SUMMARY

A proposed design for concentrating solar power receiver uese a granular material - such as sand, which is inert, inexpensive, and able to operate at relatively high temperatures, thereby increasing thermodynamic efficiency - as the heat transfer and energy storage medium. An early design of particle heating receivers (PHR) utilizes a falling curtain of particles which directly absorbs the concentrated solar radiation. However, falling curtain receivers have several disadvantages including significant heat and particle losses and short residence time within the irradiation zone. One design proposal which overcomes these challenges is the so called impeded flow PHR design, in which the particles flow over, around, or through a series of obstacles in the flow path. This reduces the average velocity of the particles, thereby increasing residence time in the irradiation zone of the receiver. It also reduces heat and particle losses from the receiver. However, the hydrodynamics of complex granular flows are not well understood, rendering a priori design of impeded flow PHR geometries difficult. This investigation had two main goals. First, a series of representative impeded flow PHR geometries were constructed, instrumented and tested, allowing detailed quantitative measurement of such parameters such as mass flux and particle velocity distribution within the receiver geometry. This allowed the development of performance envelopes for the various receiver geometries, which may be useful for future receiver designers. Second, numerical models of the receiver designs were developed using two different approaches - the discrete element method (DEM), which tracks individual particles and models particle collisions as small overlaps, and a two-fluid finite volume method (FVM), in which a granular flow is modeled using typical computational fluid dynamics methods. Predictions of both models were compared against experimental data. It was found that the DEM models generally described the granular flow characteristics better than the FVM models, and were generally able to run faster on parallel computing resources. However, inclusion of heat transfer may be more easily accomplished in future FVM models.