Anda di halaman 1dari 23

b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Available at www.sciencedirect.com

http://www.elsevier.com/locate/biombioe

Biomass gasification as the first hot step in clean syngas


production process e gas quality optimization and primary
tar reduction measures in a 100 kW thermal input
steameoxygen blown CFB gasifier

M. Siedlecki*, W. de Jong
Energy Technology Section, Process & Energy Department, 3mE Faculty, Delft University of Technology, Leeghwaterstraat 44,
2628CA Delft, The Netherlands

article info abstract

Article history: Syngas production based on biomass gasification is an attractive, feasible alternative to
Received 27 August 2010 fossil fuel feedstock for the production of transportation fuels. However, the product gas
Received in revised form from biomass gasification must be cleaned and tailored to comply with strict syngas
21 May 2011 quality requirements, as it consists of a wide variety of major and minor components and
Accepted 30 May 2011 impurities. The characterization of such species is important to determine downstream
Available online 2 July 2011 gas treatment steps, and to assess the efficiency of the gasification process.
This paper gives an overview of the results obtained during experiments on steameoxygen
Keywords: gasification of biomass using 100 kW maximal thermal input circulating fluidized bed gasifier
Gasification (CFBG) that have been performed at Delft University of Technology during the CHRISGAS
Circulating fluidized bed (CFB) project. The unit is also equipped with a high-temperature ceramic gas filter and downstream
Syngas reactors for upgrading of the gas.
Tar In the experiments biomass types of both woody and agricultural origin have been used.
Process conditions optimization They were represented by clean wood, demolition wood, an energy crop species (mis-
canthus) and a true residue (Dutch straw), respectively. Moreover, different bed materials
have been applied, namely quartz sand, treated and untreated olivine and magnesite.
During the experiments extensive measurements of gas composition have been carried out
throughout the integrated test rig. The gas characterization included major gas compo-
nents as well as certain minor species and tar.
The results show that with the use of magnesite as bed material, remarkable increases
of hydrogen yield were attained, as compared to sand or olivine; up to a volume fraction of
almost 40% (dry, nitrogen-free basis). Also the H2:CO ratio increased from values near or
lower than 1 to 2.3e2.6. This is near the values needed, for e.g., FischereTropsch diesel
production, indicating a potential for simplification of the gas upgrading. Furthermore, by
using magnesite tar content of the raw gas was reduced to values near 2 g m3 (STP).
Moreover, magnesite complied with the expectation to have a positive impact on
agglomeration prevention for the agricultural fuels containing alkali and chlorine in the
ash. The kind of olivine applied during the experiments did not yield the expected tar
reduction; the measured tar concentration was even higher than when quartz sand was
used as bed material. Finally kaolin proved to be an effective additive to counteract the

* Corresponding author. Tel.: 31 152783120; fax: 31 152782460.


E-mail address: m.siedlecki@tudelft.nl (M. Siedlecki).
URL: http://www.pe.tudelft.nl
0961-9534/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biombioe.2011.05.033
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S41

agglomeration when fuels with high alkali content in the ash are gasified using bed
material that is rich in silica, as it is the case with quartz sand and olivine.
2011 Elsevier Ltd. All rights reserved.

1. Introduction Gasifier (CFBG) seems to be a very logical choice for biomass


conversion. Firstly, a CFBG addresses immediately the first,
In general, biomass has been recognized as a renewable and second and fifth issues related to biomass gasification: it is
potentially sustainable energy source. However, raw biomass easily scalable between 10 and 100 MW in terms of fuel
is not very convenient to use and for some applications even thermal power input [2], it can accept fuels with a quite broad
not directly suitable. This is due to the fact that: particle size distribution range (10 mm order of magnitude)
and can easily be integrated in a process with downstream gas
 biomass is a solid and, therefore, can only be distributed as upgrading process units. Issues related to operability using
any other bulk material; high-alkali fuels and tar formation have received substantial
 biomass has a relatively low energy density (typically 9  5 attention of different research groups throughout Europe, and
MJ m3, compared to 38  5 MJ m3, both on LHV-basis for can be considered as work in progress. Here the CFBG has to
natural gas), which makes transport over long distances cope with the following phenomena:
inefficient. Besides the energy density of biomass strongly
depends on the appearance and densification methods  due to the presence of an (inert) bed material inside the
applied (pelletized or loose); reactor, the risk of interactions between the fuel ash and the
 even the same kinds of biomass (e.g., wood, straw) can have bed particles always has to be considered. Agglomeration
very different composition and appearance. Therefore, it is could be initiated by particles that are covered with low-
difficult to guarantee constant product specifications; and melting (and, therefore, sticky) eutectics formed from
 conversion of solid fuel is more complicated from a tech- alkali elements present in biomass ash with bed material
nical point of view than conversion of gas or liquid. In minerals; and
particular issues like pre-treatment (size reduction), reactor  moderate tar production (typical amounts between those of
feeding, ash removal (dedusting), etc., form main hurdles updraft and downdraft gasifiers [3]), due to the relatively low
and make the processes more complicated compared to operational temperature of this type of reactor. The opera-
homogeneous gas processes or liquid/gas processes. tional temperature is technically limited by the sintering
temperature of the bed material.
The (industrial) applications where raw biomass can be
directly converted into the final product are basically limited Over the last 20e30 years significant and increasing
to (co-)combustion. Over the last decades many different research efforts are being put into the investigations related to
technologies have been proposed to convert solid biomass the gasification of biomass using fluidized bed technology.
into a more convenient secondary energy carrier. Also in this These kind of studies are not only being performed at local or
case one of the possible routes is the thermochemical national level; international communities and authorities also
conversion, and the gasification of biomass has been devel- recognize the importance of the energy transition and are
oped already far enough to come into consideration as one of consequently willing to provide the necessary financial
the contributors to the sustainable energy well during and support. The work presented in this paper has been carried out
after the energy transition. Nevertheless there are still some within the scope of the CHRISGAS project, one of the con-
issues to be addressed before large-scale commercial intro- tracted Integrated Projects from the 6th Framework Pro-
duction of biomass gasification. Those issues, depending on gramme (FP6) call; mainly supported financially by the
the type of the gasifier, are related to: European Commission. The aim of the tasks performed at the
Process & Energy laboratory at Delft University of Technology
1. scale of operation and availability of biomass; was to identify the influence of a range of process parameters
2. size distribution of raw biomass; on the performance of the circulating fluidized bed steam-
3. operability of the gasifier with fuels containing large oxygen blown gasification process with respect to gas quality in
amounts of ash, especially if the fraction of alkali, chlorine terms of major and minor species produced, and the opera-
and sulfur is high [1]; tional behavior of the gasifier. For that purpose nearly 80
4. the formation of condensable higher hydrocarbons (tar); experiments have been carried out with the 100 kW thermal
and input CFB gasification test rig within the CHRISGAS project. The
5. cleaning and upgrading of the gas for dedicated down- main objectives of these test runs were:
stream application.
 a thorough gas characterization (major and minor compo-
In view of the abovementioned disadvantages of solid raw nents), tar measurements (both quantity and composition),
biomass and the challenges posed by to-be-solved issues and solids characterization during the gasification experi-
with the process of gasification, the Circulating Fluidized Bed ments with the fuels agreed for the project;
S42 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

