FTIR mätningar av absorptionsvätskor i Bioenergy Carbon Capture and Storage processer

Sammanfattning: The effects of global warming are well understood. In order to combat this, society must move towards net zero emissions of green house gases, where carbon dioxide (CO2) plays a key role. In several IPCC climate scenarios that meet the Paris agreement, negative emission technologies that effectively remove CO2 from the atmosphere are included. Of several different technologies, bioenergy with carbon capture and storage (BECCS) is one of the most mature. This technology utilises an absorption-desorption process where CO2 is solved in liquid, producing a rich solvent, and later desorbed, resulting in pure CO2. There are, however, still challenges to implement this technology on a large scale, and one such issue is the monitoring of process streams to gain control over process conditions and system parameters.  In this project, the absorption solvent in BECCS processes were mimicked in order to determine if FTIR spectroscopy could be used to produce process parameters that are accurate, sensitive and robust. Accuracy and sensitivity are defined as the ability to correctly predict the presence and amount of species of interest in the absorption liquid. Robustness on the other hand is defined as the ability to produce precise measurements in the presence of pollutants. To evaluate how accurate and sensitive the measurements are, two different numerical models were developed and calibrated using prepared samples mimicking an absorption solvent. One model was solely based on the least square method, whereas the other was based on principal component analysis (PCA). These models were then tested on clean validation samples, as well as pilot plant samples from Stockholm Exergi, in a case study. An analysis of FTIR spectra from simulated absorption liquids showed that it could distinguish between the species of interest. Furthermore, the spectra showed that pollutants did not impact the readings in a major way. The results showed that both models produced accurate predictions of process parameters.