- Email: [email protected]
Effect of type of granular activated carbon on DOC biodegradation in biological activated carbon filters
Process Biochemistry 45 (2010) 355–362
Contents lists available at ScienceDirect
Process Biochemistry journal homepage: www.elsevier.com/locate/procbio
Effect of type of granular activated carbon on DOC biodegradation in biological activated carbon filters Kozet Yapsakli a,*, Ferhan C¸ec¸en b,1 a b
Environmental Engineering Department, Faculty of Engineering, Marmara University, Go¨ztepe, Istanbul, Turkey Institute of Environmental Sciences, Bogazici University, 34342 Bebek, Istanbul, Turkey
A R T I C L E I N F O
A B S T R A C T
Article history: Received 29 April 2009 Received in revised form 5 August 2009 Accepted 12 October 2009
The purpose of this paper is to clarify the effect of the two different GAC types (steam activated or chemically activated) on DOC biodegradation in biological activated carbon (BAC) columns. For this purpose, raw water taken from a surface reservoir was fed to continuous-flow lab-scale biofiltration columns which were run for more than 18,000 bed volumes. The effect of pre-ozonation on DOC removal was also evaluated. Experimental results showed that biological activity inside the BAC columns extended the service life and the choice of filter material was crucial in BAC systems. The DOC biodegradation was higher in thermally activated carbon columns compared to the chemically activated one. The ability of GAC to better adsorb and retain organic compounds increased the chance of biodegradation. Contrary to expectations, pre-ozonation did not significantly enhance DOC biodegradation. Despite the high increase in biodegradable dissolved organic carbon (BDOC) upon ozonation, overall DOC biodegradation efficiencies did not differ from raw water. Overall, the DOC biodegradation in columns was higher than in most of the studies. This observation was primarily attributed to the low specific ultraviolet absorption (SUVA) values in raw water indicating a high biodegradability. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Biological activated carbon Ozonation Chemically activated GAC Steam activated GAC Specific UV absorption Biodegradation
1. Introduction The granular activated carbon (GAC) filtration is widely applied in the production of potable water. Most organic components of water, including natural organic matter (NOM), are readily adsorbed onto activated carbon. GAC is also home to an ecosystem of bacteria and protozoans due to its well-suited surface and macropores. The rough, pitted surface of GAC provides shelter from fluid shear forces, enriches substrates, nutrients and oxygen concentrations on its surface and contains functional groups that enhance the attachment of microorganisms [1,2]. The combination of ozonation with biological activated carbon treatment is one of the most promising processes among advanced drinking water treatment processes. The GAC, which has bioactivity on its surface and removes significant amounts of DOC by biodegradation is referred to as biological activated carbon (BAC) [3]. Pre-oxidation processes are usually applied because biofiltration is usually not capable of removing biorefractory substances. Ozone is an oxidant which is frequently used for this purpose. Combination of ozonation with biological treatment has
* Corresponding author. Tel.: +90 216 348 0292; fax: +90 216 348 1369. E-mail address: [email protected] (K. Yapsakli). 1 Tel.: +90 212 359 7256; fax: +90 212 257 5033. 1359-5113/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.procbio.2009.10.005
the advantage of reduction of the biological regrowth potential because biological treatment can selectively remove biodegradable organic matter. Ozonation causes substantial structural changes in humic substances. The consequences are a strong and rapid decrease in color and UV absorbance (UV254) due to a loss of aromaticity and depolymerization [4]; a small reduction in TOC (e.g., 10% at 1 mg O3/mg C); a slight decrease in the high apparent molecular weight fraction, and a slight increase in smaller fractions; a significant increase of carboxylic functions; formation of ozonation by-products [5]; conversion of humic into non-humic material and increase in polarity [3]. The shift of molecular distribution to lower values is important from the point of enhancing the biodegradability of NOM because low molecular compounds are more easily transported across the cell membrane and are attacked by metabolic enzymes [6]. UV absorbance at 254 (UV254) represents the existence of unsaturated carbon bonds including aromatic compounds, which are generally recalcitrant to biodegradation, and a decrease in UV absorbance results in an increase in biodegradability [7,8]. The main advantage of BAC filtration is to remove biodegradable compounds comprising the most undesirable fraction of organic matter in water [9]. Moreover, recalcitrant NOM molecules may also be removed from water by first sorption onto the biofilm and then slow biodegradation due to the longer detention time within the biofilm [10].
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
356
In drinking water biofiltration the filter media supplies the necessary surface area for cell attachment and different filter media have different accessible areas. For example, the surface of the sand is smooth and non-porous, while the GAC shows a much rougher surface with widely distributed crevasses and ridges. The rougher surface of GAC has been shown to enhance bacterial attachment compared with smoother surfaces [11]. Moreover, the amorphous GAC structures could help protect newly attached bacteria from shear forces, which could have been a major hindrance for biofilm development. Therefore, it is thought that BAC filters should perform better than sand or anthracite filters in terms of biological NOM removal. In the last years, several investigations compared the efficiency of adsorptive media (GAC) and non-adsorptive media (anthracite and sand) for biological DOC removal [12,13]. On the other hand, there is very limited information on the effect of the GAC type on the biodegradation taking place in BAC filters. Therefore, it is the purpose of this paper to clarify the effect of the GACs (steam activated or chemically activated) on DOC removal using a raw water from a reservoir in Istanbul. The effect of pre-ozonation on DOC removal was also evaluated. 2. Materials and methods 2.1. Raw water characteristics ¨ merli reservoir in Istanbul, which Raw water samples were taken from the O supplies water to the city at a rate of 1,000,000 m3/d. According to the year round measurements, the dissolved organic carbon (DOC) and total organic carbon (TOC) of the source varied between 3.5–5.8 mg/L and 4.1–6.0 mg/L, respectively. The raw water was filtered through 5 mm polypropylene cartridge filter as soon as it arrived at the laboratory. These filters can be used for DOC determinations and were found not to leach any contaminants that cause erroneous DOC measurements [14]. In order to prevent biodegradation, water samples were kept at 4 8C. 2.2. Experimental set-up The scheme of the set-up is shown in Fig. 1. The GAC media occupies 50 cm of the total depth of the 1 m column with a diameter of 2 cm. The empty part of the column provided space for expansion when backwashing was required due to bacteria clogging. The empty bed contact time (EBCT) in the columns was 18 min. All columns were operated in down-flow mode. In a typical column operation the influent flow rate was 10.2 mL/min, which was equivalent to a hydraulic loading of 1.67 m/h. The columns were operated in a temperature-controlled room of 25 8C. All activated carbon samples were obtained from the Norit Company, the Netherlands and the samples were used without pretreatment. CAgran, which has a very open (meso and macro) pore structure type, is a wood based granular activated carbon. It has been produced by chemical activation using phosphoric acid. Norit GAC 1240 is produced by heat activation and is a coal-based GAC grade. Physicochemical characteristics of virgin GAC grades are summarized in Table 1.
Table 1 Properties of activated carbon grades used in the study. Activated carbon
Norit GAC 1240
CAgran
Origin Physical form Activation method Total pore volume (cm3/g) Micropore (w < 2 nm) (cm3/g) Mesopore (w < 2–50 nm) (cm3/g) Macropore (w > 50 nm) (cm3/g)
Coal Granular Thermal 0.83 0.35 0.22 0.26
Wood Granular Chemical 1.75 0.26 0.60 0.89
Batch adsorption/desorption isotherm experiments served as a basis for choosing the right GAC type to be employed in continuous-flow biofiltration experiments. Norit 1240 was one of the most efficient GAC types in terms of adsorption characteristics among the others tested. On the other hand, the chemically activated CAgran exhibited inferior adsorption compared to thermally activated carbons but its desorbability was better [15]. Desorption is a very important phenomenon in terms of bioregeneration. In their studies Walker and Weatherley [16] stated that bioregeneration can only occur with compounds that readily desorb. Hence, bioregeneration is controlled by the reversibility of adsorption [17– 20]. Because of its better desorbability, CAgran was also selected for continuousflow biofiltration experiments. Parallel GAC columns were operated with raw and ozonated water. Since two different carbon grades were used, in total four columns were operated in this study. The raw water was ozonated at a rate of 2 mg O3/mg DOC at laboratory conditions. This dosage was found to produce the maximum biodegradable ¨ merli water dissolved organic carbon (BDOC) without mineralizing the DOC in the O ¨ merli raw water [21]. by more than 5% for O 2.3. Operation of columns The influent and the effluent water were daily analyzed for DOC, NH4+–N, NO2– N, NO3–N, Total Nitrogen, pH, UV254, UV280 and BDOC. 2 h composite samples were collected in the effluent. The dissolved oxygen concentrations were measured by using HACH HQ 40d Multimeter. Prior to DOC analysis, water samples were passed through 0.45 mm syringe driven PVDF filter (Millipore). Dissolved organic carbon (DOC) was determined by the high temperature combustion method (Standard Methods 5310 B) using a Teledyne-Tekmar Apollo 9000 model TOC analyzer. UV absorbance measurements were carried out at wavelengths 254 and 280 by a UV– visible double beam spectrophotometer (Shimadzu UV-2450) with 1 cm quartz cell. In order to enhance and accelerate the start-up of biological activity inside the BAC columns, bacteria were inoculated into the system before column operation. The concept of inoculation into GAC columns to make them biologically active was practiced by some authors including Mochhidzuki and Takeuchi [22], Zhao et al. [23], Nishijima and Speitel [3], Traenckner et al. [24]. An activated sludge MLVSS was not used as inoculum in the current study because of the probability that the culture contains pathogens. Pathogens should not be inoculated to a system which ¨ merli water were serves as drinking water purposes. Therefore, bacteria in the O enriched for three months in a suspended culture batch reactor, so that the inoculum in BAC system contained bacteria that originate from the reservoir itself and were acclimated to organic carbon of the source [25]. Before the suspended growth batch culture was transferred to the columns, it could remove up to 40% of ¨ merli water by biodegradation [25]. This culture was contacted DOC the incoming O with GAC overnight and then this bacteria–GAC mixture was transferred to BAC columns. In columns at least 10 cm of free water was kept above the packed height to ensure submergence of packing media. Thus, due to the inoculation, the columns were biologically active from the first day of the operation, and biodegradation and adsorption took place simultaneously. The influent and effluent DOC concentrations are reported as a function of the number of bed volumes instead of the volume treated or the time of treatment. Under the present working conditions, 1000 bed volumes were equal to 12.5 days of operation. During operation of columns, water samples were taken from the ports located at the 0, 27 and 52 cm above the bottom. Since the height of the packed GAC column was 50 cm, the top port was 2 cm above the GAC media. Columns were also operated in the sterile mode in order to differentiate between adsorption and biodegradation. In order to assure the sterility of the columns the influent raw water was always autoclaved.
3. Results 3.1. Comparison of DOC removal in sterile GAC and BAC columns
Fig. 1. Experimental set-up used in biofiltration experiments.
