- Email: [email protected]
Performance of an anaerobic membrane bioreactor for pharmaceutical wastewater treatment
Accepted Manuscript Performance of an Anaerobic Membrane Bioreactor for Pharmaceutical Wastewater Treatment Jan Svojitka, Luká š Dvoř ák, Martin Studer, Jürg Oliver Straub, Heinz Frömelt, Thomas Wintgens PII: DOI: Reference:
S0960-8524(17)30042-1 http://dx.doi.org/10.1016/j.biortech.2017.01.022 BITE 17517
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
14 October 2016 9 January 2017 11 January 2017
Please cite this article as: Svojitka, J., Dvoř ák, L., Studer, M., Oliver Straub, J., Frömelt, H., Wintgens, T., Performance of an Anaerobic Membrane Bioreactor for Pharmaceutical Wastewater Treatment, Bioresource Technology (2017), doi: http://dx.doi.org/10.1016/j.biortech.2017.01.022
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.
1
PERFORMANCE OF AN ANAEROBIC MEMBRANE BIOREACTOR FOR
2
PHARMACEUTICAL WASTEWATER TREATMENT
3 4
Jan Svojitka1, Lukáš Dvořák2, Martin Studer3, Jürg Oliver Straub 3, Heinz Frömelt4,
5
Thomas Wintgens1*
6 7 8
1
9
and Arts Northwestern Switzerland, Gründenstrasse 40, CH-4132 Muttenz, Switzerland
Institute for Ecopreneurship, School of Life Sciences, University of Applied Sciences
10
2
11
University of Liberec, Studentská 2, 461 17, Liberec 1, Czech Republic
12
3
F.Hoffmann-La Roche Ltd, Grenzacherstrasse 124, CH-4070 Basel, Switzerland
13
4
ProRheno AG, Grenzstrasse 15, CH-4057 Basel, Switzerland
Centre for Nanomaterials, Advanced Technologies and Innovation, Technical
14 15 16
E-mail: [email protected], [email protected], [email protected],
17
[email protected], [email protected], [email protected]
18 19 20
* Corresponding author
21
1
22
Abstract
23
Anaerobic treatment of wastewater and waste organic solvents originating from the
24
pharmaceutical and chemical industries was tested in a pilot anaerobic membrane
25
bioreactor, which was operated for 580 days under different operational conditions. The
26
goal was to test the long-term treatment efficiency and identify inhibitory factors. The
27
highest COD removal of up to 97% was observed when the influent concentration was
28
increased by the addition of methanol (up to 25 g·L–1 as COD). Varying and generally
29
lower COD removal efficiency (around 78%) was observed when the anaerobic
30
membrane bioreactor was operated with incoming pharmaceutical wastewater as sole
31
carbon source. The addition of waste organic solvents (>2.5 g·L–1 as COD) to the
32
influent led to low COD removal efficiency or even to the breakdown of anaerobic
33
digestion. Changes in the anaerobic population (e.g., proliferation of the genus
34
Methanosarcina) resulting from the composition of influent were observed.
35 36
Keywords:
37
Anaerobic membrane bioreactor (AnMBR); COD removal; pharmaceutical wastewater;
38
fouling; biogas; inhibition of methanogens
39 40
1 Introduction
41
Industrial wastewater treatment by anaerobic biological processes is a proven
42
technology with several advantages compared to aerobic treatment, such as production
43
of biogas, lower energy costs and low excess sludge production (Lew et al., 2009).
44
Although modern high-rate anaerobic reactors such as Up-flow Anaerobic Sludge
45
Blanket (UASB) can achieve chemical oxygen demand (COD) removal efficiencies
2
46
over 90% (Choi et al., 2013; Delforno et al., 2014), a more widespread use of anaerobic
47
wastewater treatment is hampered by higher residual effluent pollution (Chen et al.,
48
2008) and poor retention of biomass in the reactor (Lin et al., 2013). Furthermore, the
49
granular biomass in high-rate anaerobic reactors can be negatively affected by the
50
characteristics of various industrial wastewaters.
51
Anaerobic membrane bioreactors (AnMBR) present an attractive approach for the
52
treatment of wastewater with a high content of COD and suspended solids, high salinity
53
and large variations of flow and composition, even in the presence of fat, oil and grease
54
or inhibitory compounds (Dereli et al., 2012; Diez et al., 2012; Lin et al., 2013).
55
Moreover, the membrane in an AnMBR represents a barrier for slow-growing
56
microorganisms with specialised degradation pathways, resulting in an increase in their
57
activity. For instance, Tao et al. (2012) used an AnMBR for the retention of slow-
58
growing Anammox microorganisms; installing a membrane in the system increased
59
their activity 19 times.
60
Despite the aforementioned advantages, there are still several drawbacks associated
61
with AnMBR, especially the lower filterability of the biomass leading to lower filtration
62
fluxes compared to aerobic MBRs (Lin et al., 2013).
63
AnMBRs have been successfully applied for the treatment of various industrial
64
wastewaters at both pilot and full scale. Most applications targeted wastewater from
65
food processing (e.g. Spagni et al., 2010; Wijekoon et al., 2011) since this is in general
66
highly biodegradable, contains high concentrations of organic matter and often high
67
amounts of suspended solids (Liao et al., 2006; Lin et al., 2013). AnMBRs have been
68
further used for treatment of pulp and paper industry wastewater (Gao et al., 2010;
69
Savant et al., 2006), textile industry wastewater (dos Santos et al., 2007) or polymer
3
70
synthesis effluents (Araya et al., 1999). However, there have been few studies dealing
71
with the treatment of pharmaceutical wastewater in AnMBRs to date, particularly in
72
pilot plants treating real wastewater or full scale installations (Dvořák et al., 2015).
73
Most studies focusing on the treatment of real pharmaceutical wastewaters have been
74
conducted either in aerobic MBRs or in lab-scale AnMBRs. For instance, Ng et al.
75
(2014) tested the treatment of pharmaceutical wastewater in a lab-scale AnMBR and
76
achieved COD removal of 14.7 – 47.2%. The low organic removal efficiencies were
77
caused by high salinity and complex nature of the organics in the wastewater. In another
78
study Ng et al. (2015) evaluated microbial communities and AnMBR performance
79
treating wastewater from a pharmaceutical factory in a lab scale AnMBR and
80
demonstrated a positive influence of halophilic organisms on the treatment efficiency.
81
The present study focused on the performance of an AnMBR pilot plant, which was fed
82
by real wastewater originating from pharmaceutical and chemical production. The
83
treatment feasibility and efficiency of the industrial wastewater under anaerobic
84
conditions, biogas production and process stability under highly varying composition
85
was assessed. Furthermore, concentrated waste organic solvents from the production
86
were tested as a co-substrate for the anaerobic process with regard to their degradability
87
and possible inhibition properties. The goal of the study was to evaluate whether
88
anaerobic treatment can be a reliable and economical addition or alternative to current
89
aerobic wastewater treatment.
90 91
2 Material and methods
92
2.1 Anaerobic membrane bioreactor
4
93
The AnMBR consisted of a bioreactor (50 L) and external cross-flow membrane unit
94
(Figure 1). Before entering the bioreactor, the incoming raw wastewater was pumped
95
into a buffer tank (20 L) in order to equilibrate concentration and flow peaks as well as
96
to adjust the pH to 7.0–7.5. In some phases of operation (specified in chapter 2.3) an
97
additional substrate, i.e. methanol or waste organic solvents, was added to the incoming
98
wastewater. The daily influent into the bioreactor was between 10 and 30 L.
99
The AnMBR was operated under mesophilic conditions (35–37 °C). Continuous
100
recirculation between the bioreactor and filtration step facilitated mixing of the reactor
101
and pressurization of two tubular membranes (TAMI Industries; Nyons, France). The
102
membranes were 1-inch ceramic (ZrO2–TiO2) tubes with 8 channels (MWCO 50 kDa)
103
and had a total area of 0.25 m2. Filtration was operated at a flux of 8.4 L·m–2·h–1. Since
104
the permeate flow exceeded the desired inflow rate, part of the permeate was
105
periodically recycled back to the bioreactor in order to reach the targeted hydraulic
106
retention time (HRT). The excess sludge was removed discontinuously three times per
107
week. The volume of excess sludge was adjusted to the incoming organic load, so that
108
the concentration of total solids in the bioreactor was kept above 10 g·L–1. The solids
109
retention time (SRT) was calculated as the ratio between the bioreactor volume and the
110
sum of sludge volume extracted weekly; the resulting value was between 120 and 450 d.
111 112
2.2 Industrial wastewater and waste organic solvents
113
The AnMBR was placed indoors at the industrial wastewater treatment plant (WWTP)
114
ProRheno AG in Basel (Switzerland), so that part of the incoming wastewater could be
115
pumped into the bioreactor. The industrial wastewater originated from
116
pharmaceutical/chemical production with the main constituents being methanol and
5
117
ethanol (forming on average 80–90% of the organic load); the rest was a mixture of
118
other organic solvents, e.g., tetrahydrofuran, dichloromethane, acetone, ethyl acetate,
119
tripropylamine, acetonitrile, toluene, isopropanol, acetone or dimethylacetamide. The
120
wastewater composition fluctuated greatly due to varying production with a COD
121
between 0.55 and 10.6 g·L–1 (the average value was 4.20 g·L–1). The wastewater was
122
mostly free of suspended solids.
123
Beside the wastewater, significant amounts of waste organic solvents (WOS) with
124
minor water content (max. 50%) arise from pharmaceutical production. WOS samples
125
were tested for their anaerobic biodegradability; two batches (WOS 9 and WOS 18,
126
compositions shown in Table 1) were selected and used as additional substrate for the
127
AnMBR. The composition of WOS varied largely from one batch to another. In most
128
cases methanol, ethanol and acetone were the main constituents; other components
129
included acetonitrile, dichloromethane, ethyl acetate, tetrahydrofuran, isopropyl acetate,
130
n-heptane and toluene.
131 132
2.3 Phases of AnMBR operation
133
The AnMBR was operated for a period of 580 days. The operation was divided into
134
three phases differing in influent composition and organic loading rate (Table 2).
135
Phase 1: The AnMBR was inoculated with a mixture of granular anaerobic sludge
136
taken from a UASB reactor treating food industry wastewater (high starch content) and
137
anaerobic sludge from a municipal WWTP digester. The two inocula were mixed in 1:1
138
ratio based on total solids (TS) and diluted to give a concentration of 10 g·L–1. The
139
AnMBR was then fed by the incoming industrial wastewater to test its treatability under
140
realistic conditions and varying feed quality. The operation of the AnMBR was started
6
141
by the adaptation of the biomass consisting of slow filling the reactor with wastewater
142
in the first five days with minimum effluent. Afterwards the influent rate was increased
143
from 10 to 30 L·d–1 (in 5 L·d–1 steps, each approx. 30 d) in order to increase the organic
144
loading.
145
Phase 2: To avoid variations in the wastewater composition, batches of wastewater
146
(1 m3) were collected and used as influent, each for a period of several weeks. In total
147
three batches of wastewater were used during Phase 2 with COD between 3.24 and 7.54
148
g·L–1, methanol and ethanol being the main constituents. Methanol and WOS 9 (1–
149
7 g·L–1 as COD) were added to the wastewater in order to increase organic loading in a
150
controlled way and test the treatability of the WOS 9 as a co-substrate. In later stages of
151
Phase 2 gradually increasing amounts of methanol were added to the wastewater in
152
order to increase the incoming COD concentration and test the operation of the AnMBR
153
at high organic loading. The influent rate of wastewater was kept constant at 14.3 L·d–1.
154
Phase 3: In Phase 3 the feeding of the AnMBR was switched from the batches of
155
wastewater back to continuous dosing of the incoming industrial wastewater. The COD
156
concentration in the influent was increased by a constant amount of methanol and WOS
157
18 in order to keep the organic loading high. The goal of this phase was to test the
158
operation at generally high loading combined with variations in wastewater quality and
159
test the treatability of WOS 18 as a co-substrate.
