Bioresource Technology 158 (2014) 111–118

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Anaerobic digestion of horse dung mixed with different beddingmaterials in an upflow solid-state (UASS) reactor at mesophilicconditions

http://dx.doi.org/10.1016/j.biortech.2014.02.0340960-8524/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +49 (4441) 15 706; fax: +49 (4441) 15 469.E-mail address: jboeske@gwdg.de (J. Böske).

Janina Böske a,⇑, Benjamin Wirth b, Felix Garlipp a, Jan Mumme b, Herman Van den Weghe a

a Faculty of Agricultural Sciences, Department of Animal Sciences, Division Process Engineering, Georg-August-University of Göttingen, Universitaetsstraße 7, 49377 Vechta, Germanyb Leibniz Institute for Agricultural Engineering Potsdam-Bornim, Max-Eyth-Allee 100, 14469 Potsdam, Germany

h i g h l i g h t s

�Mesophilic anaerobic digestion of horse dung and bedding materials was studied.� Horse dung and bedding materials were tested in biochemical methane potential tests.� Methane yields of horse dung and wheat straw were examined in UASS process.� The influence of three organic loading rates on UASS systems was tested.� Two operating systems (one- and two-stage system) of UASS process were compared.

a r t i c l e i n f o

Article history:Received 6 November 2013Received in revised form 7 February 2014Accepted 10 February 2014Available online 17 February 2014

Keywords:Horse manureAnaerobic digestionMethaneSolid-stateUASS

a b s t r a c t

Aim of this study was to investigate the use of upflow anaerobic solid-state (UASS) digestion for treatinghorse manure. Biochemical methane potential (BMP) tests conducted for varying mixtures of dung (hayand silage feed) and bedding material (wheat straw, flax, hemp, wood chips) showed that straw mixedwith hay horse dung has the highest potential of 235:4 LCH4 kg�1

vs . Continuous mesophilic digestion wasconducted for 238 days using a single-stage UASS reactor (27 L) and a two-stage UASS system with ananaerobic filter (AF, 21 L). Increasing the organic loading rate (OLR) from 2.5 to 4.5 gvs L�1 d�1 enhancedthe methane rate of the single-stage reactor from 0.262 to 0.391 L L�1 d�1 while the methane yielddeclined from 104:8 to 86:9 L kg�1

vs . The two-stage system showed similar yields. Thus, for solid-statedigestion of horse manure a single-stage UASS reactor appears sufficient.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The need for alternative energy sources increased drastically inthe last years. Anaerobic digestion is a proven option to producerenewable energy. The process has the potential of convertingbiodegradable organics into biogas, which mainly consists ofmethane and carbon dioxide. There are numerous abundantsources of biodegradable organic waste including animal wastefrom pigs, cattle, poultry, or horses (Yusuf et al., 2011). The equineindustry has a substantial problem with waste disposal causingdirect costs for horse owners (Mönch-Tegeder et al., 2013). Withanaerobic digestion the costs could be compensated by energy pro-duction (Wartell et al., 2012). On average basis, a horse (454 kg)produces a daily amount of excrements of 17.5–25 kg composed

of 17 kg feces (dung) and 9 L urine (Romano et al., 2006; Wartellet al., 2012; Westendorf and Krogmann, 2004; Wheeler andZajaczkowski, 2002). A horse that is kept in boxes requires 8–10 kg of bedding per day (Häussermann et al., 2002; Westendorfand Krogmann, 2004). Because of the broad variety of beddingmaterials stall cleaning systems the ratio of dung and beddingmaterial varies from farm to farm. In general, horse manure (dung,urine and bedding) accounts for up to 25.5 kg of total raw wasteper horse per day (Westendorf and Krogmann, 2006; Wheelerand Zajaczkowski, 2002).

Straw is deemed to be the classically bedding material forhorses, which was tested in solid-state anaerobic digestion asraw material and as spent bedding material in horse stall by Cuiet al. (2011). It could be shown that the highest methane yield ofspent wheat straw was 56.2% higher than that of raw wheat straw.A current overview of the methane potential of horse manure isgiven by Mönch-Tegeder et al. (2013). They showed that especially

112 J. Böske et al. / Bioresource Technology 158 (2014) 111–118

straw based horse manure could be an efficient substrate formethane production. In batch digestion tests they found methaneyields of 0:191� 0:020 Nm3 CH4 kg�1

vs , (pure straw: 0:183—0:237Nm3 CH4 kg�1

vs ). Alternative bedding materials such as flax andwood chips led to lower methane yields. These results are inaccordance with findings of Kusch et al. (2008) who presented amethane yield of 0:170 Nm3 CH4 kg�1

vs for horse manure (horsedung with straw bedding) digested at mesophilic temperatures.Anaerobic digestion of horse manure in conventional biogas plantsis already practiced but is known to causes problems in terms ofmixing and conveying due to its fibrous nature (Hashimoto,1983; Ibrahim et al., 1997; Ward et al., 2008). In order to overcomethese problems strategies like pretreatment of the manure andadaption of the reactor design are used.

One novel reactor type designed for the continuous digestion ofsolid-materials is the upflow anaerobic solid-state (UASS) reactorintroduced by Mumme et al. (2010). As described by Pohl et al.(2012, 2013) the principle of this reactor type is based on the spon-taneous solid-liquid separation caused by natural differences indensity of substrate and process liquid.

