Performance of a monolith biotrickling filter treating high concentrations of H2S from mimic biogas and elemental sulfur plugging control using pigging

Accepted Manuscript Performance of a monolith biotrickling filter treating high concentrations of H2S from mimic biogas and elemental sulfur plugging control using pigging Xintong Qiu, Marc A. Deshusses PII:

S0045-6535(17)31255-9

DOI:

10.1016/j.chemosphere.2017.08.032

Reference:

CHEM 19730

To appear in:

ECSN

Received Date: 13 June 2017 Revised Date:

4 August 2017

Accepted Date: 8 August 2017

Please cite this article as: Qiu, X., Deshusses, M.A., Performance of a monolith biotrickling filter treating high concentrations of H2S from mimic biogas and elemental sulfur plugging control using pigging, Chemosphere (2017), doi: 10.1016/j.chemosphere.2017.08.032. 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.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT 1

Performance of a Monolith Biotrickling Filter Treating High Concentrations of H2S from

2

Mimic Biogas and Elemental Sulfur Plugging Control using Pigging

3 Xintong Qiu and Marc A. Deshusses

5

Department of Civil and Environmental Engineering

6

127C Hudson Hall; Box 90287

7

Duke University

8

Durham, NC 27708-0287. USA

9

RI PT

4

10

*Corresponding author: [email protected]

11

(+1) 919-660-5480; (+1) 919-660-5219 FAX

13

SC

12

Keywords: biogas desulfurization; biotrickling filter; sulfur control; pigging

15 16

M AN U

14

Keywords: Biotrickling filter; Biofilter; Hydrogen sulfide; Biogas sweetening; Pigging;

18

Monolith biotrickling filter

TE D

17

19 20

Abstract

22

A novel biotrickling filter using a 3D-printed honeycomb-monolith as its filter bed has been

23

proposed and studied in this work and a solution to bed-clogging problems using pigging

24

was demonstrated. The inlet H2S concentration was controlled around 1000 ppmv,

25

corresponding to a loading rate of 127 g S–H2S m−3 h−1, and the empty bed gas residence time

26

(EBRT) was 41 s. The influence of different H2S/O2 ratios on the removal performance and

27

fate of sulfur end-products was investigated. The results indicated that at a H2S/O2 molar

AC C

EP

21

1

ACCEPTED MANUSCRIPT ratio of 1:2, an average removal efficiency of 95% and an elimination capacity of 122 g H2S

29

m-3 h-1 was obtained. Under all conditions investigated, elemental sulfur (rather than sulfate)

30

was the dominant end-product which mostly accumulated in the bed. However, the

31

monolith bed design reduced the risk of clogging by elemental sulfur while bed pigging was

32

shown to be an effective means to remove excess biomass and elemental sulfur accumulated

33

inside the bed and extend the life of the system indefinitely. Altogether, these findings could

34

lead to significant process improvement for biological sweetening of biogas or for removing

35

biomass in biotrickling filters at risk of plugging.

M AN U

SC

RI PT

28

36 37

1. Introduction

39

The biological conversion of organic wastes to biogas through anaerobic digestion is gaining

40

acceptance as an economical method for resources recovery at wastewater treatment plants,

41

animal farms, and other organic waste-generating facilities, since biogas is a renewable

42

energy that can be used as fuel for heating and electricity generation (López et al. 2015).

43

However, hydrogen sulfide (H2S) is inevitably formed during anaerobic digestion due to the

44

widespread presence of sulfate in the organic rich feedstock (Chaiprapat et al. 2015). H2S is

45

an odorous and corrosive gas that will cause detrimental effects on equipment including

46

pipelines, cogeneration engines, and microturbine units (Soreanu et al. 2008). The

47

combustion of H2S will also lead to the formation of SO2, which is a precursor of acid rain

48

(Chaiprapat et al. 2015). Therefore, the treatment of biogas to remove H2S is required before

AC C

EP

TE D

38

2

ACCEPTED MANUSCRIPT the final utilization of biogas. Among all the technologies used for biogas desulfurization,

50

chemical scrubbing is considered as one of the most common and effective technologies for

51

its high removal efficiency and short gas contact time (Mannucci et al. 2012). Nonetheless,

52

this method also has significant drawbacks including high operating costs, utilization of

53

hazardous chemicals, and large consumption of reagents and energy (Gabriel et al. 2003,

54

Charron et al. 2004). In this context, the application of biotechnological alternatives to

55

chemical scrubbing has received much attention in recent years since biological sweetening

56

does not require hazardous chemicals and is environmentally friendly (Wu et al. 2001,

57

Gabriel et al. 2003, Soreanu et al. 2008, Fortuny et al. 2011, Mannucci et al. 2012). Among the

58

biotechnological methods, biotrickling filters (BTFs) are emerging as the most promising

59

biological alternative to conventional methods (Montebello et al. 2014), for BTFs have been

60

shown to be as effective as chemical scrubbing in selected studies (Fortuny et al. 2008;

61

Montebello et al. 2014).

SC

M AN U

TE D

62

RI PT

49

In previous studies, BTFs have successfully controlled H2S in biogas with different

64

concentration levels and achieved removal up to 100% H2S, sometimes for gas streams with

65

thousands of ppm H2S from biogas (Wu et al. 2001, Soreanu et al. 2008, Mannucci et al.

66

2012). Microorganisms from the following genera: Thiobacillus, Hyphomicrobium, Xanthomonas,

67

and Methylophaga have been found in such BTFs operated at acidic or neutral pH conditions

68

(Chung et al. 1998, Ruokojärvi et al. 2001). Generally, mixed cultures of sulfide oxidizing

69

bacteria (SOB) are utilized for the biodesulfurization process which can be mainly described

AC C

EP

63

3

ACCEPTED MANUSCRIPT 70

by following Eqs. (1), (2), (3), and (4).

