Bioresource Technology 144 (2013) 107–114

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

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

Analysis of the stability of high-solids anaerobic digestionof agro-industrial waste and sewage sludge

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.06.074

⇑ Corresponding author. Tel.: +34 943 212 800; fax: +34 943 213 076.E-mail address: mesteban@ceit.es (M. Esteban-Gutiérrez).

E. Aymerich, M. Esteban-Gutiérrez ⇑, L. SanchoCEIT and Tecnun (University of Navarra), 15 Paseo Manuel de Lardizabal, San Sebastián 20018, Spain

h i g h l i g h t s

� Similar evolution of the parameters evaluated was observed in different experiments.� The experiments ran stably although the thresholds in the literature were exceeded.� The results reveal that the dynamics are key for imbalance and inhibition phenomena.� The results are useful in developing new control strategies for high-solids AD.

a r t i c l e i n f o

Article history:Received 10 April 2013Received in revised form 19 June 2013Accepted 20 June 2013Available online 28 June 2013

Keywords:Anaerobic co-digestionHigh-solidsStabilitySewage sludgeAgro-industrial waste

a b s t r a c t

The pilot-scale high-solids anaerobic digestion (HS-AD) of agro-industrial wastes and sewage sludge wasanalysed in terms of stability by monitoring the most common parameters used to check the perfor-mance of anaerobic digesters, i.e. Volatile Fatty Acids (VFA), ammonia nitrogen, pH, alkalinity and meth-ane production. The results reflected similar evolution for the parameters analysed, except for anexperiment that presented an unsuccessful start-up. The rest of the experiments ran successfully,although the threshold values proposed in the literature for the detection of an imbalance in wet pro-cesses were exceeded, proving the versatility of HS-AD to treat different wastes. The results evidencethe need for understanding the dynamics of a high-solids system so as to detect periods of imbalanceand to determine inhibitory levels for different compounds formed during anaerobic decomposition.Moreover, the findings presented here could be useful in developing an experimental basis to constructnew control strategies for HS-AD.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

High-solids anaerobic digestion (HS-AD) or dry anaerobic diges-tion is a technology specially developed for the treatment of wastewith high total solids (TS) content, in the range of 20–35%. In par-ticular, HS-AD is an advantageous management option comparedto wet anaerobic digestion since it involves smaller space require-ments, a high volumetric organic loading rate, lower water and en-ergy consumption, and easy handling of digested waste. Despite itsnumerous advantages, it has been mainly applied to the treatmentof Organic Fraction of Municipal Solid Waste (OFMSW) (Fernándezet al., 2008; Guendouz et al., 2010). At present there are many re-search studies available in the literature that are devoted to theanaerobic digestion treatment of waste with TS content rangingfrom 5% to 15%, although recent studies dealing with HS-AD (Daiet al., 2013; Duan et al., 2012; Kim and Oh, 2011; Shi et al.,

2013) can also be found, reflecting the fact that interest in thistechnology is rising.

Several authors have analysed anaerobic digestion processes interms of stability for different types of waste by means of monitor-ing parameters such as pH, alkalinity, the Volatile Fatty Acid (VFA)concentration, the VFA/alkalinity ratio and biogas or methane pro-duction (Björnsson et al., 2000; Nielsen et al., 2007; Raposo et al.,2009; Razaviarani et al., 2013). Even though the selection of themost convenient parameters for detecting periods of imbalance isa subject of discrepancy, certain values for the abovementionedparameters have been accepted to some extent as being indicativeof unstable performance. However, there are few studies dealingwith the stability of HS-AD processes and consequently no valueranges that are indicative of imbalance have been described sofar. With regard to the VFA that is commonly monitored in anaer-obic digestion, it is noteworthy that special attention has been paidto propionic acid, and its importance as a process imbalance indi-cator has been discussed by several authors. In a study carried outby Nielsen et al. (2007), the authors concluded that propionate is akey parameter for indicating process imbalances as well as for

Fig. 1. Layout of the anaerobic digesters: pilot-scale reactor (a) and lab-scale

108 E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114

regulating and optimising the biogas process. Similarly, in previousstudies, the propionic acid concentration was found to increaseprior to digester failure (Fischer et al., 1981; van der Berg andLentz, 1977). In contrast, Pullammanappallil et al. (2001) statedthat high propionic concentrations do not necessarily indicate pro-cess imbalance. Moreover, these authors suggest that isobutyricacid could be more interesting from the point of view of processcontrol, which is in line with the findings by Ahring et al. (1995),who proposed a combined parameter that reflects the concentra-tions of butyrate and isobutyrate. Among the substances that canhave toxic effects on anaerobic digestion, ammonia nitrogen is an-other one that has been studied (Angelidaki and Ahring, 1993; Calliet al., 2005; Gallert et al., 1998). Nonetheless, there is no clear con-sensus on the ammonia levels that are inhibitory for the anaerobicdigestion process. To a great extent, the variability in the valuesthat have been reported to cause inhibition in anaerobic digestionis due to the capability of methanogenic bacteria to adapt toincreasing concentrations of ammonia. Furthermore, althoughmany authors refer to TAN (Total Ammonia Nitrogen) when dis-cussing the performance of anaerobic reactors, the use of freeammonia nitrogen (NH3) can be considered more accurate owingto the fact that this form is commonly regarded as the truly inhib-itory one.

