Evaluation of cleaning conditions for polymeric ultrafiltration membranes after black liquor filtration

Sammanfattning

This thesis investigated the chemical cleaning of a polymeric ultrafiltration membrane after black liquor filtration. The main objective was to evaluate how cleaning temperature, cleaning- agent concentration, and cleaning time affected flux recovery after cleaning. Black liquor filtration caused a gradual decline in permeate flux during the 25 h fouling experiment, confirming that fouling developed on the membrane. After fouling, eight cleaning experiments were performed using a full factorial Design of Experiments approach. The fouling factor was not identical in all experiments, which means that the initial degree of fouling varied between membrane samples and should be considered when interpreting the cleaning results. The cleaning experiments showed that the highest flux recovery was obtained using Ultrasil 110 at 50 °C and 1 wt.%. The highest measured recovery in the experimental design was achieved after 30 min of cleaning, while the model suggested 60 min as the most favourable condition within the investigated range. Therefore, the validation experiment was performed at 50 °C, 1 wt.% Ultrasil 110, and 60 min. The validation experiment resulted in a flux recovery of 76.43%, which was slightly higher than the best result obtained in the original experimental design. The model should be treated as a screening tool rather than a strong predictive model because of the limited dataset and the absence of statistically significant effects. Membrane characterisation was performed using scanning electron microscopy with energy- dispersive X-ray spectroscopy (SEM–EDS), atomic force microscopy (AFM), and Brunauer– Emmett–Teller (BET) nitrogen adsorption–desorption analysis. SEM–EDS and AFM showed that black liquor fouling caused local surface deposits and increased surface heterogeneity. After cleaning, the surface appeared partly restored, although some deposits and irregularities remained. BET analysis showed only small differences between the conditioned, fouled, and cleaned samples, suggesting that major changes in the nitrogen-accessible pore structure were not detected. Overall, the results showed that cleaning temperature and cleaning-agent concentration were the most important parameters within the tested range, while cleaning time had a weaker effect. Chemical cleaning improved membrane performance after black liquor fouling, but additional experiments with repetitions and more factor levels would be needed to confirm the trends and improve reliability of the model.