Adsorption and biodegradation are known to be the predominant mechanisms contributing to DOC removal in BAC columns. However, it is difficult to identify their relative importance at
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
different operational stages. Under sterile conditions only, the removal by adsorption alone can be determined. In the current study, CAgran and Norit 1240 columns were examined in a system in which only adsorption took place. The comparison was made only with raw water. In order to assure that adsorption took place only and no biodegradation, the influent raw water was autoclaved. On the other hand, in the BAC column occupied by microorganisms, both adsorption and biodegradation mechanisms would be dominant for DOC removal before breakthrough. After breakthrough is reached, DOC removal can occur mostly by biodegradation. The normalized effluent DOC concentrations in the sterile GAC and BAC columns are presented in Fig. 2. Examination of Fig. 2 shows that after approximately 1000 bed volumes, complete breakthrough of DOC was observed in the sterile CAgran column. Before breakthrough was reached, 25% of total DOC was non-adsorbable. The column was further operated for 1200 bed volumes, during which no DOC removal was detected. On the other hand, in the BAC column packed with CAgran, breakthrough commenced at later times. Additionally, the effluent DOC concentrations in the initial stages of BAC operation with CAgran were shown to be less than that of the sterile GAC column, indicating the presence of active bacteria effectively removing DOC. The effluent of BAC column before breakthrough quantifies both the non-adsorbable and non-biodegradable matter which amounted to 20% of the total DOC. At later stages, the presence of bacteria resulted in 47% DOC removal in the column filled with CAgran. The breakthrough commenced at approximately after 4000 bed volumes in the sterile column filled with Norit 1240. 22% of the initial DOC was found to be non-adsorbable in the sterile column. Compared to sterile CAgran, Norit 1240 outperformed in terms of DOC adsorption, which was also shown in batch adsorption isotherm experiments [15]. In the BAC column, breakthrough did
357
not start during 220 days of column operation. It is clear from Fig. 2 that biodegradation extended the operation life of Norit 1240. It is assumed that biological activity enhanced the adsorption capacity of GAC for non-biodegradable compounds by eliminating substances that would otherwise compete for adsorption sites. The non-adsorbable and non-biodegradable fraction leaving the Norit 1240 was 20% of total DOC. 3.2. Sampling from the ports of sterile GAC columns Port sampling at different operational stages of sterile Norit 1240 column showed that oxygen concentrations along the column depth decreased at all stages of column operation (data not shown). Through the 50 cm of carbon depth, approximately 90% of the oxygen was lost in the bulk liquid. This is due to the chemisorption of oxygen onto GAC pores. Thermally activated carbons have high affinities towards oxygen and upon contact with oxygen, surface chemistry of GAC may change [18]. Since thermal activation of these carbons is carried out in the absence of oxygen, they have a more reactive surface. Contrary to this, chemically activated carbons have a surface with fully oxidized active sites so that interaction with oxygen does not affect the surface. Sampling from the ports of the sterile CAgran revealed limited oxygen uptake (not more than 3.5% decrease) along the depth of the column. 3.3. Operation of BAC columns Continuous-flow biofilters were operated for more than 220 days before shutdown. There is a decreasing trend in influent DOC which is related to seasonal changes in DOC. In the ozonated water the influent DOC concentrations were not significantly different (p = 0.64) from raw water, as determined by ANOVA test at a 95% confidence interval. This showed that during ozonation mineralization of DOC was avoided. Indeed, the percent decrease in the original DOC upon ozonation was less than 5% in all cases. 3.3.1. Operation of CAgran The fastest breakthrough of DOC was observed in the column packed with CAgran (Fig. 3). Exhaustion took place approximately at the same time (about 2000 bed volumes) for both raw water and ozonated water. Therefore, ozonation did not seem to have any effect on adsorption of NOM during continuous-flow operation. Before the exhaustion of the column filled with CAgran, adsorption and biodegradation took place simultaneously on the surface since filters were biologically active from the first day of operation. The effluent DOC concentration was 0.6 mg/L in the case of both raw and ozonated water (Fig. 3). On the other hand, after exhaustion of filter capacity, biodegradation was the dominant mechanism. The difference between the influent and the effluent
Fig. 2. Normalized DOC in the effluent of sterile GAC/BAC of Norit 1240 and CAgran columns.
Fig. 3. Influent and effluent DOC concentrations in the two continuous-flow BAC columns tested with raw and ozonated water (RW: Raw water; OW: Ozonated water).
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
358
Fig. 4. Concentrations along the length of the CAgran at different bed volumes of BAC operation.
DOC concentrations after exhaustion can be attributed mostly to biodegradation. In this case, the average DOC removal by biodegradation is 47% for the raw water, and 53% for the ozonated water. At an average influent raw water of DOC 3.65 mg/L, an effluent DOC concentration of 1.94 mg/L was generally achieved. This means that 1.71 mg/L of DOC was biodegraded in the system fed with raw water. Surprisingly, the average biodegradable DOC (BDOC) concentration in the influent raw water was as low as 0.68 mg/L. For the ozonated water, influent DOC was lowered from an average value of 3.56 mg/L to 1.62 mg/L by biodegradation, which corresponds to a removal of 1.94 mg/L. The average influent BDOC for the ozonated water was 1.76 mg/L. This indicated that the biological DOC removal in the column fed with ozonated water is slightly higher than the amount of biodegradable DOC (BDOC) measured in the influent. The BDOC measurement, in principle, is done using a suspended culture with limited bacteria concentration and poor mixing conditions. It is therefore assumed in general that the actual BDOC concentration is underestimated this test. Moreover, BAC columns are attached-growth systems which are always more efficient in contaminant removal than suspended growth systems. Additionally, GAC provides a long retention time by adsorption which increases the chance of biodegradation of slowly biodegradable substances. Sampling along the length of the reactor had the purpose of examining the DOC, NH4+–N, and dissolved oxygen profiles in the
column. Sampling from ports reflected the concentration profiles after breakthrough was reached in CAgran columns and are summarized in Fig. 4. On average, 42% of the DOC was biodegraded in the first 25 cm of the column and a further 5% biodegradation was seen in the lower part of the column fed with raw water. Therefore, the first 25 cm of the column (equivalent to an EBCT of 9 min) was primarily responsible for biodegradation. It is expected that a biomass gradient was present where the maximum biomass was retained in the first centimeters of the filters and therefore most of the DOC removal occurred during the 9 min EBCT. This result is supported by some other studies [9,26–28]. Sampling along the depth of sterile CAgran column has shown that 3.5% of the initial dissolved oxygen was lost in a carbon depth of 0.5 m (data not shown). Therefore, further decrease in oxygen concentration through the depth was attributed to the uptake of oxygen due to organic biodegradation and nitrification. In nitrification, about 4.33 mg O2/mg NH4+–N is consumed by nitrifiers [29]. In biodegradation of organic matter the corresponding value is 1 mg DO/1 mg COD. The organic carbon concentration was converted into oxygen demand using a conversion factor of 2.66. The corresponding dissolved oxygen balance is presented in Table 2. In these calculations, the oxygen requirement due to nitrite oxidation was neglected because of negligible concentration. A similar comparison was also done in a study conducted by
Table 2 Examples of the calculation of the theoretical and actual oxygen consumptions in CAgran columns. Sample Raw Raw Raw Raw
water water water water
Ozonated Ozonated Ozonated Ozonated (1) (2) (3) (4) (5) (6) (7)
BV = 10100 BV = 15300 BV = 16160 BV = 17280
BV = 10100 BV = 15300 BV = 16160 BV = 17280
DDOC (1)
DDOCOD (2)
DNH4+–N (3)
DDONH4 þ N (4)
DDOcs (5)
DDOth (6)
DDOactual (7)
1.234 1.504 1.180 1.229
3.282 4.001 3.139 3.269
0.23 0.08 0.07 0.10
0.996 0.346 0.303 0.433
0.273 0.265 0.252 0.273
4.551 4.611 3.694 3.975
4.9 4.74 4.08 3.90
1.537 1.296 1.630 1.620
4.088 3.447 4.336 4.309
0.14 0.19 0.04 0.07
0.606 0.823 0.173 0.303
0.300 0.265 0.294 0.290
4.994 4.535 4.712 4.902
4.9 4.74 4.5 4.29
DDOC: Measured DOC removal (mg C/L) DDOCOD: Calculated DO consumption in terms of COD removed (2.66 1 mg DO/1 mg COD DDOC) DNH4+–N: Measured NH4+–N removal DDONH4 þ N : Calculated DO consumption for NH4+–N (4.33 mg O2/mg NH4+–N DNH4+–N) DDOcs: Determined DO consumption due to chemisorption (assumed as 3.5% Initial DO) DDOth: Theoretical DO consumption ðDDOCOD þ DDONH4 þ N þ DDOcs Þ DDOactual: Initial DO–Final DO
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
Yu et al. [30], the purpose of which was to explain the nitrogen loss and oxygen paradox in a full-scale treatment plant. Table 2 compares the oxygen uptake resulting from organic carbon biodegradation and nitrification with the measured oxygen uptake along the column depth (Columns (6) and (7) in Table 2). Results showed that the theoretical oxygen consumption and the measured dissolved oxygen values were in good agreement with each other. Thus, it was proven that the decrease in dissolved oxygen concentrations between Port 1 and Port 3 reflected biodegradation of organic carbon and nitrification. Organic carbon removal in the filters was due to biodegradation only but not adsorption. This suggested that carefully performed DO measurements may serve as an approximation for biological DOC removal in biofilters. This is of particular interest because of the relative simplicity of DO measurements. 3.3.2. Operation of Norit 1240 Norit 1240 is a microporous coal-based and steam activated carbon and had the best adsorption performance among the GACs tested [15]. In continuous-flow column experiments, CAgran reached breakthrough within 2000 bed volumes. On the other hand, during 220 days of operation (approximately 18000 bed volumes), breakthrough did not commence in the case of Norit 1240. The average effluent DOC concentrations for the columns operated with raw water and ozonated water were 0.69 mg/L and 0.66 mg/L, respectively. Effluent DOC concentrations remained constant regardless of influent DOC concentration. The average DOC removal efficiency in the column was 81% (Fig. 3). Samples taken from the ports at different bed volumes demonstrate the DOC, ammonium and the dissolved oxygen concentration profiles at different depths (Fig. 5). Since Norit 1240 has a high reactivity towards oxygen and chemisorbs it on its surface, the oxygen consumed through the depth of the column cannot be taken as a measure of bacterial activity. In the sterile Norit 1240 column, the dissolved oxygen concentration at the bottom port was as low as 0.84 mg/L (data not shown). In the biologically active Norit 1240 column (BAC), despite biological activity, the dissolved oxygen concentration did not further decrease. The most interesting finding in port sampling experiments with Norit 1240 was that, this carbon type can decrease the DOC
359
concentration to at least 1 mg/L by biodegradation. DOC concentration in the middle port remained constant (around 1 mg/L) during a 7000 bed volume operation. 7000 bed volumes correspond approximately to 87 days of operation. A 7000 bed volume is long enough for the mass transfer zone to shift to lower depths. Therefore, the decrease in DOC between the upper and middle ports should be attributed to biodegradation only. The DOC levels in the bottom port were further reduced to 0.6 mg/L. This concentration was the typical effluent DOC concentration observed during 220 days of column operation. Therefore, it is difficult to identify the mechanism leading to this effluent concentration. The results of molecular studies revealed that bacteria were also present in the bottom port [21]. Therefore, biodegradation might have played a role in DOC reduction in this region. On the other hand, there is a probability that adsorption is still occurring between the middle and bottom ports. The average DOC concentration entering the lower port was as low as 1 mg/L. Therefore, the GAC pores near the bottom may still be not saturated. 3.3.3. Effect of ozonation on biodegradation of organics It is well established that ozonation increases the biodegradable portion of DOC. Studies up to now showed that the increase in biodegradable DOC enhanced biological DOC removal. The percent increase in BDOC upon ozonation at a dose of 2 mg O3/mg DOC was as high as 245%, whereas on the average the increase was 166% (data not shown). Despite this huge increase, the biological DOC removal efficiency was not so much affected in the filters fed with ozonated water (Fig. 3). The biological DOC removal was only 8% higher than in the CAgran column fed with raw water. In case of Norit 1240, no enhancement of DOC removal could be observed. Therefore, it was concluded that ozonation did not have a pronounced effect on DOC removal. 3.3.4. Effect of specific UV absorption (SUVA) on biodegradation SUVA provides a quantitative measure of unsaturated bonds and/or aromaticity within NOM. Increase in SUVA generally reflects higher humification, aromaticity and hydrophobicity of DOM and hence, lower biodegradability. Calculation of SUVA based on DOC and UV254 measurements showed that the average SUVA values of raw and ozonated waters were 2.5 and 1.42 m1/(mg
Fig. 5. Concentrations along the length of the Norit 1240 at different bed volumes of BAC operation.
