This study aimed at monitoring the changes of fluorescent components in wastewater samples from 22 Korean biological wastewater treatment plants and exploring their prediction capabilities for total organic carbon (TOC), dissolved organic carbon (DOC), biochemical oxygen demand (BOD), chemical oxygen demand (COD), and
[...] Read more.
This study aimed at monitoring the changes of fluorescent components in wastewater samples from 22 Korean biological wastewater treatment plants and exploring their prediction capabilities for total organic carbon (TOC), dissolved organic carbon (DOC), biochemical oxygen demand (BOD), chemical oxygen demand (COD), and the biodegradability of the wastewater using an optical sensing technique based on fluorescence excitation emission matrices and parallel factor analysis (EEM-PARAFAC). Three fluorescent components were identified from the samples by using EEM-PARAFAC, including protein-like (C1), fulvic-like (C2) and humic-like (C3) components. C1 showed the highest removal efficiencies for all the treatment types investigated here (69% ± 26%–81% ± 8%), followed by C2 (37% ± 27%–65% ± 35%), while humic-like component (
i.e., C3) tended to be accumulated during the biological treatment processes. The percentage of C1 in total fluorescence (%C1) decreased from 54% ± 8% in the influents to 28% ± 8% in the effluents, while those of C2 and C3 (%C2 and %C3) increased from 43% ± 6% to 62% ± 9% and from 3% ± 7% to 10% ± 8%, respectively. The concentrations of TOC, DOC, BOD, and COD were the most correlated with the fluorescence intensity (
Fmax) of C1 (
r = 0.790–0.817), as compared with the other two fluorescent components. The prediction capability of C1 for TOC, BOD, and COD were improved by using multiple regression based on
Fmax of C1 and suspended solids (SS) (
r = 0.856–0.865), both of which can be easily monitored
in situ. The biodegradability of organic matter in BOD/COD were significantly correlated with each PARAFAC component and their combinations (
r = −0.598–0.613,
p < 0.001), with the highest correlation coefficient shown for %C1. The estimation capability was further enhanced by using multiple regressions based on %C1, %C2 and C3/C2 (
r = −0.691).
Full article