Accepted Manuscript

Impacts of backwashing on granular activated carbon filters for advanced wastewater

treatment

Joshua Frank, Aki Sebastian Ruhl, Martin Jekel

PII: S0043-1354(15)30233-5
DOI: 10.1016/j.watres.2015.09.020
Reference: WR 11528

To appear in: Water Research

Received Date: 12 June 2015

Revised Date: 9 September 2015
Accepted Date: 10 September 2015

Please cite this article as: Frank, J., Ruhl, A.S., Jekel, M., Impacts of backwashing on granular
activated carbon filters for advanced wastewater treatment, Water Research (2015), doi: 10.1016/
j.watres.2015.09.020.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

1 Impacts of backwashing on granular activated

2 carbon filters for advanced wastewater treatment

3 Joshua Frank1, Aki Sebastian Ruhl1,* and Martin Jekel1

PT
1
4 Chair of Water Quality Control, Technische Universität Berlin, Sekr. KF4, Straße des 17. Juni 135,

RI
5 10623 Berlin, Germany.

SC
*
6 Corresponding author, Tel.: +49 30 314 25493; fax: +49 30 314 23313. E-mail: aki.s.ruhl@tu-berlin.de

7 Abstract

U
AN
8 The use of granular activated carbon (GAC) in fixed bed filters is a promising option for the

9 removal of organic micropollutants (OMP) from wastewater treatment plant effluents.

M

10 Frequent backwashing of the filter bed is inevitable, but its effect on potential filter

11 stratification is not well understood yet and thus has been evaluated in the present study for
D

12 two commercial GAC products. Backwashing of GAC filters was simulated with 10 or 100
TE

13 filter bed expansions of 20 or 100% at backwash velocities of 12 and 40 m/h, respectively.

14 Five vertical fractions were extracted and revealed a vertical stratification according to grain
EP

15 sizes and material densities. Sieve analyses indicated increasing grain sizes towards the

16 bottom for one GAC while grain sizes of the other GAC were more homogeneously
C

17 distributed throughout the filter bed. The apparent densities of the top sections were
AC

18 significantly lower than that of the bottom sections of both products. Comparative long term

19 fixed bed adsorption experiments with the top and bottom sections of the stratified GAC

20 showed remarkable differences in breakthrough curves of dissolved organic carbon, UV light

21 absorption at 254 nm wavelength (UVA254) and OMP. GAC from the upper section showed

22 constantly better removal efficiencies than GAC from the bottom section, especially for

1
 ACCEPTED MANUSCRIPT
23 weakly adsorbing OMP such as sulfamethoxazole. Furthermore correlations between UVA254

24 reductions and OMP removals were found.

25 Keywords: Advanced wastewater treatment; adsorption; trace organic contaminants;

26 pharmaceuticals; filter backwash; breakthrough curve.

PT
27 1. Introduction

RI
28 Recent studies indicate that the presence of organic micropollutants (OMP) of anthropogenic

29 origin such as pharmaceuticals, diagnostics and industrial chemicals might be a risk to

SC
30 humans and aquatic ecosystem. The origin and stabilities of OMP are well documented but

31 long term impacts and risks are not known sufficiently (Schwarzenbach et al. 2006, Ternes

32

U
2007). In the near future regulatory limits (e.g. environmental quality standards in the EU)
AN
33 might require effective methods for the elimination of OMP.
M

34 Due to the high persistence and insufficient elimination of several OMP in conventional

35 wastewater treatment plants (WWTP) (Jelic et al. 2011, Oulton et al. 2010, Verlicchi et al.
D

36 2012) advanced wastewater treatment becomes increasingly relevant (Eggen et al. 2014,
TE

37 Reemtsma et al. 2006). Besides oxidative treatment of OMP with ozone (Bahr et al. 2007)

38 and OMP retention by nano-filtration (Snyder et al. 2007) OMP adsorption onto powdered
EP

39 (PAC) or granular activated carbon (GAC) has been proven to be effective for the removal of

40 OMP from WWTP effluents (Altmann et al. 2014, Boehler et al. 2012, Margot et al. 2013).
C

41 WWTP effluents contain elevated concentrations of particles which accumulate on GAC filter
AC

42 material and cause increased head losses that necessitate frequent backwashing. Hong et

43 al. (2006) assumed that backwashing causes complete mixing of the GAC and that the

44 mixing leads to earlier breakthrough due to solid phase concentrations higher than the

45 corresponding liquid phase concentration close to the filter effluent. Shpirt and Alben (1986)

46 observed that backwashing leads to a classification of GAC grains with respect to their grain

47 size but adsorption properties were not considered. Pota and Mathews (1999) considered

2
 ACCEPTED MANUSCRIPT
48 stratification according to particle sizes in breakthrough simulations and proposed a reversed

49 stratified GAC filter with smaller grains are at the effluent end of the filter and larger grains at

50 the inflow end. Batch experiments with pulverized GAC from different strata of a frequently

51 backwashed filter revealed greater adsorption capacities for GAC from the top (Ruhl et al.

52 submitted). However, the impact of backwashing on filter stratification and corresponding

PT
53 differences in adsorption performance is not well understood yet.

54 The objectives of this study were 1) to describe the stratification of GAC depending on the

RI
55 number of backwash cycles and the backwash intensity (as percentage in filter bed

SC
56 expansion), 2) to analyze differences in OMP adsorption in continuous adsorption tests with

57 GAC from different levels of a stratified filter bed and 3) to test the correlation of UV

U
58 absorption at 254 nm wavelength (UVA254) with OMP removals as recently reported (Anumol
AN
59 et al. 2015, Zietzschmann et al. 2014a) for different GAC fractions.

