Anaerobic Digestion of Waste Water from

0 downloads 0 Views 705KB Size Report
such as digestate from biogas plants [1] and sewage sludge [2]. Besides the ..... leading to insoluble iron(II) sulfide (FeS) and iron(II) phosphate. (Fe3(PO4)2).
Applied Bioenergy Research Article • DOI: 10.2478/apbi-2013-0001 • APBI • 2013 • 1–10

Anaerobic Digestion of Waste Water from Hydrothermal Carbonization of Corn Silage Abstract This experimental work investigates anaerobic digestion of waste water from hydrothermal carbonization of maize silage comparing a continuously stirred-tank reactor (CSTR) and an anaerobic filter (AF). Both reactors were operated for 91 days at a constant organic loading rate of 1 gCOD L-1 d-1. During the first five weeks of operation both reactors showed a removal efficiency of the chemical oxygen demand of up to 80 % and a methane production rate of up to 0.25 L L-1 d-1. Consecutively lower degradation rates were assumed to be caused by a significant lack of sulfur and phosphorus due to a precipitation by ferrous iron. Over the whole time the AF proved to be more stable. Very small concentrations of phenol compounds contained in the waste water were nevertheless degraded by up to 80 %.

Benjamin Wirth1,2*, Leibniz Institute for Agricultural Engineering Potsdam-Bornim, Max-Eyth-Allee 100, 14469 Potsdam, Germany

1

Jan Mumme1 Technische Universität Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany

2

Keywords Hydrothermal carbonization • Anaerobic digestion • Waste water treatment • Biogas • Phenols ©2  013 Benjamin Wirth, Jan Mumme, licensee Versita Sp. z o. o.

Received 18 July 2013 Accepted 23 October 2013

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs license, which means that the text may be used for non-commercial purposes, provided credit is given to the author

1. Introduction Hydrothermal carbonization (HTC) is an artificial coalification process converting raw biomass into a coal-like product, socalled hydrochar. HTC is recently promoted as an upgrading technology for highly wet biomass and residual biomass streams such as digestate from biogas plants [1] and sewage sludge [2]. Besides the solid phase, it further produces a high-strength liquid phase and a gaseous phase mainly containing carbon dioxide. HTC waste water usually contains high amounts of dissolved organic compounds resulting in a chemical oxygen demand (COD) of 10 to 40 g L-1 [3]. The total organic carbon (TOC) content of HTC liquor usually ranges from 5 to 20 g L-1 [3,4]. In case of recirculating process water, the TOC can amount up to 40 g L-1 [5]. In respect to anaerobic digestion, the degradation speed of HTC products should not be limited by hydrolysis due to the fact that only minor concentrations of complex organic matter can be found on hydrochar and therefore in HTC liquor [6]. Consequently, a relatively fast degradation can be assumed providing the absence of a process inhibition. This degradation should be dominated by acetogenesis and acetate-consuming methanogenesis [7] due to its high content of formic acid and acetic acid. Phenolic compounds can often be found in waste water streams and can cause serious health and environmental risks [8,9]. The general idea of combining the HTC process with the process of anaerobic digestion was already described by [1] and

[2]. [10] report of specific biogas yields of 1.9 to 22.8 mL mL-1 of HTC waste water obtained with mesophilic batch experiments. The gas yields are highly different depending on the feedstock used for HTC and the resulting TOC within the waste water. These numbers are in accordance with a recent study by [11]. They report of maximum biogas yields of up to 21.1 L kgFM-1 also obtained with mesophilic batch experiments. The aim of this study was to analyze the feasibility of treating HTC waste water by anaerobic digestion. First of all, the general feasibility and long-term stability of the anaerobic digestion of HTC waste water as a sole substrate was tested by experimental means. The focus was placed on the process performance and the potentially inhibitory effect of phenols and other organic compounds usually contained in HTC liquor. Furthermore, in scope of the experiments two different reactor types were compared – a completely stirred-tank reactor (CSTR) and an anaerobic filter (AF).

2. Materials and Methods 2.1 Substrate Properties The HTC liquor used in this study was obtained from the company AVA-CO2, Switzerland from a HTC technical scale plant located at Karlsruhe, Germany. It is operated as a batch process. The batch from which the used HTC liquor originates was produced from corn silage treated at 220 °C and a residence time of six hours.

* E-mail: [email protected]

Bereitgestellt von | Universitaet Potsdam 1 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

B. Wirth, J. Mumme

Approximately 90 L of the liquor were homogenized and transferred into metal gasoline canisters with a capacity of 20 L each. One of the canisters was stored at 4 °C whereas the remaining substrate was frozen at -20 °C to prevent chemical changes. A sample of the liquor was taken before transferring it into the metal containers. Its chemical parameters are displayed in Table 1. The analytical methods are described in Section 2.4. The amount of acetic acid accounts for approximately 13 % of the overall TOC content and is mainly responsible for the low pH-value. The amount of propionic acid is at a critical level compared to inhibitory values within a biogas reactor [12]. Another sample taken from a metal storage canister after 14 weeks of storage (8 weeks at −20 °C, 6 weeks at 4 °C) showed only a slight decrease in TOC (3.8 %) and COD (2.8 %), respectively. The phenol sum parameter of 0.29 g L-1 indicates the presence of phenolic compounds, even though on a relatively low level. Nutrients that are important for the biogas process are contained in sufficient concentrations [13-15]. In respect to trace elements, a relatively high concentration of ferrous iron of 500 mg L-1 appears notable. Phenols and cresols were also analyzed but their concentrations were only in the range of a few micrograms (110 and 7 µg L-1, respectively).