The aim of this investigation was to test the use of continuoussolid-state digestion for the treatment of horse manure. Singleobjectives were to determine the process performance parametersof solid-state manure digestion with liquor recirculation, to com-pare a single-stage and a two-stage UASS system, and to determinethe biochemical methane potential of different mixtures of dung(hay and silage feed) and bedding materials (wheat straw, flax,hemp, wood chips).

2. Methods

2.1. Substrate properties und nutrient supplements

The substrates were obtained from different sources. The twodifferent horse manures – one from horses fed with hay and onefrom horses fed with grass silage – were collected at a horse ridingstable with 20 horses located near Vechta, Germany over a periodof 4 weeks in April 2012. Wheat straw was obtained from a stan-dard baling product from agriculture in Lower Saxony, Germanyin 2011. For batch experiments it was further chaffed to a particlesize of approximately 1–4 cm. Wheat straw, which was also usedas substrate for continuous UASS reactor systems, was choppedto a cutting length below 20 cm and did not undergo any otherpretreatments.

The other bedding materials spruce wood chips (Goldspan,Brandenburg), hemp (Hempbed, Dun Agro) and flax (Euro-Lin,Vetripharm) were obtained from a distribution for horse-ridingequipment. Their chemical properties are displayed in Table 1.

Table 1Total solid (TS), volatile solid (VS), and total carbon (C) and total nitrogen (N) contents on frdigestion (means of three single determinations with standard deviations in brackets).

Substrate TS (%FM) VS (%TS)

Hay horse manure 19.87 88.31(±0.15) (±1.08)

Silage horse manure 25.89 82.49(±0.23) (±1.09)

Wheat straw 91.55 95.71(±0.40) (±0.05)

Spruce wood chips 90.57 99.56(±0.84) (±0.02)

Hemp 87.51 81.67(±0.23) (±1.67)

Flax 88.12 98.17(±0.14) (±0.03)

The C:N ratio of every single substrate as well as for every pos-sible mixture was above the ratio of 20:1 as proposed by Weiland(2010) and also above 16:1 to 25:1 as recommend by Deublein andSteinhauser (2010). In a continuous process this could lead to ashortage of available nitrogen for microbial growth. To compensatethe low content of nitrogen the UASS reactors were additionallyfed with ammonium carbonate. Ammonium carbonate was addedon a weekly basis with a target concentration of ammonium-nitro-gen in the process liquor of 400–600 mg L�1.

As horse manure is known to also lack trace elements, espe-cially nickel, iron, cobalt, manganese, molybdenum and selenium(Mönch-Tegeder et al., 2013), the fivefold concentrated traceelement solution prepared according to medium No. 144 of the‘‘German collection of microorganisms and cell cultures’’(Brunswick, Germany) was fed to the process liquid of the UASSreactors at a dosage of 0:01 g g�1

vs;feed on a daily basis (Abdoun andWeiland, 2009).

2.2. Experimental set-up

For the biochemical methane potential (BMP) test batch exper-iments were carried out according to guideline 4630 published bythe German Engineering Association (VDI, 2006). The experimentswere conducted at mesophilic conditions (37 �C) within an incuba-tor maintaining a constant temperature of ±1 �C. Therefore, 2 Lglass bottles were inoculated with 1500 mL of reactor effluent ofprevious mesophilic biogas experiments. The applied inoculumhad a pH of 7.88, a NH4-N content of 1229 mg L�1, a total solid(TS) content of 2.88%FM and a volatile solid (VS) content of65.92%TS (1.90%FM) which is within the range recommended byVDI (2006). The overall experimental set-up is also compliant withthe VDI guideline 4630. All the bottles were connected to gas sam-pling tubes outside the incubator, which are therefore complyingwith ambient conditions. The sampling tubes are filled with a seal-ing liquid, which is saturated with NaCl salt and acidified with sul-furic acid to a pH of approximately 4. This prevents the dissolvingof carbon dioxide during the whole experiment.

Focus of the continuous experiment was the comparison of twoUASS systems with liquor recirculation. One reactor system was asingle-stage UASS reactor (UASS1, hydraulic working volume:27.36 L), the other a two-stage system, which consisted of an UASSreactor (UASS2, 27.36 L) and an anaerobic filter (AF, workingvolume 21.26 L) integrated in the process liquor recirculation asdescribed by Pohl et al. (2012). The AF was intended to preventan accumulation of volatile fatty acids (VFA). Process liquid waspumped through each reactor (including AF) from bottom to topby peristaltic pump Multifix constant MC 1000 PEC (A. SchwinherrKG, Schwäbisch Gmünd, Germany) with a volumetric flow rate of1.34 L h�1. Both reactor systems are shown in Fig. 1.

esh mater (FM) and TS basis of all substrates used for BMP tests and continuous UASS

VS (%FM) C (%TS) N (%TS) C:N ratio

17.54 46.62 1.36 34(±0.09) (±1.03) (±0.08) (±2)21.36 40.59 0.96 42(±0.33) (±1.64) (±0.04) (±1)87.63 47.35 0.50 94(±0.37) (±0.05) (±0.03) (±7)90.17 51.45 0.01 10020(±0.85) (±0.16) (±0.00) (±1148)71.47 43.88 1.14 39(±1.29) (±1.32) (±0.07) (±3)86.51 51.44 0.22 244(±0.14) (±0.14) (±0.05) (±54)

Fig. 1. Scheme of the single-stage (a) and two-stage reactor system (b) (UASS: upflow anaerobic solid-state reactor; AF: anaerobic filter).