60 2. Material and Methods

M

2.1 Adsorbents and water

D

61
TE

62 Two different commercial coal based GAC were evaluated in the present study: GAC #1

63 (AquaSorb 5000, Jacobi, 12x30 mesh) and GAC #2 (Norit 1240, 12x40 mesh). All GAC
EP

64 samples were wetted in deionized water and degassed in a vacuum chamber prior to

65 backwash and adsorption experiments.

C
AC

66 Backwashing was conducted in deionized water. Adsorption tests were conducted with Berlin

67 tap water with a comparably high DOC concentration of 4.5 mg/L. Tap water was used

68 instead of WWTP effluent for better reproducibility of the results and due to practical reasons.

69 A stock solution containing different OMP (20 mg/L each, listed in Table S1 in the Supporting

70 Information) was added to achieve target concentrations of 1 µg/L of each OMP. Here, we

71 focus on carbamazepine (CBZ) and benzotriazole (BTZ) as process indicator substances for

3
 ACCEPTED MANUSCRIPT
72 well adsorbable compounds and sulfamethoxazole (SMX) as indicator substance for

73 moderately adsorbable compounds (Jekel et al. 2015).

74 2.2 Stratification of GAC

75 The relation between flow rate and filter bed expansion was determined in an experimental

PT
76 setup as schematically shown in Figure 1a with 60.0 g pre-wetted and degassed GAC #1

77 (initially 139 mL bed volume, 405 mm bed height). A linear relationship between backwash

RI
78 velocity and filter bed expansions was observed (Figure S1). Samples of GAC #1 were 10 or

79 100 times backwashed with defined bed expansions of 20 or 100% resulting in four

SC
80 combinations (10x20%, 10x100%, 100x20% and 100x100%). The number of cycles was

U
81 varied to clarify if stratification already occurs after a few backwashes or if it is only relevant

82 in the long term. Low expansions of 20% are relevant for full scale applications (Shpirt and
AN
83 Alben 1986). Complete bed fluidization at expansions of 100% is not feasible in practical
M

84 applications but was performed to approach the maximum extent of stratification. Air was not

85 injected during backwashing.

D

86 GAC #2 was investigated in less detail. Based on the results obtained with GAC #1, only
TE

87 10x100% expansions were applied on GAC #2. After the repeated backwashing procedures,

88 5 identical volume fractions of GAC were collected at the bottom by controlling a valve at the
EP

89 top of the water filled column.

C

90 2.3 Long term adsorption tests

AC

91 Stratified GAC from the top and the bottom sections (fractions A and E) were directly

92 compared in filtrations experiments as schematically shown in Figure 1b. In order to compare

93 identical bed volumes of 24 mL, different amounts (Table S2) according to different apparent

94 densities were filled into columns (21 mm inner diameter, 350 mm length) leading to a filter

95 bed of 70 mm height. A filter velocity of 5 m/h corresponding to a flow rate of 30 mL/min and

96 an empty bed contact time (EBCT) of 0.84 min, respectively, was adjusted with a peristaltic

4
 ACCEPTED MANUSCRIPT
97 pump. The elevated water table of the influent tap water in two identical 1 m³ stainless steel

98 tanks led to full saturation during down flow operation. Samples were taken daily at the filter

99 effluents and in 3-5 day periods form the influent water. The experiments were conducted at

100 room temperature around 20°C.

101 The efficiencies of GAC filter volume elements were compared according to carbon usage

PT
102 rates (CUR10%) as defined in equation 1 with the apparent bed density ρb and the number of

103 treated bed volumes (BV10%) until a concentration of 10% of the influent concentration was

RI
104 detected in the effluent (Zietzschmann et al. 2014b).

SC
ρb
105 CUR10%= BV (Eq. 1)
10%

106 2.4 Analyses

U
AN
107 All five fractions of each backwash experiment were analyzed for volumes with a measuring
M

108 cylinder and for weight with a digital balance (both submerged in water and after drying at

109 105°C). Masses of water saturated GAC samples with identical water and GAC levels were
D

110 determined to approximate the pore volume of the bed. Bed volumes of water saturated and
TE

111 dry GAC were almost identical. The water content W was calculated from the masses of the

112 water saturated GAC bed msat and the dried GAC bed mdry, the water density ρH2O and the
EP

113 bed volume VGAC according to equation 2.

C

ሺ௠ೞೌ೟ ି௠ಸಲ಴ ሻ∙ఘೢೌ೟

114 W= (Eq. 2)
௏ಸಲ಴
AC

115 The calculated water content represents the sum of inter- and intra-granular pore volume

116 with the inter-granular pore volume being the primary share. Grain size distributions were

117 obtained by manual sieve analyses (mesh sizes are shown with the results).

118 The UVA254 was analyzed in quartz cuvettes (10 mm) with a dual beam spectral photometer

119 (Lambda 12 UV-VIS, Perkin Elmer, Germany). DOC concentrations were analyzed with size

120 exclusion chromatography and continuous UV and organic carbon detections (DOC-Labor

5
 ACCEPTED MANUSCRIPT
121 Huber, Germany). Chromatograms were analyzed using the software ChromCALC (Huber et

122 al. 2011).

123 OMP concentrations were quantified with high performance liquid chromatography coupled

124 to tandem mass spectrometry (HPLC-MS/MS) using an XSelect HSS T3 column (2.5 µm,

125 2.1·50 mm, Waters, USA) with a linear gradient (0.5 mL/min, ultra-pure water with 0.1%

PT
126 formic acid and 5% methanol versus methanol) and a TSQ Vantage (Thermo Fisher

127 Scientific, USA) using electrospray ionization. Two mass fragments were analyzed for each

RI
128 compound and recovery was considered according to deuterated internal standards. Data

SC
129 were evaluated using the software Xcalibur 2.1 (Thermo Scientific). Concentrations below

130 the limit of quantification of 0.05 µg/L for all OMP were considered to be 0.05 µg/L.