160 161
2.4 Analytical methods
162
The samples of influent and permeate were automatically collected and kept at 4°C until
163
analysis. The concentrations of chemical oxygen demand (COD), total nitrogen,
164
phosphate phosphorus and sulfides in influent and permeate samples were regularly
7
165
measured using cuvette tests (Hach-Lange, Düsseldorf, Germany). Total solids (TS)
166
were analyzed according to the Standard methods (APHA, 2012). Gas chromatography
167
(5890 Series II, Hewlett-Packard, Palo Alto, CA, USA) was used to determine the
168
methane content in the biogas samples and to analyze the concentration of organic acids
169
in the liquid samples.
170 171
2.5 FISH analyses
172
The samples for the fluorescence in situ hybridization (FISH) analyses were fixed 1:1
173
(vol./vol) with ethanol, absolute. The content of methanogenic Archaea was monitored
174
using a FISH (Snaidr et al., 1997) test kit with standardized methodology and prepared
175
solutions (VIT® Methanogenic bacteria, vermicon AG, Germany). The test kit detects
176
methanogens affiliated with the Euryarchaeota, which cover the range of Archaea
177
reported to be responsible for methanogenesis (Liu and Whitman, 2008). In situ
178
hybridization was performed according to the manual of the VIT® Methanogenic
179
bacteria test kit. In addition, a set of specific oligonucleotide probes was used to analyse
180
the samples with respect to their content of main microbial phyla either participating in
181
anaerobic digestion or being present in anaerobic environmental samples (Weinberger
182
and Höötmann, 2011). For the identification of the Euryarchaeota detected, more
183
specific oligonucleotide probes were used which target methanogens on class, family
184
and even genus level (Raskin et al., 1994). Total cell counts were determined by
185
filtration of samples on polycarbonate filter membranes, followed by using DAPI (4',6-
186
diamidino-2-phenylindole) staining. Total viable cell counts were performed by
187
hybridizing the same filters using a mixture of oligonucleotide probes specific for the
188
domains Bacteria and Archaea (Amann et al., 1995, Daims et al., 1999).
8
189
Hybridized cells were excited and examined under a Zeiss Axiostar plus fluorescence
190
microscope. The percentage of area covered with target organisms in comparison to all
191
viable cells was analysed in at least ten randomly selected captured fields and an
192
average percentage of the populations was calculated. An eyepiece with crossline
193
graticules was used for cell counts and calculation of absolute numbers.
194 195
2.6 Respirometric tests
196
The biodegradability of WOS and their single components was evaluated by
197
respirometric degradation tests using the biomass samples from the AnMBR. The
198
degradation tests of WOS 9 were carried out with two biomass samples taken at the end
199
of Phase 1 (Day 166) and beginning of Phase 2 (Day 180) before the addition of WOS 9
200
into the AnMBR. Later during Phase 2 another mixture of organic solvents (WOS 18)
201
was collected and subjected to a new series of respirometric tests with five biomass
202
samples taken between days 250 and 287. Degradation tests of the industrial wastewater
203
collected at the same day as each biomass sample were carried out parallel to the tests
204
with WOS mixtures.
205
In order to assess possible inhibitory effects of the individual components of the
206
wastewater and WOS, two additional series of respirometric tests were carried out using
207
biomass samples from day 306 and 325 respectively. Ethanol, ethyl acetate, acetonitrile,
208
tetrahydrofuran, dichloromethane, n-heptane and toluene were tested as the sole
209
substrate (1.0 g·L–1 COD) and in mixtures (0.5 + 0.5 g·L–1 COD) with methanol, which
210
served as reference substrate.
211
The original methodology based on OECD guideline No. 311 (OECD, 2006) was
212
modified as follows. A mixture (50 mL) of dilution medium, substrate, tested sludge
9
213
and water was prepared to obtain a final COD concentration of 1.0 g·L–1 and TS of
214
1.5 g·L–1. The dilution medium containing phosphate buffer and trace elements was
215
prepared in accordance with OECD guideline No. 311. The flasks were incubated at
216
35°C and pressure readings were collected every 30 minutes. At the end of the test, the
217
methane content in the head-space was measured. Based on the data from pressure
218
measurement and gas analyses, the amount of methane produced in each incubation
219
flask was calculated using the ideal gas law. The specific methane production rate
220
(SMPR) was calculated from the pressure evolution curve assuming a linear dependence
221
between the amount of methane produced and pressure. The maximum SMPR was
222
identified from each test and used as a measure for biomass activity with given substrate.
223 224
3 Results and discussion
225
3.1 Treatment efficiency of the AnMBR
226
Phase 1
227
The AnMBR was continuously fed by incoming industrial wastewater with flow rate
228
increasing in steps from 10 to 30 L·d–1. The influent in this phase was characterized by
229
highly varying COD concentrations (1.50–10.6 g·L–1; average 4.30 g·L–1). The effluent
230
COD concentrations were between 0.240 and 1.70 g·L–1 (average 0.83 g·L–1) resulting
231
in a COD removal efficiency between 44 and 94% (78% on average) (Figure 2a). Most
232
of the effluent COD constituted non-degraded compounds from the original feed, since
233
the concentration of organic acids did not exceed 90 mg·L–1 (max. 16% of total COD).
234
The highest removal efficiencies above 90% were observed in the periods of volumetric
235
loading rate above 2.5 g COD L–1·d–1 – with peak concentrations in the influent (day
236
40–47 and day 66–73, COD concentrations 8.0–10.6 g·L–1) and at the end of phase 1,
10
237
when the organic loading was elevated by a higher wastewater flow. Similarly, the
238
lowest observed removal efficiency (44% on day 136) corresponded to the lowest COD
239
feed concentration. In general, the achieved loading rate of the AnMBR was relatively
240
low compared to values typically used in AnMBRs (2–15 g COD L–1·d–1) or
241
conventional high rate anaerobic reactors (4–40 g COD L–1·d–1) (Liao et al., 2006; Lin
242
et al., 2013), therefore the energy recovery of the full scale application would be low. In
243
order to improve the economic efficiency of the process, the following phases focused
244
on increasing the wastewater concentration by the addition of concentrated waste
245
organic solvents from production (WOS).
246 247
Phase 2
248
After switching to batch-wastewater the organic loading rate of the AnMBR was
249
increased compared to phase 1 by adding methanol to the influent (COD of the resulting
250
mixture 5.4–7.6 g·L–1). These modifications did not disturb the operation and the
251
AnMBR exhibited stable COD removal efficiency of 89–93% (Figure 2b). After 10
252
days, methanol was replaced by the WOS 9 as additional substrate (COD of the
253
resulting mixture 7.1–11.3 g·L–1). Following the addition of the WOS 9 the COD
254
removal dropped to 19% within 8 days and other signs of process imbalance were
255
observed – accumulation of organic acids (up to 480 mg·L–1 acetate) and subsequently a
256
decreasing pH in the bioreactor. Consequently, the WOS 9 was replaced by methanol
257
and reactor loading was decreased by reducing the amount of methanol added to the
258
influent (resulting COD 3.6–4.2 g·L–1). These operational parameters were maintained
259
for one month (until day 240), which led to successful recovery of the system.
11
260
Since the operational parameters and composition of the wastewater was constant
261
during phase 2, the inhibition can be attributed to the addition of WOS 9. The
262
degradation tests of WOS 9 were carried out before the addition of WOS 9 into the
263
AnMBR and they indicated lower SMPR compared to the industrial wastewater, but it
264
did not lead to complete inhibition of methane generation (Table 3). This finding
265
contradicted the operational experience of the AnMBR, where imbalance in methane
266
production and changes in microbial population were observed following the dosing of
267
WOS 9 (see also chapter 3.3). The reason might be the difference between
268
concentration of the WOS 9 in the respirometric test (1 g·L–1 COD) and in the AnMBR
269
(5–7 g·L–1 COD). Therefore, inhibition may only become visible at high concentrations.
270
Between days 240 and 350 the COD of the influent mixture increased from 7 to 44 g·L–
271
1
272
loading of the bioreactor rose from 1.8 to 12.7 g COD L–1·d –1. COD removal efficiency
273
after the recovery was mostly above 90%, except for two short disturbances around days
274
260 and 350. These probably resulted from too rapid an increase in incoming COD load
275
but the reactor performance rapidly stabilized (Figure 2b).
276
The results obtained during phase 2 confirmed that anaerobic treatment process of the
277
tested wastewater can be operated efficiently and at high energy yield, provided a well
278
degradable, non-inhibited co-substrate is available in a sufficient amount.
by the addition of methanol (up to 37 g·L–1 as COD); correspondingly the organic
279 280
Phase 3
281
Reactivated pumping of wastewater from the sewer directly to the AnMBR caused
282
variations in wastewater concentration but they were dampened by constant addition of
283
methanol (25 g·L–1 as COD) in the first 40 days of phase 3 and did not affect the COD
12
284
removal efficiency (91 – 97%, Figure 2c). Starting from day 409 WOS 18 was added to
285
the wastewater (8 g COD L–1 of WOS 18 + 17 g COD L–1 of methanol), which resulted
286
in rapid deterioration of effluent quality and the dosing of WOS 18 was interrupted.
287
After recovery of the degradation process the dose of WOS 18 was increased in steps
288
(1.25 – 5.0 g COD L–1). A rapid drop of COD removal to 60% was observed when the
289
dose of WOS 18 reached 5.0 g COD L–1. The operation continued with the addition of
290
2.5 g COD L–1 WOS 18, leading to varying treatment efficiency (67 – 93%). A third
291
drop in COD removal was observed between days 500 and 511 which was not preceded
292
by an increase in WOS 18 addition. However, the organic loading rate between days
293
480 and 495 underwent vigorous changes resulting from variations in wastewater
294
concentration connected with changes in the dosing rate of external substrates, which
295
might have induced additional stress on the biomass and resulted in a massive
296
accumulation of organic acids up to 1000 mg·L–1 on Day 502. Additionally, a failure of
297
the temperature control system occurred on day 515, when the bioreactor was
298
overheated to 45°C. After this event, generally low COD removal efficiency (< 75%)
299
was observed although the addition of WOS was stopped. The FISH analysis (chapter
300
3.3) of the biomass sample from day 579 confirmed the collapse of the methanogenic
301
population and the operation of the pilot unit was terminated.
302
Although the respirometric tests showed substantially better biodegradability of WOS
303
18 compared to WOS 9 (Table 3), low performance of the AnMBR was observed when
304
the dose of this mixture reached 8.0, 5.0 and 2.5 g·L–1 COD. Again, this may be due to
305
higher concentration used in the pilot experiment compared to the respirometric test.
306
Unlike phase 2 the composition of wastewater in phase 3 varied, which might have had
307
an additional influence on performance of the bioreactor besides the addition of WOS
13
308
18. The disturbance in COD removal around day 415 (WOS dose 8.0 g·L–1 COD),
309
coincided with an elevated concentration of tetrahydrofuran in the industrial wastewater
310
(influent concentration 800 mg·L–1). This is however below the reported concentration
311
of 11.5 g·L–1 leading to significant inhibition of anaerobic process (Yao et al., 2010).
312
Elevated levels of N,N-dimethylacetamide (up to 2500 mg·L–1) were present in the
313
influent during the second disturbance in COD removal around day 460, which is
314
comparable to the value (>2000 mg·L–1) reported as having a negative effect on
315
microorganisms (EC10 value for activated sludge and EC50 for bacteria, respectively)
316
(Merck KGaA, 2013). Even though the presence of more than 2000 mg·L–1 N,N-
317
dimethylacetamide spanned a broader period than the observed disturbance, a
318
combination with the influence of the WOS 18 cannot be excluded.
319
Overall, the results showed that the WOS samples tested are not suitable as a co-
320
substrate for the full-scale treatment of industrial wastewater, since they can lead to
321
unpredictable breakdown of the process.
322 323
3.2 Organic acids
324
The total concentration of organic acids in the effluent from AnMBR was usually below
325
150 mg·L–1, with acetate being the dominant acid. Accumulation of organic acids
326
indicating anaerobic digestion imbalance was observed on a few occasions as a result of
327
the addition of WOS 9 (days 190–210), addition of WOS 18 combined with loading
328
variations (days 498–510) and reactor overheating (days 515–530) (Figure 3). Low
329
concentration of organic acids < 70 mg·L–1 was observed during the increasing loading
330
between days 230 and 400, even during the peak volumetric loading of 12 g COD L–
331
1
·d–1 and two disturbances in COD removal around days 260 and 350. This may be due
14
332
to a low share of acetoclastic metabolism in the bioreactor: Bhatti et al. (1996)
333
identified three main pathways governing the fate of methanol during anaerobic
334
digestion – acetoclastic methanogenesis through formation of acetate, hydrogenotrophic
335
methanogenesis through formation of hydrogen and direct methylotrophic conversion
336
into methane by Methanosarcina type species. The FISH analyses of biomass samples
337
confirmed that Methanosarcina were the dominant type of methanogenic population in
338
the AnMBR, although the type of metabolism could not be identified by this analysis,
339
since the group can use a variety of substrates and metabolic pathways.
