AB S T RAC T

Organic waste collection from local municipal areas with subsequent energy valorization through
CHP systems allows for a reduction of waste disposal in landfill. Pollutant emissions released
into the atmosphere are also reduced in this way. Solid oxide fuel cell (SOFC) systems are
among the most promising energy generators, due to their high electrical efficiency (>50%), even
at part loads. In this work, the local organic fraction of municipal solid waste has beendigested in
a dry anaerobic digester pilot plant and a biogas stream with methane and carbon dioxide
concentrations ranging from 60e70 and 30e40% vol., respectively, has been obtained. Trace
compounds from the digester and after the gas clean-up section have been detected by means of
a new technique that exploits the protonation reactions between the volatile compounds of
interest and the ion source. Sulfur, chlorine and siloxane compounds have been removed from
as-produced biogas through the use of commercial sorbent materials, such as activated carbons
impregnated with metals. A buffer gas cylinder tank has been inserted downstream from the
filtering section to compensate for the biogas fluctuations from the digester. The technical
feasibility of the dry anaerobic process of the organic fraction of municipal solid waste, coupled
with a gas cleaning section and an SOFC system, has been proved experimentally with an
electrical efficiency ranging from 32 to 36% for 400 h under POx conditions.

1. Introduction

The increasing demand for fossil fuels has led to intense research and development efforts
towards renewable energy sources (RES). This increased interest is reflected by the
significantincrease in the publication of patents on renewable energy (solar, wind, biomass, etc.),
see Fig. 1 [1]. In this research area, the biomass produced from a photosynthesis process, with an
almost net zero CO2 emission balance, contributes to the reduction of CO2 emissions.
Interesting results have been achieved by exploiting organic waste, and, in this context, thewaste
from the separate collection of local municipalities plays a central role. The study by Andersen et
al. (2012), for instance, shows how the environmental impact ofhome composting is generally
quite low compared to the incineration and landfilling options [2] as far as several impact
categories are concerned (especially in terms of nutrient enrichment, acidificationand eco-
toxicity in water).
The conversion of biomass into energy can be attained through different technologies:
combustion, gasification and anaerobic digestion. Combustion is the simplest and most
consolidated technical process: it involves using a burner in which the biomass is burnt. The
combusted gas is then used to heat up a traditional boiler in order to actuate a turbine in a steam
cycle.
The gasification process allows a syngas rich in hydrogen and carbon monoxide to be produced
with a relatively high LHV. This syngas is then exploited to produce energy in power devices. A
possible alternative to the production of energy with biomass is that of anaerobic digestion.
Anaerobic digestion, despite the composting process, allows not only matter but also energy to
be recovered from bio-waste. The anaerobic fermentation of organic matter is influenced to a
great extent by the pre-treatment processesthat are used to liberate the sugars needed for
fermentation [3e5]. Many types of pre-treatments have been mentioned in literature, depending
on the substrate morphology.
Among the all the possible pre-treatments, those of mechanical, chemical and biological origin
should be mentioned [3e5]. Anaerobic digestion processes of the organic fraction of municipal
solid waste (OFMSW) were first started in Europe at the beginning of the '80s in research pilot
plants [6]. Initially, such plants only exploited the wet digestion process, and were conducted
with a continuous process involving vertical cylinders. Later, the process moved from wet to dry,
due to water management problems and the nature of organic waste, as the total solids content
was above 25% when continuous systems in vertical reactors (Dranco, Valorga) or horizontal
ones (Kompogas) were used. Several plants adopted discontinuous process or batch type systems
(Arcadis, Bekon, Biofem). From 2007 to 2010, the main technologies exploited batch systems
[7]: in particular, in the middle of the 2000s, the dry-batch technique was the principal solution
that was adopted, and several plants were built in Germany, Switzerland, France and Italy.
The biogas from OFMSW considered in this paper was produced using this latter technique. This
biological process foresees the production of a biogas that can then be used in combined heat and
power generators (e.g., an internal combustion engine, a micro-turbine or a fuel cell generator).
Energy production through a traditional internal combustion engine can reach electrical
efficiencies of around 30e35% at a power scale ranging from 1 to 500 kW [8,9].