71
  →
    1

RI PT


   →
+
    2 1
  +  →   + 
 3 2


  + 2 →  +
4

SC

72

H2S is initially transferred from the gas phase into the aqueous phase or the damp biofilm by

74

absorption where it can dissociate. It is then biologically oxidized to elemental sulfur (S0),

75

sulfate (SO42-), and other sulfur species like thiosulfate and sulfite depending on the oxygen

76

availability in the biofilm or aqueous phase (Fortuny et al. 2010).

M AN U

73

TE D

77

However, even with good performance in treating H2S in biogas, biological approaches have

79

not been widely applied in industry yet, mainly because some technical hurdles still need to

80

be resolved (López et al. 2015). One of the biggest obstacles is bed clogging problems caused

81

by the accumulation of elemental sulfur particles in the system (López et al. 2015). The

82

problem is particularly acute when H2S concentrations are high, H2S biodegradation rates

83

are fast, and for BTFs using high surface area packings, which are generally required to

84

achieve high performance. A wide variety of packing materials have been used including

85

open-pore polyurethane foam (PUF) (Gabriel et al. 2003, Ramírez et al. 2009), polyester

86

fibers (Soreanu et al. 2008), pall rings (Wu et al. 2001), porous lava (Namini et al. 2008),

AC C

EP

78

4

ACCEPTED MANUSCRIPT coconut husk (Chaiprapat et al. 2015) and activated carbon (Duan et al. 2005). A common

88

observation in BTFs used for biogas sweetening is that sulfur deposition (sometimes

89

combined with excess biomass growth) becomes the main limitation of high rate treatment

90

(Mannucci et al. 2012). This accumulation worsens the issue of O2 availability for the inner

91

layers of the biofilm, which in turn increases the proportion of elemental sulfur produced as

92

the end-product. Ultimately, increased pressure drop in the reactor and reduced treatment

93

leads to process failure (Andreasen et al. 2012, Mannucci et al. 2012, Montebello et al. 2014).

SC

RI PT

87

M AN U

94

Despite the importance of this problem, only a few studies have investigated strategies to

96

reduce the accumulation of elemental sulfur in biotrickling filters and so far, none was

97

proven to address the problem over the long term. Unfortunately, sulfur is insoluble and can

98

not be easily washed; it is often intracellular and thus sticks to the packing with the biofilm,

99

subject to the packing design. One proposed approach was to oxidize S0 aerobically during

TE D

95

sulfide (gas) starvation periods (Fortuny et al. 2010). Indeed, the accumulated S0 can be

101

utilized as the substrate for SOB and be oxidized to SO42- following the stoichiometry of Eq.

102

(5). Therefore, bed clearing would be achieved through the biological consumption of the

103

solid sulfur particles. However, the process was found to be slow (and thus would require

104

large BTFs), and incomplete sulfur removal was observed when the oxygen supply was

105

insufficient (Fortuny et al. 2010).

AC C

EP

100

3   +  +
  →   +
5 2 106 5

ACCEPTED MANUSCRIPT A possible approach to avoid clogging by elemental sulfur is to optimize the filter bed.

108

Fortuny et al. (2008) suggested that a packing with a firm, open and regular structure could

109

reduce the chance of compaction and facilitate sulfur removal. Still, to date, viable solutions

110

are lacking. In this study, a 3D-printed honeycomb monolith packing was proposed as a

111

possible replacement for traditional packing materials used in BTFs. The connected and

112

straight channels in this honeycomb monolith made it possible to use the process of pigging

113

to remove excess biomass and accumulated sulfur particles thereby greatly reducing the risk

114

of bed clogging. Pigging is a simple operation typically used in pipelines and tubular

115

reactors for cleaning and for regular maintenance, including debris removal, gauging,

116

separation of products, wall cleaning, etc. During this operation, a device called a “pig” is

117

passed through the tubular vessel or pipe, generally driven by the pipeline fluid (Tiratsoo

118

1992) or by some mechanical force. Cleaning is achieved by the small clearance between the

119

pig and the pipe walls, brushes or other cleaning mechanisms on the pig.

TE D

M AN U

SC

RI PT

107

120

In this study, a laboratory-scale honeycomb monolith BTF reactor was operated for more

122

than one year to investigate the overall performance of this novel bioreactor in treating high

123

concentration H2S (1000 ppmV). The influence of the molar ratio H2S/O2 on the H2S removal

124

efficiency (RE), elimination capacity (EC), and the speciation of different sulfur

125

end-products was investigated first. Then, pigging was conducted to demonstrate the ability

126

of using this process to remove excess biomass and accumulated sulfur particles from the

127

bed of the BTF.

AC C

EP

121

6

ACCEPTED MANUSCRIPT 128

2. Materials and Methods

130

2.1 Experimental Installation and Operating Conditions

131

The laboratory-scale setup (Fig. 1) was comprised of an upflow counter-current BTF made of

132

a 3D-printed mesh honeycomb monolith as the filter bed (220 mm in height). The design

133

was made using SketchUp (Fig. 2) and printed at Duke University’s Co-Lab using Hatchbox

134

polylactic acid (PLA) filament on a Ultimaker 2 3D printer. The outside walls of the reactor

135

(6 cm OD) were painted with an epoxy resin to ensure the reactor was gas tight. The inside

136

of the monolith column comprised of 19 hexagonal channels equally distributed, each 7.8

137

mm wide (sides of 4.3 mm). The walls between channels were designed as meshes (see Fig.