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
360
4. Discussion and conclusion 4.1. Comparison of biodegradation efficiencies in the Norit 1240 and CAgran columns
Fig. 6. Influent SUVA values in raw and ozonated water.
DOC/L), respectively. SUVA, which is the UV absorbance measured at a fixed wavelength (l, in nm) divided by DOC concentration is expressed as: SUVAl ¼
UVl ðcm1 Þ 100 DOCðmg=LÞ
(1)
UVl is usually in cm1 and DOC in mg/L. A conversion factor of 100 is used to express the unit of SUVA as m1/(mg DOC/L). Fig. 6 shows an increasing trend in the SUVA values of raw and ozonated water with respect to time. SUVA values in raw water were as high as 3.3, whereas in ozonated water these values never exceeded 1.7. According to the classification made by Edzwald and Tobiason [31], a SUVA value of 4 and greater values indicates that the water sample consists mainly of humic materials and higher molecular weight substances. When this value is below 2, NOM is mainly non-humic and hydrophobicity is low. SUVA values between 2 and 4 indicate that NOM consists of a mixture of aquatic humics and other NOM. The average SUVA value of the raw water was close to 2, therefore, the aromaticity within NOM was not high, and hence a high biodegradability of raw water can be expected. In order to investigate the role of humification on DOC biodegradation, the percent DOC biodegradation was correlated to SUVA. The Pearson’s product momentum correlation coefficient (rp) for the relationship was used for linear estimation of the strength and direction of the correlation. The relationship is depicted in Fig. 7. Indeed, in raw water samples the increase in SUVA value resulted in a decrease in percent DOC biodegradation. The overall Pearson coefficient determined at 95% confidence interval (0.82) indicated a moderate correlation between biodegradation and SUVA.
Fig. 7. Relationship between SUVA and DOC biodegradation in raw water.
Norit 1240 can decrease the DOC concentration at least down to 1 mg/L by biodegradation which corresponds to 72% removal. On the other hand, 0.6 mg/L of effluent DOC concentration can be due to either biodegradation, adsorption or the combination of both. In the CAgran column fed with raw water the average DOC removal by biodegradation was 47%. The difference between the removal efficiencies may be related to the adsorptive properties of GACs. The efficiency of adsorptive media (GAC) and non-adsorptive media (anthracite and sand) for biological DOC removal are well documented in literature. For example, Liu et al. showed that GAC filters enhanced the removal of less biodegradable substances in comparison to anthracite filters. Adsorption onto GAC can provide a longer retention time in filters for slowly biodegradable components. It is considered that similar to the improvement of removal efficiency in case of adsorptive media, a better adsorbing GAC can also lead to higher DOC biodegradation. In a previous study it was shown that microporous Norit 1240 outperformed macroporous CAgran in terms of adsorption [15]. In accordance with this, also in the current study Norit 1240 was also superior in terms of biodegradation efficiency. The ability of GAC to better adsorb and retain compounds increases their chance of biodegradation by bacteria. The major advantage of Norit 1240 over CAgran in terms of biodegradation is that it can adsorb considerable amounts of both readily and slowly biodegradable organics. Biodegradation by attached bacteria can lead to continuous bioregeneration of GAC. It is thought that, probably a couple of years should pass till breakthrough is achieved in the GAC column containing Norit 1240. Thus, using this type of carbon seems to be advantageous. Another possibility can be the production of biodegradable carbon on the surface of Norit 1240. The high oxygen consumption at GAC surface can cause surface catalytic reactions to occur on GAC surface. Surface catalytic processes can convert substances which are originally non-biodegradable into biodegradable ones [34]. The observation that more biodegradation took place in the Norit 1240 column than the CAgran column can be attributed to such surface catalytic reactions since Norit 1240 chemisorbs large amounts of dissolved oxygen on its surface. Some research has suggested that DOC removal can be limited by biomass concentration [32]. The fact that microporous GAC (Norit 1240) outperformed at NOM biodegradation may also be related to the bacterial amount in biofilters. Prior to column operation, approximately equal amounts of bacteria were added to each column. In general, little biogrowth occurs in GAC micropores because their small diameter (1–100 nm) does not allow penetration of bacteria, which typically have a diameter greater than 200 nm [13]. On the other hand, the relative amounts of bacteria in each filter may change during the operation. The pore sizes of GAC can influence bacterial retention and colonization. In a study conducted by Wang et al. [33], the bacterial concentrations (per gram of GAC) on a bituminous-coal-based (microporous), a lignite-coal-based (mesoporous) and a wood based (macroporous) activated carbon were compared. Although the highest biomass (quantified by phospholipids) was measured in the lignite-based carbon, results showed that small performance differences existed in terms of DOC biodegradation among micro-, meso- and macropore GACs. The role of microorganism concentration on biodegradation is unknown in the present system since biomass concentrations were not measured. However, it is considered that bacterial concentration in BAC columns can hardly be a rate limiting step for biodegradation
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
361
Table 3 DOC removals typically achieved by BAC columns in the literature. Water source Seagahan, UK Lake Vyrnwy, UK Norsborg, Sweden River Dee, UK Model Water Plonia River, PL Grand River, USA Miyun Reservoir Huangpu River Omerli Reservoir a b
DOC (mg/L) b
NA 2.4–4.8 NAb 3.0–7.9 4.0–5.0a 7.8–11.6a 5–7 4.9–7.3 5.2–7.7 2.9–4.9
Media
O3 dose
Removal (%)
Reference
Sand Sand Sand Sand Sand GAC GAC GAC GAC GAC
3.1–4.8 mg O3/L 1.1–2.5 mg O3/L 0.2–l mg O3/mg TOC 0.5 mg O3/mg TOC 6.7 mg O3/L 1.64 mg O3/mg TOC NA 3 mg/L 2.0–2.5 mg/L No ozonation
25 26.5 20–30 28 34–40 39 13–23 33.4 31 47–72
[36] [37] [38] [39] [40] [41] [42] [43] [44] This study
In terms of TOC. NA: Data not available.
since DOC concentrations are not high in most drinking water sources. Sampling from porst have shown that most of the DOC is removed within the 25 cm of column depth and lower depths are responsible for only minor DOC removals despite the existence of sufficient bacteria. Therefore, choosing the best type of GAC in biofiltration experiments results in better performance not only in terms of adsorption but also in terms of biodegradation, which extends the service life of the GAC. The compilation of literature studies on DOC removal is depicted in Table 3. The DOC removal efficiencies in current BAC columns seemed to be higher compared to literature values [35– 43]. In order to illustrate the organic characteristics of Huangpu River, Xu et al. [43] performed a pilot-scale treatment set-up including pre-ozonation, conventional treatment, post-ozonation and BAC filter and found that 31% of DOC could be removed by ozone-biofiltration system. In another study conducted by Li et al. [42] the granular activated carbon/O3-biological activated carbon (BAC) process was employed to treat raw water and compared to O3-BAC process. They showed that the presence of GAC in the ozone contact tank improved ozone utilization and biodegradability of the effluent and DOC removal efficiency increased from 33.4% to 39.5%. In the studies of Seredynska-Sobecka et al. [40], BAC was maintained by circulating raw water through the GAC filter. DOC removal efficiency of ozone-BAC system was 39% in the presence of 10 mg/L of phenol. However, there is not much information about the SUVA values in most studies. Therefore, a comparison cannot be made among studies. In the current study, the average SUVA values in the raw water were around 2.5. This ¨ merli was not highly aromatic may indicate that the raw water of O and therefore not recalcitrant to biodegradation. Hozalski et al. [44] observed enhanced removal of TOC by biodegradation of NOM sources with a lower SUVA value. When this is the case, even a slowly biodegradable compound can be biodegraded on the surface of the carbon, because GAC can provide the necessary retention of organics due to adsorption. This may be one of the reasons of higher DOC biodegradation compared to literature. This was also supported by the results of our recently published article ¨ merli water were enriched to be used as [25]. Bacteria in the O inoculum for BAC columns. This way the inoculum in BAC system contained bacteria that originate from the reservoir itself and were acclimated to organic carbon of the source. The suspended growth batch culture could remove up to 40% of DOC of the ¨ merli water by biodegradation [25]. Even this incoming O efficiency in a batch reactor is higher than the results of most of the studies (Table 3). In addition, the molecular weight distribution of NOM is also important factor for biodegradability [44] because lower molecular weight compounds are more easily transported across cell membrane and attacked by metabolic enzymes [6].