Innovative technologies, such as SOFC (solid oxide fuel cell) systems [10,11], can instead
achieve electrical energy conversion values of up to 65e70%. Syngas from gasification and
biogas from OFMSW digestion require high grade cleaning treatments, before they can be
exploited as fuel for SOFC systems, as reported by Aravind and de Jong (2012) [12]. However,
the amount and typology of the contaminants are quite different, because of both the different
biomasssubstrate that is used and due to the conversion process that is adopted. Sulfur, chlorine,
siloxanes and tar compounds represent are the main contaminants in biosyngas [12]. The most
common anaerobic digestion (AD) biogas contaminants are sulfur compounds, terpene and
carboxyl compounds [13].
Sulfur and chlorine compounds, as shown by Sasaki et al. (2011) [14], affect fuel cell
performances through nickel deactivation, sulfur being far more dangerous than chlorine. One of
the most detrimental biogas trace compounds for nickel anode deactivation is hydrogen sulfide,
as reported in several literature studies [14]. Different techniques are adopted, considering the
gas cleaning process at high and low temperatures [16]. Although information from literature and
the results from chemical equilibrium studies are, in principle, sufficient to set up a conceptual
design for the gas cleaning system, further detailed experiments are always necessary to finalize
the design of the cleaning system [17,18]. Moreover, the choice between commercial high and
low temperature gas cleaning systems can only be made after a thorough techno-economic
evaluation of all of the available alternatives. The requirements for gas cleaning can vary to a
great extent, depending on the variability of the biogas trace compounds: thus, there are no
standard solutions. A specific combination of impurity removal methods has to be used to ensure
a fuel gas with the quality that meets the fuel cell tolerance defined by the manufacturer [19].
Some works [20,21] indicate that two or at a maximum three steps are required: a primary clean-
up step, in which a condenser and a first sorbent bed are inserted, followed by a fine guard bed
before the biogas is delivered to the fuel cell system.
The use of a condenser appears to be useful, essentially to remove water. Moreover, siloxanes,
which are generally heavier than other biogas VOCs, are removed well by means of biogaswater
condensation (Hagmann et al., 2001) [22]. Papurello et al. (2014) [23] investigated how the
breakthrough fraction is affected by the type of sulfur compounds that have to be removed: it
was demonstrated that even only 1 ppm(v) of aromatic, carbonyl and halocarbon compounds can
reduce the removal filter efficiency by 11% of the expected value obtained in the case of the
removal of a single compound [23]. A first conclusion of this investigation was that the gas
cleaning section of fuel cell related applications still needs further investigation, due to the lack
of long-term experimental data under real working conditions with the SOFC directly integrated
coupled to an anaerobic digester.
A highly precise and robust method for monitoring trace compounds that are found in biogas is
thus useful to quantify the typology and amount of contaminants that might affect the long-term
operation of the SOFC Ni anode. Direct Injection Mass Spectrometry (DIMS) offers interesting
performances, in terms of rapidity, sensitivity and the absence of pretreatments [24]. One of the
most promising DIMS techniques is certainly PTR-MS (Proton Transfer Reaction-Mass
Spectrometry): this technique is based on an efficient implementation of chemical ionization
based on proton transfer from hydronium ions [24] and allows the rapid and on-line monitoring
of most volatile compounds. It has been applied in many situations, ranging from breath analysis
to environmental monitoring, and, recently, also to issues related to waste management and
odorant emission control [25,26]. Literature studies that experimentally investigate the
production of energy with SOFCs that exploit biomass are rare. Hacket et al. (2012) studied the
performance of button cells with syngas from coal as fuel [10]. Pieratti et al. (2011) studied a
CHP system based on an SOFC stack fed with biomass syngas [27]. The testwas stopped almost
immediately, after 80 min, due to poisoning problems caused by the gas.
The goal of the present work has been to build and to test an experimental pilot plant for energy
production exploiting organic waste with SOFCs. Three main research sections on biogas
exploitability are reported: (1) starting from biogas production from an OFMSW anaerobic dry
digestion process through (2) a pilot gas cleaning section with commercial sorbents to feed (3) a
500 Wel SOFC stack. The biogas trace compounds were monitored for more than 400 h with the
PTR-MS technique; the sulfur compounds at the outlet concentration were below 2 ppm (v) and
stable energy production was achieved through the direct coupling of all the investigated sections
.
2. Materials and methods