J. Böske et al. / Bioresource Technology 158 (2014) 111–118 113

All reactors (UASS and AF) were built from stainless steel. TheUASS reactors had inspection windows consisting of acrylic glass.To ensure a stable mesophilic process temperature of 37 �C theUASS reactors as well as the AF reactor were heated by water jack-ets and thermostat Haake D8 (Thermo Haake GmbH, KarlsruheGermany). To keep the solid-state bed that forms inside the UASSreactors submerged a perforated plate was installed under the lidof the UASS reactors. The AF that was integrated into the processliquid cycle of the two-stage system was randomly packed withpolyethylene biofilm carriers (Bioflow 40, RVT Process EquipmentGmbH, Germany). The carriers have a surface area of 305 m2/m3.

2.3. Experimental procedures

Prior to the BMP test the inoculum was filled in the digestionbottles and reactivated for three days at 37�. For reference pur-poses two bottles were incubated without any substrate and twomore bottles were fed with 14.8 g cellulose to determine the con-dition of the inoculum itself. The remaining bottles were fed witheight substrate mixtures – two kinds of horse dung mixed withfour different bedding materials. To enable some statistical analy-ses each mixture was digested in triplicates. In accordance with thepractical situation and the continuous experiments, dung and bed-ding material were mixed in a FM ratio of 2:1. The requiredamount of substrate was calculated referring to the VDI guideline4630 (VDI, 2006) which states that the amount of VS addedthrough the addition of substrate should be at maximum as highas 0.5 times the VS of the inoculum. This calculation led to horsedung amounts of 22–27 g. The amount of bedding material variedbetween 11 and 14 g with a VS ratio of 1.6:1 to 2.8:1 (beddingmaterial to dung).

The batch BMP experiment was conducted over a time span of50 days. The amount of produced biogas was determined on a dailybasis until day 36. After that, only the final value was noted at day50. For normalization of the gas volumes also ambient temperatureand pressure were recorded. Gas analyses were conducted whenthe amount of collected biogas was sufficient. The inoculum refer-

ence allowed only two measurements, whereas the cellulose refer-ence and the substrates mixtures were measured five to ninetimes.

The experimental setup of the continuous experiment wasdesigned to determine the effect of different organic loading rates(OLR) of the UASS reactors (2.5, 3.5 and 4.5 gvs L�1 d�1). Theprocess temperature was fixed at 37 ± 1 �C. Both UASS reactorsthe single-stage and the two-stage setup were operated identicallythroughout the whole 238 days of the experiment. A manure mix-ture of silage horse dung and wheat straw was added daily throughthe diagonal feeding tube to the UASS reactors. The lids of thesereactors were opened to remove digestate manually once a week.It was not possible to increase the OLR beyond 4.5 gvs L�1 d�1

because of technical limitations. However, as this stage wasalready accompanied by a substantial decline in the gas yield, aneconomic efficient operation at even higher OLRs seems highlydoubtful.

2.4. Analytical methods

The determination of TS and VS was conducted according to DINstandard methods (DIN, 2001a,b). The elemental composition ofthe substrates was determined according to VDLUFA (1976). Forthe BMP batch digestion tests the composition of the produced bio-gas (CH4 and CO2) was analyzed by means of portable gas analyzerGA 2000 (ansyco GmbH, Germany) equipped with infrared detec-tors. For the continuous experiment analyzer BM 5000 (ansycoGmbH, Germany) was used to measure CH4, CO2, O2, H2S.

For process monitoring all three reactors were equipped in-pro-cess probes for pH and temperature (InPro 4260, Mettler-Toledo,USA) and a rotary drum gas meter (TG05/5 Ritter, Germany). Thesolid digestate of the UASS reactors and the effluent liquor of theUASS reactors and the AF reactor were weekly analyzed for theirchemical properties. All samples were taken on the day of weeklydigestate removal. PH, electric conductivity, TS, VS, total nitrogen,total carbon, total ammonia nitrogen, and, every four weeks, vola-tile fatty acids (VFA) were measured. Total C and total N fractions

114 J. Böske et al. / Bioresource Technology 158 (2014) 111–118

were detected by VarioMax CN (Elementar AnalysensystemeGmbH, Hanau, Germany). Volatile fatty acids (C2–C6) were mea-sured with the HPLC system 6000 of Merck-Hitachi (Merck, Darms-tadt, Germany) equipped with a Merck Prepacked Column RTPolyspher OA KC (Cat.No. 51270) and an additional Guard Column.

2.5. Calculations

The measured biogas volume V was converted to its volume atstandard temperature, standard pressure (STP) and dry conditions(VSTP,dry) using Eq. (1) from VDI guideline 4630 (VDI, 2006) withstandard conditions p0 = 1.01325 bar and T0 = 273.15 K and ambi-ent conditions for pressure (pamb) and temperature (Tamb) in thesame units.

VSTP;dry ¼ V � ðpamb � pwÞ � T0

p0 � Tambð1Þ

The vapor pressure of water pw can be calculated by differentequations each with their own range of validity. The equation cho-sen here was published by IAPWS (1992) and reads

lnppc

� �¼ Tc

Tambða1sþ a2s1:5 þ a3s3 þ a4s3:5 þ a5s4 þ a6s7:5Þ ð2Þ

with s = 1 �H and H ¼ TambT�1c . The coefficients a1 to a6 read as fol-

lows: a1 = �7.85951783, a2 = 1.84408259, a3 = �11.7866497,a4 = 22.6807411, a5 = �15.9618719, a6 = 1.80122502.

The subscript c indicates critical point conditions of water withvalues of pc = 22.064 MPa and Tc = 647.096 K.