4.2. Effect of ozone on removal efficiency The removal efficiencies in the columns fed with raw and ozonated water were not significantly different from each other. In the CAgran column, the average DOC removal by biodegradation was 47% for raw water, ozonation slightly improved this and led to 53% removal. It was assumed that the inoculated bacteria were ¨ merli water and that they produced specific well acclimated to O enzymes to treat even the hardly biodegradable organics of the reservoir. It is for sure that ozonation made them readily biodegradable, but the contribution of ozonation to the overall removal was marginal. Most of the studies up to now documented the benefits of ozone in enhancing the biodegradability of NOM and leading to higher biodegradation in BAC columns. The effect of ozonation on DOC removal was more obvious when a non-adsorptive media like sand was used. For example, Shukairy et al. [45] compared the biofiltration efficiencies in ozonated and non-ozonated biofiltration columns consisting of sand using the raw surface water of the Ohio River. The percentage of DOC removal approximately doubled when ozonation preceded biofiltration (12% vs. 25%). The same experiments were also done with artificial water, using a solution of humic substances isolated and concentrated from ground water. However, DOC removals reported for the artificial water were significantly higher than those observed for any of the previously described studies that used natural waters (34%–48% for biofiltration; 64%–72% for ozone + biofiltration). In another study conducted by Moll et al. [46], ozonation slightly enhanced TOC removal in a biofiltration column made of sand. Ozonation at a dose of 0.9 mg O3/mg TOC increased the average TOC removal in the filters treating Harsha Lake water from 16% to 25%. DeWaters and DiGiano [47] found that pre-ozonation of humic substances obtained from Dismal Swamp resulted in 43% biodegradation of TOC in a fixed bed GAC column (Ozone dose: 1 mg O3/mg TOC, EBCT = 3.9 min; feed TOC = 7 mg/L). In contrast, Glaze [48] carried out a pilot-scale study with GAC beds at Shreveport, Louisiana to investigate the combination of ozone and GAC for removal of TOC and THM precursors. The results indicated that pre-ozonation did not affect TOC removal at any time and the average TOC removal efficiency in both columns was approximately 26%. 5. Conclusion This study showed that the choice of filter material is crucial in BAC systems in enhancing the biodegradation of organic matter. The ability of GAC to better adsorb and retain organic compounds increases the chance of biodegradation. This way, slowly biodegradable substances have a better chance to be removed in columns. Therefore, the GAC type leading to best adsorption would probably be the most suitable type for biodegradation as well.
362
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
As in our case, if raw waters have low SUVA values indicating a high biodegradability, ozone application may not be necessary for conversion of DOC into BDOC in order to increase the DOC biodegradation. The high retention time in BAC columns can lead to biodegradation of slowly biodegradable organics and eliminate the usage of ozonation. This way, operational cost associated with ozone can be eliminated. Acknowledgements ¨ BI˙TAK I˙C¸TAG (C¸110) and the The finance of this study by TU Research Fund of Bog˘azic¸i University (project no. 05Y103D) is gratefully acknowledged. We also express our gratitude to Jan C van den Dikkenberg (Norit Company, the Netherlands) for providing the carbon samples and data about characteristics and to Halil I˙brahim Uzun for providing the water from the reservoir. References [1] Weber Jr WJ, Pirbazari M, Melson GL. Biological growth on activated carbon: an investigation by scanning electron microscopy. Environ Sci Technol 1978;12:817– 9. [2] Voice TC, Pak D, Zhao X, Shi J, Hickey RP. Biological activated carbon in fluidized bed reactors for the treatment of groundwater contaminated with volatile aromatic hydrocarbons. Water Res 1992;26:1389–401. [3] Nishijima W, Speitel GE. Fate of biodegradable dissolved organic carbon produced by ozonation on biological activated carbon. Chemosphere 2004;56:113–9. [4] Camel V, Bermond A. The use of ozone and associated oxidation processes in drinking water treatment. Water Res 1998;32:3208–22. [5] Graham NJD. Removal of humic substances by oxidation/biofiltration processes – A review. Water Sci Technol 1999;40:141–8. [6] Leisinger T, Cook AM, Hu¨tter R, Nu¨esch J. Microbial degradation of Xenobiotics and Recalcitrant Compounds. New York: Academic Press; 1981. [7] Lawrence J. Semi-quantitative determination of fulvic acid, tannin and lignin in natural waters. Water Res 1980;14:373–7. [8] Novak JM, Mills GL, Bertsch PM. Estimating the percent aromatic carbon in soil and aquatic humic substances using ultraviolet absorbance spectroscopy. J Environ Qual 1992;21:144–7. [9] Servais P, Billen G, Bouillot P, Benezet M. A pilot study of biological GAC filtration in drinking water treatment. Aqua 1992;41(3):163–8. [10] Carlson G, Silverstein J. Effect of molecular size and charge on biofilm sorption of organic matter. Water Res 1998;32:1580–92. [11] Hattori K. Water treatment systems and technology for the removal of odor compounds. Water Sci Technol 1988;20:237–44. [12] Liu X, Huck PM, Slawson RM. Factors affecting drinking water biofiltration. J Am Water Works Assoc 2001;93:90–101. [13] Urfer D, Huck PM, Booth SDJ, Coffrey BM. Biological filtration for BOM removal: A critical review. J Am Water Works Assoc 1997;89:83–98. [14] Karanfil T, Erdog˘an I˙, Schlautman MA. Selecting filter membranes for measuring DOC and UV254. J Am Water Works Assoc 2003;95:86–100. ¨ , Can ZS. Impact of surface properties of granular [15] Yapsakli K, C¸ec¸en F, Aktas¸ O activated carbon and pre-ozonation on adsorption and desorption of natural organic matter. Environ Eng Sci 2009;26:489–500. [16] Walker GM, Weatherley LR. A simplified predictive model for biologically activated carbon fixed beds. Process Biochem 1997;32:327–35. [17] Schultz JR, Keinath TM. Powdered activated carbon treatment process mechanisms. J Water Pollut Control Federation 1984;56:143–51. [18] Jonge RJ, de Breure AM, van Andel JG. Bioregeneration of powdered activated carbon (PAC) loaded with aromatic compounds. Water Res 1996;30:875–82. ¨ , C¸ec¸en F. Effect of type of carbon activation on adsorption and its [19] Aktas¸ O reversibility. J Chem Technol Biotechnol 2006;81:94–101. ¨ , C¸ec¸en F. Effect of activation type on bioregeneration of various activated [20] Aktas¸ O carbons loaded with phenol. J Chem Technol Biotechnol 2006;81:1081–92. [21] Yapsakli K. Application of biological activated carbon (BAC) in drinking water treatment. Turkey: Bogazici University; Ph.D. thesis; 2008. [22] Mochhidzuki K, Takeuchi Y. The effects of some inhibitory components on biological activated carbon. Water Res 1999;33:2609–16.
[23] Zhao X, Hickey RF, Voice TC. Long-term evaluation of adsorption capacity in a biological activated carbon fluidized bed reactor system. Water Res 1999;33:2983–91. [24] Traenckner J, Wricke B, Krebs P. Estimating nitrifying biomass in drinking water filters for surface water treatment. Water Res 2008;42:2574–84. [25] Yapsakli K, C¸ec¸en F. Use of enriched inoculum for determination of biodegradable dissolved organic carbon (BDOC) in drinking water. Water Sci Technol Water Supply 2009;9(2):149–57. [26] Wang ZW. Assessment of biodegradation and biodegradation kinetics of natural organic matter in drinking water biofilters. USA: University of Cincinnati; Ph.D. Thesis; 1995. [27] Miltner RJ, Summers RS, Wang JZ. Biofiltration performance: Part II effect of backwashing. J Am Water Works Assoc 1995;87:64–70. [28] Hozalski RM, Goel S, Bouwer EJ. TOC removal in biologically active sand filters: effect of NOM source and EBCT. J Am Water Works Assoc 1995;87:40– 54. [29] Metcalf & Eddy. Wastewater Engineering, Treatment and Reuse, Fourth ed., New York: McGraw-Hill; 2003. [30] Yu X, Qi Z, Zhang X, Yu P, Liu B, Zhang L, et al. Nitrogen loss and oxygen paradox in full-scale biofiltration for drinking water biofilters. Water Res 2007;41: 1455–64. [31] Edzwald JK, Tobiason JE. Enhanced coagulation: US requirements and a broader view. Water Sci Technol 1999;40:63–70. [32] Carlson KH, Amy GL. BOM removal during biofiltration. J Am Water Works Assoc 1998;90:42–52. [33] Wang JZ, Summers RS, Miltner RJ. Biofiltration performance: Part I relationship to biomass. J Am Water Works Assoc 1995;87:55–63. [34] Uhl W. Biofiltration processes for organic matter removal. In: Rehm HJ, Reed G, editors. Biotechnology. Weinheim: Wiley VCH; 2000. [35] Gould MH, Cameron DA, Zabel TF. An experimental study of ozonation followed by slow sand filtration for the removal of humic colour from water. Ozone Sci Eng 1984;6:3–15. [36] Cable CJ, Jones RG. Assessing the Effectiveness of Ozonation Followed by Slow Sand Filtration in Removing THM Precursor Material from an Upland Raw Water. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 29–37. [37] Seger A, Rothman M. Slow sand filtration with and without ozonation in nordic climate. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996 . p. 119–28. [38] Yordanov RV, Lamb AJ, Melvin MAL, Littlejohn J. Biomass characteristics of slow sand filters receiving ozonated water. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 107–18. [39] Odegaard H. The development of an ozonation/biofiltration process for the removal of humic substances. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 39–49. [40] Seredynska-Sobecka B, Tomaszewska M, Janus M, Morawski AW. Biological activation of carbon filters. Water Res 2006;40:355–63. [41] Emelko MB, Huck PM, Coffey BM, Smith EF. Effects of media, backwash, and temperature on full-scale biological filtration. J Am Water Works Assoc 2006;98:61–73. [42] Li L, Zhu W, Zhang P, Zhang Q, Zhang Z. AC/O3-BAC processes for removing refractory and hazardous pollutants in raw water. J Hazard Mater 2006;135:129–33. [43] Xu B, Gao NY, Sun XF, Xia SJ, Siminnot MO, Causserand C, et al. Characteristics of organic removal in Huangpu River and treatability with the O3-BAC process. Sep Purif Technol 2007;57:348–55. [44] Hozalski RM, Bouwer EJ, Goel S. Removal of natural organic matter (NOM) from drinking water supplies by ozone-biofiltration. Water Sci Technol 1999;40:157–63. [45] Shukairy HM, Summers RS, Miltner RJ. The impact of ozonation and biological treatment on disinfection by-products. In: Proceedings of the 4th Drinking Water Workshop Montreal; 1992. pp.1–20. [46] Moll DM, Summers RS, Fonseca AC, Matheis W. Impact of temperature on drinking water biofilter performance and microbial community structure. Environ Sci Technol 1999;33:2377–82. [47] DeWaters JE, Digiano FA. Influence of ozonated natural organic matter on the biodegradation of a micropollutant in a GAC bed. J Am Water Works Assoc 1990;82:69–80. [48] Glaze WH. 1982. Pilot-scale evaluation of biological activated carbon for the removal of THM precursors. EPA-600/S2-82-046, Environmental Protection Agency.