An important feature to consider when analysing process per-formance and stability is the configuration of the anaerobic diges-ter. In this respect, systems with leachate recirculation workingunder high-solids conditions have been investigated by severalauthors. Veeken and Hamelers (2000) concluded that the leachaterecirculation rate has a strong effect on reactor performance,whereas El-Mashad et al. (2006) studied leachate recirculationduring the fed-batch digestion of solid manure and demonstratedthat it increases the system performance (i.e. methane production)in comparison to a system without leachate recycle. With regard tosystems with mechanical mixing, some authors have analysed theeffect of mixing intensity on the performance, such as Kaparajuet al. (2008). Likewise, Stroot et al. (2001) analysed the effect ofthe mixing level on the performance of anaerobic digestion, sug-gesting a negative effect of intensive mixing under high organicloading rates. Nevertheless, it must be pointed out that mixing isimportant in order to encourage distribution of enzymes andmicroorganisms throughout the reactor.

With respect to co-digestion, it is an attractive alternative thatconsists of mixing different wastes and treating them together inthe same facilities. Although co-digestion is not a new practice,the number of publications on this issue has grown markedly inthe last 5 years. It has been reported that the addition of a co-sub-strate positively affects anaerobic digestion processes since meth-ane production can be increased, depending on the operatingconditions and the characteristics of the co-substrates mixed(Mata-Alvarez et al., 2011). Moreover, co-digestion has the addedpotential of bringing about the dilution of toxic or inhibitory com-pounds and providing a better-balanced nutrient pool (Capelaet al., 2008).

The aim of this study is to evaluate the performance of HS-AD in terms of stability by treating different organic wastes,namely sewage sludge and several agro-industrial wastes. Forthat purpose, a single-substrate experiment was conducted inaddition to various co-digestion assays by using a lab-scale reac-tor equipped with a leachate recirculation system as well as acompletely mixed pilot-scale digester. During the experiments,biogas production was on-line monitored and daily analyses ofthe biogas composition were carried out in order to followmethane production. Furthermore, VFA concentration, pH, TANand alkalinity were analysed so as to check the performance ofthe process.

2. Methods

2.1. Equipment: anaerobic digesters

In this study, a pilot-scale 300 L reactor similar to a CSTR (Con-tinuous Stirring Tank Reactor) but adapted to work under high-sol-ids conditions was used (Fig. 1). The horizontal cylinder-shapedreactor is made of stainless steel, and it was operated in batch-mode under continuous mixing conditions. The temperature waskept within the mesophilic range (36 �C) with a heating systemthat is composed of six electric resistances and thanks to temper-ature signals collected by means of a Pt100 sensor. The mixingequipment consists of a 0.25 kW motor that is responsible for turn-ing a central axle with shovel shaped arms. Moreover, the digesteris equipped with a flow meter (Bronkhorst Hi-Tec v. Low-dP, Ref. F-101D-HAD-11-E), which enabled on-line measurement of biogasproduction.

Apart from the pilot-scale reactor, an 8 L (useful volume) lab-scale reactor was employed to conduct experiment 1 (Fig. 1). Thedigester is a vertical cylinder-shaped tank that contains a leachaterecycling system. This digester was operated as a batch processand under mesophilic temperature as well. Biogas productionwas measured by means of a gas meter by using the liquid dis-placement technique.

reactor (b).

E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114 109

2.2. Analytical methods

TS (Total Solids), VS (Volatile Solids), tCOD (total Chemical Oxy-gen Demand), sCOD (soluble Chemical Oxygen Demand), TKN (To-tal Kjeldahl Nitrogen), TAN (Total Ammonia Nitrogen), bicarbonatealkalinity (BA) or partial alkalinity, total alkalinity (TA) and pHwere performed according to the Standard Methods 20th Edition(1998). The samples of the organic wastes were processed accord-ing to its characteristics by ensuring the repeatability of the meth-od. The performance of the experiments was checked by means ofbiogas composition analyses and by means of analyses of the solu-ble or filtered fraction of the content of the digester (digestate).Biogas composition analyses (GC-TCD HP6890, column SUPELCO60/80 Carboxen, Ref. 10001-2390-U) were carried out on a dailybasis in order to check methane production. The filtered or solublefraction of the digestate was analysed three times a week and itwas obtained by means of the centrifugation of a representativesample of 100–50 g at 12,000 rpm during 20–30 min followed byvacuum filtration (Whatman 1.5 lm filters). VFA concentration(acetic acid, propionic acid, isobutyric acid, butyric acid, isovalericacid and valeric acid) was determined by means of an Agilent GC-6890 gas chromatograph equipped with a FID (Flame IonizationDetector) and a capillary column (DB-FFAP, 30 m � 0.25 mm i.d.,0.25 lm film; Agilent J&W: Ref. 122-3232E). The method used toobtain the sample for VFA determination consisted of processingthe filtered fraction of the digester content by using a liquid–liquidextraction with an organic solvent, i.e. TBME (Ter-Butil-Metyl-Eter). Pivalic acid was used as an internal standard solution. Thefree ammonia was calculated by using the values of TAN as wellas the temperature (36 �C) and the pH values measured throughoutthe experiments, according to the mathematical expressions thatwere used by Calli et al. (2005). The amount of C, N and O wasdetermined by using a sample that was dried overnight (60 �C)and grinded with a grinder IKA A11. To be precise, the equipmentused to determine the amount of O and N was a LECO TC-400,whereas the proportion of elemental C was determined by usinga LECO CS-200.