Assuming that the volumetric fractions x of methane and car-bon dioxide sum up close to 100% the measured values were recti-fied using Eqs. (3) and (4), respectively (VDI, 2006). Each of the twomeasured values is multiplied by the same factor so that the sumof the two corrected measured values is 100% neglecting tracegases.

xCH4 ;corr ¼ xCH4 �100

xCH4 þ xCO2

ð3Þ

xCO2 ;corr ¼ xCO2 �100

xCH4 þ xCO2

ð4Þ

Fig. 2. Methane yields and methane concentrations in triple determinations (mean vaexperimentation. (HMH = hay horse manure, HMS = silage horse manure, S = wheat stra

Additionally, for the BMP batch tests the average amount of bio-gas and methane was corrected by the gas produced by the inocu-lum reference and for graphical presentation in Fig. 3 fitted to Eq.(5).

GðtÞ ¼ aþ b1 expð�k1tÞ þ b2 expð�k2tÞ ð5Þ

The statistical evaluation of the data from the UASS experimentwas carried out with the software SAS 9.1 (SAS Inst. Inc., Cary NC,USA). Only data of the last four weeks of each OLR step was usedassuming steady state conditions. The mean methane productionrates and the mean methane yields of all organic loading ratesand operating systems were calculated. For comparison of themethane yields between the different bedding materials (BMPbatch test) means (±SD) were calculated. To illustrate the timecourse of methane production, the individual daily values wereplotted graphically over 238 days. The analysis of variance wascomputed using the GLM procedure, which estimates the influenceof OLR and operating system and the interaction between both onthe methane production rate and methane yield. All data weretested for normal distribution (Shapiro–Wilk-Test). The leastsquare means (LSM) ± standard error (SE) were calculated for allthe variables. The significance level was P 6 0.05 (t-test).

3. Results and discussion

3.1. BMP batch experiments

The used inoculum was found in good condition as the cellulosereference produced an overall gas amount of 694:5 ð�1:7Þ L kg�1

vs ,which clearly exceeds the lower limit of 600 LCH4 kg�1

vs given byVDI guideline 4630 (2006). All methane yields and methane con-tents obtained after 50 days are displayed in Fig. 2.

The substrate mixtures with wheat straw showed the highestyields with values of 235.4 (±7.2) and 222:6ð�5:1Þ LCH4 kg�1

vs forhay horse manure (HMH) and silage horse manure (HMS),respectively. This is within the range reported by Kusch et al.(2008) for horse manure with straw bedding digested in atechnical scale solid-state batch system. They obtained yields ashigh as 277 LCH4 kg�1

vs after fitting and extrapolating their results

lues and standard deviations) for every substrate mixture after 50 days of batchw, F = flax, H = hemp, W = spruce wood chips).

Fig. 3. Course of the gas production and degradation behavior of the tested substrate mixtures during 50 days of batch experimentation. All data was fitted to Eq. (5).(HMH = hay horse manure, HMS = silage horse manure).

J. Böske et al. / Bioresource Technology 158 (2014) 111–118 115

to an infinite retention time. Additionally, they used fresh horsemanure also containing urine eventually leading to a better degra-dation behavior.

The yields obtained for the dung-straw-mixture in this study fitwell to the data reported by Mönch-Tegeder et al. (2013) with val-ues of 164 to 212 LCH4 kg�1

vs . They also used fresh dung as receiveddirectly from different barns.

The other bedding materials led to lower methane yields withwood chips showing the lowest values of 78.1 (±12.3) for HMHand 72:3ð�2:8Þ L kg�1

vs for HMS. This is slightly higher comparedto values reported by Wartell et al. (2012) of approximately60 L kg�1

vs reached after 80 days of incubation at 35 �C using freshsoftwood bedding and horse manure in a VS ratio of 2:1. Theyattributed these low yields to the high lignin content of woodymaterial.

Mönch-Tegeder et al. (2013) recorded a methane yield of114 L kg�1

vs using fresh horse dung mixed with sawdust as beddingmaterial. This higher yield can be explained by the lower particlesize of sawdust compared to wood chips. Barlaz et al. (1990)showed that the contact between micro-organisms and organicmass is increased by a greater particle-substrate surface area. Posi-tive effects of size reduction on biodegradability have also beendiscussed by Mata-Alvarez et al. (2000). Mönch-Tegeder et al.(2013) also report on flax as bedding material. They obtainedmethane yields of 150 L kg�1

vs after 35 days of incubation at 37 �C.These values are confirmed by corresponding values within thisstudy of 150.2 (±12.0) for HMH and 139:0 ð�7:1Þ L kg�1

vs for HMS.No methane yields for hemp as a bedding material are yet

reported in literature. Hemp showed a slightly better degradabilitycompared to flax with methane yields of 170.2 (±6.8) for HMH and167:1 ð�6:8Þ L kg�1

vs for HMS. The volumetric methane concentra-tions of the total produced biogas are in the close range around50% for every tested substrate mixture (Fig. 2).

As horse manure is a very inhomogeneous substrate the yieldsreported in literature are highly fluctuating. Wartell et al. (2012)reported of certain issues conducting several batch experimentsleading to very diverse methane yields. Nevertheless, all the meth-ane yields obtained in this study show a standard deviation of lessthan 10% except the HMH substrate mixture with spruce woodchips.