2.2 Experimental Set-Up For a quick start-up of both reactors active inoculum was obtained from a biogas plant processing energy crops and solid manure at mesophilic conditions located near Potsdam, Germany. The plant is operated at an organic loading rate (OLR) of 2.2 to 2.5 kgVS m-3 d-1 and its methane generation rates are approximately 368 L kgVS-1 high with an average methane concentration of 58 %

of the produced biogas. The liquid digestate used as inoculum was extracted from the post-treatment reactor. Afterwards, in order to reduce its viscosity, the digestate was sieved two times and further diluted 1:1 with untreated tap water. The final inoculum used for reactor start-up had a TS content of 2.53 %wt, a VS content of 70.07 %TS (1.77 %wt), and a COD of 30.54 g L-1. The ammonia nitrogen value of the initial inoculum was with a value of 2500 mg L-1 at the upper limit for biogas production from industrial waste water [16]. The experimental reactor set-up built up for this study mainly consisted of two glass reactors – a continuously stirred-tank reactor (CSTR) and an anaerobic filter (AF). Both reactors had a maximum volume of 3.8 L each and a heated water jacket (Figure 1). They were perfused by the same heat bath (Thermo Fisher Scientific, Inc., USA) ensuring a constant temperature of 37 °C within the reactors. In respect to the CSTR, homogenous stirring was obtained by two pairs of plastic paddles positioned at the top and bottom of the reactor, respectively. The stirrer was propelled by an electric engine (IKA-Werke GmbH & Co. KG, Germany) set to 50 rotations per minute (rpm). For the growth of biofilms, the AF was equipped with a packed-bed of 450 individual biofilm carriers made of plastic. These carriers (RVT Process Equipment GmbH, Germany) can be characterized by a high specific surface area of 437 m2 m-3. Caused by the own volume of the carriers the hydraulic working volume in the AF was 200 mL lower than in the CSTR. To avoid floating of the biofilm carriers a sieve made of steel was mounted on top of the packed-bed. This sieve was immersed the whole time. The reactor liquor was circulated by a peristaltic pump

Table 1.  Overview of the chemical properties of the obtained HTC liquor used in this study.

Parameter

Unit

Value

1

3.88

Total solids (TS)

%wt

2.80

Volatile solids (VS)

%TS

79.06

Total organic carbon (TOC)

g L-1

15.66

Chemical oxygen demand (COD)

g L-1

41.35

Acetic acid

g L-1

5.26

Propionic acid

g L-1

0.34

Phenol sum parameter

g L-1

0.29

Sulfur

mg L-1

90.80

Phosphorus

mg L-1

197.40

Ammonia nitrogen

mg L-1

229.50

Total Kjeldahl nitrogen (TKN)

mg L-1

685.50

PH-value

Figure 1. Schematic of the experimental set-up of this study (heat bath circulation not shown).

Bereitgestellt von | Universitaet Potsdam 2 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

Anaerobic Digestion of Waste Water from Hydrothermal Carbonization of Corn Silage

(Heidolph Instruments GmbH & Co. KG, Germany) from top to bottom of the reactor at a rate of ten replacements per day. For operation and control purposes, both reactor heads were equipped with an immersed feeding tube, a gas outlet, and sensors for pH and temperature. The gas flow was measured by a rotary drum gas meter and stored in foil bags for subsequent gas analysis. Both reactors were covered with a thin layer of insulation foil to reduce heat loss and to prevent the process from light irradiation. The complete experimental set-up is displayed in Figure 1.

2.3 Conducting the Experiment For start-up, both reactors were fed with dextrose monohydrate two times per week at an organic loading rate of 0.5 gVS L-1 d-1. Additionally, the reactors were supplied with a trace element solution. The trace element solution was mixed according to medium No. 144 of the “German collection of microorganisms and cell cultures” (Brunswick, Germany). In contrast to the original recipe, a fivefold concentrated solution was prepared. The solution was diluted 1:10 with distilled water. The final solution contained the following compounds (concentrations in mg L-1 are given in brackets): N(CH2COOH)3 (6400), FeCl3 (100), MnCl2 (50), CoCl2 (90), CaCl2 (50), ZnCl2 (50), CuCl2 (10), H3BO3 (10), Na2MoO4 (10), NiCl2 (10), NaCl (500), and Na2SeO3 (10). It was added in a quantity of 560 µL two times a week when feeding the dextrose. Two weeks before starting to add HTC liquor feeding of dextrose was stopped followed by a desired decrease of gas production. The main experiment was divided into two main phases of operation. There was a shock-load addition of HTC liquor at first followed by a period of continuous feeding. With the aim to get first results on the degradation speed and potential inhibition the shock-load test was conducted with a HTC waste water to process liquor ratio of 1:20. Afterwards, during the main phase of the experiment, the reactors were continuously operated for 13 weeks. Both reactors were fed daily at an OLR of 1 gCOD L-1 d-1. As the hydraulic working volume of both reactors fluctuated because digestate was only removed once a week, the added amount of HTC liquor varied between 67.7 and 77.8 mL. As a result, this led to a hydraulic retention time (HRT) of approximately 42 days. The liquid digestate was analyzed every week for its chemical characteristics.

