Control valve is one of the most widely used actuator in process control. Its reliability is of great importance for the reliable performance of control loops. Valve stiction is a common problem which control loops usually suffer from. It prevents valve from responding to the control signal accurately and immediately, and even results in further performance degradation and oscillations in control loops. In view of the shortcomings of the existing valve stiction models and compensation methods, a reliability-based stochastic valve stiction model and an improved stiction compensation signal have been proposed in this research.

In the modeling part, reliability engineering is firstly introduced and valve reliability is discussed in two dimensions: the lifetime dimension and working...[ Read more ]

Control valve is one of the most widely used actuator in process control. Its reliability is of great importance for the reliable performance of control loops. Valve stiction is a common problem which control loops usually suffer from. It prevents valve from responding to the control signal accurately and immediately, and even results in further performance degradation and oscillations in control loops. In view of the shortcomings of the existing valve stiction models and compensation methods, a reliability-based stochastic valve stiction model and an improved stiction compensation signal have been proposed in this research.

In the modeling part, reliability engineering is firstly introduced and valve reliability is discussed in two dimensions: the lifetime dimension and working-condition dimension. After that, a stochastic valve stiction model is proposed based on the working-condition dimension of a 3D reliability model. Then, its identification algorithm with given SISO (single-input single-output) process model is presented and demonstrated by simulation. Besides, an iterative optimization procedure is also proposed for simultaneous identification of both the stiction model and the process model. This procedure is demonstrated by simulations as well.

In the compensating part, an improved Knocker compensation signal is proposed. This improved signal is based on the stochastic valve stiction model presented in this study and aimed to compensate more accurately with variable amplitude and adjustable pulse width. Different compensating performances of different compensation parameters are simulated and discussed. Suggestions for compensation signal parameter tuning are provided for performance optimization.

It can be concluded that the reliability-based stochastic valve stiction model provides a more comprehensive and flexible expression of stiction problem. The improved compensation signal, based on the stochastic stiction model, is capable of compensating more accurately compared with traditional Knocker signal. Therefore, this research shows a new opportunity to solve valve stiction problem.