Contents lists available at ScienceDirect
Process Biochemistry journal homepage: www.elsevier.com/locate/procbio
Effect of type of granular activated carbon on DOC biodegradation in biological activated carbon filters Kozet Yapsakli a,*, Ferhan C¸ec¸en b,1 a b
Environmental Engineering Department, Faculty of Engineering, Marmara University, Go¨ztepe, Istanbul, Turkey Institute of Environmental Sciences, Bogazici University, 34342 Bebek, Istanbul, Turkey
A R T I C L E I N F O
A B S T R A C T
Article history: Received 29 April 2009 Received in revised form 5 August 2009 Accepted 12 October 2009
The purpose of this paper is to clarify the effect of the two different GAC types (steam activated or chemically activated) on DOC biodegradation in biological activated carbon (BAC) columns. For this purpose, raw water taken from a surface reservoir was fed to continuous-flow lab-scale biofiltration columns which were run for more than 18,000 bed volumes. The effect of pre-ozonation on DOC removal was also evaluated. Experimental results showed that biological activity inside the BAC columns extended the service life and the choice of filter material was crucial in BAC systems. The DOC biodegradation was higher in thermally activated carbon columns compared to the chemically activated one. The ability of GAC to better adsorb and retain organic compounds increased the chance of biodegradation. Contrary to expectations, pre-ozonation did not significantly enhance DOC biodegradation. Despite the high increase in biodegradable dissolved organic carbon (BDOC) upon ozonation, overall DOC biodegradation efficiencies did not differ from raw water. Overall, the DOC biodegradation in columns was higher than in most of the studies. This observation was primarily attributed to the low specific ultraviolet absorption (SUVA) values in raw water indicating a high biodegradability. ß 2009 Elsevier Ltd. All rights reserved.
Keywords: Biological activated carbon Ozonation Chemically activated GAC Steam activated GAC Specific UV absorption Biodegradation
1. Introduction The granular activated carbon (GAC) filtration is widely applied in the production of potable water. Most organic components of water, including natural organic matter (NOM), are readily adsorbed onto activated carbon. GAC is also home to an ecosystem of bacteria and protozoans due to its well-suited surface and macropores. The rough, pitted surface of GAC provides shelter from fluid shear forces, enriches substrates, nutrients and oxygen concentrations on its surface and contains functional groups that enhance the attachment of microorganisms [1,2]. The combination of ozonation with biological activated carbon treatment is one of the most promising processes among advanced drinking water treatment processes. The GAC, which has bioactivity on its surface and removes significant amounts of DOC by biodegradation is referred to as biological activated carbon (BAC) [3]. Pre-oxidation processes are usually applied because biofiltration is usually not capable of removing biorefractory substances. Ozone is an oxidant which is frequently used for this purpose. Combination of ozonation with biological treatment has
* Corresponding author. Tel.: +90 216 348 0292; fax: +90 216 348 1369. E-mail address: [email protected] (K. Yapsakli). 1 Tel.: +90 212 359 7256; fax: +90 212 257 5033. 1359-5113/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.procbio.2009.10.005
the advantage of reduction of the biological regrowth potential because biological treatment can selectively remove biodegradable organic matter. Ozonation causes substantial structural changes in humic substances. The consequences are a strong and rapid decrease in color and UV absorbance (UV254) due to a loss of aromaticity and depolymerization [4]; a small reduction in TOC (e.g., 10% at 1 mg O3/mg C); a slight decrease in the high apparent molecular weight fraction, and a slight increase in smaller fractions; a significant increase of carboxylic functions; formation of ozonation by-products [5]; conversion of humic into non-humic material and increase in polarity [3]. The shift of molecular distribution to lower values is important from the point of enhancing the biodegradability of NOM because low molecular compounds are more easily transported across the cell membrane and are attacked by metabolic enzymes [6]. UV absorbance at 254 (UV254) represents the existence of unsaturated carbon bonds including aromatic compounds, which are generally recalcitrant to biodegradation, and a decrease in UV absorbance results in an increase in biodegradability [7,8]. The main advantage of BAC filtration is to remove biodegradable compounds comprising the most undesirable fraction of organic matter in water [9]. Moreover, recalcitrant NOM molecules may also be removed from water by first sorption onto the biofilm and then slow biodegradation due to the longer detention time within the biofilm [10].
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
356
In drinking water biofiltration the filter media supplies the necessary surface area for cell attachment and different filter media have different accessible areas. For example, the surface of the sand is smooth and non-porous, while the GAC shows a much rougher surface with widely distributed crevasses and ridges. The rougher surface of GAC has been shown to enhance bacterial attachment compared with smoother surfaces [11]. Moreover, the amorphous GAC structures could help protect newly attached bacteria from shear forces, which could have been a major hindrance for biofilm development. Therefore, it is thought that BAC filters should perform better than sand or anthracite filters in terms of biological NOM removal. In the last years, several investigations compared the efficiency of adsorptive media (GAC) and non-adsorptive media (anthracite and sand) for biological DOC removal [12,13]. On the other hand, there is very limited information on the effect of the GAC type on the biodegradation taking place in BAC filters. Therefore, it is the purpose of this paper to clarify the effect of the GACs (steam activated or chemically activated) on DOC removal using a raw water from a reservoir in Istanbul. The effect of pre-ozonation on DOC removal was also evaluated. 2. Materials and methods 2.1. Raw water characteristics ¨ merli reservoir in Istanbul, which Raw water samples were taken from the O supplies water to the city at a rate of 1,000,000 m3/d. According to the year round measurements, the dissolved organic carbon (DOC) and total organic carbon (TOC) of the source varied between 3.5–5.8 mg/L and 4.1–6.0 mg/L, respectively. The raw water was filtered through 5 mm polypropylene cartridge filter as soon as it arrived at the laboratory. These filters can be used for DOC determinations and were found not to leach any contaminants that cause erroneous DOC measurements [14]. In order to prevent biodegradation, water samples were kept at 4 8C. 2.2. Experimental set-up The scheme of the set-up is shown in Fig. 1. The GAC media occupies 50 cm of the total depth of the 1 m column with a diameter of 2 cm. The empty part of the column provided space for expansion when backwashing was required due to bacteria clogging. The empty bed contact time (EBCT) in the columns was 18 min. All columns were operated in down-flow mode. In a typical column operation the influent flow rate was 10.2 mL/min, which was equivalent to a hydraulic loading of 1.67 m/h. The columns were operated in a temperature-controlled room of 25 8C. All activated carbon samples were obtained from the Norit Company, the Netherlands and the samples were used without pretreatment. CAgran, which has a very open (meso and macro) pore structure type, is a wood based granular activated carbon. It has been produced by chemical activation using phosphoric acid. Norit GAC 1240 is produced by heat activation and is a coal-based GAC grade. Physicochemical characteristics of virgin GAC grades are summarized in Table 1.
Table 1 Properties of activated carbon grades used in the study. Activated carbon
Norit GAC 1240
CAgran
Origin Physical form Activation method Total pore volume (cm3/g) Micropore (w < 2 nm) (cm3/g) Mesopore (w < 2–50 nm) (cm3/g) Macropore (w > 50 nm) (cm3/g)
Coal Granular Thermal 0.83 0.35 0.22 0.26
Wood Granular Chemical 1.75 0.26 0.60 0.89
Batch adsorption/desorption isotherm experiments served as a basis for choosing the right GAC type to be employed in continuous-flow biofiltration experiments. Norit 1240 was one of the most efficient GAC types in terms of adsorption characteristics among the others tested. On the other hand, the chemically activated CAgran exhibited inferior adsorption compared to thermally activated carbons but its desorbability was better [15]. Desorption is a very important phenomenon in terms of bioregeneration. In their studies Walker and Weatherley [16] stated that bioregeneration can only occur with compounds that readily desorb. Hence, bioregeneration is controlled by the reversibility of adsorption [17– 20]. Because of its better desorbability, CAgran was also selected for continuousflow biofiltration experiments. Parallel GAC columns were operated with raw and ozonated water. Since two different carbon grades were used, in total four columns were operated in this study. The raw water was ozonated at a rate of 2 mg O3/mg DOC at laboratory conditions. This dosage was found to produce the maximum biodegradable ¨ merli water dissolved organic carbon (BDOC) without mineralizing the DOC in the O ¨ merli raw water [21]. by more than 5% for O 2.3. Operation of columns The influent and the effluent water were daily analyzed for DOC, NH4+–N, NO2– N, NO3–N, Total Nitrogen, pH, UV254, UV280 and BDOC. 2 h composite samples were collected in the effluent. The dissolved oxygen concentrations were measured by using HACH HQ 40d Multimeter. Prior to DOC analysis, water samples were passed through 0.45 mm syringe driven PVDF filter (Millipore). Dissolved organic carbon (DOC) was determined by the high temperature combustion method (Standard Methods 5310 B) using a Teledyne-Tekmar Apollo 9000 model TOC analyzer. UV absorbance measurements were carried out at wavelengths 254 and 280 by a UV– visible double beam spectrophotometer (Shimadzu UV-2450) with 1 cm quartz cell. In order to enhance and accelerate the start-up of biological activity inside the BAC columns, bacteria were inoculated into the system before column operation. The concept of inoculation into GAC columns to make them biologically active was practiced by some authors including Mochhidzuki and Takeuchi [22], Zhao et al. [23], Nishijima and Speitel [3], Traenckner et al. [24]. An activated sludge MLVSS was not used as inoculum in the current study because of the probability that the culture contains pathogens. Pathogens should not be inoculated to a system which ¨ merli water were serves as drinking water purposes. Therefore, bacteria in the O enriched for three months in a suspended culture batch reactor, so that the inoculum in BAC system contained bacteria that originate from the reservoir itself and were acclimated to organic carbon of the source [25]. Before the suspended growth batch culture was transferred to the columns, it could remove up to 40% of ¨ merli water by biodegradation [25]. This culture was contacted DOC the incoming O with GAC overnight and then this bacteria–GAC mixture was transferred to BAC columns. In columns at least 10 cm of free water was kept above the packed height to ensure submergence of packing media. Thus, due to the inoculation, the columns were biologically active from the first day of the operation, and biodegradation and adsorption took place simultaneously. The influent and effluent DOC concentrations are reported as a function of the number of bed volumes instead of the volume treated or the time of treatment. Under the present working conditions, 1000 bed volumes were equal to 12.5 days of operation. During operation of columns, water samples were taken from the ports located at the 0, 27 and 52 cm above the bottom. Since the height of the packed GAC column was 50 cm, the top port was 2 cm above the GAC media. Columns were also operated in the sterile mode in order to differentiate between adsorption and biodegradation. In order to assure the sterility of the columns the influent raw water was always autoclaved.
3. Results 3.1. Comparison of DOC removal in sterile GAC and BAC columns
Fig. 1. Experimental set-up used in biofiltration experiments.
