Decolorization of Reactive Red 120 in a micro-reactor by linking chemical effects with cavitation dynamics through data-driven analysisPetkovšek, Martin (Avtor) Benato, Alberto (Avtor) Dular, Matevž (Avtor) Kozjek, Dominik (Avtor) Sech, Edoardo (Avtor) Šmid, Alenka (Avtor) Zupanc, Mojca (Avtor) microfluidicsflow characterizationazo dyehydrogen peroxidesalicylic acid dosimetryTextile wastewater containing azo dyes is difficult to treat due to their poor biodegradability, and although hydrodynamic cavitation has shown potential for their removal, the effect of microfluidic systems remains to be investigated. In this study, a Venturi-type cavitation micro-reactor was evaluated for the decolorization of Reactive Red 120 by systematically investigating the effects of initial dye concentration, solution's pH, H▫$_2$▫O▫$_2$▫ addition, inlet pressure, and sample volume. Cavitation dynamics were characterized by visualization and pressure measurements, •OH generation was assessed by salicylic acid dosimetry, and most important process parameters were identified by statistical analysis and machine learning. Decolorization increased with the number of cavitation passes under all conditions. Acidic conditions (pH = 2) resulted in the highest observed removal. The addition of 1700 μM H▫$_2$▫O▫$_2$▫ into 250 mL samples further improved decolorization and resulted in removal of 6.2 ± 1.2 mg after 35 cavitation passes. Pressure difference at 700 kPa showed maximum decolorization (10.2 ± 2.5 mg in 250 mL), whereas •OH generation increased up to 900 kPa (2.53 ± 0.4 μg/mL). Sample volume had the strongest effect where dye removal increased from 1.7 mg at 125 mL to 55.3 mg at 1000 mL after 35 cavitation passes at 100 mg/L. Statistical analysis confirmed that initial dye concentration, cavitation passes, pH, and H▫$_2$▫O▫$_2$▫ affected decolorization. Machine learning analysis further showed that in this study dye removal could be predicted from operating conditions with R▫$^2$▫ = 0.84, with sample volume and initial dye concentration identified as the most important variables.20262026-10-08 13:42:16Članek v reviji189533sl