138

2) with 1 mm thick strands while square hole in the mesh were 1×1 mm each and arranged

139

in a diamond pattern with 0.4 mm separation between the holes. This resulted in a total an

140

empty channel volume of 204 cm3, a bed porosity of 88% and a specific surface area of about

141

620 m2 m-3. The mesh design for the walls was conceived to provide more surface area for

142

biofilm growth and cavities for attachment. It also provided connections between the

143

channels, a feature discussed with the pressure drop investigations. The reactor was

144

inoculated with activated sludge from a local wastewater treatment plant, which led to

145

mixed SOB cultures to immobilize on the inside surfaces of the monolith as they removed

146

H2S from the biogas mimic. The system was operated continuously for about 220 days

147

(reaching a pseudo-steady state) before all the experiments in this study were conducted.

148

These lasted about 200 days.

AC C

EP

TE D

M AN U

SC

RI PT

129

7

ACCEPTED MANUSCRIPT 149 The synthetic sour biogas mimic consisted of metered streams of N2, and H2S in controlled

151

amounts to which of a minute stream of air (as needed per H2S:O2 ratio imposed) was

152

added. Using a mimic biogas that does not contain CH4 or CO2 was shown to be acceptable

153

for biogas sweetening studies before by other investigators, as H2S degraders are autotrophs

154

and are not affected by the presence or absence of CH4. The inlet H2S concentration was

155

controlled to around 1000 ± 150 ppmv, corresponding to a loading of 127 ± 18 g S–H2S m−3 h−1.

156

During the entire experimental period, the synthetic biogas was humidified by bubbling in

157

DI water and then fed to the bottom of the BTF at a constant flow rate of 0.30 L min-1, which

158

was also the gas flow rate of N2 since the flows of air and H2S were small enough that they

159

could be neglected. The N2 and air flows were regulated with Alicat mass flow controllers

160

(Alicat Scientific, Tucson, AZ), while H2S gas was metered from a Tedlar bag using a

161

peristaltic pump. The empty bed residence time (EBRT) was 41 s, which was close to the

162

range of 10-40 s based on the reference EBRT from previous studies (Namini et al. 2008,

163

Mannucci et al. 2012). During normal operation, the liquid volume in the accumulation tank

164

below the column was usually controlled to 0.4 L, and the liquid was recycled to the top of

165

the monolith column and then sprayed on the top of the column at a recirculation rate of 0.2

166

L min-1, using a peristaltic pump. This flow corresponds to a trickling liquid velocity of 13 m

167

h-1 which is a high value, but was necessary to ensure uniform distribution between the

168

channels. The nutrients required by for the growth of SOB were provided by a mineral

169

medium (in grams per liter) KH2PO4 0.5 g, MgCl2 0.1 g, CaCl2 0.05 g, NaHCO3 1 g, FeSO4 0.01

AC C

EP

TE D

M AN U

SC

RI PT

150

8

ACCEPTED MANUSCRIPT NH4Cl 1 g, and pH was adjusted to approximately 6.8. Fresh mineral medium was

171

replenished at the same flow rate of 0.2 L d-1 as the liquid that was withdrawn from the

172

accumulation tank. The liquid withdrawn was collected in a bottle stored in a refrigerator (to

173

stop any biological transformation), and together with the liquid in accumulation tank, was

174

sampled for sulfur species measurement. Throughout the experimental period, the system

175

was operated at room temperature (20-25 °C).

RI PT

170

SC

176

(figs 1 and 2 around here – revision note, Figure 2 is new)

178

M AN U

177

2.2 Analytical Methods

180

2.2.1 Gas Sampling and H2S Measurement

181

Gas samples were collected using 1 ml syringes from the inlet/outlet sampling ports and

182

then injected into 1-liter Tedlar bags for dilution. The H2S concentration was determined

183

using a portable Jerome 631-X Hydrogen Sulfide Analyzer (Arizona Instruments, Chandler,

184

AZ) which has a range of 0.003-50 ppmv of H2S. Error propagation on the 1 mL sampling,

185

1000 fold dilution and analysis revealed that the precision on H2S concentration was ±6%.

186

The performance of the BTF was reported in terms of H2S removal efficiency (RE),

187

elimination capacity (EC), as a function of the loading rate (LR) (see Eq. (6) and (7) for EC

188

and LR).

AC C

EP

TE D

179

 =  −   × /" 6 $% =  × /" 7 9

ACCEPTED MANUSCRIPT 189 Where the Cin and Cout are the inlet and outlet H2S concentrations in the gas flows (in g m-3),

191

Q is the gas flow entering the BTF (m-3 h-1), V is the empty bed volume (m3), and thus the

192

unit of LR and EC are gH2S m-3bed h-1.

RI PT

190

193

2.2.2 Sulfur Species Measurement and Sulfur Balances

195

To investigate the impact of the inlet H2S/O2 ratio on the fate of sulfur and on the production

196

of different sulfur end-products, sulfur species measurements in the liquid phase of both the

197

liquid recycle vessel and the liquid collection bottle were conducted at three different H2S/O2

198

condition of ratio 2:1, 1:1, and 1:2, the extreme values corresponding to the stoichiometry of

199

Eq. 3 and 4, respectively. No ratio smaller than 1:2 was attempted to avoid excessive dilution

200

of biogas with air. The sulfur species included sulfate (SO42-), sulfide (S2-), elemental sulfur

201

(S0), sulfite (SO32-), and thiosulfate (S2O32-) and were measured to establish a sulfur mass

202

balance over a period of time (generally 3 to 5 days). The sulfur balance was established by

203

tallying sulfur species produced (from the volume × concentration in the liquid collection

204

bottle + volume × concentration increase of the liquid recycle vessel) and comparing with the

205

S-H2S degraded during that time period. In some experiments, the weight of the monolith

206

bioreactor was measured in an attempt to quantify the accumulation of biomass and

207

elemental sulfur.