 application of different process conditions, bed materials


and in-bed additives as primary measures for gas upgrading
and improvement of stable operation of the reactor; and
 hot gas filtration at process temperature (800e850  C) using
ceramic candle filter.

This article presents the main results and the most


important conclusions drawn from the research introduced
above, focussing on the gasification part; the results of the
filtration research are presented in a separate article [4]. First,
the methodology is presented in the next Section, and
includes the description of the gasification test rig, the product
gas analysis methods, and the main process parameters. Then
the results of the Base Case and three Test Cases are presented
and discussed. This discussion focuses on the gas composi-
tion of the product gas with the focus on the combustible
species (mainly hydrogen), the amount and the kind of tar
measured during the experiments, and the effect of selected
process parameters on the yield of aforementioned species.
Also attention is given to the operability of the reactor with
alkali-rich fuels and to the measures taken to counteract the
agglomeration. In the final part of the article general conclu-
sions of this work are summarized.

2. Experimental facility: 100 kW thermal


input circulating fluidized bed gasifier
Fig. 1 e Schematic drawing of the Circulating Fluidized Bed
2.1. Basic geometry, instrumentation and functionality
Gasifier at TUD. Symbols 1e10: differential pressure
sensors; T1eT8: thermocouples; PR: absolute reactor
A schematic drawing of the working principle of a CFB is
pressure; Tin: gas temperature at the inlet of the reactor.
presented in Fig. 1, while a piping and instrumentation
diagram (P and ID) of the whole test rig is presented in Fig. 2.
The main characteristics of the rig are summarized below.
diameter and length, but the same power rating as the
 Geometry and materials: heaters described above;
 riser length: 5.5 m;  gas ducts to the filters were heated using heating cables
 riser inner diameter: 83 mm; downcomer inner diameter: supplied by Tyco Thermal Controls, type KMIN,
54 mm; 200 W m1, length 10 m each, maximum operating
 cyclone inner diameter: 102 mm; total height (excl. temperature of 1000  C;
flanges): 630 mm; inlet inner dimensions: 21  50 mm  electrical circulation heater supplied by Watlow, heating
(width  height); gas outlet inner diameter: 54 mm; solids power of 6 kW, was used to preheat the primary gaseous
outlet inner diameter: 45 mm; input stream (fluidization stream). Its maximum oper-
 material: stainless-steel AISI310, DIN 1.4845 for the parts ating temperature was 400  C, nominal operating
exposed to nominal process temperature and to contact temperature 360  C.
with the reactants and/or products; for other parts  Gas and solids supply systems:
stainless-steel AISI316, DIN 1.4404 is used.  feeding system with a maximum feed rate of approxi-
 Reactor heating system: mately 20 kg h1 of biomass and the possibility of inde-
 the riser, downcomer and the cyclone were heated using pendent co-feeding of two other kinds of solids (e.g.,
modular, semi-cylindrical ceramic fiber radiant heaters additives). The feeding system is designed to be gas tight;
supplied by ZMC Zamac (Poland). Length of each module during the operation it is continuously swept with a small
was 40 cm, maximum heating power 2.4 kW at 230 V, flow of nitrogen (0.7 kg h1) to maintain inert conditions
maximum operating temperature 1200  C. In total 20 within the bunkers. The biomass and bed material
heaters have been installed on the riser and 20 on the bunkers are equipped with lockehopper systems to allow
downcomer, grouped in segments consisting of 4 heaters. refilling during operation;
To obtain adequate power rating within each segment,  two optional fuel admission ports on the riser: one
pairs of heaters have been connected in series, while all directly above the gas distribution plate and one located
the pairs have been connected in parallel. This resulted in 900 mm above the gas distribution plate. In all experi-
a heating power of 2.4 kW per segment. For the cyclone ments described here the upper feeding port has been
a custom sized pair of heaters was ordered, with different used;
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S43

N back pulse

C.I. T T C.I.

PSG HT FILTER
fly ash removal
N back pulse
V-cone

T T F

to afterburner

G.A.

BWF HT FILTER
C.I.: Cascade Impactor
G.A.: Gas Analysis
CFB REACTOR

fly ash removal

from feeding sys.

purge flows

P F
T

from O2 bottles F
T

F P
T

from N2 supply

sender F P
P P F T
from air supply
from steam line

Fig. 2 e P & ID of the gasification test rig: CFBG and high-temperature ceramic filters.