131 3. Results
U
AN
132 3.1 Apparent densities and pore volumes
M

133 The apparent densities of the stratified fractions from GAC #1 normalized to the highest
D

134 apparent density (fraction E) are shown in Figure 2 (numerical values are listed in Table S3).
TE

135 GAC #1 without backwash (0x0%) already revealed stratification caused by different sinking

136 velocities at the first filling of the GAC into the water saturated column. Apparent densities of
EP

137 GAC #1 after intensive backwashes (100x100%) were 288 g/L for the top layer (fraction A)

138 and 519 g/L for the bottom layer (fraction E) corresponding to a decrease of 44% from the
C

139 bottom to the top. Similar stratifications with regard to bulk densities were obtained for only
AC

140 10 backwash cycles (10x100%) or with only 20% expansion (100x20%). The apparent

141 densities for GAC #1 after 100x20% expansions were 293 g/L for fraction A and 507 g/L for

142 fraction E and 314 g/L for fraction A and 546 g/L for fraction E after 10x100% expansion,

143 respectively. Both backwash intensities showed a decrease of 42% from the bottom to the

144 top.

6
 ACCEPTED MANUSCRIPT
145 Smaller grains probably filled the pore space between larger grains to a lesser extent in

146 stratified filter beds. This effect causes a decrease in overall apparent densities or an

147 increase of the total bed volume and the contact time. Greater inter-granular porosity leads to

148 lower flow velocity and thus longer contact time when preferential flow is excluded. The pore

149 volumes were estimated by comparing weights of water saturated and dried GAC. As

PT
150 expected, differences in pore volume including inter- and intra-granular pores were inversely

151 distributed but were only in the range of 20% as shown in Figure 2.

RI
152 Differences between 100x100% and 10x100% expansions were very small which indicated

SC
153 that stratification already occurred within 10 expansions and probably recurred after each

154 expansion. Based on these observations GAC #2 was expanded to 10x100% only.

155
U
Lower apparent densities in the upper layers might partly be caused by higher intra-granular
AN
156 porosity which might be connected to increased adsorption capacities. Moore et al. (2001)

157 investigated pore size distributions in a full scale GAC filter and observed greater cumulative
M

158 pore volumes in the upper layers and greater micro-pore volumes in the bottom layers. A
D

159 greater total pore volume in the upper fractions might contribute to lower apparent densities
TE

160 and higher adsorption capacities.

161 Differences of apparent densities between the fractions of GAC #2 (10x100%) were not as
EP

162 wide as for GAC #1 as shown in Figure 3. The apparent densities of fractions A and E were

163 376 g/L and 449 g/L, respectively. The apparent density of fraction A was only 16% less than
C

164 that of fraction E. The total pore volumes were even more similar with differences of less
AC

165 than 6% between the bottom and the top fractions.

166 However, direct linkages between the apparent densities or pore volumes and the adsorption

167 performances of the fractions are not possible and filtrations tests are mandatory to compare

168 removal efficiencies.

169 3.2 Grain size distribution

7
 ACCEPTED MANUSCRIPT
170 Intensive backwashing of the filter bed also caused a separation and stratification of grain

171 sizes as shown in Figures 4 and 5 (numerical values are provided in Table S4). For all GAC

172 #1 samples, significant differences were found between fractions A and E. The unexpanded

173 GAC #1 already showed stratification as a greater share of larger grains was found in lower

174 strata. Compared to the reference sample, a considerable stratification developed during the

PT
175 backwashes. While the reference sample showed wide particle size distribution in all

176 fractions the backwashed samples excluded the large grains in the top fraction and small

RI
177 grains in the bottom fraction. Fraction A from GAC #1 (100x100%) consisted of 94% grains

178 smaller than 1 mm whereas fraction E consisted of 46% grains larger than 1mm (Figure 5d).

SC
179 Although the apparent densities and pore volumes of GAC #2 (10x100%) were more

U
180 homogeneously distributed, a more pronounced stratification with regard to grain sizes was
AN
181 found as shown in Figure 5. Fraction A consisted of 96% grains smaller than 1 mm while

182 fraction E was made up 15% smaller than 1 mm and almost no grains smaller than 0.63 mm.
M

183 Photographs of the top and the bottom fractions of GAC #2 are provided in Figure 6.

184 Differences in grain size distributions of GAC #2 were observed between all fractions while
D

185 differences between fractions A, B and C of GAC #1 were negligible.

TE

186 Following expansion, larger grains migrate to the bottom section due to greater settling
EP

187 velocities. A slight abrasion of grains cannot be excluded as the cumulated weights of the

188 individual fraction indicate a slight decrease of the largest grain size (Figure S2 and data in
C

189 Table S4). Shpirt et al. (1986) also demonstrated that different grain sizes can be found after
AC

190 backwashing at different bed heights and that abrasion leads to a decrease of grain sizes in

191 the long-term.

192 3.3 Adsorptive UVA254 reduction

193 The continuous adsorption tests showed significant differences between the top and the

194 bottom fractions with regard to UVA254 reductions as shown in Figure 7 (results for other

195 backwashes are provided in Figures S3-S4). As the apparent densities varied for fractions A

8
 ACCEPTED MANUSCRIPT
196 and E, the masses for equivalent filter volumes differed accordingly. With respect to these

197 variances, the graphs show the relative concentrations versus liters of treated water per

198 gram of GAC. The top fraction A constantly reduced the UVA254 to a greater extent

199 throughout the entire experiment for all samples and for both GAC.

200 Fraction A achieved significantly greater reductions especially for filtered volumes up to 20

PT
201 L/g. GAC #1 (100x100%) started at 84% reduction with fraction A whereas fraction E only

202 reduced the UVA254 by 55%. An 80% breakthrough of UVA254 was estimated by linear

RI
203 interpolation and occurred at 11.3 L/g for fraction A and 1.9 L/g for fraction E. The

SC
204 experiment with GAC #2 (10x100%) confirmed the trend with an initial reduction of 93% for

205 fraction A and 70% for fraction E. GAC #2 (10x100%) showed an overall higher adsorption

U
206 performance compared to GAC #1 but also showed similar differences between the two
AN
207 fractions. The results clearly show that stratification leads to different adsorption efficiencies

208 for different bed depths of backwashed GAC filters.

