Experimental Validation of Model-Based Prognostics for Pneumatic Valves.

Area of Science:

• Engineering
• Prognostics and Health Management
• Mechanical Systems

Background:

• Valves are critical components in many systems, necessitating effective prognostic algorithms for reliability.
• A significant challenge in developing these algorithms is the scarcity of real-world failure data.
• Existing methods like run-to-failure experiments are often impractical due to the long operational lifespan of valves.

Purpose of the Study:

• To address the lack of failure data for valve prognostic algorithm development.
• To create a mechanism for rapidly observing the development of fault signatures in pneumatic valves.
• To facilitate the testing and validation of end-of-life estimation algorithms.

Main Methods:

• Development of a mobile testbed capable of physically injecting adjustable leakage faults into pneumatic valves.
• Modulation of fault magnitude at high speeds to accelerate fault signature development.
• Utilization of data generated from the testbed for the development and partial validation of a model-based prognostic approach.

Main Results:

• The testbed successfully simulates leakage faults, the most common failure mode in pneumatic valves.
• The ability to modulate fault magnitude allows for rapid observation of fault progression.
• The testbed's mobility enables connection to field valves, bridging the gap between lab development and real-world validation.

Conclusions:

• The developed testbed provides a viable solution for overcoming data scarcity in valve prognostics.
• It enables faster testing and validation of prognostic algorithms by accelerating fault signature development.
• The mobile nature of the testbed supports closer-to-validation processes for prognostic algorithm development.