 gas distribution plate consisting of nine tuyeres  for the flow control of the primary flows Samson pneu-
(do 6 mm, di 4 mm) with two holes (d 2 mm) each. matic control valves are used. Those valves are controlled
 Gas cleaning systems: by the signal from the PI-control loops in the PLC;
 high-temperature ceramic fiber tissue candle filter unit  purge flows, L-valve control, and other small flows are
(BWF, Germany) operating at 450  C; and controlled using mass flow controllers (M W Mass-
 high-temperature Si-SiC ceramic candle filter (Pall Fil- Stream) supplied by Bronkhorst High Tech B.V.;
tersystems GmbH, Germany) operating at a maximum  the flow of the product gas is measured using a differen-
temperature of approximately 800  C. tial pressure flowmeter (McCrometer V-cone);
 Reactor control and measurement systems:  eight thermocouples (K-type), of which seven are located
 flow measurement of all primary gaseous input streams, on the riser and one on the downcomer. Initially two
using Endress & Hauser AT70 thermal flowmeters for all more thermocouples have been used in the downcomer,
primary flows, except for steam, where an Endress & but they were removed as they were causing flow stag-
Hauser Prowirl 72 vortex flowmeter is used; nation in the pipe;
S44 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

 nine differential pressure (dp) cells measuring the pres- probe points into the direction of the flow to avoid the coarsest
sure drop over different parts of the installation; and particles from penetrating the line. The gas sampling line was
 in-house implemented Supervision, Control, And Data heated using a trace heating cable (resistance wire) from the
Acquisition (SCADA) package coupled to a Programmable company Horst (type HSS-450  C). Additionally, the particle
Logical Controller (PLC, make ABB, type SattCon200) for filter vessel was heated using a heating jacket (Tyco IJ-GL glass
test-rig control, which is operated from two PCs in the silk heating jacket). The temperature of both the sampling line
control room. Process data is logged with a frequency of and particle filter vessel is maintained at 300  C using
0.1 Hz (1 Hz is possible). temperature controllers. The Solid Phase Adsorption (SPA) tar
sampling point is located immediately downstream of the
filter outlet. An additional thermocouple is used to confirm the
2.2. Analysis methods gas temperature in the line at the SPA sampling point. The gas
is then led through a (primary) condenser to remove the
During the operation a sample flow of the product gas is condensables with the highest boiling point, predominately
extracted continuously from the main stream just down- heavy tar and some water, before carrying the gas to the
stream of the gas outlet of the cyclone. The schematic drawing control room, where the gas analysis takes place. After
of the sampling line is presented in Fig. 3. The gas analysis passing through a secondary condensation vessel and two

heated line to FTIR (170C)

dust filter with Rayton


cartridge (170C)

(300C)

SPA to gas analysis


probe tip
HT valve sample point (30C)
(800C)
top floor control room

(300C)
TC
flange condenser condenser gas pump
dust filter (primary) (secondary)
with glass wool
cartridge (300C)
cyclone

Fig. 3 e Main gas analysis line: schematic (top) and practical implementation (bottom).
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S45

impinger bottles filled with silica gel beads, the dried gas is Off-line measurements of the tar by SPA, where tar is adsor-
characterized with several analysis methods. For the analysis bed on a solid amino-phase porous block, in combination with
of the wet gas (Fourier Transform InfraRed spectrophotometer gas chromatography (GC) analysis of the eluted tar were also
and gravimetric water content measurement) a second gas performed. The complete analysis procedure of tar can be
sampling probe was installed to allow permanent connection found in [5]. The results of the analysis are divided into two
of the heated line (170  C), and simultaneous dry and wet parts: the polyaromatic hydrocarbon (PAH) species and the
measurements. The layout presented in Fig. 3 depicts the final phenolic species. As mentioned above the sampling point is
arrangement of the sampling lines directly downstream the located directly after the dust filter because the temperature
gasifier. needs to approach 300  C to avoid condensation of the higher
The complete overview of the analysis methods applied is tar components. A sample of 100 ml dry gas was drawn with
presented in Table 1. As shown in Table 1, the FTIR analysis is a syringe and directly sealed to avoid interaction with air and
applied to both dry and wet gas. Applying FTIR analysis to wet stored in the refrigerator. After the experimental campaign
gas has a number of advantages. It allows for the quantifica- the samples were sent for analysis to the Royal Institute of
tion of water-soluble species, such as NH3, and also water Technology (KTH, Stockholm, Sweden). The measured quan-
concentration can be measured directly, however, only up to tities were delivered in micrograms per sample, which can be
a volume fraction of approximately 20% because of peak recalculated to mg m3 (STP: 273.15 K, 101 325 Pa) of tar in the
saturation. In the process considered in this work the volume dry gas. Of course, it is the raw gas that is drawn through the
fraction of water in the gas can be expected to be between 40 SPA cartridge during sampling; however, water is partially
and 70%. Nevertheless it can still be determined from the dry adsorbed on the column, and the fraction that breaks through
and wet FTIR gas analysis data by comparing the respective condenses instantaneously. The volume of the droplets
concentrations of a compound, e.g., methane or ethylene. formed can be neglected in comparison to the volume of
Then, the water concentration (volumetric fraction) can be 100 ml of sample, and therefore, the tar sample is assumed to
calculated using the following simple formula: be taken on a dry basis.
The following PAH compounds have been quantit-
yi;wet atively analyzed by KTH: benzene, toluene, m/p-xylene,
yH2 O 1  (1)
yi;dry o-xylene, indan, indene, naphthalene, 2-methylnaphthalene,
1-methylnaphthalene, biphenyl, acenaphthylene, acenaph-
where the subscript i refers to a component measured both on thene, fluorene, phenanthrene, anthracene, fluoranthene and
a dry and wet basis. The approach applied here was to pyrene. The phenolic fraction consisted of phenol, o-cresol,
calculate the water concentration from methane and ethylene m-cresol, p-cresol, 2,4-xylenol, 2,5/3,5-xylenol, 2,6-xylenol,
concentrations and take the average of those two values. 2,3-xylenol, and 3,4-xylenol. Also, the non-identified peaks
could be quantified using an internal standard. The sum of the
unknowns was added to PAH and phenolic subtotals,
Table 1 e Overview of analyzed species and analysis respectively.
methods.
The concentrations of benzene, toluene and xylenes (BTX)
Technique Specification Components Aa Pb analyzed with SPA were known to show a relatively high error
Hartmann & Braun Online CO2 and CO 98%c 0.5% value compared to other species analyzed with the same
Uras10P (NDIR) method. Due to their volatility these components tended to
Hartmann & Braun Online O2 98% 0.5% evaporate from the sample before analysis could take place.
Magnos6G (PM) Therefore, the SPA results are not used for those components;
Varian CP4900 m-GC Semi-online BTX (benzene, n/a 1%
instead, the micro-GC results are used for further evaluation.
module: CP-Sil 5 toluene,
However, it has to be mentioned, that the concentration of the
CB, 4 m xylenes)
Varian CP4900 Semi-online N2, H2, CO, 98% 1% BTX compounds was greatly influenced by the type of adsor-
m-GC module: CO2, CH4 bent used in the gas sampling line (the impinger bottles).
CP-COX, 1 m Initially no adsorbent was used at all, but that still led to too
Fourier transform Off-line, dry CO2, CH4, 98% 10% high moisture contents in the gas entering the analyzers.
infrared (FTIR) C2H4, CO, When phosphorus pentoxide (P2O5) was used for desiccation,
ThermoElectron C2H2
it did not lead to BTX adsorption. However, its powdery form
Nicolet 5700
proved not to be very convenient for this application, so silica
Heated gas cell (150  C), Off-line, wet As above 98% 10%
2 m optical length, plus COS, gel beads were used instead. Unfortunately, those adsorb BTX
resol. 0.125 cm1 HCN, H2O, fully, until they become saturated e that is the reason why for
NH3 some experiments no BTX results are reported, since they
Gravimetric water Off-line H2 O n/a n/a were considered not to be reliable.
measurement
Solid Phase Off-line PAH, 94% 1.7%
Adsorption (SPA) Phenolics [5]
2.3. Fuels, bed materials, additives
a Accuracy.
b Precision or repeatability standard deviation (r.s.d.). 2.3.1. Fuels
c Limited by the accuracy in the composition of the calibration gas
It was agreed within the consortium that the same fuels
mixture.
would be used by all partners when carrying out the
S46 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