AC C

EP

TE D

M AN U

SC

194

208 209

The sulfate concentration was analyzed with the USEPA SulfaVer 4 method, which works for 10

ACCEPTED MANUSCRIPT a range of 2 to 70 mg L-1. The sulfide formed was measured spectrophotometrically using the

211

USEPA Methylene Blue method. To measure the elemental sulfur in the liquid phase, a

212

modification of the colorimetric determination method proposed by Bartlett et al. (1954) was

213

used. Chloroform (rather than petroleum ether) was used to extract elemental sulfur prior to

214

quantitative determination using the Bartlett et al. method. The measurements of both

215

sulfite and thiosulfate concentration were conducted by using test kits from Chemetrics

216

(Titrets® Titration Cells K-9602 and K-9705).

SC

RI PT

210

M AN U

217

2.3 Pigging Process

219

After about 160 days of operation, pigging of the hexagonal channels of the BTF was

220

investigated. The BTF was first taken off-line and end caps removed. Pigging was conducted

221

using a steel hexagonal bar with 4.8 mm wide (2.7 mm sides). Recall the channel width is 7.8

222

mm; thus the pig should allow a film of 1-1.5 mm of material to remain in the channels. To

223

physically remove the extra biomass and sulfur particles clinging on the inside surface of the

224

monolith column, the hexagonal pig bar was pushed twice through each channel, one

225

channel at the time, and the material pushed out was collected for analysis.

EP

AC C

226

TE D

218

227

The pressure drop through the BTF at different gas flow rates were measured by a U-shaped

228

water gauge allowing pressure drop resolution of about 0.5 mm of water column (5 Pa).

229

These differential pressure measurements were conducted before and after the pigging

230

process to evaluate the effectiveness of the pigging on clearing the channels. After the 11

ACCEPTED MANUSCRIPT pigging process, the materials removed were collected for further analysis. The dry/wet

232

mass ratio was obtained by first weighing known amounts of the wet materials removed by

233

pigging, drying the samples in a vacuum oven at 90 °C and reweighing them when totally

234

dried. To determine the amount of biomass and elemental sulfur in these materials, they

235

were first sonicated to lyse the cells and the biomass fraction was determined using the

236

concentration of the nitrogen after total Kjeldahl nitrogen analysis (TKN kit TNT828, Hach)

237

and assuming a typical biomass composition (C5H7NO2) (Montebello et al. 2014). The

238

fraction of the elemental sulfur was determined using the same colorimetric method

239

mentioned above for the elemental sulfur measurement in the liquid phase.

M AN U

SC

RI PT

231

240

3. Results and Discussion

242

3.1 Performance of the Monolith BTF and effects of H2S/O2 ratio

243

The influence of different H2S/O2 ratios on H2S removal is shown in Fig. 3. As can be seen,

244

the experiment was conducted in two rounds replicating each conditions to ensure validity,

245

with each round providing similar performance for the specific H2S/O2 supplied ratio. The

246

REs of H2S ranged from 42% to 100%, with the highest RE achieved at H2S/O2 ratio 1:2 and

247

lowest RE obtained at H2S/O2 ratio 2:1. The lower RE at ratio 2:1 can be explained by the low

248

O2 availability in the biofilm, when the dissolved oxygen was likely limiting the rate of H2S

249

removal from the biogas. At these conditions, formation of elemental sulfur is preferred (see

250

Eq. 3) and since that biotransformation is less favorable for the microorganisms (less

251

negative ΔG°), lower degradation rates are observed. The individual measurements of ECs

AC C

EP

TE D

241

12

ACCEPTED MANUSCRIPT 252

ranged from 52.9 to 134 g H2S m-3 h-1 obtained at H2S/O2 ratios of 2:1 and 1:2, respectively.

253 254

(Fig 3 around here)

RI PT

255 These results are summarized in Table 1. Here, the average EC value of 122 g H2S m-3 h-1 at a

257

H2S/O2 ratio of 2:1 is in the upper range of ECs (110–140 g H2S m-3 h-1) of previous studies

258

(see e.g., Fortuny et al. 2008). According to Fortuny et al. (2008), 500 ppmv is the highest H2S

259

concentration that can be withstood by internal combustion engines without causing

260

corrosion problems. Based on this, operation at a H2S/O2 ratio of 2:1 would just meet that

261

criterion at an inlet concentration of 1000 ppmv and 50% removal. Performance was better

262

but fairly variable at a ratio of 1:1. Thus, from the conditions tested, the H2S/O2 ratio of 2:1

263

was the best, with stable performance and both high RE and high EC.

264

266

(Table 1 around here)

EP

265

TE D

M AN U

SC

256

A brief comparison with previously published papers reveals that 90-100% RE was obtained

268

when the H2S/O2 ratio was 1:49 and the EBRT was set to 130 s (Montebello et al. (2014)),

269

while a longer EBRT of 313 s was required to get 80-90 % RE of around 2000 ppm H2S in the

270

study of Chaiprapat et al. (2011). Higher ECs (up to 250-280 g g m-3 h-1) were reported by

271

Fortuny et al. (2008), but this was for H2S concentrations 3 to 10 times higher, and at EBRTs

272

of about 3 minutes. Here, up to 100% RE of 1000 ppmv was achived at an EBRT of 41 s at a

AC C

267

13

ACCEPTED MANUSCRIPT 273

H2S/O2 ratio 1:2 by our monolith BTF, and despite the relatively low EBRT and low O2

274

supply, the process was stable over time and robust.

