Biogas Production from Thin Stillage on an Industrial Scale—experience and Optimisation

With the increasing demand for renewable energy and sustainable waste treatment, biogas production is expanding. Approximately four billion litres of bio-ethanol are produced annually for vehicle fuel in Europe, resulting in the production of large amounts of stillage residues. This stillage is energy-rich and can be used for biogas production, but is a challenging substrate due to its high levels of nitrogen and sulphate. At the full-scale biogas production plant in Norrköping, Sweden (Svensk Biogas i Linköping AB), thin grain stillage is used as a biogas substrate. This paper describes the plant operation and strategies that have been implemented to digest thin stillage successfully. High ammonia concentrations in the digester have resulted in syntrophic acetate oxidation (SAO) becoming the major pathway for acetate degradation. Therefore, a long hydraulic retention time (HRT) (40–60 days) is used to allow the syntrophic acetate-oxidising bacteria time to grow. The high sulphate levels in thin stillage result in high levels of hydrogen sulphide following degradation of protein and the activity of sulphate-reducing bacteria (SRB), the presence of which has been confirmed by quantitative polymerase chain reaction (qPCR) analysis. To optimise biogas production and maintain a stable process, the substrate is diluted with tap water and co-digested with grain residues and glycerine to keep the ammonium nitrogen (NH 4-N) concentration below 6 g L −1. Combined addition of iron, hydrochloric acid and cobalt successfully precipitates sulphides, reduces ammonia toxicity and supplies microorganisms with trace element. Mesophilic temperature (38 °C) is employed to further 5643 avoid ammonia toxicity. Together, these measures and doubling the digester volume have made it possible to increase annual biogas production from 27.7 TJ to 69.1 TJ. Nomenclature: HRT hydraulic retention time COD chemical oxygen demand OLR organic loading rate TS total solids VS volatile solids VFA volatile fatty acids DNA deoxyribonucleic acid qPCR quantitative polymerase chain reaction SAO syntrophic acetate oxidation SAOB syntrophic acetate-oxidising bacteria SRB sulphate-reducing bacteria


Introduction
Anaerobic digestion is the microbiological degradation of organic material in the absence of oxygen.The product from this degradation is biogas, which mainly consists of bio-methane and carbon dioxide.Organic wastes from industry, wastewater treatment plants or households can be treated in biogas plants to produce energy-rich bio-methane, which may be used to produce vehicle fuel or electricity and heat.A liquid residue rich in nitrogen and phosphates formed during anaerobic digestion is suitable for use as organic fertiliser [1].Therefore, anaerobic digestion is a feasible technique to treat organic wastes, produce renewable energy and recover nutrients for re-use in agriculture.As a consequence of the increased demand for renewable energy and sustainable waste treatment, the number of full-scale biogas production plants in Europe is increasing [2,3] and these require a supply of substrate, e.g., energy-rich industrial residues.Stillage is one such potential substrate.Approximately four billion litres of bio-ethanol for vehicle fuel are produced annually in Europe [4].Stillage is separated from bio-ethanol after the fermentation process, with 0.8 kg total solids (TS) stillage (5-10 L) being produced per litre of bio-ethanol [5,6].Stillage consists of materials which are not degraded during hydrolysis and bio-ethanol fermentation, such as high concentrations of proteins, amino acids, smaller amounts of sugars and yeast cells [7,8].Moreover, stillage is energy-rich (13.6 MJ kg −1 TS) and has been shown to be appropriate as feedstock for anaerobic digestion [8].Thin stillage used as feedstock for biogas production has a methane potential of 10.4 MJ kg −1 chemical oxygen demand (COD) [8].Combining bio-ethanol production with biogas production also brings economic and environmental benefits, especially at sites where the respective plants are in close proximity to each other [6,9,10].However, thin stillage presents some challenges in biogas production.
Degradation of the nitrogen-rich stillage results in high levels of total ammonium nitrogen (NH 4 -N), which can result in process disturbance.The ammonia (NH 3 -N) fraction of NH 4 -N is positively correlated with increased pH and temperature, is a toxic component and has been shown to be inhibitory to most microorganisms involved in the biogas process, but particularly aceticlastic methanogens [11,12].Inhibition of aceticlastic methanogens has been suggested to result in selection of syntrophic acetate oxidation (SAO) (at approximately 3 g NH 4 + -N L −1 ) [13,14].SAO is performed in a sensitive syntrophic cooperation between syntrophic acetate-oxidising bacteria (SAOB) and hydrogenotrophic methanogens [13,14].Co-cultures of SAOB and methanogens typically have lower growth rates than aceticlastic methanogens and introduction of the SAO pathway is thus correlated with lower gas yield [13,15].In addition to high nitrogen levels, sulphuric acid is commonly used for pH regulation in bio-ethanol fermentation processes, contributing to high sulphate concentrations in stillage [16].Sulphate is converted to sulphides by the activity of sulphate-reducing bacteria (SRB), which proliferate in different types of anaerobic digestion process [17,18].The high protein content in thin stillage also results in the release of organic sulphur as sulphides, which have various negative effects on the industrial biogas production process.These include corrosion, precipitation of trace elements, inhibition of microorganisms and consumption of organic material [mainly alcohols, volatile fatty acids (VFA) and hydrogen] that would otherwise be used for methane production [11,[18][19][20].
This paper describes results and experiences from a full-scale operation using thin grain stillage (hereafter called thin stillage) as the dominant substrate at an industrial biogas plant in Norrköping, Sweden run by Svensk Biogas i Linköping AB, a subsidiary company of the public municipal company Tekniska verken i Linköping AB (Linköping, Sweden).Major challenges and the solutions devised to cope with these are discussed and on-going work to optimise gas production efficiency is described.

