by using real machine learning with signature agents that recognize normal, abnormal and failure patterns. Prescriptive analytics tool Aspen Mtell® identifies looming issues weeks or even months in advance, allowing ample time for investigation and planning. Not only that, Mtell can see concerns other solutions cannot to prevent process-induced damage. More importantly, Mtell attaches each failure to a specific root cause which guides operations and maintenance leaders directly to the fix. 

Aspen Mtell in Action

Working with a leading steel manufacturer in Latin America, AspenTech implemented effective prescriptive maintenance tools in days -- not months, as other solution providers proposed. By applying an easy to follow, step-by-step process as well as Aspen Mtell’s machine learning algorithms, the manufacturer detected pending failures on continuous casting equipment. Using 4 months of historian data on pinch rollers and visual mapping of production anomalies and failures, Mtell created over 30,000 predictions on equipment functions. 

Mtell then learned from historical events using pattern recognition to identify impending equipment and process failures. The system gave a 23-day time to failure direction on a specific pinch roller along with a 21-day time to failure direction on a specific bending roller. These alerts allowed the steel manufacturer to schedule maintenance and avoid unplanned downtime and production losses.    

With the type of information Mtell provides, leaders can not only prevent expensive and dangerous failures. They can also ensure they have the parts for the appropriate fix and guarantee properly trained resources are available and on location to do the work. This ability to plan can deliver substantial savings – more important than ever as steel makers look to maintain strong margins as prices fall.   

To learn more about how companies are leveraging this advanced technology, take a look at the recent case study Optimizing Smelting and Refining Equipment Reliability with Prescriptive Analytics


There was a problem storing your subscription

Leave A Comment