China’s steady economic growth—as opposed to hyper growth—over the past five years has reset expectations and coincides with the country’s transition to more advanced manufacturing in multiple industries. China’s manufacturers are investing in more equipment, such as robotics for their automotive industry, and on better asset management strategies for both legacy and new plants.
This optimization push centers on upgrading data acquisition platforms and crossing the bridge to true predictive maintenance approaches in many industries in China, including harsh plant floor environments.
Recently, China Steel upgraded the legacy monitoring system at its main facility in Kaohsiung, Taiwan—and another steel production plant in Taichung City—to provide better access to real-time equipment data, such as vibration levels for roller bearings and motors in their milling production area. This continuous process application includes molten metal being produced into blooms or slabs that eventually are rolled into finished products.
“Due to the continuous nature of our production facility and the harsh environment—+300 °C temp, moisture and high vibration—it is very hard for us to maintain equipment once it gets deployed to the field,” says Zhizhong Wang in China Steel’s R&D department.
The steel manufacturer used a legacy condition monitoring system, called the Facility Online Monitoring and Diagnosis System (FOMOS), to track mill equipment vibrations for multiple mill lines within facilities. However, the original monitoring system didn’t provide real-time maintenance data—or efficiency—because of a long lag time for this data to move from the plant floor to the database. Operators would typically collect raw data locally and upload condensed analysis results to a database, for example.
Also, this legacy system created volumes of vibration data as well as a large number of false alarms caused by rudimentary monitoring coming from the mill equipment. “It’s common to find that two pieces of the same equipment at similar locations and operating conditions exhibit different vibration levels after several years of operation,” Wang says.
The original monitoring system relied on National Instruments’ PXI PC-based platform for the data acquisition. China Steel called on the automation supplier to upgrade this legacy system for all of its facilities in China—at least three steel production facilities. The upgraded monitoring platform, called FOMOS-AI, uses NI’s LabView software to better define the vibration data coming from the milling equipment and leverages the company’s PXI and CompactDAQ hardware for these plants.
As the monitoring system’s name indicates, machine learning is a vital component. The system looks for useful condition indicators from the vibration signals and creates four different patterns—in an operational context—for the monitored equipment: constant speed and stable load; constant speed and variable load; variable speed and load; and reciprocating.
“China Steel created these classifications and now they can apply a different algorithm for each asset type,” says Brett Burger, principal marketing manager for National Instruments. “A bump and shimmy on one type of motor might mean one thing, but the same thing on another motor might mean something completely different.”
One of the features of the new system includes a baseline setup for equipment conditions that are determined by observation of diverse changes of the vibration signals through multiple indicators. The system also establishes a multidimensional baseline/alarm setting using statistical analysis based on operation regime and machine behavior, according to China Steel.
One example of this baseline setting in the field was a main motor cooling fan for a high-speed machining (HSM) finish mill. This cooling fan showed no signs of deterioration by overall trends, but a rise of acceleration in a high-frequency band in early July gave a preview to a functional failure of the motor bearings in early November.
Since the introduction of the monitoring system upgrade, unscheduled maintenance is down. One recent example of this is in steel forging, where abrasion wear on sliding liners can occur during the slab sliding process and can lead to broken main beams, seized synchronizer bearings and synchronous shaft fractures. Operators identified three sliding liners in failure mode via spectrum analysis in the FOMOS-AI and, after a field inspection, technicians replaced these liners during a scheduled maintenance period.
Before adopting the FOMOS-AI, the total number of unscheduled downtime hours was 250 hours per year. After three years with the new monitoring system, it's at 65 hours per year. In one year, the biggest steel maker in Taiwan saved about $230,000 in maintenance costs for all of its production facilities where FOMOS-AI was running.
“The monitoring solution is more a gateway from the engineering side of the world, rather than the IT,” Burger says. “This system has sensors, processing and software that can run on it. And it has many networking capabilities, such as Modbus, serial and Ethernet.”
Optimization for many legacy plants is the logical move and these step changes are proving to be profitable to both the company and its workforce.