2.1. Anaerobic digester pilot plant

The digester reported in Fig. 2 consists of a parallelepiped tank with an exploitable volume of 17
m3. The digester is made up of reinforced concrete treated with resins to ensure gas tightness.
Heated strings and insulating panels were drowned inside the digester walls and floor to
achieve mesophilic temperature conditions (37 _C). Leachate from the biomass digestion is
made to flow on a cockpit grating and is then conveyed to a storage tank positioned under the
digester. This leachate is withdrawn from the storage tank, pre-heated and sprinkled over the
biomass inside the digester. This aspersion has a dual function: to feed the microbial inoculum
and to maintain the mesophilic temperature inside the digester. An internal pressure sensor was
installed to monitor the pressure inside the tank, which ranges from 2 to 10 mbar (WIKA
Instrument, LP, USA). Four gas-sampling points were installed on the digester for the continuous
chemical characterization of CO2, CH4 (infrared optical benches), O2 and H2S (electrochemical
cells) (EC 322, Eco Control Milan, Italy). The other trace compounds: sulfur, aromatic,
carbonyls, carboxyls and terpenes are detected by means of appropriate measurement campaigns,
using the innovative PTR-MS technique.

2.1.1. Mixture characteristics

The OFMSW used in the experiments came from local organic waste collection site. From
preliminary checks, the collected material showed a small amount of undesired fractions (plastic,
glass, metals, etc.). Mowed and pruned material was selected as structuring material to confer
adequate structure and porosity to the mix. These wastes were first separated with a hammer mill
(Loro e Parisini 106, Milan, Italy) and then mixed with the organic waste. Table 1 shows the pH,
total solid and volatile solid values of the raw materials used in the tests. The variability in the
measurements of the various samples analyzed over the two years of the experimentation is also
indicated. The slight variability of the physical characteristics of the OFMSW can be seen in
Table 1. Although the biomass comes from the same catchment area, it was collected in

Fig. 2. Anaerobic digester pilot plant.

