Source: Société Le Nickel (SLN), part of Groupe Eramet, and Rockwell
To increase tonnage and uptime, and improve nickel ore calcination in its kilns, Groupe Eramet’s Société Le Nickel (SLN) in New Caledonia adopted Rockwell Automation’s FactoryTalk Analytics Pavilion 8 model-predictive control (MPC) software with an intelligence layer that sits on top of automation systems, continuously assesses present and predicted operations, and resets control targets to reduce variability and boost efficiency. Using MPC in its kilns, SLN reduced product temperature errors by 6%, cut temperature profile variability by 16.1%, and improved uptime from 70% to 83%.

Unplugging nickel production

Sept. 16, 2024
Société Le Nickel’s mines implement Rockwell’s FactoryTalk Analytics Pavilion8 MPC software

The clearest, most-direct guide about what data analytics method and tools to adopt is having a specific, immediate problem to solve. For instance, to maintain its position as the world’s largest producer of ferronickel, Groupe Eramet’s Société Le Nickel recently implemented Rockwell Automation’s FactoryTalk Analytics Pavilion8 model-predictive control (MPC) software to boost tonnage and increase uptime. The 140-year-old company mines nickel at five sites in New Caledonia, about 750 miles east of Australia, but legacy, fuzzy-logic controls in its ore calcination process were too slow. Varying ore content and heating values triggered temperature spikes and frequent electrical trips because the product was too hot, which also compromised product quality. The legacy system also wasn’t user-friendly and was difficult to maintain, triggering uptime issues.

Get your subscription to Control's tri-weekly newsletter.

Seeking stability

SLN reports that processing ores requires stable control of the rotary furnaces’ temperature profiles and automating operations across operating ranges. Feed ore undergoes calcination as it’s heated along the length of the rotary kiln. Heated air is supplied for combustion, but if there’s too much, more fuel must be burned to maintain the same product temperature, which decreases energy efficiency. Excess oxygen must be minimized to a safe level which also reduces costs and greenhouse gases. However, if the calcined ore isn’t hot enough, its quality and the energy efficiency of the processing plant can be compromised.

“The fuzzy logic was unable to reduce fuel fast enough to prevent trips from occurring, and during manual operation, operators sometimes couldn’t react fast enough,” says Leslie Hii, one of the advanced process control (APC) engineers at Rockwell responsible for delivering the SLN project. “Maintaining required furnace temperature can be complex and challenging given the variables that need to be managed. Fuel types can be oil, coal or a mixture, each with unique thermal characteristics. Also, material feeding rates impact furnace temperatures, so they must be carefully managed.”

Finally, the legacy, expert system could only run when the rotary furnace was operating normally. If any instability occurred, SLN’s operators had to turn it off and take control. This is why SLN upgraded to Pavilion8 to control its rotary kilns in real-time, provide an intelligence layer on top of its automation systems, and continuously assesses current and predicted operations. The software compares these results to desired outcomes, and drives new control targets to reduce process variability, improve performance, and autonomously boost efficiency in real-time.

“MPC shows how Rockwell applies artificial intelligence (AI) to achieve better results with available data,” explains Hii. “This project also used machine learning (ML), process knowledge and other data to develop kiln models tailored to SLN’s operations.”

Minimize consumption, maximize production

SLN completed the initial phase of its upgrade by implementing Pavilion8 in five rotary furnaces in just 13 months, which was much quicker than the years it took to install the fuzzy-logic controls. The MPC software lets operators opt to minimize consumption of costly fuel oil, while maximizing inexpensive, pulverized coal during mixed-mode operations.

“The MPC application can handle variable ore feed and heating values, and prevents trips, which lets furnace run at a higher rate and operate longer,” says Mickael Montarello, process control manager at SLN. “Calcined product temperature error was reduced by 6%, while furnace temperature profile variability was reduced by 16.1%. The average uptime of Rockwell’s MPC is 83% compared to 70% with the earlier, fuzzy-logic controls. Users appreciate MPC's user-friendliness and flexibility. In the event of a problem with one element of the process, operators can easily intervene with the element in question, while allowing the MPC to continue controlling the other manipulated variables. Thanks to this tool, new opportunities for optimizing control and management at SLN are opening up that weren’t possible with the old fuzzy-logic controller. Our target for 2024 is to achieve a 90% utilization rate.”

Real-time results at the edge

Todd Montpas, commercial business manager for data and analytics industry solutions at Rockwell, adds that data analytics may have progressed from clipboards to servers and cloud-computing services, but users and manufacturers still need—and often lack—real-time information and results to optimize their processes.