340
Although the amount of methane formed by the methylotrophic metabolism does not
341
usually exceed 10% (Jiang et al., 2009), the low accumulation of organic acids during
342
the breakthrough events suggests that this type of metabolism might have prevailed in
343
the specialized biomass established after inhibition at the beginning of Phase 2, which
344
was fed with a uniform substrate.
345
On day 515 the bioreactor overheated to 45 °C due to failure of the temperature control
346
system. The elevated temperature led to another peak in concentration of organic acids
347
(up to 400 mg·L–1). The concentration of organic acids returned to below 200 mg·L–1
348
soon after the repair of the temperature control, however it remained at an elevated level
349
compared to previous phases (around 150 mg·L–1) until the end of the operation.
350 351
3.3 FISH analyses
352
Ten biomass samples from the AnMBR collected during phases 2 and 3 were analyzed
353
by FISH. The results showed a massive drop in methanogenic organisms between day
354
173 and day 201, which framed the period with WOS 9 dosing. This corresponded to
355
the breakdown observed in methane production and indicated a toxic effect of this
15
356
substrate towards the methanogens. After the dosing of WOS 9 was stopped (day 193),
357
the counts of methanogens recovered and in the sample from day 229 reached the
358
original level (Figure 4). A population shift within the methanogenic organisms was
359
observed. While the first sample contained a mixture (around 50:50) of rod-shaped and
360
cluster-forming methanogens, both types (especially the rod-shaped organisms) were
361
suppressed in samples from days 194 and 201 and starting from day 215 only the
362
cluster-forming methanogens were present in the biomass. Genus-specific probes
363
applied in samples from day 327 on identified this population as Methanosarcina. The
364
genus Methanosarcina is a common methanogen in anaerobic digesters. Several studies
365
found them to be a dominant methanogen, e.g., in reactors fed by sewage sludge (Liu et
366
al., 2016) or pharmaceutical wastewater (Chelliapan et al., 2011). Due to the versatility
367
of their metabolism and overall robustness they also dominate in anaerobic bioreactors
368
under transition conditions (De Vrieze et al., 2012).
369
The counts of methanogens roughly followed the organic loading trend of the reactor
370
and reached their peak on day 327. The last sample from day 579 contained a very low
371
share of viable cells (6% compared to 27–42% in the remaining samples) including
372
methanogens. This corresponds to the poor reactor methane production observed in that
373
period.
374
Changes within the population of other microbial groups in the biomass samples were
375
also observed during the period with the dosing of WOS 9. Whereas Firmicutes,
376
Crenarchaeota and Actinobacteria disappeared or their counts were suppressed during
377
the addition of WOS, Chloroflexi, Cytophaga-Flexibacter subphylum and
378
Alphaproteobacteria were found in higher amounts at the end of the first sampling
379
campaign. This indicates that the addition of WOS 9 affected only some microbial
16
380
groups, whereas the microorganisms less sensitive to the substances present rapidly
381
occupied the vacant habitat. The disappearance of Firmicutes and Actinobacteria, which
382
are connected with acidogenesis and acetogenesis (Levén et al., 2007), was remarkable.
383
This was probably the result of the lack of suitable substrates, since the AnMBR
384
influent contained few organic compounds other than methanol at that time.
385
In samples from day 327 onwards, Firmicutes and other bacteria were detected again in
386
gradually increasing numbers, presumably because more diversified substrate was
387
added to the AnMBR in phase 3. Relatively few viable cells were found in the last
388
sample from day 579 (after reactor overheating) with Firmicutes forming 22% of all
389
cells. Such an evolution is usually a sign of a collapsing reactor performance needing
390
new inoculation, which was also the case in this study.
391
A decreasing trend was observed for hydrolysis bacteria (Chloroflexi and Cytophaga-
392
Flexibacter-subphylum) after the population shift (from day 215 on), presumably due to
393
the specific type of substrates in the reactor, which had low hydrolytic requirements.
394 395
3.4 Biogas production and quality
396
The biogas produced contained about 40–70% methane, on average about 60%. The
397
amount of biogas formed was between 0.5 and 0.6 L·g–1 COD removed throughout all
398
phases of operation, corresponding to 0.30–0.36 L methane per g COD. Based on the
399
equilibrium calculation of known sulfide content in the effluent, the hydrogen sulfide
400
concentration in the gas was mostly below 0.03 g·m–3, but would reach approximately
401
14 g·m–3 (1% v/v) during the peak sulfide concentration detected.
402 403
3.5 Total solids and membrane performance
17
404
The total solids (TS) in the bioreactor roughly followed the trend of the organic loading
405
rate and varied between 9 and 23 g·L–1 (Figure 5). Despite faster sludge wasting
406
(corresponding SRT approx. 120 d) the concentration of TS roughly doubled from 11 to
407
23 g·L–1 during phase 2. The effect of accentuated biomass growth resulting from
408
higher organic loading probably combined with the introduction of suspended solids
409
(approx. 0.5 g·L–1) contained in the batch of wastewater used between days 290 and 366.
410
The membrane was presumably carrying a layer of foulants from the previous
411
experiments (data not shown) and a rapid increase in TMP was observed following the
412
inoculation so that chemical cleaning was necessary after 90 days of operation. In the
413
following more than 300 days only a minor decrease of permeability was observed
414
without any chemical cleaning of the membrane and despite the rapid increase of TS.
415
No fouling problems were observed up to 25 g·L–1 TS (Figure 5).
416
On the other hand, the membrane permeability decreased in the last 80 days of phase 3,
417
triggered by difficulties with maintaining the desired cross-flow velocity (3 m·s–1)
418
rather than by changes in the structure of the mixed liquor. Since the AnMBR had no
419
automatic control of the retentate flow, the filtration was repeatedly operated at low
420
cross-flow velocities close to dead-end mode for a period of several hours. This
421
probably led to accumulation of sludge in the membrane channels, as revealed by the
422
inspection of membranes at the end of the operation.
423 424
3.6 Degradation tests with the components of WOS
425
In the two series of degradation tests carried out with the components of the WOS, pure
426
methanol was the substrate with highest degradation rate (SMPR) (Figure 6), due to the
427
pre-adaptation of the biomass to this substrate. Degradation rate was substantially lower
18
428
when methanol was mixed with other substrates including well degradable organic
429
compounds (ethanol, ethyl acetate). Dichloromethane was the only substrate which
430
completely inhibited anaerobic degradation in mixture with methanol at the
431
concentration tested (0.5 g·L–1 COD, i.e. 0.88 g·L–1). This concentration was
432
comparable to that in the AnMBR influent during phase 2 (530 mg·L–1) and phase 3
433
(110 mg·L–1) and higher than the 50% inhibition concentration (IC50) in a UASB reactor
434
reported by Ozdemir et al. (2010) – 43 mg·L–1. Thus, dichloromethane could be
435
responsible for the observed breakdowns in COD removal in the AnMBR.
436
A relatively low degradation rate was observed for toluene. Toluene might have played
437
a more important role in the inhibition during phase 2 when the dosing of WOS 9 raised
438
its concentration in the influent to 160 mg·L–1. The same amount was present in the
439
respirometric test with low SMPR when degrading the toluene/methanol mixture and
440
320 mg·L–1 in the test with pure toluene. Inhibition (IC50) of anaerobic digestion was
441
reported in a similar range between 146 and 1000 mg·L–1 by Orellana (2000) and
442
Kayembe (2013). On the other hand, the concentration of toluene during the dosing of
443
WOS 18 did not exceed 30 mg·L–1.
444
In one test the degradation of acetonitrile mixed with methanol (Figure 6a) was low
445
compared to other substrates. However, acetonitrile presumably had no effect on the
446
breakdown observed in AnMBR performance since its concentration in the influent was
447
well below the dose in the degradation tests and did not exceed 30 mg·L–1.
448
Besides organic compounds, some inorganic substances, e.g. free ammonia or high
449
concentrations of salts in general, are known to inhibit anaerobic digestion (Chen et al.,
450
2008). However, they presumably had a low effect on the inhibition of COD removal in
451
the AnMBR, since the total nitrogen concentration in the bioreactor (< 300 mg·L–1) was
19
452
below the reported inhibitory limits of 1700 mg N-HH4 + L–1 (Yenigün and Demirel,
453
2013), while the salt content of 0.7 – 6 g·L–1 in the AnMBR was below the inhibitory
454
values of >13 g·L–1 reported for non-acclimated sludge (Lefebvre et al., 2007; Lefebvre
455
and Moletta, 2006).
456
Overall, it was not clear why the WOS samples, which showed reasonable
457
biodegradability in the single tests, had such a detrimental influence on the performance
458
of the AnMBR. Besides the influence of the solvents discussed above, other
459
components that are not detectable by GC – analysis in the WOS might also be
460
responsible for the negative effects. For a commercial operation, it would be important
461
to evaluate this in more detail, because the use of WOS is a key point in increasing the
462
biogas yield as well as in lowering the disposal cost for WOS.
463 464
4 Conclusions
465
The industrial wastewater alone showed good anaerobic degradability in the AnMBR,
466
however its low concentration and varying composition decreased the treatment
467
efficiency. Constant and high removal efficiency (>90%) was reached when methanol
468
was added to the influent and variations of its composition were limited. Acidification,
469
poor methane production and shifts in methanogenic population were experienced when
470
waste organic solvents were added to the influent as co-substrate. Therefore, a
471
commercially viable anaerobic treatment of the current wastewater and WOS mixtures
472
seems not possible at the moment. Well-controlled cross-flow conditions in the
473
membrane channels were necessary for stable filtration and low fouling.
474 475
Acknowledgements
20
476
We are very grateful to F. Hoffmann-La Roche for financial support, experiments and
477
access to analytical results. ProRheno AG is kindly thanked for kind cooperation and
478
support. Thanks also to students and technicians of the FHNW for pilot plant
479
supervision, analytics and technical support. Project LO1201 „National Programme for
480
Sustainability I“, and the OPR&DI project No. CZ.1.05/2.1.00/01.0005 is also
481
acknowledged.
482 483
References
484
1.Amann RI, Ludwig W, Schleifer KH. 1995. Phylogenetic identification and in situ
485
detection of individual microbial cells without cultivation. Microbiol Rev. 59(1):143-69.
486
2.APHA, 2012. Standard Methods for the Examination of Water and Wastewater, 22
487
edition. ed. American Water Works Assn, Washington, DC.
488
3.Araya, P., Aroca, G., Chamy, R., 1999. Anaerobic treatment of effluents from an
489
industrial polymers synthesis plant. Waste Management 19, 141–146.
490
doi:10.1016/S0956-053X(99)00013-6
491
4.Bhatti, Z.I., Furukawa, K., Fujita, M., 1996. Feasibility of methanolic waste treatment
492
in UASB reactors. Water Research 30, 2559–2568. doi:10.1016/S0043-1354(96)00144-
493
3
494
5.Chelliapan, S., Wilby, T., Yuzir, A., Sallis, P.J., 2011. Influence of organic loading on
495
the performance and microbial community structure of an anaerobic stage reactor
496
treating pharmaceutical wastewater. Desalination 271, 257–264.
497
doi:10.1016/j.desal.2010.12.045
21
498
6.Chen, Z., Ren, N., Wang, A., Zhang, Z.-P., Shi, Y., 2008. A novel application of
499
TPAD–MBR system to the pilot treatment of chemical synthesis-based pharmaceutical
500
wastewater. Water Research 42, 3385–3392. doi:10.1016/j.watres.2008.04.020
501
7.Choi, W.-H., Shin, C.-H., Son, S.-M., Ghorpade, P.A., Kim, J.-J., Park, J.-Y., 2013.