2.4 Analytical Methods During the experiment the pH-value and the reactor temperature were monitored online. This is done with combined pH and temperature probes (Mettler-Toledo GmbH, Germany) attached to a data logger (WTW GmbH, Germany). The calibration chosen for the probes was a two-point calibration with pH-values of 4 and 7. The produced biogas is quantified by a rotary drum gas meter TG05 (Ritter Apparatebau GmbH & Co. KG, Germany). Its nominal volumetric flow rate is 0.5 to 50 L h-1. In the beginning both gas counters were filled with distilled water. This has to be considered relevant in respect to the quantified amount of

gas within the first weeks as CO2 dissolves well in water until saturation is reached. Subsequently, the gas was collected in foil bags and analyzed by two different gas analyzers. The industrial gas analyzer SSM 6000 (Pronova Analysentechnik GmbH & Co. KG, Germany) was used to determine the concentrations of methane (CH4), carbon dioxide (CO2), oxygen (O2), hydrogen sulfide (H2S), and hydrogen (H2) by means of electrochemical and infrared sensors. For gas volumes of less than 2 liters the portable gas analyzer GA 2000 (ANSYCO GmbH, Germany) was used to determine the biogas main compounds CH4, CO2, and O2. The COD of liquid samples was measured with test cuvettes (Hach Lange GmbH, Germany). They already contain all needed reagents (potassium dichromate, sulfuric acid, and mercury sulfate) to prevent any interference with oxidizable inorganic compounds. The used COD cuvettes LCK 014 have a measuring range of 1 to 10 g L-1. The cuvettes are heated for two hours at 148 °C after adding the sample. Finally, the COD values are determined spectrophotometrically with the photometer CADAS 200 (Hach Lange GmbH, Germany). The determined value has a measurement uncertainty of 2.5 %. This can lead to an overall theoretical inaccuracy of 25 % due to the dilution. Furthermore, the measurement can still interfere with ferrous iron. In this study, the TOC of liquid samples was determined with the analyzer TOC 5050A (Shimadzu, Japan). The applied measurement results in the amount of non-purgeable organic carbon. The sample is oxidized in a first step to release the formed CO2. In a second step the sample is further oxidized and the remaining organic compounds are detected subsequently. Due to the fact that volatile organic acids can also be purged during the first step it can be expected that the measured values are slightly lower than the actual TOC of the sample. The determination of total solids (TS), volatile solids (VS), ammonia nitrogen, and total Kjeldahl nitrogen (TKN) was conducted according to standard methods [17-20]. Volatile fatty acids (C2-C6) were measured with the gas chromatograph CP-3800 by Varian, Inc., USA (column: FFAP, 30 m x 0.32 mm, film thickness 0.5 µm by Permabond, USA) equipped with a flame ionization detector. The temperatures for oven, injector, and detector are as follows: from 80 to 210 °C at 15 °C min-1, 180 °C, and 220 °C. The measuring range is 0.1 to 8 g L-1. The samples have to be pretreated to precipitate proteins that can interfere with the VFA determination. Therefore, five grams of the sample are mixed with 1 mL Carrez I (potassium ferrocyanide) and 1 mL Carrez II (zinc sulfate) solutions. Before injecting the treated sample into the gas chromatograph it was centrifuged for 10 minutes at 5000 rpm (4240 g). The supernatant was extracted through 0.2 µm fiberglass syringe prefilters (Sartorius AG, Germany). The obtained values were then condensed into an acetic acid equivalent (HAceq) value. Phenolic compounds were determined as sum parameter in a first assay. Therefore, diluted samples were added to test cuvettes (LCK 345 by Hach Lange GmbH, Germany). The determination was then done spectrophotometrically with the photometer CADAS 200 (Hach Lange GmbH, Germany). The

Bereitgestellt von | Universitaet Potsdam 3 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

B. Wirth, J. Mumme

phenol sum parameter is only used to gain information whether phenolic compounds are present or not. The obtained value has very high error and is therefore not representing the actual concentration of phenols within the sample [21]. More information on phenols was obtained by the University of Tübingen using gas chromatography-mass spectrometry (GC-MS). The model 6890 by Agilent Technologies, Inc., USA (column: DB-5MS, 26.7 m x 0.25 mm, film thickness 0.25 µm by J&W Scientific, USA) was used for the measurement. The GC oven temperature program reads as follows: 150 °C for 1 min, from 150 to 110 °C at 5 °C min-1, 110 °C for 5 min, from 110 to 310 °C at 15 °C min-1, 310 °C for 4 min. The injector temperature was constant at 315 °C. The MS program was a single monitoring program. Trace elements (Al, B, Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, P, Pb, S, V, Zn) were determined via inductively coupled plasma optical emission spectrometry (ICP-OES). The samples are digested at 250 °C for 20 minutes under the addition of hydrogen peroxide and concentrated nitric acid within the microwave MWS 640 (MLS GmbH, Germany). The actual measurement was done with the analyzer iCAP 6000 (Thermo Fisher Scientific, Inc., USA). Some samples were centrifuged for 10 to 20 minutes at 5000 rpm (4240 g) before digestion to determine the dissolved trace element concentrations. The dissolved amount of trace elements was regarded as available for the microbial metabolism. The resulting supernatant was extracted through 0.2 µm fiberglass syringe prefilters (Sartorius AG, Germany). A fast method to determine the process stability of anaerobic digestion is the ratio of total volatile fatty acids (TVFA) and total alkalinity (TA). It can be interpreted as an acids-tobuffer ratio. It is an indicator for the workload of the reactor. In normal biogas plants the value should not exceed 0.4 to prevent overacidification [22,23]. This can be ensured either by adding ammonium salts to the reactor to increase the buffer capacity or by stopping to feed the reactor. The pH-value is decreased to 5 with a 0.5 molar sulfuric acid to determine the TA value. The pH is then further decreased to 4.4 to determine the TVFA value. The determination of the ratio is done automatically by the analyzer Titrino Plus 848 (Deutsche Metrohm GmbH & Co. KG, Germany).