Results and Discussion
Despite the challenges involved in using thin stillage as a substrate for biogas production, experiences show that large-scale gas production from this material is possible.Annual biogas production at the Norrköping plant has increased year on year, from 1.4 million m 3 (27.7 TJ) in 2008 to 3.5 million m 3 (69.1 TJ) in 2011 (Figure 1a).This increase in biogas production has been achieved by doubling the digester volume, increasing the amount of substrate treated annually from 1750 to almost 4000 t volatile solids (VS) in 2011 and improving specific biogas production (VS basis) from 680 m 3 t −1 d −1 during 2008 to 860 m 3 t −1 d −1 during 2011 (Figure 1b).The methane content was relatively low, 55% ± 1%, and thus the specific methane production (VS basis) has increased from 2008 to 2011 (370 to 466 m 3 t −1 d −1 ), which is generally higher than previously reported for stillage fractions [8,10,[21][22][23].The high and increasing specific gas production during the period 2008-2011 can be explained by increased process efficiency and by the use of glycerine during 2011.Glycerine has high energy content and thus should theoretically result in higher biogas yield.From the second half of 2009, the organic loading rate (OLR) has stabilised at 2.4-3.2kg VS m −3 d −1 (Figure 1c), determined by continuous measurements of TS and VS in the substrate.There is generally a low concentration of VFA and high alkalinity, illustrating that the process is clearly stable (Figures 1d and 2).