experimental work. These fuels represent four different


Table 2 e Composition of biomass fuels used in the
categories of biomass: CHRISGAS project.
type of fuel: A- B- Miscanthus Straw 97
 clean woody biomass: A-quality wood (further referred to as
wood wood
A-wood) delivered by the company Labee Group in Moerdijk
(NL). The fuel is delivered in pelletized form (almost white Proximate analysis, as received, raw data, mass fractions in percent
Moisture 8.10 6.20 7.30 6.80
colored), 6 mm in diameter. It represents the category of
biomass that poses the least problems during thermal Ultimate analysis, as received, normalized, mass fractions in
conversion due to its low ash content. Therefore, this fuel percent
has been used as a Base Case fuel in this study; C 47.0 47.9 45.7 40.7
H 6.13 6.02 5.93 5.50
 recycled woody biomass: B-quality (demolition) wood
N 0.203 0.752 0.678 0.596
(further referred to as B-wood) also delivered by Labee S 0.0203 0.0501 0.111 0.151
Group as 12 mm brown colored pellets. This category O 46.3 44.2 44.0 41.0
represents the more sustainable category of woody Cl 0.0122 0.0421 0.225 0.368
biomass, as it concerns a recycle stream. This type of SiO2 0.0390 0.215 1.21 7.13
biomass is characterized by a higher ash content than A- Al2O3 0.00 0.0549 0.0975 0.0897
Fe2O3 0.00 0.0846 0.0463 0.0534
wood, and also the content of trace elements in the ash is
CaO 0.118 0.224 0.184 0.480
higher. These characteristics are enough to cause opera-
MgO 0.0151 0.0565 0.129 0.114
tional problems during fluidized bed conversion (i.e., Na2O 0.00 0.0216 0.0382 0.0313
agglomeration), and also the emissions of trace elements K2O 0.0354 0.0459 0.877 2.07
with the product gas are an issue, although this topic falls P2O5 0.000 0.0138 0.179 0.118
outside the scope of this paper; Total ash (analyzed) 0.344 0.983 3.33 11.7
 energy crops: Miscanthus Giganteus (further referred to as Total Fractions 100.0 100.0 100.0 100.0
LHV (a.r.a) [MJ kg1] 15.7 16.7 15.6 14.0
miscanthus) is a fast growing plant, highly suitable for
calc.
cultivations as an energy crop. This biomass was delivered
by the company Agromiscanthus in Ter Apel (NL) as 8 mm a As received.
pellets. The ash content is significantly higher compared to
B-wood. Also, the amount of alkali elements is a clear sign of
potential for agglomerate formation when used in fluidized ash constituents have been proposed in the literature. For
bed reactors. On the other hand, miscanthus ashes have example the following so called agglomeration indicators are
been observed to be highly reactive, and they could also proposed to assess the risk of agglomeration during fluidized
show some catalytic activity during the gasification process; bed operation using sand as bed material [9]:
and
Na K
 agricultural residues: 1997 Dutch straw (further referred to agglomeration indicator 1 mass basis (2)
2S Cl
as straw), delivered as 8 mm pellets. This category, although
highly sustainable, poses the most challenges with respect
Na K Si
to thermal conversion processes when compared to the agglomeration indicator 2 mass basis (3)
Ca P Mg
kinds of biomass listed above. Its very high ash content and
ash composition form a severe threat for fluidized bed If both agglomeration indicators described above are larger
conversion (agglomeration) if no countermeasures are than 1, then they predict a high potential of alkali induced
taken. agglomeration based on the fuel characteristics in combina-
tion with SiO2 bed material at temperatures 800  C [9].
The biomass fuels used in this work had the same chain of Following the ash composition given in Table 2 (for the
custody after their receipt by the investigators. Considering calculation elemental mass fractions are needed, instead of
the scope of this study no information is requested about the oxide mass fractions), both agglomeration indicators confirm
circumstances of their growth, harvest, and post-harvest expected agglomeration risk for miscanthus and straw, as
handling and storage prior to their receipt. The chemical stated earlier. However, it has to be stressed that those indi-
composition of the biomass fuels mentioned above had been cators are purely empirical, and thus the conclusions from
analyzed by Forschungszentrum Julich (DE); the results are their use should be drawn with care, as will become clear for
summarized in Table 2. the case of B-wood.
From Fig. 4 it is clear that the content of carbon and hydrogen
on a dry, ash-free basis is nearly equal for all four fuels. 2.3.2. Bed materials
However, major differences can be observed in the ash amounts For the function as bed material in the CFB the following bulk
and compositions. Considering the amounts of silica, potas- solids had been originally selected:
sium and chlorine in miscanthus and especially straw it
is obvious that these fuels will cause agglomeration-related  quartz sand: this is an abundant and cheap material, well-
problems for the operation of the fluidized bed, if no measures proven in various (C)FB applications. The possibility of
will be taken to avoid them [6e8]. To allow a comparison purchasing a closely defined grain size range, its resistance
between different fuels and to provide tools for the assessment to attrition and its availability make sand a very attractive
of the risk of agglomeration various relationships between the basic bed material for a wide range of types and sizes of the
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S47

results have already been published [10]. The magnesite


C H N S O
used during the experiments described in this work origi-
nated from the stocks of the Varnamo plant where it has
Mass fraction d.a.f [%]

