A Comparison Between the Monte Carlo Ray Trace and the FLUENT Discrete Ordinates Methods for Treating Solar Input to a Particle Receiver

Abstract

A new type of high temperature solar receiver for Brayton Cycle power towers is being designed and built in the Combustion and Solar Energy Laboratory at San Diego State University under a DOE Sunshot Award. The Small Particle Solar Receiver is a pressurized vessel with a window to admit concentrated solar radiation that utilizes a gas-particle suspension for absorption and heat transfer. As the particles absorb the radiation that enters the receiver through the window, the carrier fluid (air in this case) heats which oxidizes the particles and the flow leaves the receiver as a clear gas stream. After passing through an in-line combustor if needed, this hot gas is used to power a turbine to generate electricity. The numerical modelling of the receiver is broken into three main pieces: Monte Carlo Ray Trace (MCRT) method (written in FORTRAN), ANSYS Fluent (CFD), and the User Defined function (written in C code) for oxidation. Each piece has its advantages, disadvantages, and limitations and the three pieces are coupled to finalize the calculation. While we have successfully demonstrated this approach to obtaining the velocity and temperature fields, one big challenge to this method is that the definition of the geometry is a time consuming programming task when using MCRT. On the other hand, arbitrary geometries can be easily modelled by Computational Fluid Dynamics (CFD) codes such as FLUENT. The goal of this study is to limit the use of MCRT method to determining the appropriate input boundary condition on the outside of the window of the receiver and to use the built-in Discrete Ordinates (DO) method for all the radiation internal to the receiver and leaving the receiver due to emission. To reach the goal, this paper focuses on the DO method implemented within FLUENT. An earlier study on this subject is based and advanced. Appropriate radiation input for the DO method is extensively discussed. MIRVAL is used to simulate the heliostat field and VEGAS is used to simulate a lab-scale solar simulator; both of these codes utilize the MCRT method and provide intensity information on a surface. Output from these codes is discretized into DO parameters allowing the solution to proceed in FLUENT. Suitable benchmarks in FLUENT are used in a cylindrical geometry representing the receiver for the comparison and validation. This method will allow FLUENT to be used for a variety of problems involving concentrated solar energy.