2.5 Calculations To compare results of biogas experiments it is necessary to convert the measured biogas volumes V to the volume at standard temperature and standard pressure (STP) VSTP and dry conditions. This can be done with Equation (1) [24]. The standard conditions used here are p0 = 1.01325 bar and T0 = 273.15 K. Ambient conditions pamb and Tamb have to be specified by the same units.

VSTP = V ⋅

( p amb − p w ) ⋅ T0 p 0 ⋅ Tamb

(1)

The most important parameter in this equation is the vapor pressure of water pw as a function of the ambient temperature. It can be calculated by different equations each with their own

 p T ln    c a1  a 2 1.5  a3 3  a 4 3.5  a5 4  a6 7.5 range of validity. chosen here was published by amb  pc The  Tequation



[25]. It can be seen in Equation (2) where the subscript c indicates critical point conditions of water with values of pc = 22.064 MPa and Tc = 647.096 K.

 p  Tc    ln a1  a 2 1.5  a3 3  a 4 3.5  a5 4  a6 7.5 (2)  pc  Tamb





with τ = 1 − Θ and Θ = Tamb Tc-1. The coefficients a1 to a6 read as follows: a1 = −7.85951783, a2 = 1.84408259, a3 = −11.7866497, a4 = 22.6807411, a5 = −15.9618719, a6 = 1.80122502. Assuming that the volumetric fractions x of methane and carbon dioxide sum up close to 100 % the measured values were rectified using Equation (3) and (4), respectively [24]. Each of the two measured values is multiplied by the same factor so that the sum of the two corrected measured values comes to 100 %. Other minor gas components are neglected.

100 x x  CH 4 ,corr

CH 4

xCH4  xCO2

100 xCH4  xCO2

xCO ,corr  xCO  2

2

(3)

(4)

To compare the gas production of different reactor types at 100 different the volume specific methane production xCH4 ,corrloading  xCH4 rates  xCH4  xCO2 rates and the methane yields were calculated using COD and TOC values as a basis. The COD and TOC degradation rates were calculated based on the produced gas verified by the measured values of the liquid samples.

3. Results and Discussion 3.1 Shock-Load Experiment The determination of the HTC liquor based biogas production includes a correction by the extrapolated residual amount of biogas produced by dextrose monohydrate that was fed before. This residual gas production was already below a level of 0.1 L d-1. The zero point for the duration of this phase was set to the moment of adding HTC liquor. The gas production of the CSTR and AF increased nearly immediately in both reactors after adding 5 %vol HTC liquor equivalent to 160 and 150 mL, respectively. Methane rates increased to 12 mLSTP L-1 h-1 at the very beginning. The measured volumetric methane fraction was about 70 % within this phase. This was found to be due to the dissolving of carbon dioxide in the distilled water within the gas counters in the beginning. Because of the dissolved gases and a frequent removal of digestate for analytical purposes the apparent gas rates were highly varying. Furthermore, the nominal lower volumetric flow rate of the gas counter of at least 0.5 L h-1 was not reached throughout this phase. In total, a measurement error of 10 % was estimated for the gas production of both reactors. After reaching the peak gas flow it was observed that the gas production decreased quickly and nearly stops completely after three to four days. The amounts of volatile organic acids contained in the HTC waste water should be completely degradable within this short period of time [26]. The residual

Bereitgestellt von | Universitaet Potsdam 4 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01



Anaerobic Digestion of Waste Water from Hydrothermal Carbonization of Corn Silage

production may be due to the degradation of sugars and other more complex organic compounds. The lower gas rates of the anaerobic filter were reasoned by a slight gas leakage located at one of the tubes. This tube was exchanged after this phase. Based on these results the COD and TOC degradation rates were calculated. The CSTR reached a degree of COD removal of 46.5 % after four days and 59.6 % after eight days. The TOC removal calculated by the gas production is significantly lower compared with the COD removal due to the solution of carbon dioxide within the gas counter and reached values of 15.8 % after four days and 18.8 % after eight days, respectively. The AF showed degradation rates of only 27.1 % and 8.1 % based on COD and TOC, respectively. The lower degradation rates of the AF are due to the gas leakage already mentioned. The described degradation is also confirmed by the measured pH-value that is displayed in Figure 2. Both reactors started with an initial pH of about 7.45. After adding the acidic HTC waste water an initial drop in pH was observed. Subsequently, both reactors nearly reached their starting pH after three days. This proves the general degradation capability of organic acids contained in HTC liquor and the general feasibility of HTC waste water treatment in both reactors. The results obtained from the CSTR are comparable with results from a recent study examining the anaerobic digestion of olive mill effluent (OME). OME is even higher contaminated with phenols and has a COD of over 100 g L-1 leading to much longer

was noted every day before and after feeding so that the given biogas and methane rates in Figure 3 equal the specific amounts per day. Again, it has to be considered that the target volumetric flow of the gas counter was not reached throughout this phase and the resulting error was again estimated to be approximately 10 %. After a starting phase of only one day the methane rates stabilized at a level of approximately 0.25 L L-1 d-1. With respect to an organic loading rate of 1 gCOD L-1 d-1 this achieved value is within the range of biogas plants digesting more complex substrates like corn silage [28]. Additionally it should increase with an increasing OLR. Compared with results by [10] this is also within the reported range of 1.9 to 22.8 mL mL-1 of HTC waste water. Converted to the reported unit the continuous experiment yielded up to 19 mL mL-1. Unfortunately the studies by [10] and [11] do not report on specific biogas yields per gram of COD or TOC contained within the HTC waste water and allow therefore only a raw comparison to the conducted study.

residence times applied. Degradation rates of 48 to 53 % are reported by [27] after 100 days of incubation. These values are only achieved by an initial addition of essential nutrients.