Nitrogen
One of the challenges with thin stillage is the high nitrogen level, giving rise to high levels of NH 4 -N during degradation.Different actions were taken at the plant to handle the high nitrogen concentration in thin stillage (Table 1).First, mesophilic temperature was chosen for the operation.It is a well-known fact that thermophilic temperature results in a relatively higher level of NH 3 -N, and thus a higher degree of microbial inhibition [12].As expected, NH 4 -N concentration in the process is high, 4000-5700 mg L −1 (Figure 1e).The average ammonia-nitrogen (NH 3 -N) level during the period 2008-2011 was 252 ± 34 mg kg −1 (Figure 1f), as a combined result of the NH 4 -N concentration, pH (7.6 ± 0.0) (Figure 1g) and temperature (37.7 ± 0.1 °C).Having an operating temperature of 55 °C instead of 38 °C would hypothetically have increased the NH 3 -N concentration from 252 mg kg −1 to 620 mg kg −1 , clearly exceeding the level of NH 3 -N previously reported to cause severe inhibition [11,12].It has been shown that a temperature increase from 25 °C to 35 °C can result in higher methanogenic reaction rates despite increased ammonia concentration according to Garcia and Angenent [24], but this effect is unlikely to occur in the Norrköping plant since it involves higher pH and a larger temperature change.Higher temperature often results in higher pH too; therefore, a shift from mesophilic to thermophilic temperature would most likely further increase the ammonia concentration [25].
A second strategy to handle the high levels of nitrogen in the Norrköping biogas plant is to use a long hydraulic retention time (HRT) (45-60 days) (Table 1).Previous studies have shown that high levels of ammonia are selective for methane production via SAO, and batch degradation experiments with [2-14 C]-acetate have shown that this is also true for the Norrköping plant [14,26].Analysis of the degradation products showed 14 CH 4 and 14 CO 2 but with the latter dominating, resulting in a 14 CO 2 : 14 CH 4 ratio of 34.0 ± 3.5.Using a long HRT allows the environment to adjust and maintain the dominating slow-growing SAOB microorganisms in the process.Therefore dilution with water is kept to a minimum, leading to high NH 4 -N concentrations in the digestate (Table 1).The presence of high levels of different known SAOB, i.e., Tepidanaerobacter acetatoxydans (6.51 ± 0.16 log gene copies mL −1 ), Syntroaceticus schinkii (10.23 ± 0.06 log gene copies mL −1 ) and Clostridium ultunese (6.89 ± 0.11 log gene copies•mL −1 ) was confirmed by quantitative polymerase chain reaction (qPCR) analysis in a recent study by Sun, et al. [27].A third strategy used for managing the high NH 4 -N levels in the process is to down-regulate the pH by addition of HCl and acidic iron (Fe 3+ ) ions to the substrate (Table 1).A decrease of 0.2 pH units can enhance biogas production significantly when degrading protein-rich substrates [28], by lowering the free NH 3 -N concentration and also creating a more optimum pH for methanogens [12,28,29].Theoretically, without process additive addition to the process in the Norrköping biogas plant, the pH would be 0.2 pH units higher and the resulting NH 3 -N concentration would increase from the present level of 252 mg kg −1 to 360 mg kg −1 , a 48% increase.This would considerably increase the risk of the upper inhibitory limit being exceeded during high NH 4 -N periods (Figure 1f).Using the strategies described above allowed stable operation to be achieved during the majority of the period 2008-2011.Except for a few peaks, the VFA concentration was generally low in the period (0.79 ± 0.09 g L −1 ) (Figure 2), even though the NH 4 -N and NH 3 -N concentration was far above the value reported to cause inhibition of anaerobic digestion [11].However, during spring 2009 increasing NH 4 -N and NH 3 -N concentrations correlated with increased process instability (Figure 3).The NH 4 -N concentration increased from approximately 4000 mg kg −1 to 5700 mg kg −1 and the NH 3 -N level simultaneously increased to approximately 450 mg kg −1 .This unexpected increase in NH 4 -N may have been the result of several different factors, such as variations in the degree of degradation (organic nitrogen mineralisation), in the OLR or, most likely, in the substrate mixture due to fluctuating urea addition in the bio-ethanol production process.The increase in NH 4 -N resulted in accumulation of total VFA to a total peak of 2.3 g L −1 , consisting of 93% propionic acid, which has previously been identified as a common product during NH 3 -N inhibition [30,31].The level at which NH 3 -N caused inhibition in the process, 450 mg kg −1 , is similar to inhibitory levels reported previously [11,30,32].This process disturbance was dealt with by significantly decreasing the OLR from 3.75 g L −1 d −1 to 2.25 g L −1 d −1 , which decreased both the total VFA and the NH 4 -N and NH 3 -N concentration (Figure 3).After breakdown of the VFA, the OLR was then raised to the original OLR without any resulting process problems (Figures 1c and 2).

