Why Are MBRs Commonly Designed to Nitrify? Because Changes in Organic Foulant Properties Impact Membrane Fouling
A submerged membrane bioreactor (SMBR) was operated at a non-nitrifying solids retention time (SRT) condition (2 d) and a nitrifying SRT condition (10 d) to determine the impact of nitrification on mixed liquor properties and membrane fouling. At each SRT condition, the SMBR was operated with a mixed liquor suspended solids (MLSS) concentration of approximately 1.2 g/L, a membrane flux of 17 L/m2.h, and a coarse bubble aeration rate of approximately 0.4 L/s. Membrane fouling rates were determined for the steady state conditions and the mixed liquor properties were quantified for the protein and carbohydrate content determined in the soluble fraction (soluble microbial products or SMP) and attached to the biological flocs (extracellular polymeric substances or EPS). The mixed liquor and its components (e.g. soluble fraction and suspended solids fraction) were analyzed for their resistance to filtration, which was expressed as a modified fouling index or MFI, using a deadend filtration apparatus. Molecular weights of the influent organics, SMP, EPS, and SMBR effluent were measured using a parallel membrane fractionation method. The molecular weights were measured as protein and carbohydrate concentrations using membranes with nominal molecular weight cutoffs of 10 kDa and 1 kDa. The membrane was investigated using Fourier Transform Infrared (FTIR) to attain a fingerprint of the organic foulants relative to the clean membrane. The results showed increased concentrations of high molecular weight molecules at the 2-d SRT than at the 10-d SRT. Additionally, the results showed that the filtration resistance of both the soluble and the solids portions of the mixed liquor increased in resistance as the SRT was reduced from 10-d to 2-d. The increased mixed liquor filtration resistance at the 2-d SRT resulted in an observed loss of membrane permeability that was not due to a fouled membrane. Finally, the FTIR results indicated the presence of protein on the fouled membranes from both conditions. However, the fouled membrane from the 2-d SRT exhibited strong peaks for carbohydrates as well while the fouled membrane from the 10-d SRT did not. This is consistent with the fact that the carbohydrate was mostly low molecular weight (<1 kDa) molecules and 97% of this carbohydrate passed through the membrane at the 10-d SRT, but these carbohydrate molecules were much larger at the 2-d SRT and almost 80% were retained in the reactor by the membrane. Thus, even though the total carbohydrate SMP concentration was highest at the 10-d SRT, fouling by carbohydrate was not observed on the membrane surface by FTIR, and molecular weights supported this finding with small molecules readily passing through the membrane to the effluent. Importantly, this work demonstrates the significance of the molecular weights of mixed liquor organic foulants and provides an explanation for conflicting literature that shows carbohydrate measurements sometimes correlating well with membrane fouling and other times not.