Ramila H. Peiris

Identifying fouling events in a membrane-based drinking water treatment process using principal component analysis of fluorescence excitation-emission matrices

The identification of key foulants and the provision of early warning of high fouling events for drinking water treatment membrane processes is crucial for the development of effective countermeasures to membrane fouling, such as pretreatment. Principal foulants include organic, colloidal and particulate matter present in the membrane feed water. In this research, principal component analysis (PCA) of fluorescence excitation-emission matrices (EEMs) was identified as a viable tool for monitoring the performance of pre-treatment stages (in this case biological filtration), as well as ultrafiltration (UF) and nanofiltration (NF) membrane systems. In addition, fluorescence EEM-based principal component (PC) score plots, generated using the fluorescence EEMs obtained after just 1hour of UF or NF operation, could be related to high fouling events likely caused by elevated levels of particulate/colloid-like material in the biofilter effluents. The fluorescence EEM-based PCA approach presented here is sensitive enough to be used at low organic carbon levels and has potential as an early detection method to identify high fouling events, allowing appropriate operational countermeasures to be taken.

Reversible and irreversible low-pressure membrane foulants in drinking water treatment: Identification by principal component analysis of fluorescence EEM and mitigation by biofiltration pretreatment

With the increased use of membranes in drinking water treatment, fouling--particularly the hydraulically irreversible type--remains the main operating issue that hinders performance and increases operational costs. The main challenge in assessing fouling potential of feed water is to accurately detect and quantify feed water constituents responsible for membrane fouling. Utilizing fluorescence excitation-emission matrices (EEM), protein-like substances, humic and fulvic acids, and particulate/colloidal matter can be detected with high sensitivity in surface waters. The application of principal component analysis to fluorescence EEMs allowed estimation of the impact of surface water constituents on reversible and irreversible membrane fouling. This technique was applied to experimental data from a two year bench-scale study that included thirteen experiments investigating the fouling potential of Grand River water (Ontario, Canada) and the effect of biofiltration pre-treatment on the level of foulants during ultrafiltration (UF). Results showed that, although the content of protein-like substances in this membrane feed water (=biofiltered natural water) was much lower than commonly found in wastewater applications, the content of protein-like substances was still highly correlated with irreversible fouling of the UF membrane. In addition, there is evidence that protein-like substances and particulate/colloidal matter formed a combined fouling layer, which contributed to both reversible and irreversible fouling. It is suggested that fouling transitions from a reversible to an irreversible regime depending on feed composition and operating time. Direct biofiltration without prior coagulant addition reduced the protein-like content of the membrane feed water which in turn reduced the irreversible fouling potential for UF membranes. Biofilters also decreased reversible fouling, and for both types of fouling higher biofilter contact times were beneficial.

Assessing the role of feed water constituents in irreversible membrane fouling of pilot-scale ultrafiltration drinking water treatment systems.

Fluorescence excitation-emission matrix (EEM) approach together with principal component analysis (PCA) was used for assessing hydraulically irreversible fouling of three pilot-scale ultrafiltration (UF) systems containing full-scale and bench-scale hollow fiber membrane modules in drinking water treatment. These systems were operated for at least three months with extensive cycles of permeation, combination of back-pulsing and scouring and chemical cleaning. The principal component (PC) scores generated from the PCA of the fluorescence EEMs were found to be related to humic substances (HS), protein-like and colloidal/particulate matter content. PC scores of HS- and protein-like matter of the UF feed water, when considered separately, showed reasonably good correlations with the rate of hydraulically irreversible fouling for long-term UF operations. In contrast, comparatively weaker correlations for PC scores of colloidal/particulate matter and the rate of hydraulically irreversible fouling were obtained for all UF systems. Since, individual correlations could not fully explain the evolution of the rate of irreversible fouling, multi-linear regression models were developed to relate the combined effect of HS-like, protein-like and colloidal/particulate matter PC scores to the rate of hydraulically irreversible fouling for each specific UF system. These multi-linear regression models revealed significant individual and combined contribution of HS- and protein-like matter to the rate of hydraulically irreversible fouling, with protein-like matter generally showing the greatest contribution. The contribution of colloidal/particulate matter to the rate of hydraulically irreversible fouling was not as significant. The addition of polyaluminum chloride, as coagulant, to UF feed appeared to have a positive impact in reducing hydraulically irreversible fouling by these constituents. The proposed approach has applications in quantifying the individual and synergistic contribution of major natural water constituents to the rate of hydraulically irreversible membrane fouling and shows potential for controlling UF irreversible fouling in the production of drinking water.

