Thousand-fold speedup of discrete-particle-based computer-aided reactor design and scale-up

L. Lu, S. Benyahia
National Energy Technology Laboratory,
United States

Keywords: reactor, design, particles, rare earth elements

Summary:

Particulate matter is the second most manipulated material in industry after water and has been observed and studied by scientists and engineers since the earliest of human civilization. In fact, we commonly distinguish between the different phases of early human civilization by referring to granular matter such as the stone, copper, and iron ages. Therefore, we have always been fascinated by the behavior of granular flows that range from solid-like to gas-like depending on the concentration and energy of the system. Industrial systems that handle particulates consist mostly of fluidized beds that are efficient and safe because of high mass/heat transfer mechanisms that lead to uniform characteristics of products. In fact, fluidized beds are so efficient that “fluidized-bed reactor drove its competitors from the market,” accordingto an undergraduate chemical reaction textbook by Scott Fogler. A discrete approach to model these flows can accurately predict most observed behaviors, but suffers mainly from tracking large numbers of particles that make simulations too slow to be practical or even useful to industrial engineers. Therefore, we developed an approach to drastically increase the computational speed of simulations. First, we propose an accurate coarse-grained method to lump several grains into a computational parcel while conserving energy dissipation during collisions. Second, we use a novel hard-sphere approach that can correctly predict packed regions in fluidized beds and others systems. This method is validated by simulating a small bubbling fluidized bed and comparing the results with a much slower but highly accurate discrete particle method based on soft-sphere collision model. For large-scale systems, more particles can be packed into a parcel and the speedup increase to more than a thousand-fold while maintaining a reasonable accuracy. The method is then used to simulate a pilot-scale riser section of a circulating fluidized bed as well as the scale-up of a leaching reactor for extracting important rare Earth elements used in many high-technology systems. This work demonstrates the reliability of this technique to speed-up computer aided design and scale-up of multiphase flow reactors.