| dc.description.abstract | ABSTRACT
Improvement of nutrient removal efficiency without recycling nitrified liquor was
accomplished in this research. A step feed strategy was successfully incorporated into
the TNCU-III process for enhancing the nitrogen removal without phosphorus
elimination decline. The thesis investigated the nutrient removal characteristics,
microbial population, and denitrification behavior of microorganisms by cultivating
activated sludge under stable wastewater influent condition, molecular biotechnology,
and external carbon source addition, respectively. Moreover, microbial kinetic
parameters obtained from continuous pilot-plant operation were used to optimize each
reactive volume of TNCU-III process. Under dynamic loading, real-time effluent
water quality data acquired from monitoring instruments and soft-computing
technology in terms of artificial neural networks (ANNs) and genetic algorithm (GA)
were combined to minimize effluent nutrients concentration.
In the steady-state experiments, non-, two or three step-feeding strategy resulted in
excellent treated water quality, and effluent total nitrogen (T-N) and phosphorus (T-P)
concentration were less than 15 mg/L and 2 mg/L, respectively. Results also showed
the potential of saving energy or alkaline consumption in TNCU-III process. When
three step-feeding strategy and enough hydraulic retention time (10hrs) were
implemented, the better effluent water quality was obtained (T-N<10mg/L,
T-P<0.5mg/L); meanwhile the ratio of anaerobic-oxic-anoxic retention time was
1.5:5.5:3.0.
In the microbial population identification, bacteria of Proteobacteria group, CFB
group, Chloroflexi group, and Firmicutes group were present in non step-feeding
operation. Additionally, bacteria of Proteobacteria group, CFB group, Chloroflexi
group, Nitrospirae group, and Planctomycetales existed in three step-feeding sludge
sample. Both two activated sludge samples involved specific species for nutrient
removal, carbon utilization, and filamentous bacteria. The diversity of microbial
population response for carbon uptake in three step-feeding was higher than that in
non step-feeding mode. Filamentous bacteria (Eikelboom Type 1851) existed in both
two sludge samples, and the proportion were about 10 and 12%.
According to external carbon source addition experiments, sodium acetate addition
resulted in phosphate re-release problem. If methanol was used to be the electron
donor, microorganisms needed more adaptive time to utilize methanol; meanwhile
resulted in carbon breakthrough. When glucose was fed, the final nitrate (NO3-N)
removal efficiency was higher than former carbon sources about 10~25%. Moreover,
the final PHAs contained within the biomass were more than original level and no
PO4-P re-release was observed when methanol and glucose addition, glycogen
accumulating organisms (GAOs) might exist in TNCU-III process and increased
carbon dosage for NO3-N removal.
Furthermore, wash-out effect of mixed liquor suspended solid and nutrients were
occurred in the first four reactive zones under dynamic loading. Because of the
obvious variation of C/P ratio (C/P=20~70) in anaerobic zone, the activity of
polyphosphate accumulating organisms (PAOs) decreased. When TNCU-III process
was operated at sludge recycle ratio=0.5 and Q1:Q2:Q3=(0.7:0.2:0.1), effluent T-N and
T-P concentration were less than 7.6mg/L and 0.3mg/L, respectively. Additionally,
real-time monitoring data demonstrated pH and reduction potential in anaerobic zones
and residual dissolved oxygen (DO) concentration in the first aerobic zone well
described the variation of organic loading. All the aforementioned parameters could
be considered as the indicators in TNCU-III control strategy-making procedure.
Moreover, artificial intelligence (AI) technique by means of serial ANNs describing
metabolic behavior of microorganisms and GA were successfully integrated into the
modeling system. The modeling system simulate the dynamic characteristics of
nutrients removal and minimized effluent NO3-N concentration. Due to low root mean
squared of error and high correlation coefficient between simulated and experimental
data, AI technology offered an alternative approach to simulate and optimized nutrient
removal of an EBNR system. | en_US |