M

209 3.4 DOC removal

D

210 DOC removals indicate similar tendencies compared to UVA254 reduction as shown in
TE

211 Figure 7 (results for other backwashes are provided in Figures S5-S6). GAC #1 (100x100%)

212 showed initial removals of 82% for fraction A and 52% for fraction E. Relative effluent
EP

213 concentrations of 80% were detected after 53.1 L/g for fraction A and 20.2 L/g for fraction E.

214 GAC #2 (10x100%) started with a removal of 93% for fraction A and 58% for fraction E.
C

215 Relative effluent concentrations of 80% were found after 6.9 L/g for fraction A and at 2 L/g for
AC

216 fraction E.

217 Similar to the UVA254 reduction, LC-OCD measurements confirmed the differences between

218 the top and the bottom fraction for both GAC and all classification setups. Results for the

219 DOC characterization with different throughputs for GAC #1 (100x100%) are shown in

220 Figure 8. As drinking water was used for the experiments DOC concentrations were

221 comparably low and the DOC consisted mainly of humic substances, building blocks and low

9
 ACCEPTED MANUSCRIPT
222 molecular-weight acids and neutrals. Differences in removal performances were observed

223 between the top and the bottom fractions throughout the full range of molecule sizes

224 corresponding to different detection times. For the comparison of removals of specific

225 molecule sizes, the OC responses at specific detection times were compared. At a retention

226 time of 51 min within the main peak, fraction A removed 66.4% of the molecules at around

PT
227 0.1 L/g whereas fraction E only removed around half that much (36.2%). At a later retention

228 time of 59 min the differences between both fractions decreased to 53.3% for A and 37.2%

RI
229 for E. These results suggest that greater differences occurred for larger molecules. Velten et

230 al. (2011) reported that building blocks do not adsorb well onto granular activated carbon.

SC
231 Therefore differences can be expected to be smaller. These observations were verified at 6.9

232 and 7.3 L/g with similar tendencies. Humic substances as comparably large molecules

233
U
probably do not fit into GAC meso- and micro pores and adsorb rather in macro pores. DOC
AN
234 was reported to significantly reduce the adsorption capacity for target contaminants

235 (Summers et al. 1989). In wastewater treatment plant effluents pore fouling can be
M

236 considered to be more significant.

D

237 The SUVA as the ratio between UVA254 and DOC is known as a helpful parameter to
TE

238 characterize bulk organic constituents (Cecen and Aktas 2012). In the beginning of the

239 experiment the reduction of UVA254 was slightly greater than the corresponding removal of
EP

240 DOC as shown in Figure 7 and therefore the SUVA was slight decreased (2.2 L/(mg·m))

241 compared to the influent (2.4 L/(mg·m)). The development of SUVA is exemplarily shown in
C

242 Figure S7 for GAC #1 (100x100%).

AC

243 3.5 Removal of OMP

244 The removal of OMP in long term filtration experiments was evaluated for GAC #1 (100x20%

245 and 100x100%) and GAC #2 (10x100%). BTA and CBZ were selected as process indicator

246 substances for well adsorbing OMP and SMX for medium adsorbing OMP (Jekel et al. 2015).

247 Breakthrough curves of other OMP and other backwash intensities are provided in Figures

248 S8-S13.
10
 ACCEPTED MANUSCRIPT
249 A comparison between 100 cycles with 20% and 100% expansions of GAC #1 samples in

250 Figure 9 (left) shows significant and comparable differences between fractions A and E.

251 Effluent concentrations developed similarly for both backwash intensities. Differences for

252 20% and 100% expansions were comparably small. Unexpected gas inclusions were

253 observed after 25 L/g for fraction E of GAC #1 (100x20%) and probably caused shorter

PT
254 contact times (Ruhl and Jekel 2014) and thus higher relative concentrations.

255 GAC #2 (10x100%) showed greater removals of all tested OMP. However, significant

RI
256 differences were observed between the top and the bottom fraction for all OMP, too.

SC
257 Within the runtime only small differences between fractions A and E were found for BTA as

258 overall adsorption was well for all fractions and GAC samples. Fraction A of GAC #1

259
U
(10x100%) still adsorbed around 100% of BTA after 73 L/g and fraction A of GAC #2 also
AN
260 adsorbed around 100% at the end of the experiment.
M

261 Removal efficiencies for CBZ were much lower for GAC #1 (100x100%) and relative

262 differences between the top and the bottom fractions were greater. GAC #2 (10x100%)
D

263 showed very significant differences for CBZ as fraction A again revealed a 10%-
TE

264 breakthrough at 55.3 L/g compared to 2.7 L/g for fraction E. SMX as indicator OMP for

265 weakly adsorbing compounds showed similar results. Differences between activated carbons
EP

266 with regard to specific OMP have been reported for powdered activated carbon (Ruhl et al.

267 2014).
C
AC

268 Differences in CUR10% are shown in Figure 10 with great differences between fractions A and

269 E for all OMP. However, due to the low filter bed height, residence times were too short to

270 achieve great removals over long periods. CUR10% for SMX cannot be evaluated for fraction

271 E of both GAC as a relative concentration above 10% was already quantified in the first

272 sample.