2.3. Experiments and feedstock composition

In the framework of this study, five HS-AD experiments werecarried out, which are briefly described in Table 1. On the one hand,four co-digestion assays were conducted by treating two vegetablespecies waste from a canning plant, i.e. Cynara carduluncus (thistle)and Cynara scolymus (artichoke), meat and bone meal (MBM) andsewage sludge. On the other hand, sewage sludge was treatedalone, representing a single substrate anaerobic digestion process(experiment 4). The sewage sludge employed underwent a 3-dayATAD (Autothermal Thermophilic Aerobic Digestion) pre-treat-ment so as to be hygienized, and hereafter it will be referred to

Table 1Details of the five experiments that have been conducted in the study.

Experiment Digester &scale

Mixing system Organic waste (substrate)

1 8 L, lab-scale Leachaterecirculation

MBM Cynaracardunculus

2 300 L, pilot-scale

Continuousmixing

MBM Cynaracardunculus

3 300 L, pilot-scale

Continuousmixing

MBM Cynaracardunculus

4a 300 L, pilot-scale

Continuousmixing

ATADsludge

–

5 300 L, pilot-scale

Continuousmixing

ATADsludge

Cynarascolymus

a Single-substrate anaerobic digestion process.

as ATAD sludge. The main characteristics of the organic wastes thathave been treated in this study are summarised in Table 2, whereasthe data corresponding to the feedstock mixtures of the HS-ADexperiments are shown in Table 3. The proportions of the differentmaterials used were selected according to Brummeler (1992). Tobe precise, inoculum from a previous anaerobic digestion processtreating the same feedstock and working under the same operatingconditions was used as the main source of active methanogenicbiomass in a proportion that amounted to 25% on a TS basis. Inaddition, an organic amendment was added in order to balancethe start-up and to adjust the TS content to around 30%. The pro-portion of substrate (organic waste) that was used in each experi-ment was 25% on a TS basis. In experiment 1, apart from theorganic amendment, wood shavings were also added to the initialmixture so that a structure that would be adequate for facilitatingleachate recirculation was obtained. It is important to note that theanaerobic biodegradability of wood shavings can be considerednegligible due to the fact that this material is characterised by ahigh organic matter content that mainly consists of lignocellulosicsubstances that can be barely degraded under anaerobicconditions.

3. Results and discussion

The evolution of the different parameters that have been ana-lysed evidences the close relationship that exists among them aswell as the necessity of using various parameters to make a gooddiagnosis of the process performance.

3.1. VFA evolution

VFA are the main intermediate compounds in anaerobic diges-tion processes, and their levels indicate the metabolic state of theobligate hydrogen producing acetogens and the acetoclastic meth-anogens (Buyukkamaci and Filibeli, 2004). Therefore, the evolutionof the VFA can be useful in detecting process imbalance. In thisstudy, VFA evolution has been monitored for the five HS-AD exper-iments (Fig. 2). Different VFA levels have been registered, presum-ably due to the particular characteristics of the feedstock, to thetechnology used and to the organic loading rate (OLR) (data notshown). On the one hand, experiment 4 showed the lowest levelsfor the VFA that were monitored. This may be attributed to thecharacteristics of the ATAD sludge and, in particular, to the factthat it is a material that has been previously treated, resulting inthe loss of part of its biodegradable organic matter. Hence, the re-sults obtained may indicate the benefits of applying a previous aer-obic treatment to that kind of waste in achieving more stableoperating conditions. On the other hand, it can be observed thatthe maximum values correspond to the acetic acid, which is thedegradation product of several organic acids, including the rest ofthe VFA. In particular, the highest value for acetic acid concentra-tion was registered in experiment 2 (Fig. 2b), where butyric acidpresented a similar trend. Although experiments 1, 2 and 3 wereconducted with the same feedstock mixture, experiment 2 pre-sented markedly higher acetic acid levels that were maintainedfor a longer period, reflecting an organic overload and processimbalance. In contrast, experiments 1 and 3 presented practicallythe same value for acetic acid, showing no accumulation. Interest-ingly, except for experiment 2, the rest of the experiments showedsignificant similarities, such as the initial build-up of acetic and n-butyric acids and their nearly simultaneous consumption.Although the acetic acid initially produced was almost completelyconsumed within the first two to three weeks of the experiments, asecond peak of this VFA appeared later in the process. Thisphenomenon may be attributed to the consumption of other

Table 2Characteristics of the organic wastes treated in this study.

Parameter MBM ATAD sludge Cynara cardunculus (thistle) Cynara scolymus (artichoke)

TS (%) 94.9 30.3 14.9 10.5VS (%) 69.4 22.0 13.0 10.0tCOD (g kg�1 DMa) 1174.3 1392.1 1005.5 1151.8sCOD (g kg�1 DMa) 225.2 41.1 201.1 8.0pH 6.14 6.8 n.a. 5.5TKN (g N kg�1 DMa) 101.8 44.3 9.02 23.07TAN (g N kg�1 DMa) 1.3 1.9 <1 1.6Cb (%) 44.5 24.9 39.7 43.8Nb (%) 10.5 3.6 0.9 1.7Ob (%) 37.6 43.0 n.a. 50.3C/N ratio 5.0 8.7 51.5 30.5

a DM: dry matter.b The amount of C, N and O is expressed in mass percentage as g element (100 g TS)�1.