The degradation behavior itself can be seen in Fig. 3 showingthe course of the cumulative gas production along the whole

duration of the batch experiments. It can be seen that all the sub-strate mixtures led to very similar cumulative gas productioncurves. Only the course of HMH mixed with spruce wood chips isslightly different. Nevertheless, both manure-wood-mixturesshowed the fastest approach to their maximum value after 50 daysof experimentation.

3.2. UASS-reactor start-up and operation

Before the experiment was started both reactor systems weretested for gas and liquid tightness. For the inoculation the UASSreactors were filled with each 8 kg digestate from horse dungand wheat straw from a previous experiment at mesophilic condi-tions. The reaming hydraulic working volume of the UASS and AFreactors were filled with process liquor (30 L). The inoculating dig-estate had a pH of 8.40, a TS value of 12.75% and a VS value of86.96%TS. Daily feeding at an OLR of 2:5 gvs L�1 d�1 started directlyafter inoculation with the dung-straw-mixture as feedstock.

The experiment lasted 238 days meanwhile the UASS reactorswere operated with three different organic loading rates (OLR):start-up phase (day 1–63), 2:5 gvs L�1 d�1 (day 64–105),3:5 gvs L�1 d�1 (day 106–161), 4:5 gvs L�1 d�1 (day 162–210) andphase of decay when feeding was stopped (day 211–238).

In order to compensate water losses accompanied by theremoval of digestate and liquor samples, a weekly average amountof 3 L of tap water was added to each UASS reactor. Another reasonwas to prevent accumulation of high concentrations of organic andinorganic substances which can lead to an inhibition of the anaer-obic digestion process as described by Nordberg et al. (2007).

3.3. Properties of solid residue, process liquor and biogas

Following presented results refer to the analyses of solid resi-dues, process liquor and biogas in the days 64 to 210 in the periodof gradual enhancement of OLR. Throughout this experiment theconcentration of VFAs (C2–C6) in the process liquor stayed eachbelow the detection limit of 0.1 g L�1 in all UASS and AF reactorseven in the one-stage system and at the highest OLR of4:5 gvs L�1 d�1 while the concentration of VFAs in the digestateswas each below 0.4 mg g�1. These findings fit to the results of Pohlet al. (2013) who report that UASS reactors are able to handle strawfermentation up to an OLR of 6 gvs L�1 d�1 as one-stage systems.

116 J. Böske et al. / Bioresource Technology 158 (2014) 111–118

The chemical properties of the digestate were found to be stable.On average basis the digestate from the single-stage UASS (andtwo-stage system) showed a pH of 8.46 (8.43) and an NH4–N con-tent of 30.08 mg L�1 (33.70 mg L�1). The TS and VS values in thedigestates were at average about 16.35% (15.79%) and 90.72%TS

(90.64%TS).As opposed to the digestate the properties of the process liquor

changed with the progress of the experiment. The TS concentra-tions were at average about 1.10% (0.80%) but increased from0.95–1.26% (0.67–0.96%) and the VS portion of the TS increasedfrom 64.12–70.67% (60.63–67.57%), respectively. The AFs TS andVS values rose also from 0.72–0.94% and 61.27–67.20%TS. The in-crease of TS and VS in all reactors could be explained by the higheramount of fed substrate and accordingly more suspended solids inprocess liquor. Additionally salts, humic acids and microorganismsaccumulate by the time in process liquor. In the phase of decay TSvalues rose up to 1.44% (1.30%) and 1.28% in the AF. VS alsoincreased up to 69.49%TS (69.14%TS) and 68.56%TS in AFs process li-quor, which underlines the assumption of accumulation processes.The pH values were at average 7.24 (7.29) and 7.40 in AFs processliquor and increased in phase of decay to 7.60 (7.66) and 7.82 forAF which shows that the accumulations involved not acids on alarge scale. That the pH value of the AF reactor is slightly higherthan that of the UASS2 indicates that acids which accrued in UASSprocess were degraded by the AF.

As presented by Pohl et al. (2012), the electrical conductivity ofon average 7.39 mS cm�1 (7.13 mS cm�1) and 7.22 mS cm�1for theAF was dependent on the ammonium level, which was kept up at astable level to prevent any shortening. The ammonium levels wereat average at a level of 424.57 mg L�1 (466.19 mg L�1) and479.53 mg L�1 for AF. Pohl et al. (2013) showed that mono-fermentation of wheat straw would lead to an undersupply withammonium and needs additional nitrogen source for a stable pro-cess. As shown in Table 1 horse manure also contains not muchnitrogen. In practice, the addition of ammonium-nitrogen wouldnot be necessary because of horse urine containing high levels ofnitrogen (Westendorf and Krogmann, 2004; Wheeler and Zajacz-kowski, 2002).

Methane fractions of the produced biogas during the wholeexperiment were at average about 51.2% (51.2%). The remainderwas mostly carbon dioxide with an average amount of 48.8%(48.8%). The methane content of the biogas from the AF was alwayshigher than the UASS’s biogas with a mean value of 61.0%CH4 and39.0%CO2. This can be explained as follows: Firstly, favorable condi-tions for VFA degradation inside the AF are accompanied by anincrease of pH and, in consequence, a higher capacity for dissolvingCO2. Secondly, liquor recirculation carries the dissolved CO2 fromthe AF to the UASS reactor. Thirdly, VFA production inside the UASScauses parts of the dissolved CO2 to release into the gas phase. As aresult, the UASS favors production of a CO2–rich gas, whereas theAF supports production of methane enriched biogas. This featureof the UASS-AF system was reported earlier by Pohl et al. (2012).