273 3.6 Correlation between OMP removals and UVA254 reduction

11
 ACCEPTED MANUSCRIPT
274 A correlation of OMP removals from WWTP effluent, especially of carbamazepine, diclofenac

275 and iodinated x-ray contrast media with the corresponding reduction of UVA254 was reported

276 for batch tests by Zietzschmann et al. (Zietzschmann et al. 2014a) and confirmed for fixed

277 bed activated carbon adsorbers by Anumol et al. (2015). Figure 11 shows the removal of

278 BTA, SMX and CBZ in comparison to the corresponding reductions of UVA254 for the GAC

PT
279 fixed bed filter effluents. GAC #1 (100x100%) showed complete BTA removals for UVA254

280 reductions above 10% for both fractions A and E. CBZ removals around 80-90% were

RI
281 observed for UVA254 reductions above 20% for both fractions. SMX removals above 80%

282 were only obtained for corresponding UVA254 reductions above 50%. SMX removals above

SC
283 80% were only observed in the first effluent sample of fraction E.

U
284 Fraction A of GAC #2 (10x100%) revealed constant removals of BTA and UVA254 reductions
AN
285 above 15%. Fraction E also achieved BTA removals above 90% at corresponding UVA254

286 reductions above 15%. CBZ removals above 90% were obtained for UVA254 reductions
M

287 above 25% for both fractions.

D

288 UVA254 reduction is thus a helpful indicator for OMP removals in GAC fixed bed filters as
TE

289 OMP quantification by mass spectrometry cannot be performed continuously. Further

290 investigations with larger pilot plant or full scale GAC filters are required to verify the
EP

291 correlations.

292 Conclusion
C
AC

293 This study aimed at investigating the impact of GAC filter backwashing on stratification and

294 its influences on bed depth dependent removals of different OMP and DOC and reductions of

295 UVA254. Results for two commercial GAC products showed that backwashing causes

296 significant stratification according to grain size and material density. This means that the

297 middle of the filter with regard to volume does not correspond to the middle with regard to

298 GAC mass. As similar stratifications were found for 10 and 100 backwash cycles, the major

299 stratification occurred within 10 cycles with a re-arrangement in the following 90 cycles.

12
 ACCEPTED MANUSCRIPT
300 These results indicate that significant differences from top to bottom can be expected for all

301 GAC filters that were backwashed only a few times.

302 Stratification of GAC #1 was primarily observed according to different apparent densities.

303 The apparent densities ranged from 288 g/L for the top and 519 g/L for the bottom fraction

304 (100x100%). Stratification of GAC #2 after 10 cycles with 100% filter bed expansions was

PT
305 primarily according to grain size as the top fraction consisted to 96% of grains smaller than 1

306 mm while the bottom fraction consisted of only 15% grains smaller than 1 mm. Therefore the

RI
307 potential of differences along the vertical direction caused by backwashing is specific for a

SC
308 given GAC.

309 The impact of stratification on OMP removals was confirmed in continuous fixed bed

310
U
adsorption tests. The local efficiencies in a GAC filter decreased from top to bottom and thus
AN
311 a partial breakthrough at a sampling point in the middle of a filter cannot be extrapolated to

312 the whole filter bed. Batch test with material from different levels of a GAC filter without
M

313 considering gradients of bulk densities (Landwehrkamp et al. 2015) might not reflect actual
D

314 differences.
TE

315 The good correlation of OMP removals with UVA254 reduction for different fractions of a given

316 GAC implies that photometric measurements in the influent and effluent of a filter can be
EP

317 used to control the breakthrough of target OMP without expensive analyses of OMP

318 concentrations.
C
AC

319 Acknowledgements

320 The investigations were supported by the Federal Ministry of Education and Research

321 (BMBF) within the projects ASKURIS project (contract 02WRS1278A) (Jekel et al. 2013) and

322 TestTools (contract 02WRS1368A) as part of the funding measure RiSKWa. We thank F.

323 Zietzschmann for support in OMP analyses.

13
 ACCEPTED MANUSCRIPT
324 References

325 Altmann, J., Ruhl, A.S., Zietzschmann, F. and Jekel, M. (2014) Direct comparison of
326 ozonation and adsorption onto powdered activated carbon for micropollutant removal in
327 advanced wastewater treatment. Water Research 55, 185-193.

328 Anumol, T., Sgroi, M., Park, M., Roccaro, P. and Snyder, S.A. (2015) Predicting trace
329 organic compound breakthrough in granular activated carbon using fluorescence and UV

PT
330 absorbance as surrogates. Water Research 76(0), 76-87.

331 Bahr, C., Schumacher, J., Ernst, M., Luck, F., Heinzmann, B. and Jekel, M. (2007) SUVA as
332 control parameter for the effective ozonation of organic pollutants in secondary effluent.

RI
333 Water Science and Technology 55(12), 267-274.

334 Boehler, M., Zwickenpflug, B., Hollender, J., Ternes, T.A., Joss, A. and Siegrist, H. (2012)

SC
335 Removal of micropollutants in municipal wastewater treatment plants by powder-activated
336 carbon. Water Science and Technology 66(10), 2115-2121.

337 Cecen, F. and Aktas, Ö. (2012) Activated Carbon for Water and Wastewater Treatment:

U
338 Integration of Adsorption and Biological Treatment, Wiley-VCH, Weinheim.
AN
339 Eggen, R.I.L., Hollender, J., Joss, A., Scharer, M. and Stamm, C. (2014) Reducing the
340 Discharge of Micropollutants in the Aquatic Environment: The Benefits of Upgrading
341 Wastewater Treatment Plants. Environmental Science & Technology 48(14), 7683-7689.
M

342 Hong, S. and Summers, R.S. (2006) Effect of backwashing on activated carbon adsorption
343 using plug flow pore surface diffusion model. Korean Journal of Chemical Engineering 23(1),
344 57-62.
D

345 Huber, S.A., Balz, A., Abert, M. and Pronk, W. (2011) Characterisation of aquatic humic and
346 non-humic matter with size-exclusion chromatography - organic carbon detection - organic
TE

347 nitrogen detection (LC-OCD-OND). Water Research 45(2), 879-885.