Table 3Composition of the feedstock mixture used in the different anaerobic digestionexperiments (on a TS% basis) and some characteristics of the initial mixture.

Experiments

1a 2b 3b 4b 5b

ComponentOrganic amendment 40 50 50 50 45MBM 10 10 10 0 0Thistle 10 10 10 0 0Artichoke 0 0 0 0 12.5ATAD sludge 0 0 0 25 12.5Bicarbonate 0 5 5 0 5Wood shavings 15 0 0 0 0

ParameterTS (%) 30 31 32 29 31VS (%) 17 16 19 13 14

a HS-AD lab-scale experiment (8 L reactor).b Experiments conducted at pilot-scale (300 L reactor).

110 E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114

intermediary products, including other VFA. Interestingly, a secondbuild-up of acetic acid was not detected for the experiment con-ducted in the digester with leachate recirculation (Fig. 2a), whichmay be interpreted as a positive sign and could be related to theshorter duration of the process. The VFA that showed minimumlevels during the HS-AD experiments was valeric acid, which pre-sented values below 1000 mg COD L�1. Isobutyric acid and isova-leric acid evolved in parallel and were consumed shortly afterthe acetic acid and butyric acid had disappeared, possibly indicat-ing a product inhibition phenomenon caused by high acetic acidconcentrations as suggested by other authors (Pind et al., 2003).In the experiment representing an acidified process (experiment2), the evolution of these VFA showed a different pattern(Fig. 2b) that consisted of a long accumulation period of approxi-mately 70 days. The initial accumulation of VFA might be consid-ered a transitory organic overload that indicates an imbalanceamong the different biomass populations. Acidogens not only havea greater growth rate but they also withstand lower pH conditions.In contrast, acetogenic and methanogenic species have lowergrowth rates and are more sensitive to pH variations that may oc-cur in the medium. It is noteworthy that the consumption of theVFA seemed to follow a cascade pattern, particularly in the exper-iments using continuous mixing.

With regard to the propionic acid, this intermediate compoundshowed a particular evolution in the batch HS-AD experiments ofthe present study since its concentration started to increase atthe beginning and continued to rise subsequent to the consump-

tion of acetic and butyric acids (Fig. 2). The maximum valuesregistered for propionic acid were within the range of 2200–12,000 mg COD L�1, presenting the longest accumulation periodin experiment 2 (Fig. 2b), in a way that is similar to the rest ofthe VFA. Considering the evolution of propionic acid during exper-iments 1, 3, and 5, it can be stated that the leachate recirculationsystem seems to contribute to the lessening of its accumulationin comparison to continuous mixing. Therefore, the data may indi-cate that continuous mixing conditions could extend the persis-tence of propionic acid by means of disrupting the syntrophicassociations necessary to bring about its degradation, which is inaccordance with the findings of some authors (Stroot et al., 2001;Speece et al., 2006). This accumulation phenomenon has been pre-viously observed in other studies (Pullammanappallil et al., 2001;Walker et al., 2009) and it is apparently related to the difficultyin achieving the specific thermodynamic conditions required forpropionic acid metabolism. Particularly, hydrogen partial pressurein the liquid phase of the reactor must be kept within a rather nar-row window (10�4–10�6 atm) (Fukuzaki et al., 1990) in order forpropionate degradation to be thermodynamically favourable.Hence, interspecies hydrogen exchange between hydrogen-pro-ducing species and hydrogen-consuming species is one of the keypoints of propionic acid metabolism. On the contrary, the degrada-tion of other VFAs depends to a lesser extent on hydrogen concen-tration and is not thermodynamically as disfavoured as propionicconsumption, a fact which could explain their earlier consumption.According to the results discussed here, it appears that the condi-tions necessary for propionate degradation are not reached untilthe rest of the VFA have been almost completely degraded. Finally,with respect to propionic evolution, it is interesting to note that,even though some authors have demonstrated its suitability asan early indicator of process imbalance (Nielsen et al., 2007), in thisstudy the accumulation of propionic acid was not accompanied bysigns of failure. Moreover, the high concentrations measured didnot seem to cause the inhibition of methanogenesis, which is inline with the findings of Pullammanappallil et al. (2001). Likewise,the accumulation of propionic acid could be also attributed to apossible inhibition due to high acetic acid concentrations (Fuku-zaki et al., 1990).

It must be pointed out that the VFA concentration values quan-tified in the HS-AD co-digestion experiments are much superior tothose considered to be inhibitory by other authors working withlow-solids content (Wang et al., 2009). However, in a recent studyby Shi et al. (2013), acetic acid reached levels close to18,000 mg L�1 while digesting corn stover under high-solids condi-tions at thermophilic temperature. Moreover, the authors detected

0

5000

10000

15000

20000

25000

0 10 20 30 40 50 60

mg

CO

D L

-1

Time (d)

0

5000

10000

15000

20000

25000

30000

35000

0 10 20 30 40 50 60 70 80 90 100

mg

CO

D L

-1

Time (d)

0

5000

10000

15000

20000

25000

30000

0 10 20 30 40 50 60 70 80 90 100

mg

CO

D L

-1

Time (d)

0

500

1000

1500

2000

2500

0 10 20 30 40 50 60

mg

CO

D L

-1

Time (d)

0

2500

5000

7500

10000

12500

15000

17500

20000

0 10 20 30 40 50

mg

CO

D L

-1

Time (d)

(a)

(c)

(e)

(b)

(d)

Fig. 2. VFA evolution during the HS-AD lab-scale experiment (experiment 1) (a) and the HS-AD pilot-scale experiments: experiment 2 (acidified process) (b), experiment 3(c), experiment 4 (single-substrate digestion) (d) and experiment 5 (e). (d) Acetic acid, (s) propionic acid, (N) butyric acid, (4) isobutyric acid, (j) valeric acid, (h) isovalericacid.