3.2 Long-Term Process Behavior After the shock-load test both reactors – the CSTR and AF – were operated for 13 weeks with a constant organic loading rate of 1 gCOD L-1 d-1 with HTC waste water. The gas counter reading

Figure 2. D  evelopment of the pH-value during the shock load experiment at the very beginning of the experiment.

Figure 3. S  pecific gas and methane rates during 13 weeks of continuous operation with HTC liquor as sole substrate.

Bereitgestellt von | Universitaet Potsdam 5 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

B. Wirth, J. Mumme

Within the sixth week of operation there was a significant decrease in gas production in both reactors. This was assumed to be due to a lack of nutrients. Trace elements were fed regularly every day starting on day 42 – after exact one HRT. Both reactors recovered to a certain extent within one week. Another decline a few days later could not be reversed only by the addition of trace elements. The nutrient concentrations within the reactors showed a lack in sulfur and phosphorus available for the microorganisms despite sufficient concentrations within the used HTC liquor of 91 mg L-1 and 197 mg L-1, respectively. For that reason small amounts (450 mg) of elemental sulfur were added showing, however, no significant effect. Afterwards, phosphoric buffer solution (PBS) was added to increase the phosphorus level. Again, no clear effect on the gas production was observed. As a side effect, the addition of PBS led to a doubling of the chloride ion concentration within both reactors up to 1 g L-1 and a subsequent increase of the electric conductivity of 15 %. In order to prevent an over-salting, further P additions were conducted by means of phosphoric acid. Both reactors did not fully recover until the end of the experiment as measured by the gas production rates and the concentrations of available phosphorus. The gas production stabilized after six weeks indicating a bottleneck in degradation. The minor fluctuations of the AF compared to the CSTR are also noticeable and might be favored by immobilized microorganisms and sludge formation. After initially high volumetric methane concentrations of 70 %vol caused by the selective solution of carbon dioxide in distilled water contained in the gas counters the methane concentrations stabilized at about 60 to 65 %vol. In respect to the CSTR, temporary drops to 50 %vol during days 41-47 and days 64-68 were assumed to be caused by the lack of essential nutrients already mentioned and a subsequent increase of VFAs. As these drops were only observed for the CSTR the AF can be considered

to be more stable. Additional measurements of hydrogen and hydrogen sulfide in the produced biogas showed a complete absence of both gas compounds within the detection limit of 4 ppm. This is very unlikely for untreated biogas and further confirms the assumed lack of available sulfur within the liquid phase. The decline of the gas production during continuous operation was accompanied by an accumulation of volatile fatty acids, especially within the CSTR. The concentration of propionic acid intermittently reached a high value of 0.7 g L-1 in the CSTR. This development was also observed via the TVFA/TA-ratio. The ratio reached values of over 1.2 at the CSTR. The increase was mainly due to an increasing TVFA value. The AF hereby proved its stability with significantly lower levels of volatile organic acids in the digestate. The TVFA/TA-ratio at the AF did not even extent a value of 0.4 during the whole experiment. COD as well as acetic acid equivalent concentrations of the effluent of both reactors are displayed in Figure 4. Despite the increase of volatile fatty acids the COD effluent concentration remained quite constant. It was observed that HTC liquor had a dilutive effect during the first three weeks of operation. The better mixing within the CSTR led to higher COD effluent concentrations – another indicator for sludge formation and settlement within the AF. This sludge can be very beneficial for long-term operation of biogas reactors. Opposed to that it can also cause clogging in pipes and other liquid handling devices requiring down time of the reactors. Since there was a nutrient deficiency already mentioned the VFA concentrations within the CSTR remained at a very high and inhibitory level from the sixth week on. The pH-value was noted every day before feeding and logged in an hourly interval. Both reactors showed a slight decline in pH from initial 7.5 to 7.2-7.3 during the first weeks of operation. Afterwards, the AF showed stable pH-values of over

 OD and acetic acid equivalent (HAceq) effluent concentrations of both reactors during the whole experiment. Figure 4. C Bereitgestellt von | Universitaet Potsdam 6 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

Anaerobic Digestion of Waste Water from Hydrothermal Carbonization of Corn Silage