Adsorption and biodegradation are known to be the predominant mechanisms contributing to DOC removal in BAC columns. However, it is difficult to identify their relative importance at
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
different operational stages. Under sterile conditions only, the removal by adsorption alone can be determined. In the current study, CAgran and Norit 1240 columns were examined in a system in which only adsorption took place. The comparison was made only with raw water. In order to assure that adsorption took place only and no biodegradation, the influent raw water was autoclaved. On the other hand, in the BAC column occupied by microorganisms, both adsorption and biodegradation mechanisms would be dominant for DOC removal before breakthrough. After breakthrough is reached, DOC removal can occur mostly by biodegradation. The normalized effluent DOC concentrations in the sterile GAC and BAC columns are presented in Fig. 2. Examination of Fig. 2 shows that after approximately 1000 bed volumes, complete breakthrough of DOC was observed in the sterile CAgran column. Before breakthrough was reached, 25% of total DOC was non-adsorbable. The column was further operated for 1200 bed volumes, during which no DOC removal was detected. On the other hand, in the BAC column packed with CAgran, breakthrough commenced at later times. Additionally, the effluent DOC concentrations in the initial stages of BAC operation with CAgran were shown to be less than that of the sterile GAC column, indicating the presence of active bacteria effectively removing DOC. The effluent of BAC column before breakthrough quantifies both the non-adsorbable and non-biodegradable matter which amounted to 20% of the total DOC. At later stages, the presence of bacteria resulted in 47% DOC removal in the column filled with CAgran. The breakthrough commenced at approximately after 4000 bed volumes in the sterile column filled with Norit 1240. 22% of the initial DOC was found to be non-adsorbable in the sterile column. Compared to sterile CAgran, Norit 1240 outperformed in terms of DOC adsorption, which was also shown in batch adsorption isotherm experiments [15]. In the BAC column, breakthrough did
357
not start during 220 days of column operation. It is clear from Fig. 2 that biodegradation extended the operation life of Norit 1240. It is assumed that biological activity enhanced the adsorption capacity of GAC for non-biodegradable compounds by eliminating substances that would otherwise compete for adsorption sites. The non-adsorbable and non-biodegradable fraction leaving the Norit 1240 was 20% of total DOC. 3.2. Sampling from the ports of sterile GAC columns Port sampling at different operational stages of sterile Norit 1240 column showed that oxygen concentrations along the column depth decreased at all stages of column operation (data not shown). Through the 50 cm of carbon depth, approximately 90% of the oxygen was lost in the bulk liquid. This is due to the chemisorption of oxygen onto GAC pores. Thermally activated carbons have high affinities towards oxygen and upon contact with oxygen, surface chemistry of GAC may change [18]. Since thermal activation of these carbons is carried out in the absence of oxygen, they have a more reactive surface. Contrary to this, chemically activated carbons have a surface with fully oxidized active sites so that interaction with oxygen does not affect the surface. Sampling from the ports of the sterile CAgran revealed limited oxygen uptake (not more than 3.5% decrease) along the depth of the column. 3.3. Operation of BAC columns Continuous-flow biofilters were operated for more than 220 days before shutdown. There is a decreasing trend in influent DOC which is related to seasonal changes in DOC. In the ozonated water the influent DOC concentrations were not significantly different (p = 0.64) from raw water, as determined by ANOVA test at a 95% confidence interval. This showed that during ozonation mineralization of DOC was avoided. Indeed, the percent decrease in the original DOC upon ozonation was less than 5% in all cases. 3.3.1. Operation of CAgran The fastest breakthrough of DOC was observed in the column packed with CAgran (Fig. 3). Exhaustion took place approximately at the same time (about 2000 bed volumes) for both raw water and ozonated water. Therefore, ozonation did not seem to have any effect on adsorption of NOM during continuous-flow operation. Before the exhaustion of the column filled with CAgran, adsorption and biodegradation took place simultaneously on the surface since filters were biologically active from the first day of operation. The effluent DOC concentration was 0.6 mg/L in the case of both raw and ozonated water (Fig. 3). On the other hand, after exhaustion of filter capacity, biodegradation was the dominant mechanism. The difference between the influent and the effluent
Fig. 2. Normalized DOC in the effluent of sterile GAC/BAC of Norit 1240 and CAgran columns.
Fig. 3. Influent and effluent DOC concentrations in the two continuous-flow BAC columns tested with raw and ozonated water (RW: Raw water; OW: Ozonated water).
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
358
Fig. 4. Concentrations along the length of the CAgran at different bed volumes of BAC operation.
DOC concentrations after exhaustion can be attributed mostly to biodegradation. In this case, the average DOC removal by biodegradation is 47% for the raw water, and 53% for the ozonated water. At an average influent raw water of DOC 3.65 mg/L, an effluent DOC concentration of 1.94 mg/L was generally achieved. This means that 1.71 mg/L of DOC was biodegraded in the system fed with raw water. Surprisingly, the average biodegradable DOC (BDOC) concentration in the influent raw water was as low as 0.68 mg/L. For the ozonated water, influent DOC was lowered from an average value of 3.56 mg/L to 1.62 mg/L by biodegradation, which corresponds to a removal of 1.94 mg/L. The average influent BDOC for the ozonated water was 1.76 mg/L. This indicated that the biological DOC removal in the column fed with ozonated water is slightly higher than the amount of biodegradable DOC (BDOC) measured in the influent. The BDOC measurement, in principle, is done using a suspended culture with limited bacteria concentration and poor mixing conditions. It is therefore assumed in general that the actual BDOC concentration is underestimated this test. Moreover, BAC columns are attached-growth systems which are always more efficient in contaminant removal than suspended growth systems. Additionally, GAC provides a long retention time by adsorption which increases the chance of biodegradation of slowly biodegradable substances. Sampling along the length of the reactor had the purpose of examining the DOC, NH4+–N, and dissolved oxygen profiles in the
column. Sampling from ports reflected the concentration profiles after breakthrough was reached in CAgran columns and are summarized in Fig. 4. On average, 42% of the DOC was biodegraded in the first 25 cm of the column and a further 5% biodegradation was seen in the lower part of the column fed with raw water. Therefore, the first 25 cm of the column (equivalent to an EBCT of 9 min) was primarily responsible for biodegradation. It is expected that a biomass gradient was present where the maximum biomass was retained in the first centimeters of the filters and therefore most of the DOC removal occurred during the 9 min EBCT. This result is supported by some other studies [9,26–28]. Sampling along the depth of sterile CAgran column has shown that 3.5% of the initial dissolved oxygen was lost in a carbon depth of 0.5 m (data not shown). Therefore, further decrease in oxygen concentration through the depth was attributed to the uptake of oxygen due to organic biodegradation and nitrification. In nitrification, about 4.33 mg O2/mg NH4+–N is consumed by nitrifiers [29]. In biodegradation of organic matter the corresponding value is 1 mg DO/1 mg COD. The organic carbon concentration was converted into oxygen demand using a conversion factor of 2.66. The corresponding dissolved oxygen balance is presented in Table 2. In these calculations, the oxygen requirement due to nitrite oxidation was neglected because of negligible concentration. A similar comparison was also done in a study conducted by
Table 2 Examples of the calculation of the theoretical and actual oxygen consumptions in CAgran columns. Sample Raw Raw Raw Raw
water water water water
Ozonated Ozonated Ozonated Ozonated (1) (2) (3) (4) (5) (6) (7)
BV = 10100 BV = 15300 BV = 16160 BV = 17280
BV = 10100 BV = 15300 BV = 16160 BV = 17280
DDOC (1)
DDOCOD (2)
DNH4+–N (3)
DDONH4 þ N (4)
DDOcs (5)
DDOth (6)
DDOactual (7)
1.234 1.504 1.180 1.229
3.282 4.001 3.139 3.269
0.23 0.08 0.07 0.10
0.996 0.346 0.303 0.433
0.273 0.265 0.252 0.273
4.551 4.611 3.694 3.975
4.9 4.74 4.08 3.90
1.537 1.296 1.630 1.620
4.088 3.447 4.336 4.309
0.14 0.19 0.04 0.07
0.606 0.823 0.173 0.303
0.300 0.265 0.294 0.290
4.994 4.535 4.712 4.902
4.9 4.74 4.5 4.29
DDOC: Measured DOC removal (mg C/L) DDOCOD: Calculated DO consumption in terms of COD removed (2.66 1 mg DO/1 mg COD DDOC) DNH4+–N: Measured NH4+–N removal DDONH4 þ N : Calculated DO consumption for NH4+–N (4.33 mg O2/mg NH4+–N DNH4+–N) DDOcs: Determined DO consumption due to chemisorption (assumed as 3.5% Initial DO) DDOth: Theoretical DO consumption ðDDOCOD þ DDONH4 þ N þ DDOcs Þ DDOactual: Initial DO–Final DO
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
Yu et al. [30], the purpose of which was to explain the nitrogen loss and oxygen paradox in a full-scale treatment plant. Table 2 compares the oxygen uptake resulting from organic carbon biodegradation and nitrification with the measured oxygen uptake along the column depth (Columns (6) and (7) in Table 2). Results showed that the theoretical oxygen consumption and the measured dissolved oxygen values were in good agreement with each other. Thus, it was proven that the decrease in dissolved oxygen concentrations between Port 1 and Port 3 reflected biodegradation of organic carbon and nitrification. Organic carbon removal in the filters was due to biodegradation only but not adsorption. This suggested that carefully performed DO measurements may serve as an approximation for biological DOC removal in biofilters. This is of particular interest because of the relative simplicity of DO measurements. 3.3.2. Operation of Norit 1240 Norit 1240 is a microporous coal-based and steam activated carbon and had the best adsorption performance among the GACs tested [15]. In continuous-flow column experiments, CAgran reached breakthrough within 2000 bed volumes. On the other hand, during 220 days of operation (approximately 18000 bed volumes), breakthrough did not commence in the case of Norit 1240. The average effluent DOC concentrations for the columns operated with raw water and ozonated water were 0.69 mg/L and 0.66 mg/L, respectively. Effluent DOC concentrations remained constant regardless of influent DOC concentration. The average DOC removal efficiency in the column was 81% (Fig. 3). Samples taken from the ports at different bed volumes demonstrate the DOC, ammonium and the dissolved oxygen concentration profiles at different depths (Fig. 5). Since Norit 1240 has a high reactivity towards oxygen and chemisorbs it on its surface, the oxygen consumed through the depth of the column cannot be taken as a measure of bacterial activity. In the sterile Norit 1240 column, the dissolved oxygen concentration at the bottom port was as low as 0.84 mg/L (data not shown). In the biologically active Norit 1240 column (BAC), despite biological activity, the dissolved oxygen concentration did not further decrease. The most interesting finding in port sampling experiments with Norit 1240 was that, this carbon type can decrease the DOC
359
concentration to at least 1 mg/L by biodegradation. DOC concentration in the middle port remained constant (around 1 mg/L) during a 7000 bed volume operation. 7000 bed volumes correspond approximately to 87 days of operation. A 7000 bed volume is long enough for the mass transfer zone to shift to lower depths. Therefore, the decrease in DOC between the upper and middle ports should be attributed to biodegradation only. The DOC levels in the bottom port were further reduced to 0.6 mg/L. This concentration was the typical effluent DOC concentration observed during 220 days of column operation. Therefore, it is difficult to identify the mechanism leading to this effluent concentration. The results of molecular studies revealed that bacteria were also present in the bottom port [21]. Therefore, biodegradation might have played a role in DOC reduction in this region. On the other hand, there is a probability that adsorption is still occurring between the middle and bottom ports. The average DOC concentration entering the lower port was as low as 1 mg/L. Therefore, the GAC pores near the bottom may still be not saturated. 3.3.3. Effect of ozonation on biodegradation of organics It is well established that ozonation increases the biodegradable portion of DOC. Studies up to now showed that the increase in biodegradable DOC enhanced biological DOC removal. The percent increase in BDOC upon ozonation at a dose of 2 mg O3/mg DOC was as high as 245%, whereas on the average the increase was 166% (data not shown). Despite this huge increase, the biological DOC removal efficiency was not so much affected in the filters fed with ozonated water (Fig. 3). The biological DOC removal was only 8% higher than in the CAgran column fed with raw water. In case of Norit 1240, no enhancement of DOC removal could be observed. Therefore, it was concluded that ozonation did not have a pronounced effect on DOC removal. 3.3.4. Effect of specific UV absorption (SUVA) on biodegradation SUVA provides a quantitative measure of unsaturated bonds and/or aromaticity within NOM. Increase in SUVA generally reflects higher humification, aromaticity and hydrophobicity of DOM and hence, lower biodegradability. Calculation of SUVA based on DOC and UV254 measurements showed that the average SUVA values of raw and ozonated waters were 2.5 and 1.42 m1/(mg
Fig. 5. Concentrations along the length of the Norit 1240 at different bed volumes of BAC operation.