275 During the entire operating period, the pressure drop in the system remained below 1 cm of

277

water column. This is a low value, but is consistent with the relatively long gas residence

278

time. Even at the end of the experiment, no signs of major clogging could be observed. This

279

contrasts with the study of Fortuny et al. (2008) who observed a pressure drop of 10 cm of

280

water column and found that elemental sulfur particles gradually accumulated inside the

281

PUF packing material leading to complete clogging of the BTF system. Thus the application

282

of a honeycomb monolith as the packing bed in this study shows a large advantage by

283

greatly reducing the risk of bed-clogging when treating high concentrations H2S. As will be

284

discussed later, the straight channel configuration of the monolithic bed further allows for

285

pigging to remove eventual blockages of the channels making it an attractive bed

286

configuration for sustained operation.

SC

M AN U

TE D

EP

287

RI PT

276

3.2 Sulfur Species Measurement in Liquid Phase

289

Sulfur species, including sulfate (SO42-), sulfide (S2-), elemental sulfur (S0), sulfite (SO32-), and

290

thiosulfate (S2O32-), were measured in the liquid phase to determine the fate of sulfur in the

291

system. The fractions of each sulfur species produced at different H2S/O2 supplied ratios are

292

shown in Fig. 4. Sulfate and elemental sulfur were the two predominant sulfur species in the

293

liquid phase; the sum of these two species accounted for up to 93% of all end-products

AC C

288

14

ACCEPTED MANUSCRIPT measured, while the sulfide was less than 1% in all three H2S/O2 supplied conditions. Sulfite

295

and thiosulfate as intermediate sulfur species were also detected in all three H2S/O2

296

conditions and made up to 3-13% of the total end-products recovered in the liquid. A study

297

by Van den Bosch et al. (2008) indicated that the formation of thiosulfate can be influenced

298

by the pH, where a higher pH could lead to an increase of the selectivity of thiosulfate

299

formation. This was later confirmed by two studies which showed minimal thiosulfate

300

production when treating H2S at acidic condition (Montebello et al. 2014, Chaiprapat et al.

301

2015). Thus, the detection of thiosulfate in our study could be explained by the pH setting

302

(kept around 7 because of buffer in the liquid) in the system and the low oxygen availability

303

in the aqueous phase. Fig. 4 also shows the trend of increasing sulfate fraction and

304

decreasing elemental sulfur fraction in the liquid phase when the supplied H2S/O2 decreases

305

from 2 to 0.5, which is consistent with the reaction mechanisms that higher oxygen levels

306

favor the formation of sulfuric acid/sulfate.

307

309

(Fig 4 around here)

AC C

308

EP

TE D

M AN U

SC

RI PT

294

310

Closure of the sulfur balance was attempted (see methods for details) initially using only the

311

measurement on different sulfur species in the liquid phase. A low recovery of sulfur species

312

in the liquid phase was observed with only around 3% of the total S-H2S removed from the

313

gas phase being recovered in the liquid. One possible explanation is that most of the H2S

314

oxidized ended immobilized as elemental sulfur, and accumulated inside the channels of the 15

ACCEPTED MANUSCRIPT filter bed. To verify this hypothesis, the increase in weight of the BTF was determined over

316

time and balances were calculated assuming the added weight had a sulfur content of 62.1%

317

(as determined in Section 3.3.3). The results shown in Table 2 reveal that while excellent

318

closure of the S balance was achieved at a ratio of 1:1, either excess or deficit of S was

319

observed for the other two conditions. It should be noted that these measurements include a

320

fairly large experimental uncertainty. Still, they are consistent with the liquid measurements

321

findings that most of the H2S was converted to elemental sulfur which accumulated in the

322

system and that the trend of sulfur production was inverse proportional to the oxygen

323

supply.

M AN U

SC

RI PT

315

324 325

(Table 2 around here)

TE D

326

3.3 Performance of Pigging Process in Cleaning Up Filter Bed

328

3.3.1 Pigging Process Observations

329

As described in the methods section, a 4.8 mm wide hexagon steel bar was used at the end of

330

the BTF operation to demonstrate the feasibility of using pigging for removing excess

331

biomass and elemental sulfur accumulated inside the system. The pictures of the filter bed

332

before the initial startup, on day 165 before and after the pigging process are presented in

333

Fig. S1. Observing the top and bottom sections of the column at day 165 before the pigging

334

process, the blue surface of the honeycomb monolith was fully covered with yellowish solids,

335

which was assumed to consist mainly of elemental sulfur and biomass. When looking

AC C

EP

327

16

ACCEPTED MANUSCRIPT through the honeycomb channels inside the monolith, some solids could also be observed

337

partly obstructing the view through the channels. Comparing the top and bottom sections of

338

the filter bed, the bottom section had more accumulated sulfur than the top section

339

indicating that either the desulfurizing process mainly happened at the bottom section of the

340

filter bed, close to the H2S inlet port, and/or that deposits were carried downward in the bed

341

and accumulated at the bottom. The former explanation is more likely and is consistent with

342

previous studies where the highest sulfur removal rate took place near the gas inlet port of

343

the reactor, (Fortuny et al. 2008, Montebello et al. 2014) although accumulation of elemental

344

sulfur washed from the upper column is also very plausible.

M AN U

SC

RI PT

336

345

During the pigging process, all 19 channels were pigged with the hexagonal bar and total 7.8

347

g of wet deposits was collected from this process. The weights of deposits from each channel

348

are presented in Fig. S2. The results showed that the accumulated elemental sulfur and

349

biomass were not evenly distributed and removed from the monolith column. The biggest

350

difference between the weight of material removed from two channels was 0.83 g or a factor

351

of about 5. One possible reason for this result is the unevenly distribution of the recycle

352

liquid and thus uneven wetting of the bed. Ensuring equal distribution of liquid to every

353

channel is very difficult at the small scale, but for large scale systems, industrial spray

354

nozzles exist that can easily provide uniform wetting.

