MODELING AND OPTIMIZATION OF A MICROPARTICLES SEPARATION-CLASSIFICATION PROCESS

CECILIA I PAULO; EUGENIA BORSA; M. SOLEDAD DIAZ; MIRTA R. BARBOSA

La Plata

Workshop; Pan-American Advanced Studies Institute (PASI 2014) on Frontiers in Particulate Media: From Fundamentals to Applications; 2014

National Science Foundation (USA), Department of Energy (USA), New Jersey Institute of Tecnology, CONICET La Plata, Universidad Tecnológica Nacional - Facultad Regional La Plata.

In this paper a nonlinear programming problem (NLP) was developed to optimize the microparticles separation and classification process, for a diameters range of 0.1-1,000 µm. The units included in the process are a powder feeder, a high efficiency cyclone, a bag filter and a fan that drives the movement of air and particulates from the feeder to the filter. The cases studied involve two particulate materials: dolomite and limestone, each one with three different particles size distributions; fifteen initials powder concentrations, varying between 2 and 20,000 g/m3 of air feeding the process; and three different types of design models for high efficiency cyclones: Stairmand, Swift and Echeverri Londoño. Different models of cyclone separation efficiency were analyzed and compared. Equations for the filter efficiency, fan and powder feeder operations were also added to the NLP model. Mass and energy balances were formulated and solve for a 2.4 m3/s air flow process. The mass fractions expressions corresponding to the outlet streams of cyclone and filter were derived taking into account the equipments efficiencies. These nonlinear expressions and the design constraints of each model design were incorporated into the optimization problem.The NLP formulated for each case, including 104 continuous variables and 83 equations, was implemented in GAMS and solved with CONOPT. The numerical results show that the separation efficiency increases with the initial microparticles concentration feeding to the process, for all the cases analyzed. The cyclone Swift proved the optimal cyclone efficiency value 93.3%, for the maximum particle concentration studied, reaching an overall process efficiency of 96.7%. The proposed nonlinear model has provided useful information for better understanding of the overall microparticles separation-classification process. Moreover, it is a valuable tool for the optimal design of these processes, due to its versatility to work at different operating conditions and different materials. The use of computational tools and constraint-based optimization has proven to be an effective method to adequately predict the behavior of the system.