Pilot-scale GAC column testing in Canada

Multibarrier treatment system in a water treatment plant (WTP) in southern Ontario, Canada, includes the use of UV–H₂O₂ as an advanced oxidation treatment for T&O control (Fig. 2). The studied Canadian WTP is using GAC contactor to quench the residual H₂O₂ (Fig. 2). This pilot study was conducted to verify the performance of virgin and aged GAC (from quenching H₂O₂) at different depths for the removal of geosmin and MIB. Figure 3 demonstrates the schematic representation and the complete installation of the pilot setup that was installed in the studied Canadian WTP. The conditions of the columns are presented in Table 1.

Figure 2.

Schematic representation of a UF and advanced oxidation plant in Ontario, Canada, and fate of algal cells and their by-products during the treatment: red square highlights the UV–H₂O₂ process that causes the residual H₂O₂ in the water post-AOP

Notes: *T&O, T&O compounds; **UV, ultraviolet advanced oxidation; and ***GAC, granular activated carbon.

Figure 3.

Pilot setup at the Canadian plant: (A) schematic representation of the GAC columns and (B) general view of the fully built functional pilot and details of the sampling ports.

Table 1.

Filter column conditions: pilot-scale (Canada) and laboratory-scale (Australia) parameters.

The GACs were selected based on the material (Table 2) used for their preparation and also the current application of a particular type at the studied Canadian WTP. The characteristics of all GAC types are presented in Table 2. Treatment at the Canadian WTP (Fig. 2) comprised prechlorination in water temperatures >12 °C, ultrafiltration (UF), UV disinfection, injection of H₂O₂ during the algal bloom season (September to November), adsorption on GAC, and postchlorination. The plant draws water from the Canadian side of Lake Ontario, with average total organic carbon of 2.5 mg-C/L and 8.0 pH. There was negligible chlorine remaining by the time the water was directed onto the GAC pilot setup. The GAC contactors had been in service for approximately 1 year (130,000 bed volumes) and 2 years (260,000 bed volumes). The contactors had only been exposed to H₂O₂ on one occasion from September 2012 to November 2012 (approximately 32,000 bed volumes) with an average concentration of 4 mg/L H₂O₂ during that period. Therefore, the pilot GAC columns were aged using 4 mg/L of H₂O₂ for 3 and 6 months, 33,000 and 65,000 bed volumes, respectively.

Table 2.

Detailed characteristics of the GAC used.

An array of 16 pilot-scale glass columns containing virgin GACs was used for this test (Fig. 3). The 16 columns allowed for the testing of the GAC in duplicate. The columns were each 2.5 cm in diameter and 128 cm in depth to simulate the GAC depth of the full-scale plant. Quality control tests confirmed that such a diameter led to perfect plug flow conditions with no short-circuiting as a result of wall effects. Thus, these control tests confirmed that adsorption kinetics were consistent with the plug flow within the columns. Water obtained post UF membrane and prior to UV disinfection at the WTP was directed through pilot-scale GAC columns (Fig. 3) with empty bed contact time (EBCT) of 4.1 minutes (similar to the full-scale plant operational EBCT). Per each GAC type, two columns were aged using only the membrane permeate, and another two columns were aged using membrane permeate spiked with 4 mg/L of H₂O₂. The columns were run in a continuous downflow operation, with no provision for backwash, as the quality of the UF membrane permeate feed led to no observable accumulation of head loss over the period of operation. At times 0 month (virgin), after 3 months, and 6 months of aging, the capacity of GACs to remove 100 ng/L of geosmin and MIB was evaluated (Table 3). At these predetermined time intervals (ie, 0, 3, and 6 months), the aging process was stopped, and membrane permeate was spiked with 100 ng/L of geosmin and MIB, which is the highest detected value in Lake Ontario, Canada. The geosmin and MIB containing membrane permeate were filtered through the columns for equivalent of 1-week bed volume. Geosmin and MIB concentrations were measured along the depth of the column (ie, inlet; 25 cm, 50 cm, 75 cm, and 100 cm; and outlet).

Table 3.

Range (including analytical method bias) of concentrations used for spiking periods, T = 0, 3, and 6 months, for geosmin and MIB.