AC C

EP

TE D

346

355 356

After the pigging process, the channels inside the monolith column appeared much cleaner 17

ACCEPTED MANUSCRIPT than before. When operation of the BTF was resumed and the liquid recirculation pump was

358

turned on again, more solids was observed to detach from the inside surfaces of the channels.

359

This is probably because pigging loosened the materials possibly pushing it through the

360

open mesh that made up the wall of the monolith reactor. The amount of wet solid thus

361

collected from the bottom cap of the reactor was 16.52 g. Its aspect (a yellow-gray slurry)

362

was very similar to material removed during pigging albeit slightly more liquid. Thus,

363

throughout the entire pigging and flushing process, total of 24.32 g of wet solids were

364

obtained. Recall that the total empty bed volume is 204 mL, and thus the material removed

365

constituted a significant fraction (>10%) of the entire bed volume. The dry/wet ratio of these

366

deposits was found to be 0.47, and thus this corresponded to 11.4 g of dry matter. The mixed

367

deposits were then subjected to further analysis to determine their biomass and elemental

368

sulfur contents (results discussed in Section 3.3.3).

SC

M AN U

TE D

369

RI PT

357

3.3.2 Pressure Drops Before and After Pigging Process

371

During the entire experiment, the pressure drop of the biotrickling filter was always lower

372

than 1 cm of water column, indicating that there was no significant clogging of the reactor.

373

As mentioned earlier, this is an improvement over earlier studies which typically report

374

increasing pressure drops, sometimes to values greater than 10 cm of water column, leading

375

to the eventual shutdown of the system (Fortuny et al. 2008). The relationship between the

376

pressure drop of the filter column and gas flow rate was measured before and after the

377

pigging process hoping to illustrate the effect of pigging. Increasing the gas flow rate was to

AC C

EP

370

18

ACCEPTED MANUSCRIPT obtain larger pressure drops since very low values were obtained at the nominal gas flow

379

due to the long gas residence time. The results presented in Fig. S3 show that there were no

380

large differences between the pressure drop before and after pigging. It is only for a few

381

points that the pressure drop slightly decreased after the pigging process but differences are

382

not statistically significant. This observation is largely attributed to the straight channels of

383

the monolith bioreactor which offer very low resistance to gas flow, but could also be

384

accentuated by the fact that the walls of the monolith were made from a mesh, thereby

385

allowing connection between the channels. The low pressure drop shows that even after 165

386

days of operation, the BTF was not clogged. Further, applying pigging in this monolith BTF

387

was proven effective in removing the accumulated biomass and elemental sulfur allowing to

388

extend stable operation indefinitely.

TE D

389

M AN U

SC

RI PT

378

3.3.3 Pigging Deposits Analysis

391

The analysis of the collected solids from the pigging process revealed a high content (62.13%

392

of dry mass) of elemental sulfur, while the TKN analysis revealed a low content of 7.03% of

393

(dry) biomass in the deposits. The remainder of the dry mass is unknown but could be

394

inorganic material such as CaSO4 which is not very soluble. Trace metal ions like iron and

395

sodium as well as other sulfur species were detected in one previous study (Montebello et al.

396

2014). The results are in-line with the low recovery rate of sulfur in the liquid phase and also

397

support the previous hypothesis that the inlet H2S was mostly oxidized to elemental sulfur

398

that accumulated inside the system. The low biomass content indicates a low growth rate

AC C

EP

390

19

ACCEPTED MANUSCRIPT and biomass yield for the SOB in the system, which was discussed by others (Mannucci et al.

400

2012, Montebello et al. 2014). The high percentage of elemental sulfur in the pigging deposits

401

is favored in this system, since the generation of elemental sulfur does not change the pH in

402

the trickling liquid (see Eqs. 2 and 3), while the process of producing sulfate will reduce the

403

pH, thereby adding costs for pH control. Moreover, elemental sulfur is a valuable raw

404

material, and the high sulfur-content deposits could be further purified and used for

405

commercial purposes, or used as is as a fertilizer.

SC

RI PT

399

M AN U

406

3.3.4 System Recovery After Pigging Process

408

The performance of the system, including RE and EC was recorded (at a H2S/O2 ratio of 1:2)

409

prior to and after pigging and the results are shown in Fig. 5. After pigging, the system went

410

through gradual declines of both RE (from 100 to 13%) and EC (from 132 to 15.7 g H2S m-3h-1)

411

with the lowest RE and EC observed on the third day (at 69 hours) after the pigging process.

412

A recovery trend was then observed starting on the fourth day, an even though a second

413

drop in performance was observed during a short maintenance period following a crack in

414

the reactor body, eventually the RE reached 49% and the EC recovered to about 63 g H2S

415

m-3h-1. These drastic changes in performance post-pigging indicate that the pigging process

416

removed most of the active SOB, a claim supported by the observation of solids detachment

417

by the recycle liquid mentioned before. With the re-growth of SOB in the reactor, the RE and

418

EC was found to slowly increase.

AC C

EP

TE D

407

419 20

ACCEPTED MANUSCRIPT 420

(Fig. 5 around here)

421 However, the recovery of the system was lower and slower than expected. Thus, there is no

423

doubt that the pigging process did have a negative effect on the H2S removal performance of

424

the system. Since all 19 channels were pigged at once, reducing the numbers of channels

425

being pigged at one time is likely a good method to mitigate the negative effects of pigging

426

on the performance of the system. Possibly, in a large biotrickling filter system, a robot could

427

conduct pigging of a few channels at regular intervals thereby minimizing the impacts on

428

the overall performance of the system. The size of the pig and the control of its position

429

inside the channel during pigging are expected to be two critical factors in the application of

430

pigging of BTFs.