E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114 111

the accumulation of non-acetate VFA (i.e. propionate and buty-rate), at both mesophilic and thermophilic conditions, which wasattributed to a lack of sufficient syntrophic acetogenesis. In spiteof the evolution of the VFA, the process evolved satisfactorily. Itcan be stated that the ability of the AD process to perform underthe high VFA concentrations registered in this study may be ex-plained by the fact that in HS-AD processes chemical compoundsdiffuse differently. Likewise, it is likely that inhibition phenomenamay be avoided under high-solids conditions due to a different dy-namic behaviour in which the methanogenic zones or methano-genic centres are most likely to play a key role by acting as adefensive barrier for methanogenic bacteria. Consequently, the re-sults presented here suggest that different inhibitory levels for thetypical VFA monitored in anaerobic digestion experiments shouldbe established case-by-case.

3.2. Ammonia nitrogen evolution

In anaerobic digesters, ammonia groups are typically releasedas a result of the bacterial degradation of proteins and amino acids.Total ammonia nitrogen (TAN) is a parameter that is commonlyused for the analysis of process stability owing to the fact that itmay cause inhibitions above certain levels. It is noteworthy thatin the present study TAN evolved similarly in experiments 1–4,with a pronounced increase at the beginning of the assay(Fig. 3a). With regard to the maximum levels registered for thisparameter, the values corresponding to experiments 1 and 3 arevery similar, presenting concentrations close to 4500 mg N L�1 oreven higher. Although ammonia inhibition was reported to occurin the range of 1500–3000 mg N L�1, the important role ofacclimation in adapting the methanogenic biomass to higher

0

1000

2000

3000

4000

5000

6000

0 10 20 30 40 50 60 70 80 90 100

TAN

(mg

N L

-1)

Time (d)

0

250

500

750

1000

1250

1500

0 10 20 30 40 50 60 70 80 90 100

NH

3(m

g N

L-1

)

Time (d)

5

6

7

8

9

10

0 10 20 30 40 50 60 70 80 90 100

pH

Time (d)

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80 90

CH

4%

Time (d)

0

5

10

15

20

25

0 10 20 30 40 50 60 70 80 90 100

Dai

ly C

H4

yiel

d (L

kg-

1VS

adde

d)

Time (d)

(a)

(c)

(e)

(b)

(d)

Fig. 3. Evolution of the TAN (a), free ammonia nitrogen (NH3) (b), pH (c), methane percentage (d) and daily methane yield (e) during the HS-AD lab-scale experiment (d,experiment 1) and the HS-AD pilot-scale experiments: j, experiment 2; h, experiment 3; s, experiment 4; N, experiment 5.

112 E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114

concentrations has been demonstrated (Koster, 1986). One excep-tion was observed in experiment 5, where TAN showed a slight andslow increase, as well as the minimum values compared to the restof the assays. The differences in the concentration of TAN mea-sured in the assays are in line with the composition of the organicwaste that was added in the initial mixture. In particular, the highTAN levels corresponding to experiments 1, 2 and 3 can be attrib-uted to the fact that MBM, a material rich in proteins, was added inthe feedstock mixture. With respect to sewage sludge, it is usuallyrich in fats and proteins as opposed to vegetable waste, which isprincipally composed of carbohydrates. Accordingly, the anaerobicdigestion of ATAD sludge (experiment 4) ran under higher TANconcentration in comparison with the ATAD sludge and artichokewaste co-digestion process (experiment 5). However, the proteincontent of the sludge can be assumed to be considerably lowerthan MBM since the TAN values of experiment 4 are significantly

lower than the values registered for experiments where MBMwas treated (1–3). It is noteworthy the fact that the TAN levelreached in the acidified process (experiment 2) was below the levelin the experiments treating the same feedstock, possibly suggest-ing that the hydrolysis of nitrogen containing compounds was par-tially inhibited.

Even though TAN has been discussed in studies dealing with theperformance of anaerobic digesters, the form considered to reallyhave an inhibitory effect on methanogenic bacteria is free ammo-nia. The evolution of free ammonia corresponding to the experi-ments is shown in Fig. 3b. It can be seen that the concentrationreached in experiments 4 and 5 is notably lower than in experi-ments 1 and 3, which presented considerably high maximum con-centrations of free ammonia: 1379 mg N L�1 and 1133 mg N L�1,respectively. Such values are regarded as inhibitory according toprevious research studies carried out under low-solids

E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114 113

concentration (Angelidaki and Ahring, 1993; Gallert and Winter1997). Furthermore, in a recent study conducted under high-solidsconditions, free ammonia was reported to have an inhibitory effectwhen its concentration was around 1000 mg N L�1, showing stableoperation under concentrations as high as 700 mg N L�1 due toadaptability (Kim and Oh, 2011), whereas methane yield increasedconcurrently with free ammonia concentration up to a value of548 ± 33.9 mg L�1 in the co-digestion of sewage sludge with steril-ized solid slaughterhouse waste (Pitk et al., 2013). Therefore, theexperiments conducted in the present study may reflect the capa-bility of methanogenic biomass to adapt to different operationalconditions and different levels of species that have a toxic poten-tial, such as free ammonia and VFA, proving the versatility of thesekinds of processes. Thus, the role of biomass acclimation is also afactor that one must consider when analysing the stability of aHS-AD process. This statement is in accordance with the results re-ported by Shi et al. (2013) with respect to the dynamic trend ob-served for the microbial communities during the HS-AD or solid-state AD of corn-stover (Shi et al., 2013).