7 whereas the CSTR showed fluctuating values between 6.7 and 7.0. The pH-value was about 0.2 lower at the days after a digestate removal due to a lower actual reactor volume of work. The stronger initial drops observed for the AF compared to the CSTR were due to the missing stirring. The decreasing pH during the last days of operation was due to the addition of phosphoric acid. As expected, the TKN concentration decreased over time and approached the values of the fed substrate (0.68 g L-1) asymptotically. In the end, the CSTR and AF showed a 14 and 4 % higher TKN value compared to the HTC liquor. The concentration of ammonia nitrogen in both reactors was nearly twice as high compared to the HTC waste water (0.23 g L-1). The AF showed a 10 % higher ammonia nitrogen concentration compared to the CSTR. It was also observed that the VS concentration of the AF digestate was always 5-10 % lower than that of the CSTR. This can be explained by the visible sedimentation of solid particles inside the AF. However, based on the observed discrepancy between the S and P concentrations in the HTC liquor and in the digestate also a ferrous iron based precipitation was assumed leading to insoluble iron(II) sulfide (FeS) and iron(II) phosphate (Fe3(PO4)2). The assumed precipitation was validated through a chemical analysis of the precipitated sludge of the AF at the end of the experiment. It revealed very high concentrations of sulfur (335 mg L-1) and phosphorus (305 mg L-1) and a TS content of the sludge of 3.8 %. An additional source of iron besides the original HTC liquor was found to be the metal gasoline canisters that started to corrode during storage of the HTC liquor. This increased the Fe concentration from 478 to 719 mg L-1. However, at the end of the experiment the initial ferrous iron concentration still accounted for two thirds of the concentration within the digestate. In order to reduce the Fe concentrations the HTC liquor should be stored in acid proof containers and all parts

within an HTC plant that are in contact with the acidic waste water should be made of stainless steel. Having a look on the development of the concentrations of S, P, and Fe during the whole experiment shows an increase of ferrous iron contained in both reactors (Figure 5). The subsequent precipitation of sulfur and phosphorus was validated by three times lower concentrations within the AF digestate compared to the CSTR. In order to determine the amount of dissolved S and P some digestate samples were centrifuged before ICP analysis. The obtained concentrations were beneath 25 and 10 mg L-1, respectively, which are suspected to be insufficient [13]. After adding S and P (as described before) a clear decrease of the ferrous iron concentration was observed. This further proved the precipitation of available ferrous iron by S and P. The precipitation of sulfur could also have been forced by so-called bacterial anaerobic corrosion of metal installation parts within biogas reactors [29]. In a wet environment bacteria are able to corrode metal parts by removing the thin protective coating of molecular hydrogen. Part of the ferrous iron is subsequently precipitated with hydrogen sulfide to form the water insoluble FeS. Possibly affected could be the stirrer shaft in the CSTR and the restraint sieve in the AF, both made of steel. Both installations showed no clear signs of corrosion at the end of the experiment. During the last weeks of continuous operation when adding phosphoric acid to the feeding syringes it was observed that the black tinge of the substrate vanished to a certain extent. This can be explained by the solution of residual FeS due to a lower pH. It illustrates again the complex reactions within both reactors. The nutrient requirements of digesting HTC waste water could be fulfilled by co-digesting another nutrient-rich substrate such as animal slurries. A recent study by [27] showed that OME benefits from the co-digestion with laying hen litter leading to a 90 % increased biogas production.

 evelopment of the concentrations of ferrous iron, sulfur, and phosphorus in the liquid digestate during 13 weeks of continuous operation. Figure 5. D Bereitgestellt von | Universitaet Potsdam 7 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

B. Wirth, J. Mumme

3.3 Degradation of Critical Substances

3.4 Process Performance Parameters

Substances such as phenols, furfural, and PAHs are known to be formed during the production of hydrochar. Small amounts of these substances can also be found in HTC liquor and may be harmful for the microbiological community involved in anaerobic digestion. Weekly samples of the CSTR were analyzed for phenols and nitrophenols by GC-MS. The measured values were compared with theoretical calculated values based on the assumption that the compounds are not degradable and enter the reactor only via feeding and leave it during the digestate removal. Exemplary, the courses of phenol and p-cresol are displayed in Figure 6. Phenol seems to be degraded to a certain extent (up to 83 %). As opposed to that, the measured concentrations of p-cresol are substantially higher than the calculated values. Furthermore, p-cresol reached values about 17-times higher than in the HTC waste water. This can be at least partly attributed to a hypothesized formation of p-cresol by binding phenol and methane together and releasing hydrogen forced by dedicated microorganisms. A potential sorption of phenolic compounds within the reactor could, however, not account for the observed increase of p-cresol. Inhibiting substances could be removed by chemical pre-treatment. A study by [30] examined the phenol removal efficiency of aluminum and iron salts added to raw OME rich in phenols with concentrations of up to 10 g L-1. After acid

Because of the inhibition assumed for the second part of the experiment the average methane yields as well as the degree of COD and TOC removal were calculated for two periods – the first five weeks and weeks six to 13 of operation (Table 2).

cracking and adding these salts the concentrations of phenol diminished by up to 50 %. All treatments improved the anaerobic degradability of OME. This could also be an interesting option as a pre-processing step for HTC waste water. Other inhibiting substances such as furfural and hydroxymethylfurfural were not determined within this study but can have significant concentrations in the range of g L-1. These concentrations might also have an inhibitory effect on the anaerobic digestion of HTC waste water. Overall, any possible inhibition by organic compounds except VFAs could not be detected within this study due to the superimposition with the nutrient problems already described before.

Figure 6. Concentrations of phenol and p-cresol of CSTR samples compared with their theoretical calculated concentration if they would act as inert components.