different periods of the year, and there is therefore some variability
of the parameters. The structuring material shows a higher variability,
depending on whether the material was used immediately
or stored elsewhere. In particular, the volatile solid contents are
much higher in the fresh material e test a), compared to the
material stored in cumulus for a few weeks e tests b) and c). The
most appropriate mixture (30% w. OFMSW, 20%wwood and 50%w.
digestate) was identifed from several experimental campaigns
which focused on biogas yields. The mixtures adopted for the three
different tests are reported in Table 1. The total solids and volatile
solids of the three different adopted mixtures are similar. The pH
value is the parameter that shows the highest variability from batch
to batch, and it reached the highest value in test c (8.24).
2.1.2. Test setup and pre-treatments
The parameter of infuence in order to reduce the initial lag
phase of the digestion process and rapidly start the anaerobic
digestion phase, is the biomass temperature: the temperature
required to reduce the extent of this phase is about 50 _C. This
temperature increase is due to the forced aeration, which is able to
trigger aerobic bacteria and thus reduce the anaerobic lag phase
[28e32]. Coupled with the temperature increase, a subsequent
treatment with calcium hydroxide was carried out to increase the
pH of the mixture [33]. The OFMSW was homogenized with the
structuring material by means of an auger mixer (capacity 12 m 3)
(Zago King, Campo San Martino PD, Italy). The mixer was able to
pulp and tear the eco-bags. The biomass mixture was loaded into
the digester by means of a mechanical shovel (New Holland, USA).
The total weight of the input biomass was about 15.8 t, 14.9 t and
12.6 t for the three different loads (named a, b and c). When the
loading phase had been completed, the digester tailgate was closed
and a frst aerobic/anaerobic combined phase started. In just a few
hours, the temperature of the mass reached the mesophilic condition
(about 37 _C), due to the presence of the heated coils
drowned inside the walls and foor. Two k-type thermocouple
probes (Tersid, Milan, Italy), inserted inside the biomass matter,
constantly monitored the process temperature. The leachate irrigation
system provided both biomass humidifcation and inoculum
of the bacteria to maintain the digestion process. The leachate was
maintained under mesophilic conditions inside the storage tank
placed below the digester. The duration of the three tests was
different, depending on the aerobic pre-treatment. The frst test
required at least forty days to accomplish complete digestion of the
biomass. The tests were stopped when the methane concentration
fell below 40% vol. Post processing of the discharged matter was
accomplished by measuring the dynamic respirometric index (IR)
using a dynamic adiabatic respirometer (Costech International spa,
Milan Italy).
2.2. Gas cleaning section
A gas cleaning section was built according to the scheme reported
in Fig. 3. The biogas line was connected to the cleaning flter
through a 550 Wel blower (bio-komp e biokomp Srl, Sarcedo (VI)
Italy). The flter was flled with commercial activated carbon Carbox
(Air dep, San Bonifacio (VR) Italy). A 5000 g activated carbon
mass was loaded into the reactor. A pressure controller (Brkert
GmbH & Co. KG, Germany) was inserted into the gas line in order to
control an electro-valve (Brkert GmbH & Co. KG, Germany)
through a pressure sensor (WIKA Instrument, LP, USA). The electrovalve
only opens the line to the flter section when the digester
pressure is above 3 mbar, in order to maintain the microbiological
activity inside the digester. A manual valve is mounted onto the
flter to discharge the condensate. Two sampling points were used
to fll Nalophan bags before and after the biogas purifcation section.
Biogas samples were collected in Nalophan bags (available
volume 8 l) and sealed with Tefon stoppers. The time between the
sampling and analyses was kept short (30 min) in order to minimize
VOC losses from the sampling bags. Prior to each PTR-MS
analysis, and no longer than 30 min after the gas sampling, the
biogas was diluted with nitrogen gas at a volume ratio of 1:5 and
incubated at 35 _C for 30 min before the samples were submitted to
further analyses using a thermostatic bath (Techne Ltd, Cambridge,
UK).
2.3. PTR-MS for monitoring contaminants
A PTR-MS instrument was used to monitor the volatile organic
sulfur compounds (VSCs) and the other organic compounds. PTRMS
has already been successfully used to investigate the removal
performance of adsorbent materials employed for gas fltration
from gas cylinders [13,19,21]. PTR-MS is a high sensitivity mass
analytical instrument that is particularly suitable for the concentration
measurement of trace gases down to parts-per-quadrillion
volume (ppq(v)) [34]. The basic principle of PTR-MS is chemical
ionization with hydronium ions, H3O. It is a direct injection mass
spectrometric technique and does not require sample preparation
[35]. A description of the PTR-QMS 300 apparatus (Ionicon Analytik
GmbH, Innsbruck, Austria) is given in Ref. [35]. The exiting gas is
injected directly into the drift tube of the instrument via a heated
(80 _C) PEEK tube. The sampling time is close to 1e5 s for each mass
to charge ratio up to m/z 200. The drift tube conditions are 600 V,
2.01 mbar and 80 _C, which make it possible to achieve an E/N ratio
of the electric feld to the number density of the gas in the drift tube
of the PTR-MS of about 145e155 Td (Td Townsend,
1Td 10_17 Vcm2mol_1). Ten spectra were acquired for each
sample before proceeding to the next sample. The sampling time
was every 1 h during the frst day of test, and every 24 h for the
remaining days till the end of the experiment. It is possible, in
principle, to estimate the VOC concentration in the gas phase,
without any calibration, for given drift conditions, via the following
equation:
R_
_
RH
_
_
ktH3O_
_ (1)
Eq. (1) relates the unknown number of a neutral molecule, [R],
to the measured intensity of the parent ion in its protonated form
via the rate coeffcient (k), and the average time (t) spent by the
ions in the reaction region. Eq. (1) is useful to estimate gas