Likewise, one must bear in mind the fact that in a high-solidsmedium, toxic agents like free ammonia diffuse differently andinhibition may be avoided due to different dynamic behaviour.With regard to experiment 2, the concentration of free ammoniain this assay was kept close to zero for a rather long period, mostlikely due to the persistence of an acid pH caused by VFA accumu-lation. The level of the free ammonia form in all the assays evolvedin parallel with pH, showing the lowest values at the beginning ofthe experiments.

3.3. Alkalinity, pH and methane production

Alkalinity is considered to be an important parameter in study-ing the evolution of an anaerobic digestion process since it can beuseful for detecting periods of imbalance. To be precise, VFA tobicarbonate alkalinity (VFA/BA), or intermediate alkalinity to bicar-bonate alkalinity (IA/BA) ratios have been used to evaluate the per-formance of anaerobic digestion processes (Raposo et al., 2009;Razaviarani et al., 2013). VFA are acid substances that may alterthe pH and can cause digester failure when excessively accumu-lated, whereas bicarbonate is the major species that provides alka-linity in anaerobic digestion. Therefore, the VFA/BA ratiorepresents a good approach to estimating the capability of ananaerobic digester to prevent pH acidification as well as for theearly detection of an unbalanced period where VFA are not effi-ciently consumed. For low-solids anaerobic digestion processes, athreshold value of 0.8 for VFA/BA has been fixed, and above this va-lue the process is believed to be working under unstable condi-tions. The trend followed by this ratio in the experiments wassimilar in all cases (data not shown), with an initial steep increaseand the maximum values far exceeding the aforementionedthreshold value. In particular, the highest value corresponded tothe experiment conducted in the 8 L lab-scale digester with leach-ate recirculation (experiment 1), which increased up to 28. Simi-larly, the value registered for the acidified process (experiment 2)was close to 23. Although no measures were taken earlier in theprocess, it is most likely that the value was even higher in previousdays. On the contrary, the experiment carried out with ATADsludge as single substrate registered the lowest maximum value(1.5).

The data obtained for the pH are gathered in Fig. 3c. A commontrend can be observed in the experiments, where there is a drop inthe pH in the first week, followed by a stabilization of this param-eter. However, some remarkable differences in the evolution of thepH were observed. Firstly, the pH decrease was least pronouncedin experiment 4, whereas in experiment 2 it was more pronounced,falling from an initial value of 8.5 to a value as low as 6.4. Regard-

ing the rest of the experiments, a moderate decrease was observedfor this parameter. Secondly, the pH in the digester stabilized atdifferent values and at a different time. Interestingly, in experi-ments 1 and 3 this parameter evolved in parallel and its value un-der the stable period was very similar. Despite presenting a notablylonger period of acid pH, the value corresponding to the stable per-iod in experiment 2 was nearly the same as the one in the latterexperiments. One must bear in mind that experiments 1, 2 and 3were realised using the same feedstock, which consisted of a mix-ture containing MBM. The TAN released in the degradation ofnitrogenous compounds in MBM is responsible for a rise in thepH due to its buffer capacity. Accordingly, the highest pH valuescorresponded to the experiments where MBM was treated. None-theless, it is noteworthy that in the three cases the process perfor-mance did not seem to be affected by high pH, which is most likelydue to the fact that the biomass was acclimated to high pH condi-tions in consecutive experiments. These results may suggest that,when treating the same feedstock, pH is a parameter that dependson the start-up, rather than on the configuration of the process.

