Wednesday, November 11, 2015: 1:54 PM
251E (Salt Palace Convention Center)
Recently, Solid Oxide Fuel Cells (SOFCs) are being considered to be used as power source on ships due to the high energy efficiency and low waste emission. However, the exposure to several air contaminants like gaseous impurities and particulates is detrimental to the performance of fuel cells. In the ocean environment, sea salt particles are the most common particulates and therefore, their removal from air stream fed into cathode side of SOFC systems is highly essential. Currently, filtration is the most widely applied method due to its high efficiency and low costs. In the past, a lot of different filtration materials have been developed and used as particulates filter media, including polymer, metal, ceramic and glass material. However, they all have shown certain disadvantages under various situations. In this work, different composites were manufactured using wet-laid method for salt particles filtration performance measurement. These composites consist of a three dimensional network of sinter-locked micron size fibers with high surface area. The diameter of the fibers is in the range of 3 to 15 microns and the fiber materials can be polymer, metal, activated carbon or their combinations. As is known, filtration efficiency, pressure drop and particles loading capacity are the most important parameters during filter media performance evaluation. This work mainly focused on the theoretical part and therefore, modeling of filtration efficiency, pressure drop across the composite media and particles loading capacity were studied. Further modified PMP (Porous media permeability) equation for both laminar flow and turbulent flow were developed for pressure drop estimation. Diffusion and interception model for filtration efficiency were modified as well according to various composite media. Moreover, a new approach is proposed and experimentally verified to evaluate loading capacity of various composites for salt particles according to particles loading experiments on bench scale test rig. RMS (Root mean square) errors for all the modeling work were calculated and they were smaller than 6%, which suggests that all the models could be able to represent the experiment results very well. The discussions will be presented in depth.