Laboratory-scale GAC column testing in Australia

Figure 4A shows the schematic representation of the column setup for the Australian trials (without advanced oxidation prior to GAC), and Figure 4B is a photograph of the columns in the laboratory at a WTP located in South Australia. Four laboratory-scale columns were set up and run at the Australian WTP, fed with water from the settled water ducts. The performance of two types of GAC for the removal of geosmin and MIB was verified using the laboratory-scale experiment over a period of 6 months. This setup provides the opportunity to conduct the test in duplicate. Column conditions and GAC details are presented in Tables 1 and 2, respectively. Spiking trials were conducted at 0, 3, and 6 months after commissioning to determine the removal of cyanobacterial metabolites. Two inlet concentrations of metabolites were used for the conducted trials. During the spiking trials, for 1–2 weeks, metabolites were dosed in continuously, and samples were collected from the influent and filtrate daily for analysis. The percent removal values and the mean removal values for each inlet concentration (errors associated: standard deviation) were calculated. Table 3 presents the target concentration for the spiking trials.

Figure 4.

Laboratory setup at the Australian plant: (A) schematic representation of the columns and (B) general view of the functional columns.

Cyanobacterial metabolite materials and water quality analysis

Geosmin and MIB were purchased from commercial suppliers (Novachem Pty Ltd. and Ultrafine Chemicals) and dissolved in ultrapure water (Millipore Pty Ltd.) to prepare a stock solution. Aliquots from the stock solution were then dosed into waters at the required target concentrations.

Samples for dissolved organic Carbon (DOC) and UV absorbance were filtered through 0.45 µm prerinsed (ultrapure water) membranes. DOC was measured using a Sievers 900 Total Organic Carbon Analyzer (GE Analytical Instruments). UV absorbance at 254 nm was measured through a 1 cm quartz cell using an Evolution 60 Spectrophotometer (Thermo Scientific).

All analyses were conducted in triplicate for each sample. Bench-scale experiment samples for geosmin and MIB analyses were concentrated using a solid-phase micro extraction syringe fiber (Gerstel) and analyzed on a 7890A Gas Chromatograph System with 5975C VL Series Mass Selective Detector (Agilent Technologies) against quantified labeled internal standards (Ultrafine Chemicals). Full details of this method have been documented by Graham and Hayes.⁴¹ The reporting limits for MIB and geosmin were 4 and 2 ng/L, respectively. Significant difference between the removals of T&O compounds using GAC exposed to different aging processes was statistically assessed.

Two methods were used to monitor the biological activity: (1) adenosine triphosphate (ATP) and (2) scanning electron microscopy (SEM). In Canada, total ATP (in nanogram/gram) is measured using a commercially available deposit and surface analysis (DSM™) test kit (Luminultra). In Australia, ATP was measured (in nanogram/gram) using a method outlined by McDowall.⁴² Samples of the filter media were taken from the top of each column prior to and immediately after the spiking events.

For SEM analysis, samples of the filter media were taken before and after the 6-month laboratory- and pilot-scale trails. They were fixed for 1 hour using EM fixative (4% paraformaldehyde/1.25% glutaraldehyde in PBS, +4% sucrose, and pH 7.2), washed in buffer (Phosphate-buffered saline (PBS) and +4% sucrose), post-fixed with 2% osmium tetroxide in water for 30 minutes, and then dehydrated with increasing ethanol solutions (70%, 90%, and 100%). Subsequently, the samples underwent critical point drying and were mounted on a stub and coated with carbon ready for analysis on a Field Emission Scanning Electron Microscope (FESEM). Samples of the virgin filter media were also mounted and coated ready for analysis. Photographs were obtained using a Philips XL30 FESEM.

Impact of advanced UV–H₂O₂ oxidation and H₂O₂ residual quenching on GAC capacity to remove T&O compounds

AOP used in the studied Canadian WTP (Fig. 2) includes H₂O₂ dosage prior to UV oxidation to generate very strong OH^· radical oxidant.^43,44 The elementary reaction in the UV–H₂O₂ advanced oxidation system is as follows (Equation 1):

where hv is the UV irradiation.