348 Jekel, M., Dott, W., Bergmann, A., Duennbier, U., Gnirss, R., Haist-Gulde, B., Hamscher, G.,
349 Letzel, M., Licha, T., Lyko, S., Miehe, U., Sacher, F., Scheurer, M., Schmidt, C.K.,
EP

350 Reemtsma, T. and Ruhl, A.S. (2015) Selection of functional organic indicator substances for
351 the anthropogenically influenced water cycle. Chemosphere 125, 155-167.

352 Jekel, M., Ruhl, A.S., Meinel, F., Zietzschmann, F., Lima, S., Baur, N., Wenzel, M., Gnirß, R.,
C

353 Sperlich, A., Dunnbier, U., Bockelmann, U., Hummelt, D., van Baar, P., Wode, F., Petersohn,
354 D., Grummt, T., Eckhardt, A., Schulz, W., Heermann, A., Reemtsma, T., Seiwert, B.,
AC

355 Schlittenbauer, L., Lesjean, B., Miehe, U., Remy, C., Stapf, M. and Mutz, D. (2013)
356 Anthropogenic organic micro-pollutants and pathogens in the urban water cycle:
357 assessment, barriers and risk communication (ASKURIS). Environmental Sciences Europe
358 25(1), 20.

359 Jelic, A., Gros, M., Ginebreda, A., Cespedes-Sanchez, R., Ventura, F., Petrovic, M. and
360 Barcelo, D. (2011) Occurrence, partition and removal of pharmaceuticals in sewage water
361 and sludge during wastewater treatment. Water Research 45(3), 1165-1176.

362 Landwehrkamp, L., Kouchaki-Shalmani, A., Forner, C., Hobby, R., Eduful, J. and Wagner, C.
363 (2015) Development of efficient characterization parameters for activated carbon used in
364 drinking water treatment. Journal of Water Supply: Research and Technology - Aqua 64(6),
365 688-696.

14
 ACCEPTED MANUSCRIPT
366 Margot, J., Kienle, C., Magnet, A., Weil, M., Rossi, L., de Alencastro, L.F., Abegglen, C.,
367 Thonney, D., Chèvre, N., Schärer, M. and Barry, D.A. (2013) Treatment of micropollutants in
368 municipal wastewater: Ozone or powdered activated carbon? Science of the Total
369 Environment 461-462(0), 480.

370 Moore, B.C., Cannon, F.S., Westrick, J.A., Metz, D.H., Shrive, C.A., DeMarco, J. and
371 Hartman, D.J. (2001) Changes in GAC pore structure during full-scale water treatment at
372 Cincinnati: a comparison between virgin and thermally reactivated GAC. Carbon 39(6), 789-
373 807.

PT
374 Oulton, R.L., Kohn, T. and Cwiertny, D.M. (2010) Pharmaceuticals and personal care
375 products in effluent matrices: A survey of transformation and removal during wastewater
376 treatment and implications for wastewater management. Journal of Environmental Monitoring
377 12(11), 1956-1978.

RI
378 Pota, A.A. and Mathews, A.P. (1999) Effects of particle stratification on fixed bed absorber
379 performance. Journal of Environmental Engineering-Asce 125(8), 705-711.

SC
380 Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonzalez, S., Barcelo, D., Ventura, F.
381 and Knepper, T.P. (2006) Polar pollutants entry into the water cycle by municipal
382 wastewater: A European perspective. Environmental Science & Technology 40(17), 5451-

U
383 5458.
AN
384 Ruhl, A.S. and Jekel, M. (2014) Degassing, gas retention and release in Fe(0) permeable
385 reactive barriers. Journal of Contaminant Hydrology 159, 11-19.

386 Ruhl, A.S., Zietzschmann, F., Altmann, J., Meinel, F., Sperlich, A. and Jekel, M. (submitted)
M

387 Stratification of granular activated carbon filters for advanced wastewater treatment.

388 Ruhl, A.S., Zietzschmann, F., Hilbrandt, I., Meinel, F., Altmann, J., Sperlich, A. and Jekel, M.
D

389 (2014) Targeted testing of activated carbons for advanced wastewater treatment Chemical
390 Engineering Journal 257, 184-190.
TE

391 Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten,
392 U. and Wehrli, B. (2006) The challenge of micropollutants in aquatic systems. Science
393 313(5790), 1072-1077.
EP

394 Shpirt, E.A. and Alben, K.T. (1986) Changes in Particle-Size Distributions on a Fixed-Bed of
395 Granular Activated Carbon. Water Science and Technology 18(1), 31-42.
C

396 Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J.,
397 Wert, E.C. and Yoon, Y. (2007) Role of membranes and activated carbon in the removal of
AC

398 endocrine disruptors and pharmaceuticals. Desalination 202(1-3), 156-181.

399 Summers, R.S., Haist, B., Koehler, J., Ritz, J., Zimmer, G. and Sontheimer, H. (1989) The
400 Influence of Background Organic-Matter on Gac Adsorption. Journal American Water Works
401 Association 81(5), 66-74.

402 Ternes, T. (2007) The occurrence of micopollutants in the aquatic environment: a new
403 challenge for water management. Water Science and Technology 55(12), 327-332.

404 Velten, S., Knappe, D.R.U., Traber, J., Kaiser, H.-P., von Gunten, U., Boller, M. and Meylan,
405 S. (2011) Characterization of natural organic matter adsorption in granular activated carbon
406 adsorbens. Water Research 45, 3951-3959.

15
 ACCEPTED MANUSCRIPT
407 Verlicchi, P., Al Aukidy, M. and Zambello, E. (2012) Occurrence of pharmaceutical
408 compounds in urban wastewater: Removal, mass load and environmental risk after a
409 secondary treatment-A review. Science of the Total Environment 429, 123-155.