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
360
4. Discussion and conclusion 4.1. Comparison of biodegradation efficiencies in the Norit 1240 and CAgran columns
Fig. 6. Influent SUVA values in raw and ozonated water.
DOC/L), respectively. SUVA, which is the UV absorbance measured at a fixed wavelength (l, in nm) divided by DOC concentration is expressed as: SUVAl ¼
UVl ðcm1 Þ 100 DOCðmg=LÞ
(1)
UVl is usually in cm1 and DOC in mg/L. A conversion factor of 100 is used to express the unit of SUVA as m1/(mg DOC/L). Fig. 6 shows an increasing trend in the SUVA values of raw and ozonated water with respect to time. SUVA values in raw water were as high as 3.3, whereas in ozonated water these values never exceeded 1.7. According to the classification made by Edzwald and Tobiason [31], a SUVA value of 4 and greater values indicates that the water sample consists mainly of humic materials and higher molecular weight substances. When this value is below 2, NOM is mainly non-humic and hydrophobicity is low. SUVA values between 2 and 4 indicate that NOM consists of a mixture of aquatic humics and other NOM. The average SUVA value of the raw water was close to 2, therefore, the aromaticity within NOM was not high, and hence a high biodegradability of raw water can be expected. In order to investigate the role of humification on DOC biodegradation, the percent DOC biodegradation was correlated to SUVA. The Pearson’s product momentum correlation coefficient (rp) for the relationship was used for linear estimation of the strength and direction of the correlation. The relationship is depicted in Fig. 7. Indeed, in raw water samples the increase in SUVA value resulted in a decrease in percent DOC biodegradation. The overall Pearson coefficient determined at 95% confidence interval (0.82) indicated a moderate correlation between biodegradation and SUVA.
Fig. 7. Relationship between SUVA and DOC biodegradation in raw water.
Norit 1240 can decrease the DOC concentration at least down to 1 mg/L by biodegradation which corresponds to 72% removal. On the other hand, 0.6 mg/L of effluent DOC concentration can be due to either biodegradation, adsorption or the combination of both. In the CAgran column fed with raw water the average DOC removal by biodegradation was 47%. The difference between the removal efficiencies may be related to the adsorptive properties of GACs. The efficiency of adsorptive media (GAC) and non-adsorptive media (anthracite and sand) for biological DOC removal are well documented in literature. For example, Liu et al. showed that GAC filters enhanced the removal of less biodegradable substances in comparison to anthracite filters. Adsorption onto GAC can provide a longer retention time in filters for slowly biodegradable components. It is considered that similar to the improvement of removal efficiency in case of adsorptive media, a better adsorbing GAC can also lead to higher DOC biodegradation. In a previous study it was shown that microporous Norit 1240 outperformed macroporous CAgran in terms of adsorption [15]. In accordance with this, also in the current study Norit 1240 was also superior in terms of biodegradation efficiency. The ability of GAC to better adsorb and retain compounds increases their chance of biodegradation by bacteria. The major advantage of Norit 1240 over CAgran in terms of biodegradation is that it can adsorb considerable amounts of both readily and slowly biodegradable organics. Biodegradation by attached bacteria can lead to continuous bioregeneration of GAC. It is thought that, probably a couple of years should pass till breakthrough is achieved in the GAC column containing Norit 1240. Thus, using this type of carbon seems to be advantageous. Another possibility can be the production of biodegradable carbon on the surface of Norit 1240. The high oxygen consumption at GAC surface can cause surface catalytic reactions to occur on GAC surface. Surface catalytic processes can convert substances which are originally non-biodegradable into biodegradable ones [34]. The observation that more biodegradation took place in the Norit 1240 column than the CAgran column can be attributed to such surface catalytic reactions since Norit 1240 chemisorbs large amounts of dissolved oxygen on its surface. Some research has suggested that DOC removal can be limited by biomass concentration [32]. The fact that microporous GAC (Norit 1240) outperformed at NOM biodegradation may also be related to the bacterial amount in biofilters. Prior to column operation, approximately equal amounts of bacteria were added to each column. In general, little biogrowth occurs in GAC micropores because their small diameter (1–100 nm) does not allow penetration of bacteria, which typically have a diameter greater than 200 nm [13]. On the other hand, the relative amounts of bacteria in each filter may change during the operation. The pore sizes of GAC can influence bacterial retention and colonization. In a study conducted by Wang et al. [33], the bacterial concentrations (per gram of GAC) on a bituminous-coal-based (microporous), a lignite-coal-based (mesoporous) and a wood based (macroporous) activated carbon were compared. Although the highest biomass (quantified by phospholipids) was measured in the lignite-based carbon, results showed that small performance differences existed in terms of DOC biodegradation among micro-, meso- and macropore GACs. The role of microorganism concentration on biodegradation is unknown in the present system since biomass concentrations were not measured. However, it is considered that bacterial concentration in BAC columns can hardly be a rate limiting step for biodegradation
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
361
Table 3 DOC removals typically achieved by BAC columns in the literature. Water source Seagahan, UK Lake Vyrnwy, UK Norsborg, Sweden River Dee, UK Model Water Plonia River, PL Grand River, USA Miyun Reservoir Huangpu River Omerli Reservoir a b
DOC (mg/L) b
NA 2.4–4.8 NAb 3.0–7.9 4.0–5.0a 7.8–11.6a 5–7 4.9–7.3 5.2–7.7 2.9–4.9
Media
O3 dose
Removal (%)
Reference
Sand Sand Sand Sand Sand GAC GAC GAC GAC GAC
3.1–4.8 mg O3/L 1.1–2.5 mg O3/L 0.2–l mg O3/mg TOC 0.5 mg O3/mg TOC 6.7 mg O3/L 1.64 mg O3/mg TOC NA 3 mg/L 2.0–2.5 mg/L No ozonation
25 26.5 20–30 28 34–40 39 13–23 33.4 31 47–72
[36] [37] [38] [39] [40] [41] [42] [43] [44] This study
In terms of TOC. NA: Data not available.
since DOC concentrations are not high in most drinking water sources. Sampling from porst have shown that most of the DOC is removed within the 25 cm of column depth and lower depths are responsible for only minor DOC removals despite the existence of sufficient bacteria. Therefore, choosing the best type of GAC in biofiltration experiments results in better performance not only in terms of adsorption but also in terms of biodegradation, which extends the service life of the GAC. The compilation of literature studies on DOC removal is depicted in Table 3. The DOC removal efficiencies in current BAC columns seemed to be higher compared to literature values [35– 43]. In order to illustrate the organic characteristics of Huangpu River, Xu et al. [43] performed a pilot-scale treatment set-up including pre-ozonation, conventional treatment, post-ozonation and BAC filter and found that 31% of DOC could be removed by ozone-biofiltration system. In another study conducted by Li et al. [42] the granular activated carbon/O3-biological activated carbon (BAC) process was employed to treat raw water and compared to O3-BAC process. They showed that the presence of GAC in the ozone contact tank improved ozone utilization and biodegradability of the effluent and DOC removal efficiency increased from 33.4% to 39.5%. In the studies of Seredynska-Sobecka et al. [40], BAC was maintained by circulating raw water through the GAC filter. DOC removal efficiency of ozone-BAC system was 39% in the presence of 10 mg/L of phenol. However, there is not much information about the SUVA values in most studies. Therefore, a comparison cannot be made among studies. In the current study, the average SUVA values in the raw water were around 2.5. This ¨ merli was not highly aromatic may indicate that the raw water of O and therefore not recalcitrant to biodegradation. Hozalski et al. [44] observed enhanced removal of TOC by biodegradation of NOM sources with a lower SUVA value. When this is the case, even a slowly biodegradable compound can be biodegraded on the surface of the carbon, because GAC can provide the necessary retention of organics due to adsorption. This may be one of the reasons of higher DOC biodegradation compared to literature. This was also supported by the results of our recently published article ¨ merli water were enriched to be used as [25]. Bacteria in the O inoculum for BAC columns. This way the inoculum in BAC system contained bacteria that originate from the reservoir itself and were acclimated to organic carbon of the source. The suspended growth batch culture could remove up to 40% of DOC of the ¨ merli water by biodegradation [25]. Even this incoming O efficiency in a batch reactor is higher than the results of most of the studies (Table 3). In addition, the molecular weight distribution of NOM is also important factor for biodegradability [44] because lower molecular weight compounds are more easily transported across cell membrane and attacked by metabolic enzymes [6].