Table 2. C  alculated methane yields in LSTP gCOD/TOC-1 d-1 and COD and TOC degradation rates based on the measured gas volume and its composition on a weekly basis (mean values, standard deviation in parentheses). The results are averaged for the first five weeks of operation and compared with averaged values for weeks 6 to 13 of continuous operation. Measurement

Unit

CSTR

AF

Averaged methane yield per gram of COD (weeks 1-5)

LSTP g-1

0.236 (±0.016)

0.219 (±0.029)

per gram of COD (weeks 6-13)

LSTP g

0.163 (±0.028)

0.178 (±0.007)

per gram of TOC (weeks 1-5)

-1

LSTP g

0.624 (±0.042)

0.577 (±0.078)

per gram of TOC (weeks 6-13)

LSTP g-1

0.431 (±0.073)

0.470 (±0.019)

COD (weeks 1-5)

%

74.96 (±5.03)

69.37 (±9.35)

COD (weeks 6-13)

%

51.76 (±8.78)

56.46 (±2.24)

TOC (weeks 1-5)

%

53.84 (±2.97)

48.94 (±7.30)

TOC (weeks 6-13)

%

41.53 (±5.94)

40.43 (±1.81)

-1

Averaged degradation rate

Bereitgestellt von | Universitaet Potsdam 8 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

Anaerobic Digestion of Waste Water from Hydrothermal Carbonization of Corn Silage

The COD and TOC results were compared with the results of the chemical analysis of the digestate (Figure 4) showing an average accordance of 95-100 % in respect to COD and 75-85 % in respect to TOC. The TOC value of the digestate was always lower than expected. As described before, this can be attributed to losses of CO2 caused by dissolving, leaking through some of the silicone hoses over time, and volatile losses during the TOC measurement. Consequently, the TOC degradation rates can be expected to be somewhat higher. The degradation rates for COD and TOC show nearly the same behavior for both reactors. Again, the CSTR showed higher fluctuations after five weeks of operation whereas the AF was more stable from then on. Based on the gas production, a COD balance is often more reliable than a TOC balance due to the fact that there are no experience values concerning the proportion of TOC that is needed for regenerating the microorganisms. Additionally, carbon dioxide has a higher solubility and this value can change easily with a varying temperature or pH in the reactor. Furthermore, many tubes cannot ensure that carbon dioxide is leaking over time. This can lead to higher TOC degradation rates than originally calculated or expected. Results published by [31] for the continuous operation of an anaerobic sludge blanket reactor with OME indicated COD degradation rates of up to 90 % within 10 days. The pH of the substrate was adjusted to a neutral level. The study further achieved OLRs of over 20 kgCOD m-3 d-1 indicating the advantage of immobilized microorganisms when handling highly contaminated substrates. The conducted experiments could be further optimized. Starting with the applied measurement, a mounted probe measuring the redox potential could indicate changes within the reactors faster. This is due to the fact that the pH-value reacts with a certain time lag. The anaerobic filter should also be complemented by a discharge system for settled sludge to prevent any clogging. An automated feeding system would facilitate feeding and subsequently would homogenize the daily

load. An adequate feeding system has to be capable to handle very small volumetric flows to be applied to this experimental set-up. Having a look on its application in practice the anaerobic digestion of HTC waste water on-site of an industrial-scale HTC plant would generate certain benefits. The additionally generated biogas could increase the economic feasibility and would further increase the overall energetic efficiency. Furthermore, a pretreatment of the generated waste water could reduce the effort spent on an aerobic waste water treatment. Based on the results presented here an anaerobic filter should be the first choice.

4. Conclusions Hydrothermal carbonization (HTC) is an increasingly interesting treatment of biomass. Its high-strength waste water was proven to be suitable for anaerobic digestion. The conducted experiment showed a good degradability of COD (up to 75 %) and TOC (up to 54 %). Furthermore, the degradation was relatively fast compared to the digestion of conventional organic wastes. No inhibition was yet detected in connection with organic inhibitors except volatile fatty acids. However, problems can be caused by the lack of trace elements and an iron-based precipitation of S and P. Future research should focus on optimal process conditions and post-treatment requirements.

Acknowledgements This work was funded by the German Federal Ministry of Education and Research (BMBF) in cooperation with the Project Agency Jülich (PtJ). The authors would like to thank E. Janiszewski and the whole laboratory team of the ATB for their chemical support. Furthermore, the authors would like to thank B. Stahl from the University of Tübingen for conducting the GCMS analyses.

References [1]

[2]

[3]

[4]

Mumme J., Eckervogt L., Pielert J., Diakité M., Rupp F., Kern, J., Hydrothermal carbonization of anaerobically digested maize silage, Bioresour. Technol., 2011, 102, 9255-9260 Libra J.A., Ro K.S., Kammann C., Funke A., Berge N.D., Neubauer Y., et al., Hydrothermal carbonization of biomass residuals: a comparative review of the chemistry, processes, and applications of wet and dry pyrolysis, Biofuels, 2011, 2, 71-106 Berge N.D., Ro K.S., Mao J., Flora J.R.V., Chappell M.A., Bae S., Hydrothermal Carbonization of Municipal Waste Streams, Environ. Sci. Technol., 2011, 45, 5696-5703 Koon M., Recovery of Carbon and Nutrients in Lignocellulosic Biomass during Hydrothermal Carbonization, Master’s thesis, University of Hamburg, Hamburg, Germany, 2011

[5]

[6]

[7]

[8]