410 Zietzschmann, F., Altmann, J., Ruhl, A.S., Dünnbier, U., Dommisch, I., Sperlich, A., Meinel,
411 F. and Jekel, M. (2014a) Estimating organic micro-pollutant reduction potential of activated
412 carbons using UV absorption and carbon characteristics. Water Research 56, 48-55.

413 Zietzschmann, F., Müller, J., Sperlich, A., Ruhl, A.S., Meinel, F., Altmann, J. and Jekel, M.
414 (2014b) Rapid small-scale column testing of granular activated carbon for organic micro-

PT
415 pollutant removal in treated domestic wastewater. Water Science & Technology 70(7), 1271-
416 1278.

417

RI
418

SC
419

U
AN
M
D
TE
C EP
AC

16
 ACCEPTED MANUSCRIPT
420 Figure 1: Experimental setups a) for backwashing of GAC with repeated expansions of the

421 GAC filter bed and b) for breakthrough adsorption tests for the comparison of fractions A and

422 E from the stratified filter bed.

423

PT
424 Figure 2: Normalized apparent densities (left) and water content in the filter bed or pore volume (right)

425 for stratified fractions of GAC #1 after different backwash intensities and number of backwash cycles.

RI
426

SC
427 Figure 3: Normalized apparent densities (left) and water content in the filter bed or pore volume (right)

428 of stratified fractions of GAC #2 after 10 backwash cycles with 100% filter bed expansions.

429
U
AN
430 Figure 4: Grain size compositions of GAC #1 in different strata a) without backwashing, b) after
M

431 100x20% expansion, c) after 10x100% expansion backwashes and d) after 100x100% expansion of

432 the filter bed.

D

433
TE

434 Figure 5: Grain size distribution for the five fractions of GAC #2 after 10 backwash cycles with 100%
EP

435 filter expansion.

436
C

437 Figure 6: Photographs of grains from a) fraction A and b) fraction E of GAC #2 after 10 backwashes
AC

438 with 100% filter bed expansion.

439

440 Figure 7: Relative UVA254 (left) and relative DOC concentrations (right) in dependence of throughputs

441 in liter of treated water per gram of fraction A and E of GAC #1 after 100 backwashes with 100%

442 expansions (a, b) and after 10x100% expansions of GAC #2 (c, d).

443
17
 ACCEPTED MANUSCRIPT
444 Figure 8: Size exclusion chromatograms of influent and two effluent samples for fraction A

445 (left) and fraction E (right) of GAC #1 (100x100%).

446

447 Figure 9: Relative effluent OMP concentrations of BTA (top row), CBZ (middle) and SMX (bottom)

PT
448 versus throughputs in liters of treated water per gram of GAC after 100x100% (large symbols) and

449 100x20% expansions (small symbols) for fraction A and E of GAC #1 (left) and after 10x100%

RI
450 expansions of GAC #2 (right).

451

SC
452 Figure 10: Calculated CUR10% for the indicative OMP for a) GAC #1 after 100x100% expansions and

U
453 b) GAC #2 after 10x100% expansions.
AN
454
M

455 Figure 11: OMP removal versus UVA254 reduction from filter effluents of fractions A (left) and E (right)

456 of GAC #1 after 100x100% (top row) and GAC #2 after 10x100% expansions.
D
TE
C EP
AC

18
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
AC
C EP
TE
D
M
AN
US
C RI
PT
AC
C EP
TE
D
M
AN
US
C RI
PT
 ACCEPTED MANUSCRIPT
Highlights

• Systematic variation of backwash intensities for granular activated carbon

• Significant vertical gradients of grain sizes, bulk densities and pore volumes
• Earlier breakthroughs of different parameters in material from lower strata
• Correlations between pollutant removal and UVA254 reduction

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Table S4: Masses of grain size fractions of stratified fractions.

1.4-2.0 mm [g] 1.0-1.4 mm [g] 0.63-1.0 mm [g] < 0.63 mm [g]

GAC #1, 10x100%
A 0.14 0.39 6.35 0.64
B 0.23 0.55 9.91 0.66
C 0.20 0.59 11.09 0.60
D 1.09 2.07 9.23 0.23
E 2.78 3.91 10.50 0.25

PT
GAC #1, 100x20%
A 0.30 0.59 6.55 0.52
B 0.23 0.74 9.90 0.51
C 0.49 1.41 10.99 0.26

RI
D 1.17 2.12 9.65 0.22
E 2.69 3.23 7.96 0.21
GAC #1, 100x100%

SC
A 0.12 0.33 5.97 0.59
B 0.15 0.46 9.93 0.65
C 0.16 0.51 11.35 0.56
D 0.56 1.62 10.48 0.17

U
E 2.92 4.17 8.25 0.14
GAC #1, 0x0%
AN
A 0.29 0.79 7.08 0.45
B 0.83 1.39 9.37 0.48
C 0.85 1.36 10.39 0.45
M

D 1.90 1.38 9.90 0.08

E 1.75 2.79 8.53 0.19
GAC #2, 10x100%
A 0.00 0.44 6.94 4.44
D

B 0.18 3.38 10.34 0.90

C 0.77 5.62 6.70 0.21
TE

D 1.90 8.40 4.70 0.08

E 5.61 8.95 2.35 0.03
C EP
AC
 ACCEPTED MANUSCRIPT

Supporting Information

Impacts of backwashing on granular activated

PT
carbon filters for advanced wastewater treatment

RI
SC
Joshua Frank1, Aki Sebastian Ruhl1,*, Martin Jekel1

U
AN
1
Chair of Water Quality Control, Technische Universität Berlin, Sekr. KF4, Straße des 17. Juni 135,

10623 Berlin, Germany.

M

*
Corresponding author, Tel.: +49 30 314 25493; fax: +49 30 314 23313. E-mail: aki.s.ruhl@tu-berlin.de
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Table S1: Analyzed compounds with abbreviations and fields of application.