100
been delivered by the company NBc Trading, Lerum (SE);
80  olivine: magnesium iron silicate ((Mg Fe)2SiO4), one of the
most common minerals on Earth. Its properties and chem-
60 ical composition can be regarded as an intermediate
5,7 5,7 5,7 5,8
between quartz sand and magnesite. The attrition resis-
40 tance of olivine is much better than that of magnesite and
51,4 51,7 51,3 50,2 its silica content half of that of sand, though still high
20
enough to cause agglomeration. Olivine is successfully used
as bed material in the Gussing (FICFB) gasifier in Austria.
0
Laboratory studies with model tar compounds carried out in
A wood B wood Miscanth. Straw '97
the past had shown that thermal pre-treatment signifi-
Fuel cantly enhances the activity of olivine as tar-cracking
catalyst. During this project that approach has been tested
in the gasifier under real gasification conditions. The olivine
SiO2 K2O Cl total ash (analysed) used in this work originates from the company Askania AB
14 (Sweden), brand name Olivin33.
Mass fraction dry [%]

12
10 2.3.3. Additives
8 One of the greatest advantages of the fluidized bed reactor is
the possibility to use in situ additives to enhance the process.
6 Different additives are commonly used in gasification; sulfur
4 getters, alkali getters, and tar conversion agents are the most
2 well known. For the research described in this report the
following additives were used:
0
A wood B wood Miscanth. Straw '97  kaolin (kaolinite): is a clay mineral with the chemical
Fuel composition Al2Si2O5(OH)4; kaolin used in this work was
delivered by the company Creavisie (the Netherlands).
Fig. 4 e CHNSO (top) and partial ash analysis (bottom) of During the project it has been successfully shown, that it
the fuels used in the CHRISGAS project. can be effectively used as an alkali getter to counteract
agglomeration phenomena. When during biomass conver-
sion alkali are released, and also silica is present either from
the bed material or biomass ash itself, then the formation of
reactors based on fluidization technology. However, silica alkali-silicates (K2O-SiO2) can be expected. Those
easily forms low-melting eutectics with alkali compounds, compounds have an eutectic point of about 770  C, while the
which are often found in biomass, as shown in the section eutectic point of K2O-CaO-SiO2 structures is even lower.
above. Those eutectics, when melting, form a sticky layer Introduction of alumina-rich compounds, such as kaolin,
around bed material particles, causing them to form bigger may result in the formation of alkali aluminum silicates
lumps agglomerates. Therefore, precautions need to be (K2O-Al2O3-SiO2), which have a much higher melting
taken when sand bed is used during conversion of biomass temperature [8]. Therefore, during most experiments with
that is rich in alkali; fuels rich in alkali kaolin was used to allow stable operation
 magnesite: this is a mineral consisting mainly of MgCO3 during the test;
(MgO in case of calcined magnesite), and of smaller fractions  magnesite: mentioned in the above section as a possible bed
of CaO, Fe2O3 and silica. Its low silica content and acceptable material, it has been also tested as an additive. Addition of
price, being 274V t1 (M-85, Integra Group, Slovakia) versus magnesite during gasification to a sand bed resulted in
161V t1 for sand (Filcom B.V., The Netherlands) make it an significant change in gas composition and tar concentration.
attractive bed material, as it should be more resistant to the
formation of agglomerates. Its value as a bed material for
a fluidized bed has been successfully demonstrated in the 2.4. Reaction conditions
Varnamo plant in 1990s, where different difficult biomass
feedstocks have been gasified with air without extensive Process or reaction conditions are typically governed by
defluidization/agglomeration issues. A high amount of a number of characteristic variables. These variables can in
metal oxides in magnesite and some similarity with dolo- theory have very wide ranges, but from practical and scientific
mite make it also a potential catalyst in the gasification interest points of view mostly carefully predefined ranges
process. This has been observed and investigated at our have been investigated here [11e13]. A summary of those
institute during the CHRISGAS project, and some of the ranges can be found in Table 3.
S48 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

2.4.3. Pressure
Table 3 e Summary of relevant process conditions.
Although pressurized operation puts significant additional
Parameter Range requirements on the design and operation of a gasifier, it is
l [e] 0.20 0.40 often a desirable route. Firstly, higher pressures result in
SB* [e] 0.50 1.5 lower volumetric gas flow rates, which means smaller size of
Temperature [ C] 800 850 the reactor and downstream gas cleaning and upgrading
Fluidization velocity [m s1] 3.0 4.5 equipment. Secondly, many downstream processes using the
produced syngas require pressurized conditions for the reac-
tions (e.g., FischereTropsch), and the fact is that it is easier to
2.4.1. Stoichiometric oxygen ratio (l) pressurize the reactants separately (lockehopper system for
The stoichiometric oxygen ratio is commonly used for the the solids, compressors for the gases), than to compress hot,
identification of different oxidation regimes during a ther- combustible, hydrogen-rich product gas. However, the 100 kW
mochemical fuel conversion process. This parameter, also thermal input CFB gasifier in our laboratory does not facilitate
called equivalence ratio, air factor or air ratio, is repre- real pressurized operation. The maximum pressure at which
sented by the symbol l (lambda) and its formula is given the reactor can be operated is 1500 hPa (1.5 bar). Such pressure
below: increase in the system can only be caused by the back pres-
sure of the hot gas filters present downstream the reactor, or
some kind of a blockage in the pipeline downstream the
external O2 supply=fuel supply d:a:f:
l reactor.
stoichiometric O2 requirement=unit of fuel input d:a:f:
(4)
2.4.4. Steam-to-biomass ratio
Given the formula it is clear that l > 1 refers to combustion The most common fluidization and oxidation medium used
processes, l 0 to pyrolysis, and 0 < l < 1 to gasification. With in gasification processes is air. Although air is cheap and
the aim to produce a gas suitable for transportation fuels abundant, the fact that the nitrogen present in air cannot
synthesis application, high yields of H2 and CO are required; be easily separated from the product gas is a significant
this can be achieved in low lambda regions. On the other drawback of that gasification medium. Air gasification
hand, partial oxidation of the fuel is necessary to generate produces gas of low calorific value, and approximately 50%
heat to drive the mostly endothermic gasification reactions, of the volume of the product is the inert nitrogen. There-
and allow the reactor to work in the autothermal mode. Based fore, it would be better to use a combination of gases that
upon own experience, literature data and practical consider- either will react to form useful products or will be easily
ations the lambda range to be studied in this work was chosen separated from the final product stream. That gasification
between 0.2 and 0.4, so the effect of that parameter on the agent will typically consist of a gas that provides the
process could be investigated. Later on, when other process necessary oxygen for partial oxidation of the fuel (as
variables were investigated, the lambda range was narrowed explained in Section 2.4.1) and a gas that will act as
down to values between 0.30 and 0.35. a moderator/fluidization medium. A mixture of pure
oxygen and steam fulfils the criteria mentioned above and
2.4.2. Temperature both gases are very common in process industry. Besides
Temperature is an important process parameter in thermo- acting as a fluidization medium steam is also a reactant in
chemical fuel conversion. The temperature range relevant for many gasification reactions, therefore, its presence and
biomass gasification in a fluidized bed lies between approxi- amount have an influence on the product gas composition.
mately 650  C and 950  C. Higher temperature will increase The amount of steam supplied to the process is often
the carbon conversion efficiency and reduce the amount of tar related to the amount of biomass feed in a so-called steam-
produced, however, in the case of a fluidized bed reactor the to-biomass ratio:
maximum operating temperature is limited by the melting
 