Pilot-scale investigation of drinking water ultrafiltration membrane fouling rates using advanced data analysis techniques.

A pilot-scale investigation of the performance of biofiltration as a pre-treatment to ultrafiltration for drinking water treatment was conducted between 2008 and 2010. The objective of this study was to further understand the fouling behaviour of ultrafiltration at pilot scale and assess the utility of different foulant monitoring tools. Various fractions of natural organic matter (NOM) and colloidal/particulate matter of raw water, biofilter effluents, and membrane permeate were characterized by employing two advanced NOM characterization techniques: liquid chromatography - organic carbon detection (LC-OCD) and fluorescence excitation-emission matrices (FEEM) combined with principal component analysis (PCA). A framework of fouling rate quantification and classification was also developed and utilized in this study. In cases such as the present one where raw water quality and therefore fouling potential vary substantially, such classification can be considered essential for proper data interpretation. The individual and combined contributions of various NOM fractions and colloidal/particulate matter to hydraulically reversible and irreversible fouling were investigated using various multivariate statistical analysis techniques. Protein-like substances and biopolymers were identified as major contributors to both reversible and irreversible fouling, whereas colloidal/particulate matter can alleviate the extent of irreversible fouling. Humic-like substances contributed little to either reversible or irreversible fouling at low level fouling rates. The complementary nature of FEEM-PCA and LC-OCD for assessing the fouling potential of complex water matrices was also illustrated by this pilot-scale study.

Fluorescence analysis of NOM degradation by photocatalytic oxidation and its potential to mitigate membrane fouling in drinking water treatment

This study examined the photocatalytic oxidation of natural organic matter (NOM) as a method to mitigate membrane fouling in drinking water treatment. ZnO and TiO2 photocatalysts were tested in concentrations ranging from 0.05 g L(-1) to 0.5 g L(-1). Fluorescence peaks were used as the primary method to characterize the degradation of three specific NOM components - fulvic acid-like humic substances, humic acid-like humic substances, and protein-like substances during photocatalytic oxidation. Fluorescence peaks and Liquid Chromatography-Organic Carbon Detection (LC-OCD) analysis indicated that higher NOM degradation was obtained by photocatalytic oxidation with ZnO than with TiO2. Treatment of the feed water by ZnO photocatalytic oxidation was successful in reducing considerably the extent of hydraulically reversible and irreversible membrane fouling during ultrafiltration (UF) compared to feed water treatment with TiO2. Fouling during UF of water subjected to photocatalytic oxidation appeared to be caused by low molecular weight constituents of NOM generated during photocatalytic oxidation.

Characterizing natural colloidal/particulate-protein interactions using fluorescence-based techniques and principal component analysis.

Characterization of the interactions between natural colloidal/particulate- and protein-like matter is important for understanding their contribution to different physiochemical phenomena like membrane fouling, adsorption of bacteria onto surfaces and various applications of nanoparticles in nanomedicine and nanotoxicology. Precise interpretation of the extent of such interactions is however hindered due to the limitations of most characterization methods to allow rapid, sensitive and accurate measurements. Here we report on a fluorescence-based excitation-emission matrix (EEM) approach in combination with principal component analysis (PCA) to extract information related to the interaction between natural colloidal/particulate- and protein-like matter. Surface plasmon resonance (SPR) analysis and fiber-optic probe based surface fluorescence measurements were used to confirm that the proposed approach can be used to characterize colloidal/particulate-protein interactions at the physical level. This method has potential to be a fundamental measurement of these interactions with the advantage that it can be performed rapidly and with high sensitivity.