These results are in accordance with findings of Pohl et al.(2012) for the digestion of wheat straw (41.6–60.8%CH4) and Kuschet al. (2008) who reported 51.1–53.5% of methane in solid-statedigestion of horse manure. The conducted BMP tests led to meth-ane fractions of 50% for the HMS and 51% for the HMH strawmixture.

3.4. Influence of organic loading rate and solid retention time

During regular operation the height of the solid-state bed form-ing under the top sieve was measured daily. The height was ataverage for UASS1 (and UASS2) 0.34 m (0.33 m) during OLR 2.5,0.41 m (0.42 m) during OLR 3.5 and 0.53 m (0.54 m) during OLR4.5 caused by the higher amount of fed substrate at higher OLR.

Clogging of the solid-state beds was not observed during the wholeexperiment.

To determine the solid retention time (SRT) inside the UASSreactors plastic tracer (packing material) were fed with the dailyamount of substrate. At OLR 2.5 the mean SRT in UASS1 (UASS2)was 9 (10) days. At OLR 3.5 the SRT decreased slightly to 8 (8) daysand at OLR 4.5 to 7 (7) days. This shows that the OLR has an effecton the SRT. At higher OLRs the substrate has consequently a short-er retention time in the UASS reactors. This can also be seen in Ta-ble 2.

Thamsiriroj et al. (2012), who modeled digestion of grass silageat upflow anaerobic sludge blanket (UASB), reported lower meth-ane fractions at shorter SRTs. In this investigation this effect couldnot be observed which is in accordance with the results of Pohlet al. (2013) who observed the digestion of wheat straw in UASSreactors at thermophilic conditions.

The methane production rates of UASS1 showed significanteffects (P < 0.05) for the influence of OLR. They increased from0:262 LCH4 L�1 d�1 (OLR 2.5) to 0:314 LCH4 L�1 d�1 (OLR 3.5) and0:391 LCH4 L�1 d�1 (OLR 4.5). The methane production rate of thetwo-stage system increased with OLR as well, but only the differ-ence between OLR 2.5 and 3.5 was significant. The whole data isshown in Table 2 and Fig. 4 for all reactors. As a consequence ofthe shorter SRT at higher OLRs methane yields were also influencedby OLR. The methane yield of the one-stage system decreased from104:8 LCH4 kg�1

vs to 89:6 LCH4 kg�1vs and 86:9 LCH4 kg�1

vs ; respectively(Fig. 5, Table 2). This represents only 39–47% of the methanepotential found in the BMP test after 50 days of incubationð222:6 LCH4 kg�1

vs Þ. However, looking at the BMP yield achieved inan equally short duration (Fig. 3) reveals similar values. In respectto OLR 2.5 the SRT of 9.5 d corresponds to a batch test yield of107 LCH4 kg�1

vs (48% of BMP).The findings from the UASS test are in accordance with results

from mesophilic UASS digestion of wheat straw reported by Pohlet al. (2012). They achieved 96 LCH4 kg�1

vs at an OLR of2:5 gvs L�1 d�1 and a SRT of 14-21 days. In the same investigationPohl et al. (2013) found considerably higher methane yields of161 LCH4 kg�1

vs when wheat straw was fermented at thermophilictemperatures. Other results found in literature show methaneyields of 191 LCH4 kg�1

vs in a mesophilic batch essay for straw basedhorse manure (Mönch-Tegeder et al., 2013) and 200 LCH4 kg�1

vs fordigestion of horse dung alone at a SRT of 45–60 days (Wartellet al., 2008). Fischer et al. (2013) found higher methane yields of328 LCH4 kg�1

vs when horse manure was digested in mesophilicplug-flow reactors at an SRT of up to 100 days. In a mesophilicbatch digestion investigation from Wartell et al. (2012) methaneyields of 53—231 LCH4 kg�1

vs were observed depending on the dura-tion of the batch runs (33–79 days). This shows the limitationscomparing batch reactors with continuous systems. However,based on the results and the state of knowledge it can be concludedthat the methane rate and the methane yield of horse manuredigestion in a UASS reactor mainly depend on OLR, SRT and tem-perature. In order to increase the methane yield to a maximumthe SRT should be maximized and the use of thermophilic insteadof mesophilic temperatures should be considered. For practicalapplication the consideration of optimization efforts should alsoinclude their economic effects.

3.5. Performance of a single- and two-stage UASS system

Aim of the endorsement of the UASS reactor by an AF was tostabilize the anaerobic digestion process by degrading VFAsespecially at higher OLRs. At the tested OLRs in this investigationthere was no positive effect of the two-stage system detected.Methane production rate as well as methane yield of the combinedtwo-stage system were nearly equally to the one-stage system

Table 2Influence of organic loading rate (OLR) on solid retention time (SRT), height of the solid-state bed, methane production rate (n = 167) and methane yield (n = 167) (least squaremeans ± standard error) of the upflow anaerobic solid-state (UASS) reactors and anaerobic filter (AF).

Operating system OLR Period Mean Mean height solid-state bed Methane production rate Methane yieldSRT

ðgvs L�1UASS d�1Þ (d) (d) (cm) (L L�1 d�1) ðL kg�1

vs Þ

UASS1 2.5 64–105 9 34 0.26(±0.01)a 104.8(±3.34)a

(27.36 L) 3.5 106–161 8 41 0.31(±0.01)b 89.6(±3.28)b,c

4.5 162–210 7 53 0.39(±0.01)c 86.9(±3.28)b

UASS2 + AF 2.5 64–105 10 33 0.26(±0.01)a 98.6(±3.28)a,c

(27.36 + 21.26 L) 3.5 106–161 8 42 0.32(±0.01)b,d 87.7(±3.28)b

4.5 162–210 7 54 0.35(±0.01)d 74.5(±3.28)d

a–d LSM (least squares means; GLM procedure) within the production rate and yield without common letters differ significantly (P < 0.05).