Hydrogen Sulphide
In addition to its high nitrogen concentration, thin stillage typically contains high levels of sulphur in amino acids but also in the form of sulphate, originating from the sulphuric acid used for pH regulation during bio-ethanol production.Irrespective of source, sulphur causes problems during anaerobic digestion as it is converted to hydrogen sulphide (H 2 S), either during the degradation of amino acids or through the activity of SRB [11].In addition to the malodorous and corrosive properties of H 2 S gas, it is also toxic to microorganisms and can cause precipitation of trace elements required for microbial growth and activity [11,20,33].Furthermore, the H 2 S formed have to be removed to meet Swedish legal requirements on vehicle fuel and to reduce negative effects on anaerobic digestion.The thin stillage used at the Norrköping biogas plant contains approximately 44 kg sulphate t −1 TS, resulting in high levels of H 2 S due to the reduction of sulphate by SRB, leaving only trace amounts of sulphate in the digestate [17].SRB have been shown to exist in most types of biogas processes and to increase in number in response to increasing sulphate levels [18].qPCR analysis performed during a five-month period clearly revealed a stable population of SRB at the Norrköping plant between 1.8 × 10 5 and 7.3 × 10 5 gene copies mL -1 .
However, the levels were no higher than in biogas plants running with substrates containing lower levels of sulphate [18].To cope with the issues related to high levels of H 2 S, e.g., precipitation of trace elements, toxicity to microorganisms and cost associated with upgrading of the biogas, the process was supplemented with continuous addition of iron (Table 1).This strategy resulted in a rather low sulphide concentration (89 ± 6 µL L −1 ) through precipitation of sulphides as iron sulphide.To achieve this low concentration of sulphide, iron needed to be added to the digester at an average concentration of 2814 ± 72 mg kg −1 .Since there has been no period without addition of iron, the inherent concentration of H 2 S is difficult to predict.However, judging from the iron demand to keep the H 2 S concentration low, the inherent concentration in the gas would be 8,000-10,000 µL L −1 .There is a negative correlation (R 2 = 0.34) between the H 2 S and iron concentrations in the digester during operation, showing that iron is indeed necessary (Figure 4).

Trace Elements and Co-Digestion
Trace element addition, in particular cobalt and nickel, or co-digestion with other substrates has previously been reported to be necessary for successful anaerobic digestion of stillage fractions [10,34,35].In addition, trace element addition during anaerobic digestion has been proven to be important in supplying microorganisms with the necessary trace elements, particularly when digesting NH 3 -N-rich substrates [34,[36][37][38].Addition of a combination of trace elements and iron has also been shown to enhance the capacity for degradation of VFA, particularly propionic acid, in biogas processes with methanogenesis mainly performed via SAO [37,39].A combined additive containing cobalt and iron has therefore been developed for use in the Norrköping biogas plant.This has most likely enhanced process performance considerably and may be the reason why the VFA concentration, in particular propionic acid, is relatively low despite the high nitrogen concentration.Inclusion of other substrates (grain residues and glycerine) is also used to provide a more complete nutrient composition of the feedstock.The C:N ratio of these substrates is lower than that of thin stillage, contributing to lower nitrogen in the feedstock and thus decreasing the nitrogen inhibitory effect.Glycerine also contributes to higher gas yields due to its high energy content.

Further Measures
In summary, a number of measures have been introduced to optimise the anaerobic digestion process at the Norrköping biogas plant.These measures have been governed by the high nitrogen and sulphate concentrations in the feedstock (Table 1).Further measures are being planned to decrease the NH 4 -N concentration, reduce the concentrations of sulphides and supply a more complete set of nutrients to the process.This can be accomplished by mixing in other substrates, thus increasing the C:N and C:S ratio [35].It has been shown in an experimental simulation of the Norrköping plant that the HRT could be reduced to 25 days while keeping OLR at 4.6 kg VS m −3 d −1 without any process disturbance, as a result of inclusion of household waste (up to 75% of total OLR) (unpublished data).Inclusion of household waste is planned at the Norrköping plant and most of the challenges coupled with nitrogen and sulphides are predicted to be reduced by this action.Furthermore, negative effects of temporary process disturbance owing to VFA accumulation (Figure 1c) can be better handed in the future, since a post-digester has now been installed (in 2012).With this post-digester, the outgoing VFA from the primary digesters will be degraded, increasing the efficiency of the plant during high VFA peaks.The post-digester is also predicted to increase methane production at the plant by ~5%, even in absence of VFA, as a result of increased degree of degradation.This would reduce methane emissions from digestate storage and increase the level of mineralised NH 4 -N in the residue used as fertiliser [40].

