This article summarizes a detailed scientific study of a particle classification system to be used in a fully integrated regenerative calcium looping (FICaL) system for CO2 capture. In conventional calcium looping, the required calcination heat is provided by a separate oxyfuel combustor, which needs an air separation unit (ASU) to provide the required oxygen for the process. The ASU demands a considerable amount of power, which gives an energy penalty of typically 5 % to the power plant. However, in the FICaL system, the required heat for the calcination is instead supplied indirectly from the main combustor in the power plant, so there is a no need for an ASU in the system. Based on process simulation studies done with Aspen Plus®, this may reduce the energy penalty to values in the order of 1-2 %. The indirect heat transfer is done by using inert heat transfer (HT) particles that are heated up in the combustor and then transferred to the calciner where the sorbent material is heated up after mixing with the hotter inert particles. Thereby the sorbent material is calcined. However, after calcination, the sorbent and heat transfer particles have to be separated. Hence, an efficient classifier is required. The current work has therefore focused on designing, constructing and doing experiments in a novel particle classification system. A novel cross-flow fluidized bed classifier was designed, using computational fluid dynamics (CFD) simulations as a tool, in order to separate the sorbent and HT particles exiting from the calciner. The classifier, which has no mechanical moving parts exposed to very high temperature, can be operated under the required high-temperature conditions prevailing in a full-scale plant. Two different cold-flow lab-scale versions of the classifier was built and used for a large number of experiments. In the classification, the aim is to minimize the loss of sorbent particles via the bottom exit from the classifier, and also to minimize the loss of HT particles via the top exit from the classifier. The second and improved version was able to classify very well a mixture of down-scaled sorbent particles (zirconia) and down-scaled HT particles (steel). The experiments with the improved classifier version gave particle losses in the order of 2-3 %, values that are close to what can be seen as acceptable in a full-scale hot-flow system. Extensive CFD simulations were carried out with the commercial software Barracuda® 17.1 to investigate in detail how the different particle types behave in the classifier. Even if the exact particle losses were not well predicted, Barracuda was able to predict the general gas-solids flow behavior and proved to be a useful tool in the design process. Different drag models were used to reproduce the experimental findings and validate the CFD model. Barracuda was also used to simulate the classification process under hot-flow conditions and indicated that the classifier will also perform well under such conditions. The results of the corresponding research work are promising as the classifier is able to give a high degree of purity of the particle streams leaving the classifier. The iterative design and modelling effort from this research work has produced a functional, high-efficiency classification concept.
Read full abstract