“We need to close this loop at the edge, close to the process, using the automation investment customers have already made. The analytics and AI tools here need better links to the cloud and software as a service (SaaS) to analyze data from different systems across multiple plants to build larger insights,” says Montpas. “We advise using the power of smart devices, many of which have contextualized data, to feed these analytics systems.”

Instead of building dedicated applications that get data from PLCs and create dashboards, Montpas reports that Rockwell has developed an industrial data operations platform called FactoryTalk DataMosaix that lets users access this siloed data, contextualize it, and provide access across the organization.

“A traditional, dedicated analytics application requires users to understand which device parameters are relevant, and then program each PLC to find these tags and pull all their pieces of data together. Each time a new insight is required, this process needs to be repeated,” explains Montpas. “DataMosaix provides access to these intelligence devices, and as additional data is needed, it updates the data models without needing additional programming at the PLC level.”

Easier extraction

Montpas adds this data architecture also pulls information from the middle layer, which then feeds upper-level applications like PowerBi and other analytics software applications. In fact, Rockwell just released an analytics application for its PlantPAx distributed control system, which includes data models built into the DataMosaix layer for quick out-of-the-box configuration and other tasks.

“Previously, users had to write code to get access and pull the necessary data for many analytics functions. Now they can pull operations data from applications like Rockwell’s Plex manufacturing execution system (MES) to provide energy-per-unit-manufactured,” says Montpas. “Users need access to structured, unstructured and smart device data, so DataMosaix builds knowledge graphs about how they relate without requiring programming to make this data useful. Additional value is realized by using pre-built applications, building your own, or using other charting and analytics tools.

For example, FactoryTalk Energy Manager software application and DataMosaix were implemented at several Rockwell facilities in September 2023. Where it used to take weeks or months to configure and start up plants like this, Montpas reports that using these applications with DataMosaix ability to combine and coordinate multiple Rockwell platforms, allows complete configuration and start up in days.

“Making data truly meaningful for process analytics means having the ability to digitize engineering content like P&IDs, engineering diagrams, and other hard documents and static pieces, so they can be turned into data models and pulled into other systems,” explains Montpas. “For instance, all the material associated with a pump and its support equipment can be pre-embedded, and then subsequent details like installation locations can be added later. This data is combined with AI applications like FactoryTalk Analytics GuardianAI to detect pump issues like cavitation, and direct maintenance people on how and where to fix the problems before failure.

“This isn’t control, but it does enable remote planning, or ties into applications like Energy Manager to optimize pump operations. Many users are building their own apps, which can look at individual processes, whole facilities, or fleets of assets and enterprises. DataMosaix software as a service (SaaS), makes it easier and quicker to scale up functions and visibility, such as presenting real-time monitoring data in context.”

For example, monitoring batch performance requires knowing flow rates and sequences, and having a time-series database that’s tied to the batch, can talk to the MES, identify peak batches, and determine ideal pump settings. These tasks are traditionally very time-consuming, which means they may get neglected over time, and eventually generate stale or unusable results. Software tools like DataMosaix can streamline these tasks, making it more likely they’ll continue to get done over the long-term. Because they can integrate helpful information sources like energy use, operators and engineers can better adjust flow rates to maximize energy efficiency and savings.

“In the past, one software program would be written to run an ideal batch, while someone else would write an application to look at energy use. As a result, these two programs would risk conflicting and even fighting with each other, and be unable to achieve a single source of truth,” adds Montpas. “Relaying device and MES data to a platform like DataMosaix lets users add more contextual pieces, and send that context back to the MES and device levels. This allows them to make better decisions and coordinate their efforts to advise the loop, even as their PLC maintains control of process operations. Applications like GuardianAI and Energy Manager need to fit together, so they can pull edge devices and loop data, build models to verify operations are OK, and relay adjustments back to controls and edge device. Consequently, when process parameters change, their model can be updated, and build in new predictive parameters to further improve performance and value.”

About the Author

Jim Montague | Executive Editor

Jim Montague is executive editor of Control. 

Sponsored Recommendations

Make Effortless HMI and PLC Modifications from Anywhere

The tiny EZminiWiFi is a godsend for the plant maintenance engineers who need to make a minor modification to the HMI program or, for that matter, the PLC program. It's very easy...

The Benefits of Using American-Made Automation Products

Discover the benefits of American-made automation products, including stable pricing, faster delivery, and innovative features tailored to real-world applications. With superior...

50 Years of Automation Innovation and What to Expect Next

Over the past 50 years, the automation technology landscape has changed dramatically, but many of the underlying industry needs remain unchanged. To learn more about what’s changed...

Manufacturing Marvels Highlights Why EZAutomation Is a Force to Be Reckoned With

Watch EZAutomation's recent feature on the popular FOX Network series "Manufacturing Marvels" and discover what makes them a force to be reckoned with in industrial automation...