502
Anaerobic treatment of palm oil mill effluent using combined high-rate anaerobic
503
reactors. Bioresource Technology, Challenges in Environmental Science and
504
Engineering (CESE-2012) 141, 138–144. doi:10.1016/j.biortech.2013.02.055
505
8.Daims, H., A. Brühl, R. Amann, K.-H. Schleifer, and M. Wagner. 1999. The domain-
506
specific probe EUB338 is insufficient for the detection of all Bacteria: development and
507
evaluation of a more comprehensive probe set. Syst. Appl. Microbiol. 22:434–444.
508
9.De Vrieze, J., Hennebel, T., Boon, N., Verstraete, W., 2012. Methanosarcina: The
509
rediscovered methanogen for heavy duty biomethanation. Bioresource Technology 112,
510
1–9. doi:10.1016/j.biortech.2012.02.079
511
10.Delforno, T.P., Moura, A.G.L., Okada, D.Y., Varesche, M.B.A., 2014. Effect of
512
biomass adaptation to the degradation of anionic surfactants in laundry wastewater
513
using EGSB reactors. Bioresource Technology 154, 114–121.
514
doi:10.1016/j.biortech.2013.11.102
515
11.Dereli, R.K., Ersahin, M.E., Ozgun, H., Ozturk, I., Jeison, D., van der Zee, F., van
516
Lier, J.B., 2012. Potentials of anaerobic membrane bioreactors to overcome treatment
517
limitations induced by industrial wastewaters. Bioresource Technology, Membrane
518
Bioreactors (MBRs): State-of-Art and Future 122, 160–170.
519
doi:10.1016/j.biortech.2012.05.139
520
12.Diez, V., Ramos, C., Cabezas, J.L., 2012. Treating wastewater with high oil and
521
grease content using an Anaerobic Membrane Bioreactor (AnMBR). Filtration and
22
522
cleaning assays. Water Science and Technology 65, 1847–1853.
523
doi:10.2166/wst.2012.852
524
13.dos Santos, A.B., Cervantes, F.J., van Lier, J.B., 2007. Review paper on current
525
technologies for decolourisation of textile wastewaters: Perspectives for anaerobic
526
biotechnology. Bioresource Technology 98, 2369–2385.
527
doi:10.1016/j.biortech.2006.11.013
528
14.Dvořák, L., Gómez, M., Dolina, J., Černín, A., 2015. Anaerobic membrane
529
bioreactors—a mini review with emphasis on industrial wastewater treatment:
530
applications, limitations and perspectives. Desalination and Water Treatment 0, 1–15.
531
doi:10.1080/19443994.2015.1100879
532
15.Gao, W.J.J., Lin, H.J., Leung, K.T., Liao, B.Q., 2010. Influence of elevated pH
533
shocks on the performance of a submerged anaerobic membrane bioreactor. Process
534
Biochemistry 45, 1279–1287. doi:10.1016/j.procbio.2010.04.018
535
16.Jiang, N., Wang, Y., Dong, X., 2009. Methanol as the primary methanogenic and
536
acetogenic precursor in the cold zoige wetland at tibetan plateau. Microbial Ecology 60,
537
206–213. doi:10.1007/s00248-009-9602-0
538
17.Kayembe, K., 2013. Comparative study of benzyl and phenyl compounds chemical
539
structure effects on the inhibition of methane production by digested pig manure
540
methanogens. International Research Journal of Pure and Applied Chemistry 3, 48–58.
541
doi:10.9734/IRJPAC/2013/2489
542
18.Lefebvre, O., Moletta, R., 2006. Treatment of organic pollution in industrial saline
543
wastewater: A literature review. Water Research 40, 3671–3682.
544
doi:10.1016/j.watres.2006.08.027
23
545
19.Lefebvre, O., Quentin, S., Torrijos, M., Godon, J.J., Delgenès, J.P., Moletta, R., 2007.
546
Impact of increasing NaCl concentrations on the performance and community
547
composition of two anaerobic reactors. Applied Microbiology and Biotechnology. 75,
548
61–69. doi:10.1007/s00253-006-0799-2
549
20.Levén, L., Eriksson, A.R.B., Schnürer, A., 2007. Effect of process temperature on
550
bacterial and archaeal communities in two methanogenic bioreactors treating organic
551
household waste. FEMS Microbiology Ecology 59, 683–693. doi:10.1111/j.1574-
552
6941.2006.00263.x
553
21.Lew, B., Tarre, S., Beliavski, M., Dosoretz, C., Green, M., 2009. Anaerobic
554
membrane bioreactor (AnMBR) for domestic wastewater treatment. Desalination 243,
555
251–257. doi:10.1016/j.desal.2008.04.027
556
22.Liao, B.-Q., Kraemer, J.T., Bagley, D.M., 2006. Anaerobic membrane bioreactors:
557
applications and research directions. Critical Reviews in Environmental Science and
558
Technology 36, 489–530. doi:10.1080/10643380600678146
559
23.Lin, H., Peng, W., Zhang, M., Chen, J., Hong, H., Zhang, Y., 2013. A review on
560
anaerobic membrane bioreactors: Applications, membrane fouling and future
561
perspectives. Desalination 314, 169–188. doi:10.1016/j.desal.2013.01.019
562
24.Liu Y, Whitman WB. 2008. Metabolic, phylogenetic, and ecological diversity of the
563
methanogenic archaea. Ann N Y Acad Sci. 1125:171-89.
564
25.Liu, C., Li, H., Zhang, Y., Si, D., Chen, Q., 2016. Evolution of microbial community
565
along with increasing solid concentration during high-solids anaerobic digestion of
566
sewage sludge. Bioresource Technology 216, 87–94.
567
doi:10.1016/j.biortech.2016.05.048
568
26.Merck KGaA, 2013. N,N-Diethylacetamide for synthesis. Safety data sheet.
24
569
27.Ng K.K., Shi X., Tang M.K.Y., Ng H.Y. 2014. A novel application of anaerobic bio-
570
entrapped membrane reactor for the treatment of chemical synthesis-based
571
pharmaceutical wastewater. Separation and Purification Technology 132, 634–643.
572
doi:10.1016/j.seppur.2014.06.021
573
28.Ng, K.K., Shi, X., Ng, H.Y., 2015. Evaluation of system performance and microbial
574
communities of a bioaugmented anaerobic membrane bioreactor treating pharmaceutical
575
wastewater. Water Research 81, 311–324. doi: 10.1016/j.watres.2015.05.033
576
29.OECD, 2006. Test No. 311: Anaerobic Biodegradability of Organic Compounds in
577
Digested Sludge: by Measurement of Gas Production. OECD, Paris.
578
30.Orellana, R.I., 2000. Effects of various substrates on anaerobic biodegradation of
579
benzene and toluene. PhD thesis, The George Washington University, Washington, DC.
580
31.Ozdemir, C., Sen, N., Dursun, S., Kalipci, E., 2010. Removal of dichloromethane in
581
up-flow anaerobic sludge bed reactors and methane production. Asian Journal of
582
Chemistry 22, 6423–6436.
583
32.Raskin L, Stromley JM, Rittmann BE, Stahl DA. 1994. Group-specific 16S rRNA
584
hybridization probes to describe natural communities of methanogens. Appl Environ
585
Microbiol. 60(4):1232-40.
586
33.Savant, D.V., Abdul-Rahman, R., Ranade, D.R., 2006. Anaerobic degradation of
587
adsorbable organic halides (AOX) from pulp and paper industry wastewater.
588
Bioresource Technology 97, 1092–1104. doi:10.1016/j.biortech.2004.12.013
589
34.Snaidr J, Amann R, Huber I, Ludwig W, Schleifer KH. 1997. Phylogenetic analysis
590
and in situ identification of bacteria in activated sludge. Appl Environ Microbiol.
591
63(7):2884-96.
25
592
35.Spagni, A., Casu, S., Crispino, N.A., Farina, R., Mattioli, D., 2010. Filterability in a
593
submerged anaerobic membrane bioreactor. Desalination 250, 787–792.
594
doi:10.1016/j.desal.2008.11.042
595
36.Tao, Y., Gao, D.-W., Fu, Y., Wu, W.-M., Ren, N.-Q., 2012. Impact of reactor
596
configuration on anammox process start-up: MBR versus SBR. Bioresource
597
Technology 104, 73–80. doi:10.1016/j.biortech.2011.10.052
598
37.Weinberger G, Höötmann U. 2011. Wastewater treatment plants – a look into a black
599
box. IPW 11-12/2011:18-21. www.ipwonline.de.
600
38.Wijekoon, K.C., Visvanathan, C., Abeynayaka, A., 2011. Effect of organic loading
601
rate on VFA production, organic matter removal and microbial activity of a two-stage
602
thermophilic anaerobic membrane bioreactor. Bioresource Technology, Special Issue on
603
Challenges in Environmental Science and Engineering, CESE-2010: Technological
604
Advances in Waste Treatment for a Sustainable Future 102, 5353–5360.
605
doi:10.1016/j.biortech.2010.12.081
606
39.Yao, Y., Guan, J., Tang, P., Jiao, H., Lin, C., Wang, J., Lu, Z., Min, H., Gao, H.,
607
2010. Assessment of toxicity of tetrahydrofuran on the microbial community in
608
activated sludge. Bioresource Technology 101, 5213–5221.
609
doi:10.1016/j.biortech.2010.02.051
610
40.Yenigün, O., Demirel, B., 2013. Ammonia inhibition in anaerobic digestion: A
611
review. Process Biochemistry 48, 901–911. doi:10.1016/j.procbio.2013.04.012
612
26
613
Figure 1 Flow scheme of the pilot AnMBR.
614 615
27
616
Figure 2 COD concentration and COD removal during phase 1 (a), phase 2 (b) and
617
phase 3 (c). The areas in the diagram indicate the target dose of methanol (grey) and
618
WOS (yellow), both expressed as COD.
619
620
621 622 28
623
Figure 3 Organic acids in the effluent and organic loading of the AnMBR. The yellow
624
area in the diagram indicates target dose (expressed as COD) of WOS 9 (Phase 2) and
625
WOS 18 (Phase 3), respectively.
626 627
29
628
Figure 4 Comparison of biomass samples according to the absolute cell counts per mL
629
of each population.
630 631
30
632
Figure 5 Permeability (K) and total solids over AnMBR operation.
633 634
31
635
Figure 6 Maximum SMPR during respirometric tests of anaerobic biodegradability
636
with individual components of the WOS mixed with methanol (a) and as sole substrates
637
(b). Respirometric tests a) and b) were conducted with two different sludge samples.
638
MeOH – methanol, CH2Cl2 – dichloromethane, THF – tetrahydrofuran.
639
640 641
32
642
Table 1 Composition of WOS 9 and WOS 18 added to the pharmaceutical wastewater
643
during Phases 2 and 3. Parameter pH Water Methanol Ethanol Acetone Dichloromethane Ethyl acetate Tetrahydrofuran n-Heptane Toluene Ammonia nitrogen
WOS 9 (Phase 2) 11.3 14.7% 62.5% 6.4% 0.5% 6.9% 2.8% 0.4% – 2.1% 2.7%
WOS 18 (Phase 3) 10.9 6.0% 73.6% 2.0% 9.6% 1.0% 3.0% 4.4% 0.5% – –
644 645
33
646
Table 2 Technological parameters during different operational phases. Duration Wastewater Additional Phase [d] collection substrate 1
175
2
191
3
214
Sludge Volumetric HRT concentration loading [d] [g TS·L–1] [g COD·L–1·d–1]
no 7.3–13.1 co-substrate methanol, Batch 10.0–24.0 WOS 9 methanol, Continuous 18.3–26.4 WOS 18 Continuous
0.6–4.0
1.7–5
1.0–12.7
3.5
4.5–14.2
3–3.5
647 648
34
649
Table 3 Maximum SMPR during respirometric tests of anaerobic degradability with
650
WOS 9, WOS 18 and wastewater. Substrate
WOS 9
WOS 18
Wastewater
Maximum SMPR [mg COD g-1 h-1 ]
10.5 ± 3.5
37.9 ± 8.8
47.4 ± 6.0
651 652
35
653
Highlights
654
1.
Wastewater from pharmaceutical industry can be treated by anaerobic MBR technology.
655
2.
Varying wastewater composition had negative influence on the treatment efficiency.
656
3.
Addition of waste organic solvents caused inhibition of the anaerobic degradation.