Organic micropollutants Abbreviation Field of application

Acesulfame ACS Artificial sweetener

Benzotriazole BTA Corrosive and frost protection agent

Carbamazepine CBZ Anticonvulsant

PT
Diclofenac DCF Painkiller

RI
Formylaminoantipyrine FAA Painkiller

Gabapentin GAB Anticonvulsant

SC
Iomeprol IOM Radio-opaque substance

U
Methylbenzotriazole MBT Corrosive and frost protection agent
AN
Metoprolol MTP Beta blocker

Primidone PRI Anticonvulsant

M

Sulfamethoxazole SMX Antibiotic

D
TE
C EP
AC

Figure S1: Relationship between backwash velocity and percentaged filter bed expansion.
 ACCEPTED MANUSCRIPT

Table S2: Masses of the individual GAC fractions used in the long term experiments.

PT
Fraction A [g] Fraction E [g]

RI
GAC #1, 100x20% 7.0 12.2

GAC #1, 100x100% 6.9 12.5

SC
GAC #2, 10x100% 9.0 10.8

U
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
Table S3: Numerical values of all stratified fractions.

wet weight dry weight wet bed dry bed pore apparent
[g] [g] volume [mL] volume [mL] volume [%] density [g/L]
GAC #1, 10x20%
A+B 65.29 20.19 57 55 69 367
C 35.58 13.06 30 28 63 466
D 35.30 13.33 29 27 62 494
E 36.67 14.35 29 26/27 61 542

PT
GAC #1, 10x100%
A 27.17 7.53 25 24 72 314
B 33.43 11.36 29 27 66 421
C 34.35 12.50 29 28 64 446

RI
D 31.59 12.62 25 25 60 505
E 41.81 17.46 35 32 58 546
GAC #1, 100x20%

SC
A 29.97 7.92 28 27 73 293
B 34.02 11.37 29 29 67 392
C 36.03 13.14 30 29 64 453
D 33.11 13.16 28 27 60 487

U
E 36.27 14.19 30 28 61 507
GAC #1, 100x100%
AN
A 26.68 6.93 24 24 74 289
B 34.06 11.11 29 28 67 397
C 35.51 12.49 30 29 65 431
M

D 33.20 12.74 28 26 62 490

E 37.44 15.33 30 29/30 59 520
GAC #1, 0x0%
A 27.82 8.67 24 23 69 377
D

B 34.44 12.12 39 28 65 433

C 35.18 13.10 29/30 29 63 452
TE

D 32.95 12.20 27/28 27 63 452

E 34.97 13.20 27 28 62 472
GAC #2, 10x100%
A 35.05 11.10 30 29/30 68 376
EP

B 41.80 13.91 36 35 67 397

C 38.58 13.23 33 32 66 413
D 42.86 15.04 36 35/36 65 424
C

E 45.53 16.18 38 36 64 449

AC
 ACCEPTED MANUSCRIPT

90
80
Percentage [%] 70
60
50
GAC #1, 0
0x0%
40

PT
30 GAC #1, 1
10x100%

20 GAC #1, 1
100x100%
10

RI
0
1.4‐2.0 1.0‐1.4 0.63‐1.0 <0.63

SC
n [mm]
Fraction

Figure S2:
S Cumula
ated percenttage of fracctions before
e and after backwashin
ng

U
AN
M
D
TE
C EP
AC

Figure S3:
S Relative
e effluent UV
VA254 versu er per gram of GAC forr
us liters of trreated wate
GAC #1
1 (100x20%
%).
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure S4:
S Relative
e effluent UV
VA254 versu
us liters of trreated wate
er per gram of GAC forr GAC
#1 (10xx100%).

U
AN
M
D
TE
C EP
AC

Figure S5:
S Relative
e effluent DOC concen
ntrations verrsus liters of treated wa
ater per gra
am of
GAC for GAC #1 (1
100x20%).
 ACCEPTED MANUSCRIPT

PT
RI
SC
Figure S6:
S Relative
e effluent DOC concen
ntrations verrsus liters of treated wa
ater per gra
am of
GAC for GAC #1 (1
10x100%).

U
AN
M
D

SUVA
3
TE

2,5
SUVA [L/(mg∙m)]

2
EP

1,5

1
C

0,5
AC

Fraction E Fracttion A
0
0 50 100 150
Litters of water ttreated per grram of GAC [LL/g]

Figure S7:
S SUVA versus
v literss treated wa
ater per gram
m of GAC for
f GAC #1 (100x100%
%).
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

Figure S8:
S Relative
e effluent diclofenac, fo
ormylaminoantipyrine, primidone, metoprolol
EP

concenttrations for GAC #2 (10

0x100%).
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

Figure S9:
S Relative
e effluent diclofenac, fo
ormylaminoantipyrine, primidone, metoprolol
EP

concenttrations for GAC #1 (10

00x100 %).
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

Figure S10:
S Relativve effluent diclofenac,
d f
formylaminoantipyrine, primidone, metoprolo
ol
EP

concenttrations for GAC #1 (10

00x20%)
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
Figurre

U SC
AN
M
D
TE
C EP
AC

Figure S11:
S Relativve effluent acesulfame
a , benzotriaz
zole, carbam
mazepine, iomeprol,
methylb
benzotriazolle, sulfametthoxazole concentratio
c C #1 (100x20%)
ons for GAC
 ACCEPTED MANUSCRIPT

PT
RI
SC
,

U
AN
M
D
TE
EP

Figure S12:
S Relativve effluent diclofenac,
d f
formylaminoantipyrine, primidone, metoprolo
ol
concenttrations for GAC #1 (10
0x100%).
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP
AC

Figure S13:
S Relativve acesulfame, benzottriazole, carrbamazepine, iomeproll,
methylb
benzotriazolle, sulfametthoxazole concentratio
c C #1 (10x100%).
ons for GAC