With regard to methane production, it is interesting to note thatthe trend observed in all the experiments was similar in terms ofmethane percentage, except for experiment 2, which was consid-ered to have an unsuccessful start-up (Fig. 3d). According to Dear-man and Bentham (2007), methane content in the biogas superioror equal to 60% is indicative of equilibrium conditions in an anaer-obic digestion process. In the different experiments, the achieve-ment of the aforementioned value coincides with the degradationof the VFA that accumulated during the initial phase (the start-up) of the process, where acidogenesis prevailed over methanogen-esis. Nonetheless, there exist some differences with respect to theachievement of a methane percentage of 60% or higher among theassays included in this study. As shown in Fig. 3d, in experiment 4,60% methane content was reached within approximately 8 days incontrast to experiment 3, where a period of 15 days was necessary.Moreover, it is noteworthy that the methane content in the biogasmeasured during experiment 2 remained at low values for a longperiod (i.e. approximately 2 months) in comparison with the restof the experiments, indicating a period of instability and the pre-dominance of acidogenic activity over methanogenesis. Despitethe variability observed in the initial stage of the process, minordifferences were registered for the maximum methane percentagein the experiments, with values within a narrow range (70–75%).The daily methane yield of all the HS-AD experiments is repre-sented in Fig. 3e. Some meaningful differences can be observedwith respect to the occurrence and the number of peaks that havebeen registered. Nonetheless, it must be pointed out that the evo-lution of the daily methane yield is closely related to the evolutionof the VFA, in such a way that the maximum peaks detected coin-cide with the consumption of the latter. As a matter of fact, inexperiment 2 two maximum peaks (17 and 11.3 L CH4 kg�1 -VSadded) can be observed, that appeared with a considerable delayin comparison to the peaks of the rest of experiments, in accor-dance with the late consumption of VFA. Both in experiments 4and 5 the methane production peaked at day 26, with maximumvalues of 15 L CH4 kg�1 VSadded and 20 L CH4 kg�1 VSadded, respec-tively. Although the initial mixture employed in experiment 3was the same mixture that was used in experiment 2, the evolutionof the daily methane production in the former was considerablydifferent. A first peak of nearly 13 L CH4 kg�1 VSadded was detectedat day 4, with a second peak of a similar value appearing on day 20.In this experiment, the daily methane yield peaked at day 27 with avalue of 7.5 L CH4 kg�1 VSadded. With respect to the experimentconducted with leachate recirculation at lab-scale (experiment1), no clear peaks were registered for the daily methane produc-tion. Moreover, the lowest values for the methane yield corre-sponded to this experiment, since all the data registered were

114 E. Aymerich et al. / Bioresource Technology 144 (2013) 107–114

below 8 L CH4 kg�1 VSadded. The results suggest that the methaneyield of an anaerobic digestion process is notably influenced bythe configuration of the reactor. Especially, the mixing systemseems to influence the performance of the process, which is mostlikely due to the importance of ensuring a good contact betweenthe anaerobic biomass and the biodegradable organic matter.

4. Conclusions

The results of the study suggest that the understanding of thedynamic behaviour and the diffusion pattern corresponding to ahigh-solids system is key for determining the inhibitory levels ofdifferent compounds such as VFA or ammonia nitrogen, and fordetecting periods of imbalance. Moreover, the adjustment of themixing frequency seems to be an important factor for the controlof HS-AD processes since it may shorten the treatment time bykeeping propionate from accumulating. Finally, the results ob-tained could be useful in developing an experimental basis thatwould gather valuable information for constructing new controlstrategies for HS-AD.

Acknowledgements

The authors would like to thank the Spanish Ministry of Scienceand Innovation (NOVEDAR_Consolider CDS2007-00055) for finan-cial support as well as the financial funding received from ProjectNumber 10833 from the old CAN Foundation, which now belongsto the Banca Cívica Group. Moreover, the authors wish to expresstheir gratitude to the Gutarra-Grupo Riberebro canning plant andNILSA (Navarra de Infraestructuras Locales, S.A.) for providing thewastes used in this research.

References

Ahring, B.K., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acids as indicators ofprocess imbalance in anaerobic digestors. Appl. Microbiol. Biotechnol. 43, 559–565.

Angelidaki, I., Ahring, B.K., 1993. Thermophilic anaerobic digestion of livestockwaste: the effect of ammonia. Appl. Microbiol. Biotechnol. 38, 560–564.

APHA–AWWA–WEF, 1998. Standard Methods for the Examination of Water andWastewater. American Public Health Association–American Water WorksAssociation–Water Environment Federation, Washington, DC, USA.

Björnsson, L., Murto, M., Mattiasson, B., 2000. Evaluation of parameters formonitoring an anaerobic co-digestion process. Appl. Microbiol. Biotechnol. 54,844–849.

Brummeler, E.T., 1992. Dry anaerobic digestion of solid waste in the Biocel processwith a full scale unit. In: Proceedings of the First International IWA Symposiumon Anaerobic Digestion of Solid Waste, Venice, Italy, pp. 557–570.

Buyukkamaci, N., Filibeli, A., 2004. Volatile fatty acid formation in an anaerobichybrid reactor. Process Biochem. 39, 1491–1494.

Calli, B., Mertoglu, B., Inanc, B., Yenigun, O., 2005. Effects of high free ammoniaconcentrations on the performances of anaerobic bioreactors. Process Biochem.40, 1285–1292.

Capela, I., Rodrigues, A., Silva, F., Nadais, H., Arroja, L., 2008. Impact of industrialsludge and cattle manure on anaerobic digestion of the OFMSW undermesophilic conditions. Biomass Bioenergy 32, 245–251.

Dai, X., Duan, N., Dong, B., Dai, L., 2013. High-solids anaerobic co-digestion ofsewage sludge and food waste in comparison with mono digestions: stabilityand performance. Waste Manage. 33, 308–316.

Dearman, B., Bentham, R.H., 2007. Anaerobic digestion of food waste: comparingleachate exchange rates in sequential batch systems digesting food waste andbiosolids. Waste Manage. 27, 1792–1799.

Duan, N., Dong, B., Wu, B., Dai, X., 2012. High-solid anaerobic digestion of sewagesludge under mesophilic conditions: feasibility study. Bioresour. Technol. 104,150–156.

El-Mashad, H.M., van Loon, W.K.P., Zeeman, G., Bot, G.P.A., Lettinga, G., 2006. Effectof inoculum addition modes and leachate recirculation on anaerobic digestionof solid cattle manure in an accumulation system. Biosyst. Eng. 95, 245–254.