Optimization of a Membrane Filtration Process for Drinking Water Treatment Using Fluorescence-Based Measurements

Abstract Membrane fouling control is of paramount importance for sustainable operation of membrane-based drinking water treatment processes. Natural organic matter (NOM) is considered as the major membrane foulant and therefore its characterization is important for implementing fouling control strategies. This study proposes a fluorescence-based modeling approach for estimating and predicting the fouling dynamics in a bench-scale ultrafiltration (UF) membrane cross flow set-up for drinking water treatment. Principal component analysis (PCA) was used to extract the information that is relevant for membrane fouling from fluorescence excitation-emission matrix measurements captured during UF operation. PCA extracted principal components (PCs) that were related to major NOM membrane foulants. The model predictions were based on PC scores of retentate and permeate captured at time = 15 min of the UF experiments. The proposed fluorescence-based modeling approach is able to forecast different fouling behaviours with good accuracy. This proposed approach was then used for optimization of the UF process in which membrane back-washing times were estimated in order to achieve minimum energy consumption while ensuring maximum production of drinking water.

Fluorescence spectroscopy and principal component analysis of soy protein hydrolysate fractions and the potential to assess their antioxidant capacity characteristics.

The potential of intrinsic fluorescence and principal component analysis (PCA) to characterize the antioxidant capacity of soy protein hydrolysates (SPH) during sequential ultrafiltration (UF) and nanofiltration (NF) was evaluated. SPH was obtained by enzymatic hydrolysis of soy protein isolate. Antioxidant capacity was measured by Oxygen Radical Absorbance Capacity (ORAC) and Folin Ciocalteau Reagent (FCR) assays together with fluorescence excitation-emission matrices (EEM). PCA of the fluorescence EEMs revealed two principal components (PC1-tryptophan, PC2-tyrosine) that captured significant variance in the fluorescence spectra. Regression models between antioxidant capacity and PC1 and PC2 displayed strong linear correlations for NF fractions and a weak linear correlation for UF fractions. Clustering of UF and NF fractions according to ORACFPCA and FCRFPCA was observed. The ability of this method to extract information on contributions by tryptophan and tyrosine amino acid residues to the antioxidant capacity of SPH fractions was demonstrated.

Optimization of a Membrane Filtration Process for Drinking Water Production Using On-Line Fluorescence and Permeate Flux Measurements

Abstract The optimization of membrane filtration processes for controlling fouling is essential for the sustainable application of membrane processes in drinking water treatment applications. Natural organic matter (NOM) and colloidal/particulate matter are considered as the major membrane foulants and therefore their characterization is essential for implementing optimization strategies. In a previous work by the authors, a fluorescence-based modeling approach was developed for prediction of the fouling dynamics and for optimization of a bench-scale ultrafiltration (UF) membrane cross flow set-up for drinking water treatment. In this study, this models predictive ability was improved by updating the model parameters based on current process measurements. The Extended Kalman Filter (EKF) approach was used to achieve this objective. The EKF approach was implemented to accomplish online-adaptive estimation of key model parameters based on either current (time = t) UF flux measurements or principal component (PC) scores related to current fluorescence measurements of membrane permeate. The model predictions and the corresponding experimental UF flux data of different membrane fouling situations revealed that on-line permeate flux-based parameter adaptation result in improved model predictions as compared to PC scores’ based adaptation. The resulting model based estimator was then employed in the optimization of the UF process in which membrane back-washing times were estimated in order to achieve minimum energy consumption while ensuring maximum production of drinking water.