In this plant, GAC contactors are used to remove the residual H₂O₂ post AOP from water prior to chlorination. Previous publications^31,36 have demonstrated that for all types of GAC (Table 2) used in this study, the H₂O₂ quenching rate decreased significantly up to 25,000 bed volumes and then stabilized to a constant rate (±5 variation) for over 2 years of operation. These results are in accordance with the performance of GAC in 3 full-scale plants in North America and Europe to remove H₂O₂ for over 6 years without media replacement.³¹

During the operation of the Canadian plant, H₂O₂ is only dosed during the bloom season (September to November). Therefore, the AOP is only available during this time period as a strong oxidation barrier against T&O compounds in water. For the rest of the operational time, adsorption on GAC is the only treatment process for the removal of these compounds. The pilot study was used to study the impact of H₂O₂ quenching on GAC capacity to adsorb geosmin and MIB to ensure the efficiency of available treatment barriers at all times. For geosmin treatment, 25–50 cm of virgin GACs were sufficient for the removal of 100 ng/L to below detection limit, which is 2 ng/L (Fig. 5). However, after 3 months of quenching 4 mg/L of H₂O₂ with an EBCT of 4.1 minutes, the entire 128-cm column depth was required to remove >93% of the geosmin. No significant difference was observed between the geosmin removal capacity of columns with (Fig. 5) and without (data not shown) H₂O₂ quenching.

Figure 5.

Geosmin removal using virgin carbon, and after 3 and 6 months of H₂O₂ quenching. Similar geosmin concentrations were measured in water samples from columns (at all depths) without H₂O₂ quenching (data not shown).

Note: *BDL, below detection limit.

All GAC types and ages, even virgin GAC, were less efficient for the removal of MIB compared to geosmin. Furthermore, after 3 months of pilot operation, MIB centration in column permeates reached >10 ng/L. However, similar to geosmin removal, the GAC capacities for MIB removal were not significantly affected by the H₂O₂ quenching, as similar removal values were observed with (Fig. 6) and without (data not shown) H₂O₂ quenching. Ndiongue et al.⁴⁵ demonstrated that an increase in GAC depth or EBCT would significantly improve the removal of geosmin and MIB from water. Furthermore, other studies have also observed better (minimum 13%) geosmin removal compared to MIB removal by GAC.^46,47

Figure 6.

MIB removal using virgin carbon, and after 3 and 6 months of H₂O₂ quenching. Similar MIB concentrations were measured in water samples from columns (at all depths) without H₂O₂ quenching (data not shown).

All four types of GAC provided similar performance for the removal of geosmin with and without advanced oxidation (Fig. 5). Centaur and F300 provided the best MIB removal in both the absence and presence of H₂O₂ quenching aging process (Fig. 6). Independent of the GAC type, the top 25 cm of GAC was the most effective for the removal of geosmin (Fig. 5). In the meantime, the top 75 cm of GAC removed the greatest percentage of MIB (Fig. 6).

While biofilm formation covers the adsorbent surface of carbon, the biological activity of microorganisms could lead to biodegradation of T&O compounds.^48–495051 SEM is a qualitative method that can be used to investigate the existence and composition of a biofilm on filter media. Although SEM does not allow the identification of individual organisms, it can be used to show the abundance and diversity of organisms within the biofilm. SEM analysis provides the opportunity for visual observation of GAC surface during AOPs and adsorption experiences (Fig. 7). SEM images from the Canadian pilot study for six virgin GACs and six GAC aged with membrane permeate and H₂O₂ after 33,000 bed volumes are shown in Figures 7A–7H. These SEMs demonstrate that a significant biomass has accumulated during the trials. The observation of biofilm formation on GAC surface after aging was also reported in the study by Yu et al.³⁴

Figure 7.

SEM data: Centaur (A) virgin, and (B) aged with NOM at 45,000 bed volumes; F300 (C) virgin, and (D) aged with NOM at 33,000 bed volumes; HD3000 (E) virgin, and (F) aged with NOM at 33,000 bed volumes; TN5 (G) virgin, and (H) aged with NOM at 33,000 bed volumes (.Figs. 7A to 7H scale is 5 µm); AND GA1000N (AWQC) particle size 0.5–0.7 mm diameter (I) virgin (Fig. 7.I.1 scale is 50 µm and Fig. 7.I.2 scale is 10 µm), and (J) 6 months after commissioning (Figs. 7.j.1 and 7.j.2 are taken from different angels at 5 µm scale).

