Malgorzata Peszynska,Ralph E. Showalter
Abstract:We develop a practical discrete model of hysteresis based on nonlinear play and generalized play, for use in first-order conservation laws with applications to adsorption-desorption hysteresis models. The model is easy to calibrate from sparse data, and offers rich secondary curves. We compare it with discrete regularized Preisach models. We also prove well-posedness and numerical stability of the class of hysteresis operators involving all those types, describe implementation and present numerical examples using experimental data.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to develop a practical discrete model to approximate and calibrate the hysteresis function \(u \to w = H(u)\) for numerically solving partial differential equations (PDEs), especially first - order conservation laws. Specifically, the research aims to:
1. **Establish a simple and robust algorithm**: This algorithm can approximate and calibrate the hysteresis operator \(H(u)\) and can be used in numerical schemes, especially in flow and transport problems in porous media.
2. **Handle the calibration problem of sparse data**: The traditional Preisach model requires detailed data, but in practical applications, experimental data are usually sparse. Therefore, this paper proposes a model that can be calibrated from sparse data to generate smooth output curves.
3. **Compare different types of hysteresis models**: Including the generalized Play model, the discrete regularized Preisach model, and the K - nonlinear Play model. By comparing the advantages and disadvantages of these models, select the model most suitable for a specific application scenario.
4. **Prove numerical stability and well - posedness**: For the class involving all these types of hysteresis operators, prove its well - posedness and numerical stability, describe its implementation method, and provide numerical examples.
### Key problems and background
The hysteresis phenomenon is a nonlinear phenomenon in which the output of a process depends not only on the independent variable but also on the history of the process and is independent of the rate. This phenomenon is widely present in electromagnetism, plasticity, phase change, flow of multiphase fluids in porous media, etc. In particular, the adsorption - desorption hysteresis phenomenon is especially important in carbon sequestration and wood science and engineering.
### Research motivation
The main motivation of this research comes from flow and transport problems in porous media, especially the adsorption - desorption hysteresis phenomenon. In carbon sequestration and wood science and engineering, accurate modeling of the hysteresis phenomenon is crucial for predictive simulation. However, experimental data are usually sparse, which limits the application of the traditional Preisach model. Therefore, developing a practical hysteresis model that can be calibrated from sparse data is of great significance.
### Main contributions
1. **Proposed a new robust and simple algorithm**: This algorithm can calibrate the hysteresis function from sparse data and generate a smooth output curve.
2. **Extended existing results**: Extended previous research results to a broader class of hysteresis models.
3. **Provided detailed calibration methods**: Explained how to calibrate the model from experimental data and provided details of the solver.
4. **Compared the advantages and disadvantages of different models**: Verified the performance of different models through numerical experiments, providing guidance for practical applications.
### Mathematical model
Consider the following partial differential equation model:
\[
\frac{\partial}{\partial t}(a(u)+H(u)) + A(u)=f,
\]
where \(u\) is the unknown function, \(f\) is the external source term, \(A\) is the transport operator (convection and/or diffusion, generally nonlinear), and \(a(\cdot)\) is a strongly monotonic function. This problem is posed in the sense of distribution, defined in an appropriate functional space, and with boundary and initial conditions.
### Conclusion
This paper solves the problem of calibrating the hysteresis function from sparse data by developing a new robust and simple algorithm, and provides detailed calibration methods and numerical experiment results. This is of great significance for modeling flow and transport problems in porous media, especially the adsorption - desorption hysteresis phenomenon.