M AN U

SC

RI PT

422

TE D

431

4. Conclusion

433

Overall, the results presented and discussed herein indicate that the monolith biotrickling

434

filter is a promising alternative for replacing traditional packing materials in applications

435

with significant risks of plugging by either biomass growth or solids deposits, such as

436

treating high concentration H2S from biogas. Our monolith BTF reactor was stably operated

437

for more than five months and was able to effectively treat the synthetic biogas

438

contaminated with 1000 ppmv H2S. ECs exceeding 120 g H2S m-3 h-1 EC with near complete

439

removal at H2S/O2 supplied ratios of 1:2 were observed. The dominant product, elemental

440

sulfur, was primarily accumulated inside the filter bed. While such accumulation is

AC C

EP

432

21

ACCEPTED MANUSCRIPT generally a cause for system upset, the monolith design of the filter bed greatly reduced the

442

risk of the bed-clogging problems when compared with other packing materials. Further,

443

pigging was shown to efficiently remove deposited sulfur and biomass thereby indefinitely

444

extending the life of such biotrickling filters. This constitute the first demonstration of

445

pigging in biotrickling filters; possibly the method can be applied for other biotrickling filter

446

applications where significant biomass or plugging is observed. The high sulfur-content

447

deposits pigged from the reactor could possibly be reused as fertilizers or to produce raw

448

elemental sulfur. Further studies could address the optimization of the pigging process,

449

including the optimization of the pig geometry and the mode or frequency of pigging, or

450

further application at larger scale.

451

M AN U

SC

RI PT

441

Acknowledgments

453

The authors would like to thank Michelle Moffa (Deshusses lab) for designing and

454

3D-printing the biotrickling filter with the help of Chip Bobbert (Duke’s Innovation Co-Lab)

455

and Nazia Khatoon (Deshusses lab) for initial experiments with the biotrickling filter. .

EP

TE D

452

AC C

456 457

References

458

Andreasen R. R., Nicolai R. E. and Poulsen T. G. 2012. Pressure drop in biofilters as related

459

to dust and biomass accumulation. Journal of Chemical Technology & Biotechnology

460

87(6): 806-816.

461 462 463 464

Bartlett J. K. and Skoog D. A. 1954. Colorimetric Determination of Elemental Sulfur in Hydrocarbons. Analytical Chemistry 26(6): 1008-1011. Chaiprapat S., Charnnok B., Kantachote D. and Sung S. 2015. Bio-desulfurization of biogas using acidic biotrickling filter with dissolved oxygen in step feed recirculation. 22

ACCEPTED MANUSCRIPT 465 466

Bioresource Technology 179: 429-435. Chaiprapat S., Mardthing R., Kantachote D. and Karnchanawong S. 2011. Removal of

467

hydrogen sulfide by complete aerobic oxidation in acidic biofiltration. Process

468

Biochemistry 46(1): 344-352. Charron I., Fëliers C., Couvert A., Laplanche A., Patria L. and Requieme B. 2004. Use of

470

hydrogen peroxide in scrubbing towers for odor removal in wastewater treatment

471

plants. Water Science and Technology 50(4): 267-274.

472

RI PT

469

Chung Y.-C., Huang C., Pan J. R. and Tseng C.-P. 1998. Comparison of autotrophic and

473

mixotrophic biofilters for H2S removal. Journal of Environmental Engineering 124(4):

474

362-367.

475

Duan H., Yan R. and Koe L. C. C. 2005. Investigation on the mechanism of H2S removal by biological activated carbon in a horizontal biotrickling filter. Applied microbiology and

477

biotechnology 69(3): 350-357.

478

SC

476

Fortuny M., Baeza J. A., Gamisans X., Casas C., Lafuente J., Deshusses M. A. and Gabriel D. 2008. Biological sweetening of energy gases mimics in biotrickling filters. Chemosphere

480

71(1): 10-17.

M AN U

479 481

Fortuny M., Gamisans X., Deshusses M. A., Lafuente J., Casas C. and Gabriel D. 2011.

482

Operational aspects of the desulfurization process of energy gases mimics in

483

biotrickling filters. water research 45(17): 5665-5674.

Fortuny M., Guisasola A., Casas C., Gamisans X., Lafuente J. and Gabriel D. 2010. Oxidation

485

of biologically produced elemental sulfur under neutrophilic conditions. Journal of

486

chemical technology and biotechnology 85(3): 378-386.

488 489

Gabriel D. and Deshusses M. A. 2003. Performance of a full‐scale biotrickling filter treating H2S at a gas contact time of 1.6 to 2.2 seconds. Environmental progress 22(2): 111-118. López L., Dorado Castaño A. D., Mora M., Prades Martell L., Gamisans Noguera J., Lafuente

EP

487

TE D

484

J. and Gabriel D. 2015. Modelling biotrickling filters to minimize elemental sulfur

491

accumulation during biogas desulfurization under aerobic conditions. EMChIE 2015

492

Conference Proceedings: 325-330.

493

AC C

490

Mannucci A., Munz G., Mori G. and Lubello C. 2012. Biomass accumulation modelling in a

494

highly loaded biotrickling filter for hydrogen sulphide removal. Chemosphere 88(6):

495

712-717.

496

Montebello A. M., Mora M., López L. R., Bezerra T., Gamisans X., Lafuente J., Baeza M. and

497

Gabriel D. 2014. Aerobic desulfurization of biogas by acidic biotrickling filtration in a

498

randomly packed reactor. Journal of Hazardous Materials 280: 200-208.