Experimental Section
The full-scale biogas plant in Norrköping, Sweden, has been operating since 2007 using thin stillage from bio-ethanol production as the main substrate to produce biogas.Initially, the plant was operated with one top-stirred continuously stirred tank reactor (1800 m 3 ), but in 2009 the plant was expanded with another digester (2000 m 3 ) (Figure 5).The plant is operated at mesophilic conditions (38 °C) and receives a mixture of thin stillage (TS ≈ 10%, VS ≈ 9%) and condensed thin stillage (TS ≈ 30%, VS ≈ 27%), occasionally combined with hammer-milled grain residues (TS ≈ 90%, VS ≈ 88%; corresponding to on average 17% of OLR) and glycerol (TS ≈ 80%, VS ≈ 70%; corresponding to 10% of OLR, 2011 only).The stillage originates from a grain-based bio-ethanol plant (Lantmännen Agroetanol AB, Norrköping, Sweden) in close proximity to the biogas facility and is kept in a storage tank (Figure 5).The different substrates are mixed with tap water in a feeder tank, which is operated in batch mode.The OLR (measured as VS) is on average 3.0 ± 0.1 kg m −3 d −1 , resulting in a HRT in the digesters of 45-60 days.The average sulphate content (measured as amount of TS) is 44 g kg −1 and the Kjeldahl nitrogen concentration is 36 g kg −1 .A process additive named BDP, developed at Tekniska verken i Linköping AB and sold by Kemira Oyj (Helsinki, Finland), has been used since 2008.It contains iron (11%, a mixture of Fe 2+ /Fe 3+ ), cobalt and hydrochloric acid [41].The iron dose is regulated to keep the hydrogen sulphide concentration low and is on average 2814 ± 72 mg kg −1 .Cobalt is added to achieve a concentration of approx.0.5 mg L −1 in accordance with other studies [34,38] and earlier laboratory simulations of the plant, the concentration was on average 0.49 ± 0.01 mg L −1 .The liquid outflow from the plant (digestate) is collected in a covered digestate lagoon and distributed to local farmers via satellite storage.The digestate is certified according to SPCR120 [42] and is also accepted as an organic fertiliser by the Swedish association for organic farming (KRAV).The biogas produced is upgraded with a water scrubbing technique (Malmbergs, Åhus, Sweden) to ≥97% methane (Figure 5).The upgraded gas is then compressed (200-250 bar), stored, distributed and sold as vehicle fuel for urban buses, taxis and private cars.