657
4.
Dichloromethane exhibited the strongest inhibitory effect on anaerobic degradation.
658
5.
Changes in the biocenosis reflected the substrate pattern and inhibition events.
659 660
661
36
S0960-8524(17)30042-1 http://dx.doi.org/10.1016/j.biortech.2017.01.022 BITE 17517
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
14 October 2016 9 January 2017 11 January 2017
Please cite this article as: Svojitka, J., Dvoř ák, L., Studer, M., Oliver Straub, J., Frömelt, H., Wintgens, T., Performance of an Anaerobic Membrane Bioreactor for Pharmaceutical Wastewater Treatment, Bioresource Technology (2017), doi: http://dx.doi.org/10.1016/j.biortech.2017.01.022
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.
1
PERFORMANCE OF AN ANAEROBIC MEMBRANE BIOREACTOR FOR
2
PHARMACEUTICAL WASTEWATER TREATMENT
3 4
Jan Svojitka1, Lukáš Dvořák2, Martin Studer3, Jürg Oliver Straub 3, Heinz Frömelt4,
5
Thomas Wintgens1*
6 7 8
1
9
and Arts Northwestern Switzerland, Gründenstrasse 40, CH-4132 Muttenz, Switzerland
Institute for Ecopreneurship, School of Life Sciences, University of Applied Sciences
10
2
11
University of Liberec, Studentská 2, 461 17, Liberec 1, Czech Republic
12
3
F.Hoffmann-La Roche Ltd, Grenzacherstrasse 124, CH-4070 Basel, Switzerland
13
4
ProRheno AG, Grenzstrasse 15, CH-4057 Basel, Switzerland
Centre for Nanomaterials, Advanced Technologies and Innovation, Technical
14 15 16
E-mail: [email protected], [email protected], [email protected],
17
[email protected], [email protected], [email protected]
18 19 20
* Corresponding author
21
1
22
Abstract
23
Anaerobic treatment of wastewater and waste organic solvents originating from the
24
pharmaceutical and chemical industries was tested in a pilot anaerobic membrane
25
bioreactor, which was operated for 580 days under different operational conditions. The
26
goal was to test the long-term treatment efficiency and identify inhibitory factors. The
27
highest COD removal of up to 97% was observed when the influent concentration was
28
increased by the addition of methanol (up to 25 g·L–1 as COD). Varying and generally
29
lower COD removal efficiency (around 78%) was observed when the anaerobic
30
membrane bioreactor was operated with incoming pharmaceutical wastewater as sole
31
carbon source. The addition of waste organic solvents (>2.5 g·L–1 as COD) to the
32
influent led to low COD removal efficiency or even to the breakdown of anaerobic
33
digestion. Changes in the anaerobic population (e.g., proliferation of the genus
34
Methanosarcina) resulting from the composition of influent were observed.
35 36
Keywords:
37
Anaerobic membrane bioreactor (AnMBR); COD removal; pharmaceutical wastewater;
38
fouling; biogas; inhibition of methanogens
39 40
1 Introduction
41
Industrial wastewater treatment by anaerobic biological processes is a proven
42
technology with several advantages compared to aerobic treatment, such as production
43
of biogas, lower energy costs and low excess sludge production (Lew et al., 2009).
44
Although modern high-rate anaerobic reactors such as Up-flow Anaerobic Sludge
45
Blanket (UASB) can achieve chemical oxygen demand (COD) removal efficiencies
2
46
over 90% (Choi et al., 2013; Delforno et al., 2014), a more widespread use of anaerobic
47
wastewater treatment is hampered by higher residual effluent pollution (Chen et al.,
48
2008) and poor retention of biomass in the reactor (Lin et al., 2013). Furthermore, the
49
granular biomass in high-rate anaerobic reactors can be negatively affected by the
50
characteristics of various industrial wastewaters.
51
Anaerobic membrane bioreactors (AnMBR) present an attractive approach for the
52
treatment of wastewater with a high content of COD and suspended solids, high salinity
53
and large variations of flow and composition, even in the presence of fat, oil and grease
54
or inhibitory compounds (Dereli et al., 2012; Diez et al., 2012; Lin et al., 2013).
55
Moreover, the membrane in an AnMBR represents a barrier for slow-growing
56
microorganisms with specialised degradation pathways, resulting in an increase in their
57
activity. For instance, Tao et al. (2012) used an AnMBR for the retention of slow-
58
growing Anammox microorganisms; installing a membrane in the system increased
59
their activity 19 times.
60
Despite the aforementioned advantages, there are still several drawbacks associated
61
with AnMBR, especially the lower filterability of the biomass leading to lower filtration
62
fluxes compared to aerobic MBRs (Lin et al., 2013).
63
AnMBRs have been successfully applied for the treatment of various industrial
64
wastewaters at both pilot and full scale. Most applications targeted wastewater from
65
food processing (e.g. Spagni et al., 2010; Wijekoon et al., 2011) since this is in general
66
highly biodegradable, contains high concentrations of organic matter and often high
67
amounts of suspended solids (Liao et al., 2006; Lin et al., 2013). AnMBRs have been
68
further used for treatment of pulp and paper industry wastewater (Gao et al., 2010;
69
Savant et al., 2006), textile industry wastewater (dos Santos et al., 2007) or polymer
3
70
synthesis effluents (Araya et al., 1999). However, there have been few studies dealing
71
with the treatment of pharmaceutical wastewater in AnMBRs to date, particularly in
72
pilot plants treating real wastewater or full scale installations (Dvořák et al., 2015).
73
Most studies focusing on the treatment of real pharmaceutical wastewaters have been
74
conducted either in aerobic MBRs or in lab-scale AnMBRs. For instance, Ng et al.
75
(2014) tested the treatment of pharmaceutical wastewater in a lab-scale AnMBR and
76
achieved COD removal of 14.7 – 47.2%. The low organic removal efficiencies were
77
caused by high salinity and complex nature of the organics in the wastewater. In another
78
study Ng et al. (2015) evaluated microbial communities and AnMBR performance
79
treating wastewater from a pharmaceutical factory in a lab scale AnMBR and
80
demonstrated a positive influence of halophilic organisms on the treatment efficiency.
81
The present study focused on the performance of an AnMBR pilot plant, which was fed
82
by real wastewater originating from pharmaceutical and chemical production. The
83
treatment feasibility and efficiency of the industrial wastewater under anaerobic
84
conditions, biogas production and process stability under highly varying composition
85
was assessed. Furthermore, concentrated waste organic solvents from the production
86
were tested as a co-substrate for the anaerobic process with regard to their degradability
87
and possible inhibition properties. The goal of the study was to evaluate whether
88
anaerobic treatment can be a reliable and economical addition or alternative to current
89
aerobic wastewater treatment.
90 91
2 Material and methods
92
2.1 Anaerobic membrane bioreactor
4
93
The AnMBR consisted of a bioreactor (50 L) and external cross-flow membrane unit
94
(Figure 1). Before entering the bioreactor, the incoming raw wastewater was pumped
95
into a buffer tank (20 L) in order to equilibrate concentration and flow peaks as well as
96
to adjust the pH to 7.0–7.5. In some phases of operation (specified in chapter 2.3) an
97
additional substrate, i.e. methanol or waste organic solvents, was added to the incoming
98
wastewater. The daily influent into the bioreactor was between 10 and 30 L.
99
The AnMBR was operated under mesophilic conditions (35–37 °C). Continuous
100
recirculation between the bioreactor and filtration step facilitated mixing of the reactor
101
and pressurization of two tubular membranes (TAMI Industries; Nyons, France). The
102
membranes were 1-inch ceramic (ZrO2–TiO2) tubes with 8 channels (MWCO 50 kDa)
103
and had a total area of 0.25 m2. Filtration was operated at a flux of 8.4 L·m–2·h–1. Since
104
the permeate flow exceeded the desired inflow rate, part of the permeate was
105
periodically recycled back to the bioreactor in order to reach the targeted hydraulic
106
retention time (HRT). The excess sludge was removed discontinuously three times per
107
week. The volume of excess sludge was adjusted to the incoming organic load, so that
108
the concentration of total solids in the bioreactor was kept above 10 g·L–1. The solids
109
retention time (SRT) was calculated as the ratio between the bioreactor volume and the
110
sum of sludge volume extracted weekly; the resulting value was between 120 and 450 d.
111 112
2.2 Industrial wastewater and waste organic solvents
113
The AnMBR was placed indoors at the industrial wastewater treatment plant (WWTP)
114
ProRheno AG in Basel (Switzerland), so that part of the incoming wastewater could be
115
pumped into the bioreactor. The industrial wastewater originated from
116
pharmaceutical/chemical production with the main constituents being methanol and
5
117
ethanol (forming on average 80–90% of the organic load); the rest was a mixture of
118
other organic solvents, e.g., tetrahydrofuran, dichloromethane, acetone, ethyl acetate,
119
tripropylamine, acetonitrile, toluene, isopropanol, acetone or dimethylacetamide. The
120
wastewater composition fluctuated greatly due to varying production with a COD
121
between 0.55 and 10.6 g·L–1 (the average value was 4.20 g·L–1). The wastewater was
122
mostly free of suspended solids.
123
Beside the wastewater, significant amounts of waste organic solvents (WOS) with
124
minor water content (max. 50%) arise from pharmaceutical production. WOS samples
125
were tested for their anaerobic biodegradability; two batches (WOS 9 and WOS 18,
126
compositions shown in Table 1) were selected and used as additional substrate for the
127
AnMBR. The composition of WOS varied largely from one batch to another. In most
128
cases methanol, ethanol and acetone were the main constituents; other components
129
included acetonitrile, dichloromethane, ethyl acetate, tetrahydrofuran, isopropyl acetate,
130
n-heptane and toluene.
131 132
2.3 Phases of AnMBR operation
133
The AnMBR was operated for a period of 580 days. The operation was divided into
134
three phases differing in influent composition and organic loading rate (Table 2).
135
Phase 1: The AnMBR was inoculated with a mixture of granular anaerobic sludge
136
taken from a UASB reactor treating food industry wastewater (high starch content) and
137
anaerobic sludge from a municipal WWTP digester. The two inocula were mixed in 1:1
138
ratio based on total solids (TS) and diluted to give a concentration of 10 g·L–1. The
139
AnMBR was then fed by the incoming industrial wastewater to test its treatability under
140
realistic conditions and varying feed quality. The operation of the AnMBR was started
6
141
by the adaptation of the biomass consisting of slow filling the reactor with wastewater
142
in the first five days with minimum effluent. Afterwards the influent rate was increased
143
from 10 to 30 L·d–1 (in 5 L·d–1 steps, each approx. 30 d) in order to increase the organic
144
loading.
145
Phase 2: To avoid variations in the wastewater composition, batches of wastewater
146
(1 m3) were collected and used as influent, each for a period of several weeks. In total
147
three batches of wastewater were used during Phase 2 with COD between 3.24 and 7.54
148
g·L–1, methanol and ethanol being the main constituents. Methanol and WOS 9 (1–
149
7 g·L–1 as COD) were added to the wastewater in order to increase organic loading in a
150
controlled way and test the treatability of the WOS 9 as a co-substrate. In later stages of
151
Phase 2 gradually increasing amounts of methanol were added to the wastewater in
152
order to increase the incoming COD concentration and test the operation of the AnMBR
153
at high organic loading. The influent rate of wastewater was kept constant at 14.3 L·d–1.
154
Phase 3: In Phase 3 the feeding of the AnMBR was switched from the batches of
155
wastewater back to continuous dosing of the incoming industrial wastewater. The COD
156
concentration in the influent was increased by a constant amount of methanol and WOS
157
18 in order to keep the organic loading high. The goal of this phase was to test the
158
operation at generally high loading combined with variations in wastewater quality and
159
test the treatability of WOS 18 as a co-substrate.
160 161
2.4 Analytical methods
162
The samples of influent and permeate were automatically collected and kept at 4°C until
163
analysis. The concentrations of chemical oxygen demand (COD), total nitrogen,
164
phosphate phosphorus and sulfides in influent and permeate samples were regularly
7
165
measured using cuvette tests (Hach-Lange, Düsseldorf, Germany). Total solids (TS)
166
were analyzed according to the Standard methods (APHA, 2012). Gas chromatography
167
(5890 Series II, Hewlett-Packard, Palo Alto, CA, USA) was used to determine the
168
methane content in the biogas samples and to analyze the concentration of organic acids
169
in the liquid samples.