Fernández, J., Pérez, M., Romero, L.I., 2008. Effect of substrate concentration on drymesophilic anaerobic digestion of organic fraction of municipal solid waste(OFMSW). Bioresour. Technol. 99, 6075–6080.

Fischer, J.R., Ianotti, E.L., Sievers, D.M., 1981. Anaerobic digestion of swine manuresat various influent solids concentrations. Agric. Wastes 6, 157–166.

Fukuzaki, S., Nishio, N., Shobayashi, M., Nagai, S., 1990. Inhibition of thefermentation of propionate to methane by hydrogen, acetate, and propionate.Appl. Environ. Microbiol. 56, 719–723.

Gallert, C., Winter, J., 1997. Mesophilic and thermophilic anaerobic digestion ofsource-sorted organic wastes: effect of ammonia on glucose degradation andmethane production. Appl. Microbiol. Biotechnol. 48, 405–410.

Gallert, C., Bauer, S., Winter, J., 1998. Effect of ammonia on the anaerobicdegradation of protein by a mesophilic and thermophilic biowastepopulation. Appl. Microbiol. Biotechnol. 50, 495–501.

Guendouz, J., Buffière, P., Cacho, J., Carrère, M., Delgenes, J., 2010. Dry anaerobicdigestion in batch mode: design and operation of a laboratory-scale, completelymixed reactor. Waste Manage. 30, 1768–1771.

Kaparaju, P., Buendía, I., Elegaard, L., Angelidaki, I., 2008. Effects of mixing onmethane production during thermophilic anaerobic digestion of manure: lab-scale and pilot-scale studies. Bioresour. Technol. 99, 4919–4929.

Kim, D.-H., Oh, S.-E., 2011. Continuous high-solids anaerobic co-digestion of organicsolid wastes under mesophilic conditions. Waste Manage. 31, 1943–1948.

Koster, I.W., 1986. Characteristics of the pH-influenced adaptation of methanogenicsludge to ammonium toxicity. J. Chem. Technol. Biotechnol. 36, 445–455.

Mata-Alvarez, J., Dosta, J., Macé, S., Astals, S., 2011. Co-digestion of solid wastes: areview of its uses and perspectives including modeling. Crit. Rev. Biotechnol. 31,99–111.

Nielsen, H.B., Uellendahl, H., Ahring, B.K., 2007. Regulation and optimization of thebiogas process: propionate as a key parameter. Biomass Bioenergy 31, 820–830.

Pind, P.F., Angelidaki, I., Ahring, B.K., 2003. Dynamics of the anaerobic process:effect of volatile fatty acids. Biotechnol. Bioeng. 82, 791–801.

Pitk, P., Kaparaju, P., Palatsi, J., Affes, R., Vilu, R., 2013. Co-digestion of sewage sludgeand sterilized slaughterhouse waste: methane production efficiency andprocess limitation. Bioresour. Technol. 134, 227–232.

Pullammanappallil, P.C., Chynoweth, D.P., Lyberatos, G., Svoronos, S.A., 2001. Stableperformance of anaerobic digestion in the presence of a high concentration ofpropionic acid. Bioresour. Technol. 78, 165–169.

Raposo, F., Borja, R., Martín, M.A., Martín, A., de la Rubia, M.A., Rincón, B., 2009.Influence of inoculum–substrate ratio on the anaerobic digestion of sunfloweroil cake in batch mode: process stability and kinetic evaluation. Chem. Eng. J.149, 70–77.

Razaviarani, V., Buchanan, I.D., Malik, S., Katalambula, H., 2013. Pilot-scaleanaerobic co-digestion of municipal wastewater sludge with biodiesel wasteglycerine. Bioresour. Technol. 133, 206–212.

Shi, J., Wang, Z., Stiverson, A.J., Yu, Z., Li, Y., 2013. Reactor performance andmicrobial community dynamics during solid-state anaerobic digestion of cornstover at mesophilic and thermophilic conditions. Bioresour. Technol. 136, 574–581.

Speece, R.E., Boonyakitsombut, S., Kim, M., Azbar, N., Ursillo, P., 2006. Overview ofanaerobic treatment: thermophilic and propionate implications. Water Environ.Res. 78, 460–473.

Stroot, P.G., McMahon, K.D., Mackie, R.I., Raskin, L., 2001. Anaerobic codigestion ofmunicipal solid waste and biosolids under various mixing conditions—I.Digester performance. Water Res. 35, 1804–1816.

Van der Berg, L.A., Lentz, C.P., 1977. Anaerobic digestion of pear waste: laboratoryequipment design and preliminary results. J. Can. Inst. Food Technol. 4, 159–165.

Veeken, A.H.M., Hamelers, B.V.M., 2000. Effect of substrate-seed mixing andleachate recirculation on solid state digestion of biowaste. Water Sci. Technol.41, 255–262.

Walker, L., Charles, W., Cord-Ruwisch, R., 2009. Comparison of static, in-vesselcomposting of MWS with thermophilic anaerobic digestion and combinationsof the two processes. Bioresour. Technol. 100, 3799–3807.

Wang, Y., Zhang, Y., Wang, J., Meng, L., 2009. Effects of volatile fatty acidconcentrations on methane yield and methanogenic bacteria. BiomassBioenergy 33, 848–853.