(736a) CFD-DEM: Modeling the Small to Understand the Large | AIChE

(736a) CFD-DEM: Modeling the Small to Understand the Large

Authors 

Fullmer, W. - Presenter, National Energy Technology Laboratory
Cocco, R., Particulate Solid Research, Inc. (PSRI)
Liu, P., University of Colorado at Boulder
Hrenya, C. M., University of Colorado at Boulder
Gas-solid flows in industrial devices present a challenge for prediction, scale-up and optimization with numerical simulations. Direct numerical simulation (DNS), which is essentially free of closures, is limited by the computational cost associated with solving for the motion of every particle and the fluid flow around the particles. At present, DNS can reach O(104) particles, many orders of magnitude below most industrial problems. Coarse-grain methods, such as filtered two-fluid models (TFMs) and multi-phase particle in cell (MP-PIC) method. Coarse-grained methods are able to solve larger problems by using continuum grid cells or discrete parcels each containing up to ~O(105) particles. However, the closures, assumptions and simplifications required by coarse-grain methods may lead to large sources of uncertainty in model predictions. Coupled computational fluid dynamics-discrete element method (CFD-DEM) offers a compromise between computational overhead and sources of uncertainty. Though DEM has seen a recent explosion among academic researchers, as evidenced by the number of relevant papers doubling roughly every four years (a rate of about half of Mooreâ??s law), a large potential exists to extend the application of DEM to industries. The challenge still lies in the relatively high computational demand for full industrial systems, which commonly contains up to O(1014) particles, exceeding the state-of-art of DEM simulations of O(107). To guide the future enhancement of DEM solvers toward better utilization in industry, a gauge on current and future industrial perspective on DEM was conducted by a survey of CFD champions working at 34 PSRI member companies. Interesting trends and conclusions drawn from the 18 responses will be presented, in terms of user demographics, expected expenditure on personnel, acceptable time commitment, urgency in improvement of code functionality, among others. Additionally, a few examples of how CFD-DEM can be used in a manner to provide insightful, value-added information will also be presented.