Dynamic Reduced Order Models for Simulating Bubbling Fluidized Bed Adsorbers

Abstract

Spatially distributed first-principles process models provide an accurate physical description of chemical processes, but lead to large-scale models whose numerical solution can be challenging and computationally expensive. Therefore, fast reduced order models are required for model-based real-time applications, such as advanced process control and dynamic real-time optimization. In this paper, we focus on the model reduction of a bubbling fluidized bed (BFB) adsorber, which is a key component of a postcombustion carbon capture system. From a temporal aspect, dynamic reduced models are generated using the nullspace projection and eigenvalue analysis method, with the basic idea of quasi-steady state approximation for the states with fast dynamics. From a spatial aspect, dynamic reduced models are developed using orthogonal collocation and proper orthogonal decomposition to reduce the size of the rigorous model. Finally, a computationally efficient and accurate dynamic reduced model is developed for the BFB...