170 171
2.5 FISH analyses
172
The samples for the fluorescence in situ hybridization (FISH) analyses were fixed 1:1
173
(vol./vol) with ethanol, absolute. The content of methanogenic Archaea was monitored
174
using a FISH (Snaidr et al., 1997) test kit with standardized methodology and prepared
175
solutions (VIT® Methanogenic bacteria, vermicon AG, Germany). The test kit detects
176
methanogens affiliated with the Euryarchaeota, which cover the range of Archaea
177
reported to be responsible for methanogenesis (Liu and Whitman, 2008). In situ
178
hybridization was performed according to the manual of the VIT® Methanogenic
179
bacteria test kit. In addition, a set of specific oligonucleotide probes was used to analyse
180
the samples with respect to their content of main microbial phyla either participating in
181
anaerobic digestion or being present in anaerobic environmental samples (Weinberger
182
and Höötmann, 2011). For the identification of the Euryarchaeota detected, more
183
specific oligonucleotide probes were used which target methanogens on class, family
184
and even genus level (Raskin et al., 1994). Total cell counts were determined by
185
filtration of samples on polycarbonate filter membranes, followed by using DAPI (4',6-
186
diamidino-2-phenylindole) staining. Total viable cell counts were performed by
187
hybridizing the same filters using a mixture of oligonucleotide probes specific for the
188
domains Bacteria and Archaea (Amann et al., 1995, Daims et al., 1999).
8
189
Hybridized cells were excited and examined under a Zeiss Axiostar plus fluorescence
190
microscope. The percentage of area covered with target organisms in comparison to all
191
viable cells was analysed in at least ten randomly selected captured fields and an
192
average percentage of the populations was calculated. An eyepiece with crossline
193
graticules was used for cell counts and calculation of absolute numbers.
194 195
2.6 Respirometric tests
196
The biodegradability of WOS and their single components was evaluated by
197
respirometric degradation tests using the biomass samples from the AnMBR. The
198
degradation tests of WOS 9 were carried out with two biomass samples taken at the end
199
of Phase 1 (Day 166) and beginning of Phase 2 (Day 180) before the addition of WOS 9
200
into the AnMBR. Later during Phase 2 another mixture of organic solvents (WOS 18)
201
was collected and subjected to a new series of respirometric tests with five biomass
202
samples taken between days 250 and 287. Degradation tests of the industrial wastewater
203
collected at the same day as each biomass sample were carried out parallel to the tests
204
with WOS mixtures.
205
In order to assess possible inhibitory effects of the individual components of the
206
wastewater and WOS, two additional series of respirometric tests were carried out using
207
biomass samples from day 306 and 325 respectively. Ethanol, ethyl acetate, acetonitrile,
208
tetrahydrofuran, dichloromethane, n-heptane and toluene were tested as the sole
209
substrate (1.0 g·L–1 COD) and in mixtures (0.5 + 0.5 g·L–1 COD) with methanol, which
210
served as reference substrate.
211
The original methodology based on OECD guideline No. 311 (OECD, 2006) was
212
modified as follows. A mixture (50 mL) of dilution medium, substrate, tested sludge
9
213
and water was prepared to obtain a final COD concentration of 1.0 g·L–1 and TS of
214
1.5 g·L–1. The dilution medium containing phosphate buffer and trace elements was
215
prepared in accordance with OECD guideline No. 311. The flasks were incubated at
216
35°C and pressure readings were collected every 30 minutes. At the end of the test, the
217
methane content in the head-space was measured. Based on the data from pressure
218
measurement and gas analyses, the amount of methane produced in each incubation
219
flask was calculated using the ideal gas law. The specific methane production rate
220
(SMPR) was calculated from the pressure evolution curve assuming a linear dependence
221
between the amount of methane produced and pressure. The maximum SMPR was
222
identified from each test and used as a measure for biomass activity with given substrate.
223 224
3 Results and discussion
225
3.1 Treatment efficiency of the AnMBR
226
Phase 1
227
The AnMBR was continuously fed by incoming industrial wastewater with flow rate
228
increasing in steps from 10 to 30 L·d–1. The influent in this phase was characterized by
229
highly varying COD concentrations (1.50–10.6 g·L–1; average 4.30 g·L–1). The effluent
230
COD concentrations were between 0.240 and 1.70 g·L–1 (average 0.83 g·L–1) resulting
231
in a COD removal efficiency between 44 and 94% (78% on average) (Figure 2a). Most
232
of the effluent COD constituted non-degraded compounds from the original feed, since
233
the concentration of organic acids did not exceed 90 mg·L–1 (max. 16% of total COD).
234
The highest removal efficiencies above 90% were observed in the periods of volumetric
235
loading rate above 2.5 g COD L–1·d–1 – with peak concentrations in the influent (day
236
40–47 and day 66–73, COD concentrations 8.0–10.6 g·L–1) and at the end of phase 1,
10
237
when the organic loading was elevated by a higher wastewater flow. Similarly, the
238
lowest observed removal efficiency (44% on day 136) corresponded to the lowest COD
239
feed concentration. In general, the achieved loading rate of the AnMBR was relatively
240
low compared to values typically used in AnMBRs (2–15 g COD L–1·d–1) or
241
conventional high rate anaerobic reactors (4–40 g COD L–1·d–1) (Liao et al., 2006; Lin
242
et al., 2013), therefore the energy recovery of the full scale application would be low. In
243
order to improve the economic efficiency of the process, the following phases focused
244
on increasing the wastewater concentration by the addition of concentrated waste
245
organic solvents from production (WOS).
246 247
Phase 2
248
After switching to batch-wastewater the organic loading rate of the AnMBR was
249
increased compared to phase 1 by adding methanol to the influent (COD of the resulting
250
mixture 5.4–7.6 g·L–1). These modifications did not disturb the operation and the
251
AnMBR exhibited stable COD removal efficiency of 89–93% (Figure 2b). After 10
252
days, methanol was replaced by the WOS 9 as additional substrate (COD of the
253
resulting mixture 7.1–11.3 g·L–1). Following the addition of the WOS 9 the COD
254
removal dropped to 19% within 8 days and other signs of process imbalance were
255
observed – accumulation of organic acids (up to 480 mg·L–1 acetate) and subsequently a
256
decreasing pH in the bioreactor. Consequently, the WOS 9 was replaced by methanol
257
and reactor loading was decreased by reducing the amount of methanol added to the
258
influent (resulting COD 3.6–4.2 g·L–1). These operational parameters were maintained
259
for one month (until day 240), which led to successful recovery of the system.
11
260
Since the operational parameters and composition of the wastewater was constant
261
during phase 2, the inhibition can be attributed to the addition of WOS 9. The
262
degradation tests of WOS 9 were carried out before the addition of WOS 9 into the
263
AnMBR and they indicated lower SMPR compared to the industrial wastewater, but it
264
did not lead to complete inhibition of methane generation (Table 3). This finding
265
contradicted the operational experience of the AnMBR, where imbalance in methane
266
production and changes in microbial population were observed following the dosing of
267
WOS 9 (see also chapter 3.3). The reason might be the difference between
268
concentration of the WOS 9 in the respirometric test (1 g·L–1 COD) and in the AnMBR
269
(5–7 g·L–1 COD). Therefore, inhibition may only become visible at high concentrations.
270
Between days 240 and 350 the COD of the influent mixture increased from 7 to 44 g·L–
271
1
272
loading of the bioreactor rose from 1.8 to 12.7 g COD L–1·d –1. COD removal efficiency
273
after the recovery was mostly above 90%, except for two short disturbances around days
274
260 and 350. These probably resulted from too rapid an increase in incoming COD load
275
but the reactor performance rapidly stabilized (Figure 2b).
276
The results obtained during phase 2 confirmed that anaerobic treatment process of the
277
tested wastewater can be operated efficiently and at high energy yield, provided a well
278
degradable, non-inhibited co-substrate is available in a sufficient amount.
by the addition of methanol (up to 37 g·L–1 as COD); correspondingly the organic
279 280
Phase 3
281
Reactivated pumping of wastewater from the sewer directly to the AnMBR caused
282
variations in wastewater concentration but they were dampened by constant addition of
283
methanol (25 g·L–1 as COD) in the first 40 days of phase 3 and did not affect the COD
12
284
removal efficiency (91 – 97%, Figure 2c). Starting from day 409 WOS 18 was added to
285
the wastewater (8 g COD L–1 of WOS 18 + 17 g COD L–1 of methanol), which resulted
286
in rapid deterioration of effluent quality and the dosing of WOS 18 was interrupted.
287
After recovery of the degradation process the dose of WOS 18 was increased in steps
288
(1.25 – 5.0 g COD L–1). A rapid drop of COD removal to 60% was observed when the
289
dose of WOS 18 reached 5.0 g COD L–1. The operation continued with the addition of
290
2.5 g COD L–1 WOS 18, leading to varying treatment efficiency (67 – 93%). A third
291
drop in COD removal was observed between days 500 and 511 which was not preceded
292
by an increase in WOS 18 addition. However, the organic loading rate between days
293
480 and 495 underwent vigorous changes resulting from variations in wastewater
294
concentration connected with changes in the dosing rate of external substrates, which
295
might have induced additional stress on the biomass and resulted in a massive
296
accumulation of organic acids up to 1000 mg·L–1 on Day 502. Additionally, a failure of
297
the temperature control system occurred on day 515, when the bioreactor was
298
overheated to 45°C. After this event, generally low COD removal efficiency (< 75%)
299
was observed although the addition of WOS was stopped. The FISH analysis (chapter
300
3.3) of the biomass sample from day 579 confirmed the collapse of the methanogenic
301
population and the operation of the pilot unit was terminated.
302
Although the respirometric tests showed substantially better biodegradability of WOS
303
18 compared to WOS 9 (Table 3), low performance of the AnMBR was observed when
304
the dose of this mixture reached 8.0, 5.0 and 2.5 g·L–1 COD. Again, this may be due to
305
higher concentration used in the pilot experiment compared to the respirometric test.
306
Unlike phase 2 the composition of wastewater in phase 3 varied, which might have had
307
an additional influence on performance of the bioreactor besides the addition of WOS
13
308
18. The disturbance in COD removal around day 415 (WOS dose 8.0 g·L–1 COD),
309
coincided with an elevated concentration of tetrahydrofuran in the industrial wastewater
310
(influent concentration 800 mg·L–1). This is however below the reported concentration
311
of 11.5 g·L–1 leading to significant inhibition of anaerobic process (Yao et al., 2010).
312
Elevated levels of N,N-dimethylacetamide (up to 2500 mg·L–1) were present in the
313
influent during the second disturbance in COD removal around day 460, which is
314
comparable to the value (>2000 mg·L–1) reported as having a negative effect on
315
microorganisms (EC10 value for activated sludge and EC50 for bacteria, respectively)
316
(Merck KGaA, 2013). Even though the presence of more than 2000 mg·L–1 N,N-
317
dimethylacetamide spanned a broader period than the observed disturbance, a
318
combination with the influence of the WOS 18 cannot be excluded.
319
Overall, the results showed that the WOS samples tested are not suitable as a co-
320
substrate for the full-scale treatment of industrial wastewater, since they can lead to
321
unpredictable breakdown of the process.
322 323
3.2 Organic acids
324
The total concentration of organic acids in the effluent from AnMBR was usually below
325
150 mg·L–1, with acetate being the dominant acid. Accumulation of organic acids
326
indicating anaerobic digestion imbalance was observed on a few occasions as a result of
327
the addition of WOS 9 (days 190–210), addition of WOS 18 combined with loading
328
variations (days 498–510) and reactor overheating (days 515–530) (Figure 3). Low
329
concentration of organic acids < 70 mg·L–1 was observed during the increasing loading
330
between days 230 and 400, even during the peak volumetric loading of 12 g COD L–
331
1
·d–1 and two disturbances in COD removal around days 260 and 350. This may be due
14
332
to a low share of acetoclastic metabolism in the bioreactor: Bhatti et al. (1996)
333
identified three main pathways governing the fate of methanol during anaerobic
334
digestion – acetoclastic methanogenesis through formation of acetate, hydrogenotrophic
335
methanogenesis through formation of hydrogen and direct methylotrophic conversion
336
into methane by Methanosarcina type species. The FISH analyses of biomass samples
337
confirmed that Methanosarcina were the dominant type of methanogenic population in
338
the AnMBR, although the type of metabolism could not be identified by this analysis,
339
since the group can use a variety of substrates and metabolic pathways.