Analyses
Data on full-scale process parameters, amount of biogas produced, biogas composition, substrate mixtures and quantities was taken from the SCADA system (Cactus Uniview AB, Mölndal, Sweden).All data presented in this paper refer to the period 1 July 2008-31 December 2011.All volumetric gas data were converted to standard conditions at pressure 1.01325 bar and temperature 273.2 K. Confidence intervals were calculated for all means according to the students' T-test (p < 0.05) using values from the whole of study period.
For process monitoring purposes, TS and VS are analysed on each batch of incoming substrate to calculate the correct OLR.Representative sampling of the digesters was performed at dedicated sampling points in the biogas plant.pH, alkalinity and VFA were monitored and analysed in order to assess process stability and process efficiency.Nitrogen fractions were determined to predict nitrogen inhibition.Hydrogen sulphide was analysed to determine the necessary iron addition and metals in order to confirm correct metal addition.qPCR analysis was performed on frozen samples (stored at −20 °C), while chemical analyses were performed on fresh samples.VFA (C2-C7) were analysed with a Clarus 550 gas chromatograph (Perkin Elmer, Waltham, MA, USA) with a packed Elite-FFAP column (Perkin Elmer) for acidic compounds according to Jonsson and Borén [43].Ammonium nitrogen (NH 4 -N) was analysed as the sum of NH 4 -N (aq) + NH 3 -N (aq) by distillation (Kjeltec 8200, FOSS in Scandinavia, Hillerød, Denmark) in an acidic solution (H 3 BO 3 ).NH 4 -N was determined in a second step by titration with HCl (Titro 809, Metrohm, Herisau, Switzerland).
Kjeldahl nitrogen was analysed with the same procedure and equipment as for the NH 4 -N analysis, with the exception that the samples were pre-treated with H 2 SO 4 and subsequently heated to 410 °C for 1 h.The ammonia nitrogen (NH 3 -N) concentration was calculated from the NH 4 -N, pH and temperature according to Hansen, et al. [12].Bicarbonate alkalinity was measured by titration with hydrochloric acid to pH 5.4 (Titro 809, Metrohm).pH was measured with a potentiometric pH meter with a Hamilton electrode (WTW Inolab GmbH, Weilheim, Germany) at 25 °C.TS was measured by oven-drying at 105 °C for 20 h.VS was subsequently measured by combusting the TS sample at 550 °C for 3 h.The content of hydrogen sulphide in the biogas produced was analysed with a micro IV sensor (GfG GmbH, Dortmund, Germany).For qPCR analysis, samples from the reactors were withdrawn and frozen.Deoxyribonucleic acid (DNA) was extracted from small aliquots (300 µL) using the FastDNA® SPIN kit for soil (MP Biomedicals, Solon, OH, USA) according to the manufacturer's instructions, with small adjustments according to Moestedt, et al. [18].For quantification of SRB, the primers DSRp2060F-GC (5' CAA CAT CGT YCA YAC CCA GGG 3') and DSR4R (5' GTG TAG CAG TTA CCG CA 3') [44] were used according to the protocol described by Dar, et al. [45], with adjusted initial touchdown protocol using 10 cycles instead of 20 for the step-wise decrease of annealing temperature from 65 °C to 55 °C.Quantification was performed in triplicate for each extraction and the optimal dilution (1:10, 1:50 and 1:100) was evaluated to observe and avoid inhibiting effects of the amplification reaction.To determine the pathway of acetate degradation, samples from the plant were collected and treated according to Schnürer and Nordberg [14].In short, aliquots of digester liquid (20 mL) were incubated with [1-14 C]-acetate (final concentration 0.11 mCi mL −1 ; Larodan Fine Chemicals, Malmö, Sweden).The degradation of [2-14 C]-acetate was detected by scintillation counting in a xylene-based scintillation fluid.A 14 CO 2 : 14 CH 4 ratio less than 1 indicates aceticlastic methanogenesis, while a ratio above 1 indicates dominance of SAO [14,26].

Conclusions
Full-scale anaerobic digestion of thin stillage is associated with multiple problems, such as high NH 4 -N concentrations (low C:N ratio) and high sulphide concentrations (low C:S ratio).Combined addition of hydrochloric acid, iron and cobalt, together with long HRT, are necessary for successful biodigestion of thin stillage.Ammonia and sulphide toxicity is counteracted by lowering the pH, precipitation of hydrogen sulphides and addition of cobalt.In addition, mesophilic temperature is applied to avoid NH 3 -N inhibition, while the long HRT avoids the risk of wash-out of slow-growing syntrophic acetate-oxidising bacteria.However, operating problems can still arise, e.g., a temporary increase in the nitrogen content of thin stillage increased the NH 3 -N level to 450 mg L −1 , resulting in stress to the microorganisms.

Figure 4 .
Figure 4. Negative correlation between hydrogen sulphide (-) and iron concentration in the digester at the Norrköping biogas plant (◊).

Figure 5 .
Figure 5. Schematic diagram of process flow at the Norrköping biogas plant.Black arrows represent organic material flows, while grey arrows represent gas flows.Grey boxes represent upflow processes, white boxes represent main production process and boxes with crosshatched lines represent downflow processes.

Table 1 .
Summary of the measures introduced to optimise anaerobic digestion of thin stillage at the Norrköping biogas plant, the underlying cause and the mode of action.