The biofilms are composed of a variety of organisms, including bacteria, protozoa, fungi, and diatoms, and it is held together by extracellular polymeric material. It is not possible to identify individual microorganisms with analytical methods used in this study, and so no differences can be established between the different GAC types. However, it is clear that abundance and diversity of microorganisms exist within the thriving biofilms. The level of biological activity of the established biofilm would determine the potential for the biodegradation of T&O compounds. ATP is present in all active microorganisms; therefore, ATP concentrations can be used as a measure of biological activity and an indicator for the presence of an active biofilm on filter media. Samples of the filter media were taken prior to the spiking trials and directly after collecting the final effluent from the spiking trial to look at the impact of the spiked settled water on the biological activity of the filter media. GAC appeared to have negligible ATP activity before and after the spiking trials.

GAC capacity to remove T&O compounds in the absence of advanced oxidation barriers

To compare the geosmin and MIB removal with GAC depth with and without advanced oxidation, the results from the Canadian study were compared to the laboratory-based column study conducted in Australia. H₂O₂ removal on GAC was not the focus of the study conducted in Australia. In all Australian assays, geosmin was removed to below detection limit (data not shown). Figure 8 demonstrates that for MIB removal, the F300 GAC was performed as well as the GA1000N over the trial period. An application of analysis of variance (P > 0.1) to the GAC data showed that there was no significant decrease in the removal by the GAC over the trial period. As for the percent removals for the two inlet concentrations (low and high), similar results were obtained for GAC aged for 3 and 6 months.

Figure 8.

MIB removal by two types of GAC during the Australian experiment.

Note: *BDL, below detection limit.

The results for the duplicate columns over 6 months demonstrate good (<7% error) reproducibility of the column data over the trial. GAC was very effective for the removal of MIB and geosmin over the trial period with an average of 98% and 100% removal, respectively. This result supports the previous work by Ho,⁴⁸ who investigated the removal of MIB and geosmin by GAC at the laboratory and pilot scales. Such high removals had been shown previously to be attributable to physical adsorption only^50,51; however, as the removal values were high throughout the test period, it was not possible to differentiate between physical adsorption and biological degradation in this study.

Figures 7I and 7J show SEM images of the initial GAC media as new, prior to the column trials and after 6 months in contact with settled water from the Australian WTP. However, similar to the Canadian pilot study, no significant biological activity was recorded. This result was unexpected, based on previous publications.⁵² GAC surface is known to promote significant biofilm growth within 4–6 weeks of filter commissioning.⁵³ The low value of ATP on the GAC suggested that the adsorbent interfered in some way with the ATP analysis. These results suggest that no biological removal had occurred.

This study has focused on the fate of dissolved geosmin and MIB during advanced oxidation and adsorption and/or biodegradation by GAC. The multibarrier treatment systems that are in use in the studied conditions (Fig. 2) prior to GAC contactors reduce significantly the risk of breakthrough of intact cells and intracellular T&O compounds. However, plants with conventional treatment barriers, which are coagulation–clarification and filtration, are vulnerable to breakthrough of intact cells and intracellular and dissolved compounds.^{4,14,28,54–5556} The removal of cell-bound compound within intact cells is the preferred situation for conventional treatment plants, while plants equipped with advanced oxidation and/or adsorption barriers would be able to handle the presence of dissolved compounds.

However, recent publications^14,57 have demonstrated that breakthrough of cells and their intracellular compounds, even at plants with low risk of cyanobacterial presence at their source, is a major challenge. Also, lysis of accumulated cells within sludge bed or filter media^20,57,58 and release of geosmin and MIB prior to chlorination in the absence of strong oxidant agents (eg, O₃) would lead to breakthrough of these compounds into the potable water. Hence, the intra/extracellular ratio of geosmin and MIB is a key factor for their successful removal and optimization of treatment processes.²⁵ Several factors, including bloom age, producers present, water quality conditions and available nutrients, and geographical location, could influence the ratio of intra/extracellular compounds.^19,21–2223 Table 4 provides a useful summary of existing knowledge about the removal of cyanobacterial T&O compounds by conventional and advanced treatment barriers. However, this ratio and the factor that would influence its variations and the impact of these variations on conventional and advanced full-scale treatment barriers require further investigation.

Table 4.

Treatment processes for the removal of cyanobacterial cells and cell-bound T&O compounds.^{1,4,10,20,21,23,28–293031,46–47484950,56,57,59,60}