Introduction
Polymeric seals are used in increasingly difficult environments with ever-increasing demands on service life (several years of continuous operation without unplanned replacement). In the majority of instances, service life is proven through extensive endurance testing. This requires advanced test rigs capable of continuous operation for several years while testing the seal under application-representative temperature, pressure, and movement.
The project involved the development of a test rig which met a range of requirements for testing the performance of the Polymeric seals. We made the rig frame out of a stiffened steel box section to ensure that its natural vibration frequencies were well outside the operating frequencies specified in the seal test schedule. We used an air-cooled shaker to produce small, high-frequency movements (0.1 mm at up to 200 Hz) under closed-loop control and a servo motor lead screw assembly to produce large axial movements of ±6 mm.
We housed the piston, cylinder, and seals under test in an environmental chamber capable of maintaining setpoint test temperatures in the range of -50 °C to 300 °C. With liquid nitrogen, we achieved temperatures below ambient.
The system we developed can cycle the seals within the range of the mechanical and thermal envelope, which is fully representative of the real-life seal operating conditions.
To replicate operational conditions, we used a programmable pressure regulator to vary the pressure inside the seals. With a solenoid-operated ball valve, the seals were kept under pressure while we performed a leak test to record the pressure decay as a function of time. We used CompactRIO to control automated pressure decay tests repeated at set time intervals to ascertain the level of seal performance degradation.
Control System Overview
The test sequence was controlled by a PC-based human machine interface (HMI) connected to an NI cRIO-9074 control system via Ethernet. The PC also sent commands to the servo motor controller via a USB interface. The temperature controller of the environmental chamber was connected to CompactRIO via RS232 using the Modbus communication protocol, so CompactRIO could provide basic supervisory temperature control while concentrating processing power on more advanced control and acquisition tasks.
We also equipped the system with an analog input module to acquire pressure and seal position. We used an eddy-current-based, noncontact displacement sensor to record small movements induced by the shaker and provide feedback for the closed-loop control carried out by the FPGA.
The system measured large movements of the seal piston using a linear magnetic encoder connected to a digital input module. The FPGA processed the encoder signals.
We used an analog output module to set the programmable pressure regulator and a thermocouple module to record temperature from eight K-type thermocouples. All control settings and instrument calibration factors were stored on the cRIO-9074 so we could replace the PC without affecting system performance.
Although the HMI provided top-level test-sequence management, the cRIO-9074 controlled the rig and data acquisition. This device was essentially two separate systems in a single chassis, including a 400 MHz real-time processor and an FPGA connected to the processor via PCI bus. We used the LabVIEW FPGA Module to rapidly develop the deterministic FPGA code that performed hardware-level I/O tasks, including data acquisition and control.
Data Acquisition
Sensors connected to NI C Series input modules on the FPGA acquired the data and transferred it to the real-time controller by direct memory access (DMA) first-in-first-out memory buffers. We created a customizable front end with the C Series modules. CompactRIO acquired and processed the data and sent it to the HMI for display and logging. The system acquired and logged summary data at user-defined rates in the range of 0.1 Hz to 10 Hz, while waveform data was acquired and logged at 20 kHz.
PID and Adaptive Control
We used the FPGA to implement a proportional integral derivative (PID) algorithm for closed-loop shaker control. The shaker setpoint was supplied by a Setpoint Generator Loop operating in parallel on the FPGA. In practice, we found we could achieve adequate control using PI control only. System noise due to both vibration and measurement sources led derivative control to produce stability issues, so we did not use the derivative term.
Implementing closed control on the FPGA gave us the high loop rate and deterministic control we needed for the shaker. Additionally, we implemented high-speed adaptive control algorithms on the real-time system, including gain scheduling of control parameters and amplitude/offset adjustment, in addition to the PI control running on the FPGA.
We implemented adaptive control of the amplitude/offset by processing each complete cycle waveform and using an iterative control algorithm to update the FPGA setpoint generator to achieve the desired setpoints. For gain scheduling of the control parameters, we used the setpoint test frequency to select the required PI gain settings from a lookup table, updating the FPGA control settings as necessary.
Conclusion
Using a combination of in-house expertise in rig design, frequency, finite element analysis, and vibration with commercial off-the-shelf CompactRIO hardware and LabVIEW software, we designed and manufactured a powerful piston seal test rig capable of continuous running for prolonged periods of time. The rig was successfully commissioned and currently generates test data to determine an acceptable endurance life for polymeric seals.
CompactRIO was the ideal platform for implementing FPGA-based deterministic PI control of the shaker system alongside advanced adaptive real-time control, ensuring that the system could cope with the full range of requirements specified by the customer. Lower-cost, PC-based DAQ devices and programmable logic controllers did not offer the flexibility or performance required to implement the adaptive control system, and custom-designed solutions, such as a DSP-based system, would have resulted in unacceptable development times and increased development costs.
Print
PDF