Fig. 4. Methane production rates of the upflow anaerobic solid-state (UASS) reactorsystems and the anaerobic filter (AF) analyzed at start up (days 1–63) three organicloading rates (OLR), days 64–105 OLR2.5, days 106–161 OLR3.5, days 162–210OLR4.5 and phase of decay (days 211–238); a: UASS1, UASS2 and AF, b: UASS1 andcombined system (UASS2 + AF).

Fig. 5. Methane yields of the upflow anaerobic solid-state (UASS) reactor systemsand the anaerobic filter (AF) analyzed at three different organic loading rates (OLR),days 64–105 OLR2.5, days 106–161 OLR3.5, days 162–210 OLR4.5 (a: UASS1, UASS2and AF2, b: UASS1 and combined system UASS2 + AF).

J. Böske et al. / Bioresource Technology 158 (2014) 111–118 117

(Table 2). The expected limit of OLR where the liquor becomes acidand methanization process is inhibited was not achieved at OLR4.5. These findings agree with Pohl et al. (2013) who presented thatan accumulation of VFAs in the UASS reactor was not observedbefore reaching an OLR of 8 gvs L�1 d�1. They concluded that forOLRs up to this limit the UASS works as one-stage system and anAF is not necessary. Figs. 4 and 5 show the methane productionrates and methane yields of each reactor and the combined system.UASS2 (without AF) shows compared to UASS1 lower methaneyields but this lack is nearly totally balanced by the producedmethane of AF. This tendency could indicate that at least at lowOLRs the second stage can have a disadvantageous effect on theUASS reactor of the two-stage system compared to the one-stageUASS reactor due to dilutive effects. Suspended solids are not onlymetabolized in the UASS reactor but also displaced and broken

down in the AF of a two-stage system. Because of that it is essentialto sum up the produced methane of UASS2 and AF as a combinedsystem. Contrary to OLR 2.5 and 3.5 at OLR 4.5 a significant effect ofthe operating system could be observed on methane productionrate (P = 0.013) and also on methane yield (P = 0.008) This differ-ence between the one-stage system (UASS1) and the combinedsystem (UASS2 + AF) could be explained by the higher amount ofprocess liquor in a two-stage system, which is temporary storedin liquid tanks. These tanks are necessary because the total volumeof liquid varies in time and depends on the day of digestate re-moval. Organic material is partly washed out from the substrate– lesser from straw but especially from horse dung. This is clarifiedby the rising value of TS and VS in process liquid during the time ofoperation and by an increasing OLR leading to higher amounts ofsubstrate fed to the reactors. Suspended solids can also be

118 J. Böske et al. / Bioresource Technology 158 (2014) 111–118

metabolized in these storage tanks. Kusch et al. (2009) detectedthat up to 21% of the total methane generation originated fromthe recirculated liquid. The higher the amount of easily hydrolysa-ble substrate (maize silage) the higher was the methane generationwithin the process water tank from batch operated anaerobic drydigestion. The two-stage system presented in this study had twoprocess water tanks, whereas the one-stage system included justone of these. Accordingly, the volume of stored process liquidand also the amount of lost methane was higher in the two-stagesystem. That this effect first occurred in OLR 4.5 could be explainedby the higher amount of fed substrate and subsequent increase ofsuspended solids.

4. Conclusions

The study proved the feasibility to digest horse manure by amesophilic UASS process. A mixture with straw was found to pos-sess the highest methane potential. The OLR with its influence onthe SRT was distinguished as the most crucial factor for the processperformance, whereas integration of an anaerobic filter showed noadvantages. In order to increase the methane yield to a maximum,the SRT should be maximized independently of the OLR and theuse of thermophilic instead of mesophilic temperatures shouldbe considered. Further topics for future research are processcontrol and use of other animal manures.

Acknowledgements

The work was funded by the German federal agency for renew-able resources FNR (Fachagentur für Nachwachsende Rohstoffee.V.). For their technical and analytical support the authors wouldlike to thank H.-J. Technow, H. Liebenow, H. Siebenand and U.Vehlow at Georg-August-University of Göttingen and G. Rehdeand L. Herklotz at Leibniz Institute for Agricultural EngineeringPotsdam-Bornim.

References

Abdoun, E., Weiland, P., 2009. Optimization of monofermentation from renewableraw materials by the addition of trace elements. Bornimer AgrartechnischeBerichte 68, 69–78.

Barlaz, M.A., Ham, R.K., Schaefer, D.M., 1990. Methane production from municipalrefuse: a review of enhancement techniques and microbial dynamics. CRC Crit.Rev. Environ. Control 19, 557–584.

Cui, Z., Shi, J., Li, Y., 2011. Solid-state anaerobic digestion of spent wheat straw fromhorse stall. Bioresour. Technol. 102, 9432–9437.

Deublein, D., Steinhauser, A., 2010. Biogas from Waste and Renewable Resources:An Introduction. Wiley-VCH Verlag GmbH & Co, KGaA, Weinheim.

DIN, 2001a. DIN EN 12879:2001-02 – Characterization of Sludges – Determinationof The Loss on Ignition of Dry Mass. Deutsches Institut für Normung (DIN),Beuth, Berlin.