4.2. Effect of ozone on removal efficiency The removal efficiencies in the columns fed with raw and ozonated water were not significantly different from each other. In the CAgran column, the average DOC removal by biodegradation was 47% for raw water, ozonation slightly improved this and led to 53% removal. It was assumed that the inoculated bacteria were ¨ merli water and that they produced specific well acclimated to O enzymes to treat even the hardly biodegradable organics of the reservoir. It is for sure that ozonation made them readily biodegradable, but the contribution of ozonation to the overall removal was marginal. Most of the studies up to now documented the benefits of ozone in enhancing the biodegradability of NOM and leading to higher biodegradation in BAC columns. The effect of ozonation on DOC removal was more obvious when a non-adsorptive media like sand was used. For example, Shukairy et al. [45] compared the biofiltration efficiencies in ozonated and non-ozonated biofiltration columns consisting of sand using the raw surface water of the Ohio River. The percentage of DOC removal approximately doubled when ozonation preceded biofiltration (12% vs. 25%). The same experiments were also done with artificial water, using a solution of humic substances isolated and concentrated from ground water. However, DOC removals reported for the artificial water were significantly higher than those observed for any of the previously described studies that used natural waters (34%–48% for biofiltration; 64%–72% for ozone + biofiltration). In another study conducted by Moll et al. [46], ozonation slightly enhanced TOC removal in a biofiltration column made of sand. Ozonation at a dose of 0.9 mg O3/mg TOC increased the average TOC removal in the filters treating Harsha Lake water from 16% to 25%. DeWaters and DiGiano [47] found that pre-ozonation of humic substances obtained from Dismal Swamp resulted in 43% biodegradation of TOC in a fixed bed GAC column (Ozone dose: 1 mg O3/mg TOC, EBCT = 3.9 min; feed TOC = 7 mg/L). In contrast, Glaze [48] carried out a pilot-scale study with GAC beds at Shreveport, Louisiana to investigate the combination of ozone and GAC for removal of TOC and THM precursors. The results indicated that pre-ozonation did not affect TOC removal at any time and the average TOC removal efficiency in both columns was approximately 26%. 5. Conclusion This study showed that the choice of filter material is crucial in BAC systems in enhancing the biodegradation of organic matter. The ability of GAC to better adsorb and retain organic compounds increases the chance of biodegradation. This way, slowly biodegradable substances have a better chance to be removed in columns. Therefore, the GAC type leading to best adsorption would probably be the most suitable type for biodegradation as well.
362
K. Yapsakli, F. C¸ec¸en / Process Biochemistry 45 (2010) 355–362
As in our case, if raw waters have low SUVA values indicating a high biodegradability, ozone application may not be necessary for conversion of DOC into BDOC in order to increase the DOC biodegradation. The high retention time in BAC columns can lead to biodegradation of slowly biodegradable organics and eliminate the usage of ozonation. This way, operational cost associated with ozone can be eliminated. Acknowledgements ¨ BI˙TAK I˙C¸TAG (C¸110) and the The finance of this study by TU Research Fund of Bog˘azic¸i University (project no. 05Y103D) is gratefully acknowledged. We also express our gratitude to Jan C van den Dikkenberg (Norit Company, the Netherlands) for providing the carbon samples and data about characteristics and to Halil I˙brahim Uzun for providing the water from the reservoir. References [1] Weber Jr WJ, Pirbazari M, Melson GL. Biological growth on activated carbon: an investigation by scanning electron microscopy. Environ Sci Technol 1978;12:817– 9. [2] Voice TC, Pak D, Zhao X, Shi J, Hickey RP. Biological activated carbon in fluidized bed reactors for the treatment of groundwater contaminated with volatile aromatic hydrocarbons. Water Res 1992;26:1389–401. [3] Nishijima W, Speitel GE. Fate of biodegradable dissolved organic carbon produced by ozonation on biological activated carbon. Chemosphere 2004;56:113–9. [4] Camel V, Bermond A. The use of ozone and associated oxidation processes in drinking water treatment. Water Res 1998;32:3208–22. [5] Graham NJD. Removal of humic substances by oxidation/biofiltration processes – A review. Water Sci Technol 1999;40:141–8. [6] Leisinger T, Cook AM, Hu¨tter R, Nu¨esch J. Microbial degradation of Xenobiotics and Recalcitrant Compounds. New York: Academic Press; 1981. [7] Lawrence J. Semi-quantitative determination of fulvic acid, tannin and lignin in natural waters. Water Res 1980;14:373–7. [8] Novak JM, Mills GL, Bertsch PM. Estimating the percent aromatic carbon in soil and aquatic humic substances using ultraviolet absorbance spectroscopy. J Environ Qual 1992;21:144–7. [9] Servais P, Billen G, Bouillot P, Benezet M. A pilot study of biological GAC filtration in drinking water treatment. Aqua 1992;41(3):163–8. [10] Carlson G, Silverstein J. Effect of molecular size and charge on biofilm sorption of organic matter. Water Res 1998;32:1580–92. [11] Hattori K. Water treatment systems and technology for the removal of odor compounds. Water Sci Technol 1988;20:237–44. [12] Liu X, Huck PM, Slawson RM. Factors affecting drinking water biofiltration. J Am Water Works Assoc 2001;93:90–101. [13] Urfer D, Huck PM, Booth SDJ, Coffrey BM. Biological filtration for BOM removal: A critical review. J Am Water Works Assoc 1997;89:83–98. [14] Karanfil T, Erdog˘an I˙, Schlautman MA. Selecting filter membranes for measuring DOC and UV254. J Am Water Works Assoc 2003;95:86–100. ¨ , Can ZS. Impact of surface properties of granular [15] Yapsakli K, C¸ec¸en F, Aktas¸ O activated carbon and pre-ozonation on adsorption and desorption of natural organic matter. Environ Eng Sci 2009;26:489–500. [16] Walker GM, Weatherley LR. A simplified predictive model for biologically activated carbon fixed beds. Process Biochem 1997;32:327–35. [17] Schultz JR, Keinath TM. Powdered activated carbon treatment process mechanisms. J Water Pollut Control Federation 1984;56:143–51. [18] Jonge RJ, de Breure AM, van Andel JG. Bioregeneration of powdered activated carbon (PAC) loaded with aromatic compounds. Water Res 1996;30:875–82. ¨ , C¸ec¸en F. Effect of type of carbon activation on adsorption and its [19] Aktas¸ O reversibility. J Chem Technol Biotechnol 2006;81:94–101. ¨ , C¸ec¸en F. Effect of activation type on bioregeneration of various activated [20] Aktas¸ O carbons loaded with phenol. J Chem Technol Biotechnol 2006;81:1081–92. [21] Yapsakli K. Application of biological activated carbon (BAC) in drinking water treatment. Turkey: Bogazici University; Ph.D. thesis; 2008. [22] Mochhidzuki K, Takeuchi Y. The effects of some inhibitory components on biological activated carbon. Water Res 1999;33:2609–16.
[23] Zhao X, Hickey RF, Voice TC. Long-term evaluation of adsorption capacity in a biological activated carbon fluidized bed reactor system. Water Res 1999;33:2983–91. [24] Traenckner J, Wricke B, Krebs P. Estimating nitrifying biomass in drinking water filters for surface water treatment. Water Res 2008;42:2574–84. [25] Yapsakli K, C¸ec¸en F. Use of enriched inoculum for determination of biodegradable dissolved organic carbon (BDOC) in drinking water. Water Sci Technol Water Supply 2009;9(2):149–57. [26] Wang ZW. Assessment of biodegradation and biodegradation kinetics of natural organic matter in drinking water biofilters. USA: University of Cincinnati; Ph.D. Thesis; 1995. [27] Miltner RJ, Summers RS, Wang JZ. Biofiltration performance: Part II effect of backwashing. J Am Water Works Assoc 1995;87:64–70. [28] Hozalski RM, Goel S, Bouwer EJ. TOC removal in biologically active sand filters: effect of NOM source and EBCT. J Am Water Works Assoc 1995;87:40– 54. [29] Metcalf & Eddy. Wastewater Engineering, Treatment and Reuse, Fourth ed., New York: McGraw-Hill; 2003. [30] Yu X, Qi Z, Zhang X, Yu P, Liu B, Zhang L, et al. Nitrogen loss and oxygen paradox in full-scale biofiltration for drinking water biofilters. Water Res 2007;41: 1455–64. [31] Edzwald JK, Tobiason JE. Enhanced coagulation: US requirements and a broader view. Water Sci Technol 1999;40:63–70. [32] Carlson KH, Amy GL. BOM removal during biofiltration. J Am Water Works Assoc 1998;90:42–52. [33] Wang JZ, Summers RS, Miltner RJ. Biofiltration performance: Part I relationship to biomass. J Am Water Works Assoc 1995;87:55–63. [34] Uhl W. Biofiltration processes for organic matter removal. In: Rehm HJ, Reed G, editors. Biotechnology. Weinheim: Wiley VCH; 2000. [35] Gould MH, Cameron DA, Zabel TF. An experimental study of ozonation followed by slow sand filtration for the removal of humic colour from water. Ozone Sci Eng 1984;6:3–15. [36] Cable CJ, Jones RG. Assessing the Effectiveness of Ozonation Followed by Slow Sand Filtration in Removing THM Precursor Material from an Upland Raw Water. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 29–37. [37] Seger A, Rothman M. Slow sand filtration with and without ozonation in nordic climate. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996 . p. 119–28. [38] Yordanov RV, Lamb AJ, Melvin MAL, Littlejohn J. Biomass characteristics of slow sand filters receiving ozonated water. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 107–18. [39] Odegaard H. The development of an ozonation/biofiltration process for the removal of humic substances. In: Graham N, Collins R, editors. Advances in Slow Sand and Alternative Biological Filtration. Chichester: John Wiley & Sons; 1996. p. 39–49. [40] Seredynska-Sobecka B, Tomaszewska M, Janus M, Morawski AW. Biological activation of carbon filters. Water Res 2006;40:355–63. [41] Emelko MB, Huck PM, Coffey BM, Smith EF. Effects of media, backwash, and temperature on full-scale biological filtration. J Am Water Works Assoc 2006;98:61–73. [42] Li L, Zhu W, Zhang P, Zhang Q, Zhang Z. AC/O3-BAC processes for removing refractory and hazardous pollutants in raw water. J Hazard Mater 2006;135:129–33. [43] Xu B, Gao NY, Sun XF, Xia SJ, Siminnot MO, Causserand C, et al. Characteristics of organic removal in Huangpu River and treatability with the O3-BAC process. Sep Purif Technol 2007;57:348–55. [44] Hozalski RM, Bouwer EJ, Goel S. Removal of natural organic matter (NOM) from drinking water supplies by ozone-biofiltration. Water Sci Technol 1999;40:157–63. [45] Shukairy HM, Summers RS, Miltner RJ. The impact of ozonation and biological treatment on disinfection by-products. In: Proceedings of the 4th Drinking Water Workshop Montreal; 1992. pp.1–20. [46] Moll DM, Summers RS, Fonseca AC, Matheis W. Impact of temperature on drinking water biofilter performance and microbial community structure. Environ Sci Technol 1999;33:2377–82. [47] DeWaters JE, Digiano FA. Influence of ozonated natural organic matter on the biodegradation of a micropollutant in a GAC bed. J Am Water Works Assoc 1990;82:69–80. [48] Glaze WH. 1982. Pilot-scale evaluation of biological activated carbon for the removal of THM precursors. EPA-600/S2-82-046, Environmental Protection Agency.