ASSESSMENT OF MULTI-STAGE MECHANICAL VAPOUR RECOMPRESSION PROCESS FOR DECARBONIZING STEAM GENERATION IN AN ALUMINA REFINERY

Mechanical vapour recompression (MVR) is one of the key decarbonization technologies applicable to alumina refineries for recovering water vapour that is presently wasted, which can potentially be reused in the digestion step. Since MVR can potentially be powered by renewable electricity, this path offers potential to avoid CO2 emissions resulting from steam generation, which is presently derived from fossil-fuel combustion and accounts for 70% of alumina refining emissions. This study assesses the technical performance, including the required input electric power and coefficient of performance (COP) of the MVR compression process. A multi-stage MVR model was developed in Aspen Plus, with which four scenarios were analysed, involving the compression of steam from 0.1 bar to 5 bar (Scenario 1), 10 bar (Scenario 2), 15 bar (Scenario 3) and 60 bar (Scenario 4) at an outlet flow-rate of 100 t/hr. The study reveals that some 8 to 18 stages are needed to compress steam from 0.1 bar, depending on the outlet pressure. Implementing high-speed compressors can reduce the number of stages by 3~6 compared with that using only low-speed compressors, but requires higher standards of particle removal. The compression process for the 4 scenarios utilizes a total input electric power of 19 to 42 MWe, with a calculated COP of 1.9 to 3.5, depending on the overall compression duty and isentropic efficiency. The scenario with the highest compression duty, compressing steam from 0.1 to 60 bar, has the highest input power at 42 MWe and the lowest COP of 1.9 under 70% isentropic efficiency. The economic analysis estimates that the total installation cost is 15% and 13% higher when only implementing low-speed compressors for Scenarios 2 and 4, as more stages are needed when performing the same compression duty due to their low compression ratio. However, they result in a reduction in levelized cost of steam compression by 4% and 6%, respectively. Further analysis is needed to explore alternative assumptions and increase understanding of the impact of the variations in MVR performance from various manufacturers.