340
Although the amount of methane formed by the methylotrophic metabolism does not
341
usually exceed 10% (Jiang et al., 2009), the low accumulation of organic acids during
342
the breakthrough events suggests that this type of metabolism might have prevailed in
343
the specialized biomass established after inhibition at the beginning of Phase 2, which
344
was fed with a uniform substrate.
345
On day 515 the bioreactor overheated to 45 °C due to failure of the temperature control
346
system. The elevated temperature led to another peak in concentration of organic acids
347
(up to 400 mg·L–1). The concentration of organic acids returned to below 200 mg·L–1
348
soon after the repair of the temperature control, however it remained at an elevated level
349
compared to previous phases (around 150 mg·L–1) until the end of the operation.
350 351
3.3 FISH analyses
352
Ten biomass samples from the AnMBR collected during phases 2 and 3 were analyzed
353
by FISH. The results showed a massive drop in methanogenic organisms between day
354
173 and day 201, which framed the period with WOS 9 dosing. This corresponded to
355
the breakdown observed in methane production and indicated a toxic effect of this
15
356
substrate towards the methanogens. After the dosing of WOS 9 was stopped (day 193),
357
the counts of methanogens recovered and in the sample from day 229 reached the
358
original level (Figure 4). A population shift within the methanogenic organisms was
359
observed. While the first sample contained a mixture (around 50:50) of rod-shaped and
360
cluster-forming methanogens, both types (especially the rod-shaped organisms) were
361
suppressed in samples from days 194 and 201 and starting from day 215 only the
362
cluster-forming methanogens were present in the biomass. Genus-specific probes
363
applied in samples from day 327 on identified this population as Methanosarcina. The
364
genus Methanosarcina is a common methanogen in anaerobic digesters. Several studies
365
found them to be a dominant methanogen, e.g., in reactors fed by sewage sludge (Liu et
366
al., 2016) or pharmaceutical wastewater (Chelliapan et al., 2011). Due to the versatility
367
of their metabolism and overall robustness they also dominate in anaerobic bioreactors
368
under transition conditions (De Vrieze et al., 2012).
369
The counts of methanogens roughly followed the organic loading trend of the reactor
370
and reached their peak on day 327. The last sample from day 579 contained a very low
371
share of viable cells (6% compared to 27–42% in the remaining samples) including
372
methanogens. This corresponds to the poor reactor methane production observed in that
373
period.
374
Changes within the population of other microbial groups in the biomass samples were
375
also observed during the period with the dosing of WOS 9. Whereas Firmicutes,
376
Crenarchaeota and Actinobacteria disappeared or their counts were suppressed during
377
the addition of WOS, Chloroflexi, Cytophaga-Flexibacter subphylum and
378
Alphaproteobacteria were found in higher amounts at the end of the first sampling
379
campaign. This indicates that the addition of WOS 9 affected only some microbial
16
380
groups, whereas the microorganisms less sensitive to the substances present rapidly
381
occupied the vacant habitat. The disappearance of Firmicutes and Actinobacteria, which
382
are connected with acidogenesis and acetogenesis (Levén et al., 2007), was remarkable.
383
This was probably the result of the lack of suitable substrates, since the AnMBR
384
influent contained few organic compounds other than methanol at that time.
385
In samples from day 327 onwards, Firmicutes and other bacteria were detected again in
386
gradually increasing numbers, presumably because more diversified substrate was
387
added to the AnMBR in phase 3. Relatively few viable cells were found in the last
388
sample from day 579 (after reactor overheating) with Firmicutes forming 22% of all
389
cells. Such an evolution is usually a sign of a collapsing reactor performance needing
390
new inoculation, which was also the case in this study.
391
A decreasing trend was observed for hydrolysis bacteria (Chloroflexi and Cytophaga-
392
Flexibacter-subphylum) after the population shift (from day 215 on), presumably due to
393
the specific type of substrates in the reactor, which had low hydrolytic requirements.
394 395
3.4 Biogas production and quality
396
The biogas produced contained about 40–70% methane, on average about 60%. The
397
amount of biogas formed was between 0.5 and 0.6 L·g–1 COD removed throughout all
398
phases of operation, corresponding to 0.30–0.36 L methane per g COD. Based on the
399
equilibrium calculation of known sulfide content in the effluent, the hydrogen sulfide
400
concentration in the gas was mostly below 0.03 g·m–3, but would reach approximately
401
14 g·m–3 (1% v/v) during the peak sulfide concentration detected.
402 403
3.5 Total solids and membrane performance
17
404
The total solids (TS) in the bioreactor roughly followed the trend of the organic loading
405
rate and varied between 9 and 23 g·L–1 (Figure 5). Despite faster sludge wasting
406
(corresponding SRT approx. 120 d) the concentration of TS roughly doubled from 11 to
407
23 g·L–1 during phase 2. The effect of accentuated biomass growth resulting from
408
higher organic loading probably combined with the introduction of suspended solids
409
(approx. 0.5 g·L–1) contained in the batch of wastewater used between days 290 and 366.
410
The membrane was presumably carrying a layer of foulants from the previous
411
experiments (data not shown) and a rapid increase in TMP was observed following the
412
inoculation so that chemical cleaning was necessary after 90 days of operation. In the
413
following more than 300 days only a minor decrease of permeability was observed
414
without any chemical cleaning of the membrane and despite the rapid increase of TS.
415
No fouling problems were observed up to 25 g·L–1 TS (Figure 5).
416
On the other hand, the membrane permeability decreased in the last 80 days of phase 3,
417
triggered by difficulties with maintaining the desired cross-flow velocity (3 m·s–1)
418
rather than by changes in the structure of the mixed liquor. Since the AnMBR had no
419
automatic control of the retentate flow, the filtration was repeatedly operated at low
420
cross-flow velocities close to dead-end mode for a period of several hours. This
421
probably led to accumulation of sludge in the membrane channels, as revealed by the
422
inspection of membranes at the end of the operation.
423 424
3.6 Degradation tests with the components of WOS
425
In the two series of degradation tests carried out with the components of the WOS, pure
426
methanol was the substrate with highest degradation rate (SMPR) (Figure 6), due to the
427
pre-adaptation of the biomass to this substrate. Degradation rate was substantially lower
18
428
when methanol was mixed with other substrates including well degradable organic
429
compounds (ethanol, ethyl acetate). Dichloromethane was the only substrate which
430
completely inhibited anaerobic degradation in mixture with methanol at the
431
concentration tested (0.5 g·L–1 COD, i.e. 0.88 g·L–1). This concentration was
432
comparable to that in the AnMBR influent during phase 2 (530 mg·L–1) and phase 3
433
(110 mg·L–1) and higher than the 50% inhibition concentration (IC50) in a UASB reactor
434
reported by Ozdemir et al. (2010) – 43 mg·L–1. Thus, dichloromethane could be
435
responsible for the observed breakdowns in COD removal in the AnMBR.
436
A relatively low degradation rate was observed for toluene. Toluene might have played
437
a more important role in the inhibition during phase 2 when the dosing of WOS 9 raised
438
its concentration in the influent to 160 mg·L–1. The same amount was present in the
439
respirometric test with low SMPR when degrading the toluene/methanol mixture and
440
320 mg·L–1 in the test with pure toluene. Inhibition (IC50) of anaerobic digestion was
441
reported in a similar range between 146 and 1000 mg·L–1 by Orellana (2000) and
442
Kayembe (2013). On the other hand, the concentration of toluene during the dosing of
443
WOS 18 did not exceed 30 mg·L–1.
444
In one test the degradation of acetonitrile mixed with methanol (Figure 6a) was low
445
compared to other substrates. However, acetonitrile presumably had no effect on the
446
breakdown observed in AnMBR performance since its concentration in the influent was
447
well below the dose in the degradation tests and did not exceed 30 mg·L–1.
448
Besides organic compounds, some inorganic substances, e.g. free ammonia or high
449
concentrations of salts in general, are known to inhibit anaerobic digestion (Chen et al.,
450
2008). However, they presumably had a low effect on the inhibition of COD removal in
451
the AnMBR, since the total nitrogen concentration in the bioreactor (< 300 mg·L–1) was
19
452
below the reported inhibitory limits of 1700 mg N-HH4 + L–1 (Yenigün and Demirel,
453
2013), while the salt content of 0.7 – 6 g·L–1 in the AnMBR was below the inhibitory
454
values of >13 g·L–1 reported for non-acclimated sludge (Lefebvre et al., 2007; Lefebvre
455
and Moletta, 2006).
456
Overall, it was not clear why the WOS samples, which showed reasonable
457
biodegradability in the single tests, had such a detrimental influence on the performance
458
of the AnMBR. Besides the influence of the solvents discussed above, other
459
components that are not detectable by GC – analysis in the WOS might also be
460
responsible for the negative effects. For a commercial operation, it would be important
461
to evaluate this in more detail, because the use of WOS is a key point in increasing the
462
biogas yield as well as in lowering the disposal cost for WOS.
463 464
4 Conclusions
465
The industrial wastewater alone showed good anaerobic degradability in the AnMBR,
466
however its low concentration and varying composition decreased the treatment
467
efficiency. Constant and high removal efficiency (>90%) was reached when methanol
468
was added to the influent and variations of its composition were limited. Acidification,
469
poor methane production and shifts in methanogenic population were experienced when
470
waste organic solvents were added to the influent as co-substrate. Therefore, a
471
commercially viable anaerobic treatment of the current wastewater and WOS mixtures
472
seems not possible at the moment. Well-controlled cross-flow conditions in the
473
membrane channels were necessary for stable filtration and low fouling.
474 475
Acknowledgements
20
476
We are very grateful to F. Hoffmann-La Roche for financial support, experiments and
477
access to analytical results. ProRheno AG is kindly thanked for kind cooperation and
478
support. Thanks also to students and technicians of the FHNW for pilot plant
479
supervision, analytics and technical support. Project LO1201 „National Programme for
480
Sustainability I“, and the OPR&DI project No. CZ.1.05/2.1.00/01.0005 is also
481
acknowledged.
482 483
References
484
1.Amann RI, Ludwig W, Schleifer KH. 1995. Phylogenetic identification and in situ
485
detection of individual microbial cells without cultivation. Microbiol Rev. 59(1):143-69.
486
2.APHA, 2012. Standard Methods for the Examination of Water and Wastewater, 22
487
edition. ed. American Water Works Assn, Washington, DC.
488
3.Araya, P., Aroca, G., Chamy, R., 1999. Anaerobic treatment of effluents from an
489
industrial polymers synthesis plant. Waste Management 19, 141–146.
490
doi:10.1016/S0956-053X(99)00013-6
491
4.Bhatti, Z.I., Furukawa, K., Fujita, M., 1996. Feasibility of methanolic waste treatment
492
in UASB reactors. Water Research 30, 2559–2568. doi:10.1016/S0043-1354(96)00144-
493
3
494
5.Chelliapan, S., Wilby, T., Yuzir, A., Sallis, P.J., 2011. Influence of organic loading on
495
the performance and microbial community structure of an anaerobic stage reactor
496
treating pharmaceutical wastewater. Desalination 271, 257–264.
497
doi:10.1016/j.desal.2010.12.045
21
498
6.Chen, Z., Ren, N., Wang, A., Zhang, Z.-P., Shi, Y., 2008. A novel application of
499
TPAD–MBR system to the pilot treatment of chemical synthesis-based pharmaceutical
500
wastewater. Water Research 42, 3385–3392. doi:10.1016/j.watres.2008.04.020
501
7.Choi, W.-H., Shin, C.-H., Son, S.-M., Ghorpade, P.A., Kim, J.-J., Park, J.-Y., 2013.
502
Anaerobic treatment of palm oil mill effluent using combined high-rate anaerobic
503
reactors. Bioresource Technology, Challenges in Environmental Science and
504
Engineering (CESE-2012) 141, 138–144. doi:10.1016/j.biortech.2013.02.055
505
8.Daims, H., A. Brühl, R. Amann, K.-H. Schleifer, and M. Wagner. 1999. The domain-
506
specific probe EUB338 is insufficient for the detection of all Bacteria: development and
507
evaluation of a more comprehensive probe set. Syst. Appl. Microbiol. 22:434–444.