DIN, 2001b. DIN EN 12880:2001 – Characterization of Sludges – Determination ofDry Residue and Water Content. Deutsches Institut für Normung (DIN), Beuth,Berlin.

Fischer, E., Powrosnik, A.M., Beil, C., 2013. Assessment of process stability andbiogas yield for the anaerobic digestion of horse dung in lab-scale. Landtechnik68 (4), 248–251.

Hashimoto, A.G., 1983. Conversion of straw-manure mixtures to methane atmesophilic and thermophilic temperatures. Biotechnol. Bioeng. 25 (1), 185–200.

Häussermann, A., Beck, J., Jungbluth, T., 2002. Litter materials in horse husbandry.Landtechnik 57 (1), 50–51.

IAPWS, 1992. Revised Supplementary Release on Saturation Properties of OrdinaryWater Substance. The International Association for the Properties of Water andSteam (IAPWS).

Ibrahim, I.I., Guowei, G., Jinfu, Z., Tayel, S.A., Khairy, M.F.A., El-Shimi, S.A., 1997.Anaerobic digestion for waste water poultry manure by UBF reactor. J. Environ.Sci. 9 (2), 149–161.

Kusch, S., Oechsner, H., Jungbluth, T., 2008. Biogas production with horse dung insolid phase digestion systems. Bioresour. Technol. 99, 1280–1292.

Kusch, S., Oechsner, H., Kranert, M., Jungbluth, T., 2009. Methane generation fromrecirculated liquid phase in batch operated anaerobic dry digestion. Bull.UASVM Agric. 66 (2), 110–115.

Mata-Alvarez, J., Macé, S., Llabrés, P., 2000. Anaerobic digestion of organic solidwastes. An overview of research achievements and perspectives. Bioresour.Technol. 74, 3–16.

Mönch-Tegeder, M., Lemmer, A., Oechsner, H., Jungbluth, T., 2013. Investigation ofthe methane potential of horse manure. Agric. Eng. Int. 15 (2), 161–172.

Mumme, J., Linke, B., Tölle, R., 2010. Novel upflow anaerobic solid-state (UASS)reactor. Bioresour. Technol. 101, 592–599.

Nordberg, A., Jarvis, A., Stenberg, B., Mathisen, B., Svensson, B.H., 2007. Anaerobicdigestion of alfalfa silage with recirculation of process liquid. Bioresour.Technol. 98, 104–111.

Pohl, M., Mumme, J., Heeg, K., Nettmann, E., 2012. Thermo- and mesophilicanaerobic digestion of wheat straw by the upflow anaerobic solid-state (UASS)process. Bioresour. Technol. 124, 321–327.

Pohl, M., Heeg, K., Mumme, J., 2013. Anaerobic digestion of wheat straw –performance of continuous solid-state digestion. Bioresour. Technol. 146, 408–415.

Romano, P.V., Krogmann, U., Westendorf, M.L., Strom, P.F., 2006. Reduction of fecalStreptococci and Strongyle eggs and release of Aspergillus fumigatus duringsmallscale composting of horse manure mixed with wood shavings. CompostSci. Util. 14 (2), 132–141.

Thamsiriroj, T., Nizami, A.S., Murphy, J.D., 2012. Use of modeling to aid design of atwophase grass digestion system. Bioresour. Technol. 110, 379–389.

Verein Deutscher Ingenieure (VDI) 4630, 2006. Fermentation of Organic Materials –Characterization of the Substrate, Sampling, Collection of Material Data,Fermentation Tests. Beuth Verlag, Berlin.

VDLUFA, 1976. VDLUFA Method Book III. The Chemical Analysis of Feedstuffs.Association of German Agricultural Analytic and Research Institutes (VDLUFA),Darmstadt.

Ward, A.J., Hobbs, P.J., Holliman, P.J., Jones, D.L., 2008. Optimisation of the anaerobicdigestion of agricultural resources. Bioresour. Technol. 99 (17), 7928–7940.

Wartell, B., Krumins, V., George, R., Alt, J., Schwab, B., Kang, K., Fennell, D.E., 2008.Anaerobic digestion of equine stall waste. ASABE Paper No. 084253. ConferenceProceedings Paper Presented at the 2008 ASABE Annual International Meeting.June 29–July 2, 2008. Providence, RI.

Wartell, B.A., Krumins, V., Alt, J., Kang, K., Schwab, B.J., Fennell, D.E., 2012. Methaneproduction from horse manure and stall waste with softwood bedding.Bioresour. Technol. 112, 42–50.

Weiland, P., 2010. Biogas production: current state and perspectives. Appl.Microbiol. Biotechnol. 85, 849–860.

Westendorf, M., Krogmann, U., 2004. Horses and manure. rutgers cooperativeextension. Fact sheet #036. (Accessed online June 2013 at http://www.esc.rutgers.edu/publications/stablemgt/FS036.htm).

Westendorf, M., Krogmann, U., 2006. Horses manure management: bedding use.Fact sheet #537. (Accessed online June 2013 at http://www.esc.rutgers.edu/publications/stablemgt/fs537.htm).

Wheeler, E., Zajaczkowski, J. S., 2002. Horse facilities 3: Horse manure stablemanagement. Pennsylvania State University. University Park, PA. (Accessedonline June 2013 at http://extension.psu.edu/pubs/ub035).

Yusuf, M.O.L., Debora, A., Ogheneruona, D.E., 2011. Ambient temperature kineticassessment of biogas production from co-digestion of horse and cow dung. Res.Agric. Eng. 57 (3), 97–104.