Stemann J., Ziegler F., Hydrothermal carbonization (HTC): Recycling of process water, In: Proceedings of the 19th European Biomass Conference and Exhibition (6-10 June 2011, Berlin, Germany), 2011 Becker R., Dorgerloh U., Helmis M., Mumme J., Diakité M., Nehls I., Hydrothermally carbonized plant materials: patterns of volatile organic compounds detected by gas chromatography, Bioresour. Technol., 2012, 130, 621-628 Meyer H., Leistungsfähigkeit anaerober Reaktoren zur Industrieabwasserreinigung, Veröffentlichungen des Institutes für Siedlungswasserwirtschaft und Abfalltechnik der Universität Hannover, Heft 128, Hannover, 2004 Busca G., Berardinelli S., Resini C., Arrighi L, Technologies for the removal of phenol from fluid streams: A short review of recent developments, J. Hard. Mater., 2008, 160, 265-288

Bereitgestellt von | Universitaet Potsdam 9 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01

B. Wirth, J. Mumme

[9]

[10]

[11]

[12] [13]

[14] [15] [16] [17]

[18]

[19]

[20]

Chakraborty S., Bhattacharya T., Patel T., Tiwari K, Biodegradation of phenol by native microorganisms isolated from coke processing wastewater, J. Environ. Biol., 2010, 31, 293-296 Ramke H.G., Blöhse D., Lehmann H.J., Antonietti M., Fettig J., Machbarkeitsstudie zur Energiegewinnung aus organischen Siedlungsabfällen durch Hydrothermale Carbonisierung, Deutsche Bundesstiftung Umwelt (DBU), Höxter, 2010 Oliveira I., Blöhse D., Ramke H.G., Hydrothermal carbonization of agricultural residues, Bioresour. Technol., 2013, 142, 138-146 Wellinger A., Biogas-Handbuch, 2nd ed., Wirz, Aarau, 1991 Eder B., Schulz H., Biogas Praxis – Grundlagen, Planung, Anlagenbau, Beispiele, Wirtschaftlichkeit, 3rd ed., Ökobuch, Staufen bei Freiburg, 2007 Henze M.H., Anaerobic fluidized beds: ten years of industrial experience, Water Sci. Technol., 1983, 36, 415-422 Speece R.E., Anaerobic Biotechnology for Industrial Wastewaters, Vanderbilt, Archae Press, Nashville, 1996 Weiland P., Grundlagen der Methangärung – Biologie und Substrate, VDI-Bericht, 2001, 1620, 19-32 DIN 38406-E5 – Ammonium-Stickstoff; Deutsche Einheitsverfahren zur Wasser-, Abwasserund Schlammuntersuchung; Kationen (Gruppe E); Bestimmung des Ammonium-Stickstoffs (E5). Deutsches Institut für Normung (DIN), 1983 DIN EN 25663:1993-11 – Water quality; Determination of Kjeldahl nitrogen; Method after mineralization with selenium. Deutsches Institut für Normung (DIN), 1993 DIN EN 12879:2001-02 – Characterization of sludges – Determination of the loss on ignition of dry mass. Deutsches Institut für Normung (DIN), 2001 DIN EN 12880:2001 – Characterization of sludges – Determination of dry residue and water content. Deutsches Institut für Normung (DIN), 2001

[21] Licha T., Herfort M., Sauter M., Phenolindex – ein sinnvoller Parameter für die Altlastenbewertung, Grundwasser, 2001, 1, 8-14 [22] Fannin K.F., Start-up, operation, stability and control, In: Chynoweth D.P., Isaacson R. (Eds.), Anaerobic Digestion of Biomass, Elsevier, London, 1987 [23] U.S. EPA, Anaerobic sludge digestion operations manual, sect. 4-17, U.S. Environmental Protection Agency (U.S. EPA), 1976 [24] VDI 4630 – Vergärung organischer Stoffe – Substratcharakterisierung, Probenahme, Stoffdatenerhebung, Gärversuche, Verein Deutscher Ingenieure (VDI), 2006 [25] IAPWS, Revised Supplementary Release on Saturation Properties of Ordinary Water Substance, The International Association for the Properties of Water and Steam (IAPWS) 1992 [26] Vollmer G.R., Abbaugeschwindigkeit der Stoffgruppen, In: Eder, B., Schulz, H. (Eds.), Biogas Praxis – Grundlagen, Planung, Anlagenbau, Beispiele, Wirtschaftlichkeit, Ökobuch, Staufen bei Freiburg, 2007 [27] Azbar N., Keskin T., Yuruyen A., Enhancement of biogas production from olive mill effluent (OME) by co-digestion, Biomass Bioenerg., 2008, 32, 1195-1201 [28] Mumme J., Linke B., Tölle R., Novel upflow anaerobic solidstate (UASS) reactor, Bioresour. Technol., 2010, 101, 592599 [29] Hamilton W., Sulphate-Reducing Bacteria and Anaerobic Corrosion, Annu. Rev. Microbiol., 1985, 39, 195-217 [30] Azbar N., Keskin T., Catalkaya E.C., Improvement in anaerobic degradation of olive mill effluent (OME) by chemical pretreatment using batch systems, Biochem. Eng. J., 2008, 38, 379-383 [31] Azbar N., Tutuk F., Keskin T., Effect of Organic Loading Rate on the Performance of an Up-Flow Anaerobic Sludge Blanket Reactor Treating Olive Mill Effluent, Biotechnol. Bioprocess Eng., 2009, 14, 99-104

Bereitgestellt von | Universitaet Potsdam 10 Angemeldet | 141.89.210.1 Heruntergeladen am | 12.12.13 14:01