508
9.De Vrieze, J., Hennebel, T., Boon, N., Verstraete, W., 2012. Methanosarcina: The
509
rediscovered methanogen for heavy duty biomethanation. Bioresource Technology 112,
510
1–9. doi:10.1016/j.biortech.2012.02.079
511
10.Delforno, T.P., Moura, A.G.L., Okada, D.Y., Varesche, M.B.A., 2014. Effect of
512
biomass adaptation to the degradation of anionic surfactants in laundry wastewater
513
using EGSB reactors. Bioresource Technology 154, 114–121.
514
doi:10.1016/j.biortech.2013.11.102
515
11.Dereli, R.K., Ersahin, M.E., Ozgun, H., Ozturk, I., Jeison, D., van der Zee, F., van
516
Lier, J.B., 2012. Potentials of anaerobic membrane bioreactors to overcome treatment
517
limitations induced by industrial wastewaters. Bioresource Technology, Membrane
518
Bioreactors (MBRs): State-of-Art and Future 122, 160–170.
519
doi:10.1016/j.biortech.2012.05.139
520
12.Diez, V., Ramos, C., Cabezas, J.L., 2012. Treating wastewater with high oil and
521
grease content using an Anaerobic Membrane Bioreactor (AnMBR). Filtration and
22
522
cleaning assays. Water Science and Technology 65, 1847–1853.
523
doi:10.2166/wst.2012.852
524
13.dos Santos, A.B., Cervantes, F.J., van Lier, J.B., 2007. Review paper on current
525
technologies for decolourisation of textile wastewaters: Perspectives for anaerobic
526
biotechnology. Bioresource Technology 98, 2369–2385.
527
doi:10.1016/j.biortech.2006.11.013
528
14.Dvořák, L., Gómez, M., Dolina, J., Černín, A., 2015. Anaerobic membrane
529
bioreactors—a mini review with emphasis on industrial wastewater treatment:
530
applications, limitations and perspectives. Desalination and Water Treatment 0, 1–15.
531
doi:10.1080/19443994.2015.1100879
532
15.Gao, W.J.J., Lin, H.J., Leung, K.T., Liao, B.Q., 2010. Influence of elevated pH
533
shocks on the performance of a submerged anaerobic membrane bioreactor. Process
534
Biochemistry 45, 1279–1287. doi:10.1016/j.procbio.2010.04.018
535
16.Jiang, N., Wang, Y., Dong, X., 2009. Methanol as the primary methanogenic and
536
acetogenic precursor in the cold zoige wetland at tibetan plateau. Microbial Ecology 60,
537
206–213. doi:10.1007/s00248-009-9602-0
538
17.Kayembe, K., 2013. Comparative study of benzyl and phenyl compounds chemical
539
structure effects on the inhibition of methane production by digested pig manure
540
methanogens. International Research Journal of Pure and Applied Chemistry 3, 48–58.
541
doi:10.9734/IRJPAC/2013/2489
542
18.Lefebvre, O., Moletta, R., 2006. Treatment of organic pollution in industrial saline
543
wastewater: A literature review. Water Research 40, 3671–3682.
544
doi:10.1016/j.watres.2006.08.027
23
545
19.Lefebvre, O., Quentin, S., Torrijos, M., Godon, J.J., Delgenès, J.P., Moletta, R., 2007.
546
Impact of increasing NaCl concentrations on the performance and community
547
composition of two anaerobic reactors. Applied Microbiology and Biotechnology. 75,
548
61–69. doi:10.1007/s00253-006-0799-2
549
20.Levén, L., Eriksson, A.R.B., Schnürer, A., 2007. Effect of process temperature on
550
bacterial and archaeal communities in two methanogenic bioreactors treating organic
551
household waste. FEMS Microbiology Ecology 59, 683–693. doi:10.1111/j.1574-
552
6941.2006.00263.x
553
21.Lew, B., Tarre, S., Beliavski, M., Dosoretz, C., Green, M., 2009. Anaerobic
554
membrane bioreactor (AnMBR) for domestic wastewater treatment. Desalination 243,
555
251–257. doi:10.1016/j.desal.2008.04.027
556
22.Liao, B.-Q., Kraemer, J.T., Bagley, D.M., 2006. Anaerobic membrane bioreactors:
557
applications and research directions. Critical Reviews in Environmental Science and
558
Technology 36, 489–530. doi:10.1080/10643380600678146
559
23.Lin, H., Peng, W., Zhang, M., Chen, J., Hong, H., Zhang, Y., 2013. A review on
560
anaerobic membrane bioreactors: Applications, membrane fouling and future
561
perspectives. Desalination 314, 169–188. doi:10.1016/j.desal.2013.01.019
562
24.Liu Y, Whitman WB. 2008. Metabolic, phylogenetic, and ecological diversity of the
563
methanogenic archaea. Ann N Y Acad Sci. 1125:171-89.
564
25.Liu, C., Li, H., Zhang, Y., Si, D., Chen, Q., 2016. Evolution of microbial community
565
along with increasing solid concentration during high-solids anaerobic digestion of
566
sewage sludge. Bioresource Technology 216, 87–94.
567
doi:10.1016/j.biortech.2016.05.048
568
26.Merck KGaA, 2013. N,N-Diethylacetamide for synthesis. Safety data sheet.
24
569
27.Ng K.K., Shi X., Tang M.K.Y., Ng H.Y. 2014. A novel application of anaerobic bio-
570
entrapped membrane reactor for the treatment of chemical synthesis-based
571
pharmaceutical wastewater. Separation and Purification Technology 132, 634–643.
572
doi:10.1016/j.seppur.2014.06.021
573
28.Ng, K.K., Shi, X., Ng, H.Y., 2015. Evaluation of system performance and microbial
574
communities of a bioaugmented anaerobic membrane bioreactor treating pharmaceutical
575
wastewater. Water Research 81, 311–324. doi: 10.1016/j.watres.2015.05.033
576
29.OECD, 2006. Test No. 311: Anaerobic Biodegradability of Organic Compounds in
577
Digested Sludge: by Measurement of Gas Production. OECD, Paris.
578
30.Orellana, R.I., 2000. Effects of various substrates on anaerobic biodegradation of
579
benzene and toluene. PhD thesis, The George Washington University, Washington, DC.
580
31.Ozdemir, C., Sen, N., Dursun, S., Kalipci, E., 2010. Removal of dichloromethane in
581
up-flow anaerobic sludge bed reactors and methane production. Asian Journal of
582
Chemistry 22, 6423–6436.
583
32.Raskin L, Stromley JM, Rittmann BE, Stahl DA. 1994. Group-specific 16S rRNA
584
hybridization probes to describe natural communities of methanogens. Appl Environ
585
Microbiol. 60(4):1232-40.
586
33.Savant, D.V., Abdul-Rahman, R., Ranade, D.R., 2006. Anaerobic degradation of
587
adsorbable organic halides (AOX) from pulp and paper industry wastewater.
588
Bioresource Technology 97, 1092–1104. doi:10.1016/j.biortech.2004.12.013
589
34.Snaidr J, Amann R, Huber I, Ludwig W, Schleifer KH. 1997. Phylogenetic analysis
590
and in situ identification of bacteria in activated sludge. Appl Environ Microbiol.
591
63(7):2884-96.
25
592
35.Spagni, A., Casu, S., Crispino, N.A., Farina, R., Mattioli, D., 2010. Filterability in a
593
submerged anaerobic membrane bioreactor. Desalination 250, 787–792.
594
doi:10.1016/j.desal.2008.11.042
595
36.Tao, Y., Gao, D.-W., Fu, Y., Wu, W.-M., Ren, N.-Q., 2012. Impact of reactor
596
configuration on anammox process start-up: MBR versus SBR. Bioresource
597
Technology 104, 73–80. doi:10.1016/j.biortech.2011.10.052
598
37.Weinberger G, Höötmann U. 2011. Wastewater treatment plants – a look into a black
599
box. IPW 11-12/2011:18-21. www.ipwonline.de.
600
38.Wijekoon, K.C., Visvanathan, C., Abeynayaka, A., 2011. Effect of organic loading
601
rate on VFA production, organic matter removal and microbial activity of a two-stage
602
thermophilic anaerobic membrane bioreactor. Bioresource Technology, Special Issue on
603
Challenges in Environmental Science and Engineering, CESE-2010: Technological
604
Advances in Waste Treatment for a Sustainable Future 102, 5353–5360.
605
doi:10.1016/j.biortech.2010.12.081
606
39.Yao, Y., Guan, J., Tang, P., Jiao, H., Lin, C., Wang, J., Lu, Z., Min, H., Gao, H.,
607
2010. Assessment of toxicity of tetrahydrofuran on the microbial community in
608
activated sludge. Bioresource Technology 101, 5213–5221.
609
doi:10.1016/j.biortech.2010.02.051
610
40.Yenigün, O., Demirel, B., 2013. Ammonia inhibition in anaerobic digestion: A
611
review. Process Biochemistry 48, 901–911. doi:10.1016/j.procbio.2013.04.012
612
26
613
Figure 1 Flow scheme of the pilot AnMBR.
614 615
27
616
Figure 2 COD concentration and COD removal during phase 1 (a), phase 2 (b) and
617
phase 3 (c). The areas in the diagram indicate the target dose of methanol (grey) and
618
WOS (yellow), both expressed as COD.
619
620
621 622 28
623
Figure 3 Organic acids in the effluent and organic loading of the AnMBR. The yellow
624
area in the diagram indicates target dose (expressed as COD) of WOS 9 (Phase 2) and
625
WOS 18 (Phase 3), respectively.
626 627
29
628
Figure 4 Comparison of biomass samples according to the absolute cell counts per mL
629
of each population.
630 631
30
632
Figure 5 Permeability (K) and total solids over AnMBR operation.
633 634
31
635
Figure 6 Maximum SMPR during respirometric tests of anaerobic biodegradability
636
with individual components of the WOS mixed with methanol (a) and as sole substrates
637
(b). Respirometric tests a) and b) were conducted with two different sludge samples.
638
MeOH – methanol, CH2Cl2 – dichloromethane, THF – tetrahydrofuran.
639
640 641
32
642
Table 1 Composition of WOS 9 and WOS 18 added to the pharmaceutical wastewater
643
during Phases 2 and 3. Parameter pH Water Methanol Ethanol Acetone Dichloromethane Ethyl acetate Tetrahydrofuran n-Heptane Toluene Ammonia nitrogen
WOS 9 (Phase 2) 11.3 14.7% 62.5% 6.4% 0.5% 6.9% 2.8% 0.4% – 2.1% 2.7%
WOS 18 (Phase 3) 10.9 6.0% 73.6% 2.0% 9.6% 1.0% 3.0% 4.4% 0.5% – –
644 645
33
646
Table 2 Technological parameters during different operational phases. Duration Wastewater Additional Phase [d] collection substrate 1
175
2
191
3
214
Sludge Volumetric HRT concentration loading [d] [g TS·L–1] [g COD·L–1·d–1]
no 7.3–13.1 co-substrate methanol, Batch 10.0–24.0 WOS 9 methanol, Continuous 18.3–26.4 WOS 18 Continuous
0.6–4.0
1.7–5
1.0–12.7
3.5
4.5–14.2
3–3.5
647 648
34
649
Table 3 Maximum SMPR during respirometric tests of anaerobic degradability with
650
WOS 9, WOS 18 and wastewater. Substrate
WOS 9
WOS 18
Wastewater
Maximum SMPR [mg COD g-1 h-1 ]
10.5 ± 3.5
37.9 ± 8.8
47.4 ± 6.0
651 652
35
653
Highlights
654
1.
Wastewater from pharmaceutical industry can be treated by anaerobic MBR technology.
655
2.
Varying wastewater composition had negative influence on the treatment efficiency.
656
3.
Addition of waste organic solvents caused inhibition of the anaerobic degradation.
657
4.
Dichloromethane exhibited the strongest inhibitory effect on anaerobic degradation.
658
5.
Changes in the biocenosis reflected the substrate pattern and inhibition events.
659 660
661
36