Making spindle wear visible: How CTRL Engineering and partners built a custom test bench for predictive maintenance
Unexpected downtime in industrial environments often originates in rotating machinery or linear drives, where early signs of wear remain hidden until failure occurs. To address this challenge, KU Leuven’s M-Group initiated a research project led by Professor Pradeep Kundu. The goal: to develop predictive maintenance strategies based on measurable signals such as vibrations and motor currents. To support this research, CTRL Engineering, together with several partners, developed a custom spindle test bench that allows researchers to capture and analyze real motion data under controlled conditions.
CTRL Engineering and the project partners
“CTRL Engineering is an engineering company from Bruges specialized in tailor-made machines, test benches, and industrial automation,” explains Mathieu Dutré, who led the project on behalf of CTRL Engineering. “In this project we were responsible for the full design of the spindle test bench: from electrical and mechanical system design to motion control programming and the integration of motor, drives, and data acquisition. We delivered both the conceptual engineering and the concrete implementation.”
The project was realized in close collaboration with KU Leuven’s M-Group and Vansichen Linear Technology, combining research requirements, domain expertise in linear drive technology, and system integration know-how into one solution. Additional partners included Beckhoff, HIWIN, MathWorks, and CN Rood, each contributing critical technologies to the final result.
A complex challenge requiring integrated solutions
The spindle test bench needed to simulate high-speed movements with precision, while simultaneously registering extremely subtle anomalies in vibration or motor current that could indicate spindle wear. “The combination of high speeds, precise positioning, and the necessity to detect very subtle signals made the project technically challenging,” says Dutré. The integration of motion control with synchronized data acquisition was key, ensuring that every movement cycle could be linked to a full dataset.
Beckhoff delivered the PLC and motion control platform that drives the test bench, including a soft HMI interface. HIWIN provided the servo motor and drive for dynamic and accurate motion. Vansichen Linear Technology contributed their mechanical expertise in linear drive systems, essential to building a reliable test setup. MathWorks enabled researchers to process and analyze the data directly in MATLAB/Simulink, connecting measurements to predictive models without extra processing steps.
The role of NI hardware and CN Rood
For data acquisition, CTRL Engineering selected NI (National Instruments) hardware, specifically the USB-4432 devices. “NI hardware is robust, portable, and extremely well-suited for data acquisition in industrial research environments,” notes Dutré. “The USB-4432 makes measurement setup simple, and its ability to trigger externally integrates seamlessly with motion control. Combined with MathWorks support packages, the data flows directly into the research environment.”
CN Rood supported the project as technology partner, advising on the selection and configuration of NI hardware. According to Dutré, their contribution was crucial: “CN Rood’s expertise ensured we could match the right hardware with the research objectives and KU Leuven’s software environment.”
Tangible Outcomes for Predictive Maintenance
The spindle test bench can execute reproducible motion cycles while continuously capturing motor currents and acceleration signals. This data is essential for KU Leuven’s predictive models, allowing researchers to correlate anomalies with early signs of wear or defects. “The test bench enables us to assess whether maintenance is needed before downtime or damage occurs,” says Dutré. Beyond its immediate use, the system provides a validated data source for further development and testing of predictive algorithms.
Next Steps in Research
Looking forward, the project team is already considering expansions. “Future upgrades may include additional sensors, like acoustic sensors for noise analysis or distance sensors to measure spindle displacement,” explains Dutré. “We also envision simulating different load profiles and integrating real-time analysis. In the long term, this could evolve into a platform not only for research but also for industrial validation of predictive maintenance strategies.”