499

Namini M., Heydarian S., Bonakdarpour B. and Farjah A. 2008. Removal of H2S from

500

synthetic waste gas streams using a biotrickling filter. Iranian Journal of Chemical

501

Engineering 5(3): 41. 23

ACCEPTED MANUSCRIPT 502

Ramírez M., Gómez J. M., Aroca G. and Cantero D. 2009. Removal of hydrogen sulfide by

503

immobilized Thiobacillus thioparus in a biotrickling filter packed with polyurethane

504

foam. Bioresource Technology 100(21): 4989-4995. Ruokojärvi A., Ruuskanen J., Martikainen P. J. and Olkkonen M. 2001. Oxidation of gas

506

mixtures containing dimethyl sulfide, hydrogen sulfide, and methanethiol using a

507

two-stage biotrickling filter. Journal of the Air & Waste Management Association 51(1):

508

11-16.

509

RI PT

505

Soreanu G., Béland M., Falletta P., Edmonson K. and Seto P. 2008. Laboratory pilot scale

510

study for H2S removal from biogas in an anoxic biotrickling filter. Water Science and

511

Technology 57(2): 201-207.

Tiratsoo J. N. 1992. Pipeline pigging technology. Gulf Professional Publishing.

513

van den Bosch P. L., Sorokin D. Y., Buisman C. J. and Janssen A. J. 2008. The effect of pH on

514

thiosulfate formation in a biotechnological process for the removal of hydrogen sulfide

515

from gas streams. Environmental science & technology 42(7): 2637-2642.

517

Wu L., Loo Y.-Y. and Koe L. 2001. A pilot study of a biotrickling filter for the treatment of

M AN U

516

SC

512

odorous sewage air. Water Science and Technology 44(9): 295-299.

518

AC C

EP

TE D

519

24

ACCEPTED MANUSCRIPT 520 521

Captions

522 523

Table 1: Average removal efficiency and elimination capacity at different H2S/O2 molar ratios

525

Table 2: Sulfur balance and recovery rate at different H2S/O2 ratios

526

RI PT

524

Figure 1: Schematic of the laboratory-scale biotrickling filter setup. 1. Liquid collection bottle (in a fridge,

528

not shown). 2. Liquid recirculation pump. 3. Outlet gas sampling port. 4. Monolith biotrickling filter

529

reactor. 5. Liquid recycle vessel. 6. Magnetic stirrer. 7. Inlet gas sampling port. 8. Fresh mineral medium

530

supply. 9. Humidifier. 10. Peristaltic pump (for H2S gas). 11. Tedlar bag with H2S gas. 12. H2S gas

531

cylinder. 13. Mass flow controller. 14. N2 gas cylinder. 15. Metered air stream.

M AN U

SC

527

532

Figure 2: Schematic (left) and picture (right) of the 3D-printed monolith biotrickling filter;

534

note the mesh construction of the walls between the honeycomb channels (middle).

537 538 539 540

EP

536

Figure 3: Effects of different H2S/O2 molar ratios on H2S removal efficiency and elimination capacity.

AC C

535

TE D

533

Figure 4: Distribution of the different sulfur species in the liquid phase at different H2S/O2 molar ratios

Figure 5: System recovery after the pigging process

541 542 543 25

ACCEPTED MANUSCRIPT Table 1: Average removal efficiency and elimination capacity at different H2S/O2 molar ratios 2:1

1:1

1:2

Removal efficiency (RE,%)

50

63

95

Elimination capacity (EC, g m-3 h-1)

62.5

78.2

122

RI PT

H2S/O2 ratio

545 546 547

Table 2: Sulfur balance and recovery rate at different H2S/O2 ratios 1:1

1:2

0.0111

0.0166

0.0249

Sulfur production rate in liquid (g h-1)

0.0003

0.0002

0.0005

Sulfur production rate as solid (g h-1)

0.0264

0.0167

0.0192

Sulfur recovery (%)

240%

102%

79%

EP

549

AC C

548

TE D

Sulfur removal rate (g h-1)

2:1

M AN U

H2S:O2 ratio

SC

544

26

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

TE D

Figure 1: Schematic of the laboratory-scale biotrickling filter setup. 1. Liquid collection bottle (in a fridge, not shown). 2. Liquid recirculation pump. 3. Outlet gas sampling port. 4. Monolith biotrickling filter reactor. 5. Liquid recycle vessel. 6. Magnetic stirrer. 7. Inlet gas sampling port. 8. Fresh mineral

EP

medium supply. 9. Humidifier. 10. Peristaltic pump (for H2S gas). 11. Tedlar bag with H2S gas. 12. H2S

AC C

gas cylinder. 13. Mass flow controller. 14. N2 gas cylinder. 15. Metered air stream.

1

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

TE D

Figure 2: Schematic (left) and picture (right) of the 3D-printed monolith biotrickling filter;

AC C

EP

note the mesh construction of the walls between the honeycomb channels (middle).

2

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Figure 3: Effects of different H2S/O2 molar ratios on H2S removal efficiency and elimination capacity.

3

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Figure 4: Distribution of the different sulfur species in the liquid phase at different H2S/O2 molar ratios

4

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Figure 5: System recovery after the pigging process

5

ACCEPTED MANUSCRIPT

Highlights: A monolith biotrickling filter effectively treated hydrogen sulfide in biogas mimic

•

The effects of H2S/O2 ratio on treatment rate and fate of sulfur were investigated

•

Elemental sulfur was the dominant end-product which accumulated in the filter bed

•

Pigging was shown to be effective for removing the accumulated elemental sulfur

AC C

EP

TE D

M AN U

SC

RI PT

•