point of ashes or of the bed material. Also reactor construction steam mass flow
steam  to  biomass ratio SB (5)
materials can form an issue. In practice the reaction temper- fuel feed flow
ature is directly linked to l, as for higher temperature more
product gas needs to be oxidized, which in its turn reduces the In the literature it is not often mentioned whether the fuel
cold gas efficiency (heating value of the product gas compared feed is given on as received or dry (and ash-free) basis.
to the heating value of the original fuel). However, in (small) However, in case of fuels with higher moisture and/or ash
laboratory test rigs that temperature can be controlled by the content, the difference in the calculated SB will be significant,
heating elements installed, and therefore, decoupled from l. depending on the choice of the denominator. Besides, the
This is not only interesting, but also necessary, as due to moisture present in the fuel should not be neglected in the
relatively larger heat losses in a small laboratory test rig it calculation of the SB, as the resulting steam will be the first to
might be impossible to achieve the same temperature at interact with the organic part of the fuel upon devolatilization
a certain lambda value as in a big industrial unit, solely by in the reactor. Also in case of fuels with higher moisture
autothermal operation. Considering the above, and the fact content the amount of steam that originates from the fuel will
that maximum operating temperature at TUD CFBG is limited not be negligible compared to the overall steam input.
to 900  C, the operational window has been set between 800 Considering the above, a modified steam-to-biomass ratio
and 860  C. (SB*) is proposed:
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S49

maintained at all times. Subsequently all the inlet flows were


steam  to  biomass ratio SB adjusted to the desired setpoint and after a stabilization time
steam mass flow fuel moisture mass flow of 20e30 min the measurements could be started.
(6)
dry; ash  free fuel feed flow
2.6. Research questions addressed & the test matrix
In practice as reported by the literature the range for SB varies
between 0.5 and 2.0. In this work most experiments were The progress beyond the state-of-the art in any discipline
carried out at SB* between 0.9 and 1.5. requires answers to questions that mark that state-of-the
art. During the CHRISGAS project the questions to be
2.4.5. Fluidization velocity addressed were mainly related to the influence of various
During the operation of a fluidized bed reactor the superficial parameters (l, SB*, fuel, bed material, additive, etc.) on the
fluidization velocity (often shortly referred to as fluidization process of steam-O2-blown gasification of biomass in
velocity or velocity) is an important parameter that a Circulating Fluidized Bed gasifier. The quality of the
together with the information about the bed material used product gas should be then assessed in terms of hydrogen
will reflect the fluidization mode of the reactor: yield and tar concentration, while the assessment of the
stability of the gasification process was focussed on the
  Qgas m3 s1 STP,Tprocess K,1013hPa
ufl ms1 (7) agglomeration resistance during the gasification of fuels
Ariser m2 ,Pprocess;abs hPa,273:15K
with high ash and alkali metal content. The criteria for the
For bubbling bed mode the fluidization velocity has to exceed assessment and possible optimization of the process were
the minimum fluidization velocity, while for circulating imposed by the intended application for the product gas,
fluidized bed the fluidization velocity has to exceed the namely the synthesis of liquid transportation fuels. A test
transport velocity. An overview of minimum fluidization and matrix was set up to indicate the necessary experiments
transport velocities of the bed materials used in this research that should give answers to questions indicated above.
is given in Table 4. The gasifier was operated at fluidization From the beginning a Base Case was defined being gasifi-
velocities slightly exceeding the transport velocities given in cation of A-wood using a sand bed without any additives.
Table 4 and above. The maximum fluidization velocity applied The reason for defining this as a Base Case is the fact that
was below 5 m/s. Those ranges are well within the ranges gasification of A-wood, being an easy fuel in a fluidized
applied in industrial circulating fluidized bed gasifiers. bed reactor with standard bed material, should proceed
reliably before any attempts are undertaken to look for
2.5. Start-up improvements. Table 5 gives a global overview of the
variations of bed material/fuel/additive combinations
Gasification experiments have been run on the test rig during different stages of the project. A mixture of steam
described in the previous section. The operation of the gasifier and oxygen was used as the fluidization/oxidation medium
was started by applying a nitrogen flow through the gas during all the experiments reported in this paper. The
preheater allowing the primary flow to heat up to a tempera- gasification campaigns 2006 and 2008-1 were mostly
ture of about 300  C. The flow of the hot gases also caused focused on the investigations of the effects of varying
heating-up of the parts downstream the filters, that were not process parameters like l, SB and temperature, together
heated electrically, so the risk of steam condensation was with reactor operability issues. During the campaigns 2008-
anticipated in that way. When the temperature in every part 2, 2009-1 and 2009-2 parameters l and SB were kept rather
of the test rig exceeded 100  C, the fluidization agent was constant and the emphasis was put on the effects of bed
switched to steam and bed material circulation was initiated. materials different from sand on the gas composition and
The initial bed mass amounted to approximately 7 kg, to be tar production.
fed into the riser using the feeding system. Biomass feed was
started as soon as the average temperature in the reactor was
above 500  C, and as soon as CO was detected by the gas 3. Results and discussion
analysis, confirming that biomass is actually entering the
reactor, the oxygen flow was started. The strategy was to In order to present the results of the gasification experiments
reach the operational temperature (i.e., 830e850  C) as quickly in a transparent way, the results have been grouped as pre-
as possible, ensuring that substoichiometric conditions were sented below. The letter following the Test Case identifier
refers to the fuel used in that experiment: A (A-wood), B
(B-wood), M (Miscanthus), S (1997 Dutch straw).

Table 4 e Overview of minimum fluidization and  Base Case (BC): A-wood, sand as bed material;
transport velocities for the bed materials used in this  Test Case 1 (SC): sand as bed material, B-wood (SC-B) and
research. Miscanthus (SC-M) as fuel, and eventually the use of an
Sand Magnesite Olivine Olivine additive to counteract agglomeration;
(untreated) (treated)  Test Case 2 (MC): magnesite as bed material, with any of the
four fuels considered here (MC-A, MC-B, MC-M, MC-S);
Umf [m s1] 0.05 0.03 0.12 0.1
Utr [m s1] 3.1 2.8 3.9 3.7
 Test Case 3 (OC): olivine as bed material with A-wood,
B-wood and Miscanthus as fuel. For some experiments
S50 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Table 5 e Test matrix; the numbers indicate the amount of experiments carried out with a certain bed material/fuel/
additive combination in each gasification campaign.
Bed Fuel Additive Gasification campaign

2006 2008-1 2008-2 2009-1 2009-2

Oct ApreMay JuleAug JaneMay OcteDec

Sand A-wood e 8 4 4 e e
Sand B-wood e e 2 1 e e
Sand B-wood Kaolin e 1 e e e
Sand Miscanthus Kaolin e 3 2 e e
Magnesite A-wood e e e 3 2 1
Magnesite B-wood e e e e 2 e
Magnesite Miscanthus e e 3 e 3 1
Magnesite Straw Kaolina e 3 e e 2
Olivine (untr.) A-wood e e e e 2 e
Olivine (untr.) B-wood Kaolin e e e 1 1
Olivine (untr.) Miscanthus Kaolin e e e e 1
Olivine (tr.) B-wood Kaolin e e e 2 1
Olivine (tr.) Miscanthus Kaolin e e e e 1
No. of exps: Subtotal: 54 8 16 10 12 8
No. of exps: 2005: 13
No. of exps: Total: 67

a Kaolin used only during last experiment.

thermally pre-treated olivine was used.900indicates (gravimetric or FTIR) and the process data of the gasifier
a batch treatment at the temperature between 900 and (data log).
1000  C for 8 h, 1200 refers to a similar treatment, but at
1200  C. U indicates the use of untreated (fresh) olivine. So Figs. 5e9 present the key process data for the evaluation
OC(900)-M1 refers to the first experiment of the Test Case periods of Base Case and the three Test Cases introduced
3-series with olivine as bed material pre-treated at 900  C above. Table 6 presents the process data and main measure-
and Miscanthus as fuel. ment results from eight selected experiments.
All evaluation periods are in each case grouped by the kind
of fuel, and sorted by l in ascending order. The only exception
3.1. Overview of base case and test cases results is the Test Case MC-A where, in view of nearly constant
stoichiometric oxygen ratio, the sorting has been done by SB*,
A gasification experiment carried out with the test rig in order to illustrate the effect of that variable on the gas
described above, with the duration of 8 h will yield 2880 composition.
data points for each of the ca. 150 variables logged. There-
fore, it is clear that the results presented here are the 3.2. Base case
outcome of a compromise between the completeness and
the conciseness. Of the 67 experiments mentioned in Table The Base Case as defined in this work partly represents a well-
5, 61 evaluation periods have been selected for further established condition in the gasification technology (fluidized
analysis. The criteria for the selection of those evaluation bed gasification of clean wood) with the relatively novel aspect
periods were: of the use of steam and oxygen as fluidization and oxidation
media. Steameoxygen gasification itself is not new, but the
 the operation of the gasifier should be in steady state, detailed reporting on applied process conditions and
meaning that all mass and energy flows are constant over measurement results are still scarce in the open literature. The
longer time frame (here: tens of minutes order of magni- Base Case experiments can be divided in two groups: first four
tude). Short term variations (here: tens of seconds order of test are carried out at relatively constant SB* ratio, but with
magnitude) are inherent to this kind of process, and are crushed A-wood pellets; whereas the last four tests have been
caused, e.g., by variations in the particle size distribution of carried out with normal pellets while applying more variation
the fuel, screw feeding of the fuel, on-off control of the in SB*. The use of the crushed pellets was dictated by initial
reactor heating system, etc. However, those short-term problems with the flow of particles through the solids outlet of
variations do not affect the steady state as considered on the cyclone, but after a slight modification there normal pellets
long term basis; could be used. The results from the Base Case and Test Cases
 there should be a complete set of measurements taken confirm that an increase in l will lead to a decrease in the
during the considered steady state. The required data set concentrations of the combustibles. Here, with combustibles
consists of main gas components analysis (NDIR and is referred to the sum of the volume fractions of H2, CO, CH4 and
(micro-)GC), tar analysis (SPA), water content of the raw gas C2-components; light hydrocarbons and tar are considered
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S51

Fig. 5 e Main gasification parameters, gas composition and tar concentration for the BC.

separately. The highest value is 73% for l of 0.18 (BC-1). This put unambiguously as there is an important difference
effect is well known and confirmed by numerous earlier between these two groups of experiments, namely the fuel
publications. Noticeable is also the fluctuation of the amount of particle size distribution. The roughly assessed Sauter diam-
combustibles, as well as the concentration of the measured eter (d32) of the crushed pellets is approximately 50% of that of
polyaromatic (PAH) and phenolic tar compounds with the the original pellets (3.5 and 6.9 mm, respectively). It can be
variation of SB* (BC-1 to BC-4). This is according to the earlier concluded, that a decreased particle size has an adverse effect
observations that the dependence of tar concentration on on the tar concentration in the gas. This can be supported by
steam-to-biomass ratio has a hyperbola-like shape [14]. Further the theory, that the amount of tar produced during gasification
increase in SB* (BC-5 till BC-8) seems to have a positive effect on strongly depends on the interaction of big, volatile aromatic
tar reduction, see Table 7. However, this conclusion cannot be structures with the char during the pyrolysis stage. (Poly)
S52 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Fig. 6 e Main gasification parameters, gas composition and tar concentration for the SC.

aromatic chains that are given a longer residence time inside the hyperbolic relation between tar conversion and steam-to-
a char particle will be cracked into smaller molecules, leading biomass ratio, this behavior has been expected.
to lower tar concentrations in the product gas. This is clearly
observed here. Experiments BC-3 and BC-6 are carried out 3.3. Test case 1
under very similar process conditions, but the amount of
measured PAH en Phenolics is significantly lower for BC-6 (37 Test Case 1 (Sand case, SC) was defined to test the operation
and 39%, respectively). On the other side smaller particle size of the gasifier with basic bed material (sand), but with fuels
distribution will lead to a higher burnout rate (higher carbon more demanding than clean wood. Both fuels tested here, B-
conversion); this is confirmed by the mass balance of those two wood and miscanthus, have higher ash content than A-wood.
experiments (see Section 3.8). In addition, miscanthus ash is rich in alkali compounds, which
The effect of SB* > 1 is less pronounced than for the lower are known to be problematic in terms of agglomeration and
values of that variable. However, as mentioned before, due to defluidization phenomena. More on this topic is discussed in
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S53

Fig. 7 e Main gasification parameters, gas composition and tar concentration for the MC-A and MC-B.

Section 3.7. The gasification conditions applied during Test not cause any further reduction of the concentration of PAH or
Case 1 are similar to the Base Case (BC). Also the results are Phenolics. The concentration of the BTX is decreasing with
comparable, although the amount of measured combustible increasing SB*. Gasification of demolition wood (B-wood) or
components seems to be lower for SC-B than for the BC. This energy crops (miscanthus) is more beneficial from the
even when taking into account that the concentrations of the sustainability point of view than the gasification of clean wood
C2 are not available for the measurements SC-B2, SC-B3 and (A-wood). Therefore, it is very promising that the results
SC-M2. This could be explained by the fact that the B-wood regarding the yield of combustible compounds and tar are
pellets were the biggest of all fuels applied, which could have similar to those from the Base Case. The challenges, however,
led to a worse burnout characteristics. The amount of tar for are mainly posed by the ashes and their composition. The
both SC-B and SC-M is comparable with BC-5 till BC-8. In SC-M phenomena related to alkali metals will be commented upon
the first step in SB* (from 0.85 to 1.3) leads to a clear decrease in later. In addition, demolition wood might contain amounts of
the quantity of measured tar, but the second step (to 1.6) does trace elements (often heavy metals) that may cause
S54 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Fig. 8 e Main gasification parameters, gas composition and tar concentration for the MC-M and MC-S.

unacceptable emissions. This topic falls, however, outside the miscanthus without any additives even up to 11 h. With sand
scope of this research. such duration of the operation would not be possible. Next to
that, an additional benefit of the application of magnesite has
3.4. Test case 2 been observed, namely its activity as, presumably, a WGS and
hydrocarbon-converting catalyst. The main evidence of the
The main initial objective of Test Case 2 (Magnesite Case, activity of the new bed material was a clearly increased H2:CO
MC) was to test the promising properties of magnesite in ratio, from below 1 during the BC and SC to over 2 for the MC.
terms of the resistance to agglomerate formation during the The results of MC-A have already been reported by the
gasification of alkali-rich fuels. The test results indeed authors in an earlier publication [10]. In addition tests MC-A2
confirmed those expectations allowing the gasification of till MC-A5 have been performed to investigate the influence of
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S55

Fig. 9 e Main gasification parameters, gas composition and tar concentration for the OC.

the SB* ratio on the product gas composition and the amount error margin. In addition the combination of magnesite as bed
of measured tar. As can be seen in Fig. 7 the increase in SB* material and miscanthus as fuel led to the lowest PAH
does not lead to a clear change in tar concentration. The concentration (2.2 g m3 (STP)) measured during the 67 gasi-
summed concentration of PAH and Phenolics equals 3.7 g m3 fication experiments carried out at Delft University of Tech-
(STP) for MC-A1 which is a reduction of about 8% compared to nology within the CHRISGAS project (see also Fig. 10).
the average measured in MC-A2 and MC-A3. This difference
falls almost within the error limits of the measurement. A 3.5. Test case 3
measurable difference in the concentrations of PAH and
Phenolics can be observed between the MC-A and MC-M The results obtained from Test Case 2 were highly promising
cases. Table 8 presents the averaged results of light hydro- from the in situ gas upgrading and tar reduction point of view.
carbons and tar measurements carried out for the cases Nonetheless magnesite also has its drawbacks, as per the
mentioned before. The decrease is especially pronounced for earlier mentioned low attrition resistance, and its price being
the PAH compounds, while the Phenolics fluctuate within the higher than that of quartz sand. Although the authors are
S56 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Table 6 e Process parameters and main measurement results from eight selected experiments.
BC-15 SC-B04 SC-M10 MC-M03 MC-S11 OC(U)-B03 OC(900)-B01 OC(900)-M06

Date: 25-Jul- 25-Apr- 23-Jul-2008 26-May- 30-May- 28-Apr- 13-May- 07-Dec-


2008 2008 2008 2008 2009 2009 2009

Fuel: A-wood B-wood Miscanthus Miscanthus Straw B-wood B-wood Miscanthus

Bed Sand Sand Sand Magnesite Magnesite Olivine(U) Olivine(900) Olivine(900)


material

Additive e Kaolin Kaolin e e Kaolin Kaolin Kaolin


1
Steam kg h 11.4 12.3 10.8 12.6 12.3 11.7 11.4 19.5
O2 kg h1 4.4 3.2 4.4 4.4 3.8 4.6 4.6 7.09
Biomass kg h1 13.2 10.3 15.6 10.5 10.6 11.1 11.7 15.6
Air kg h1 1.2 1.3 1.2 1.8 1.7 1.7 1.4 1.9
N2 kg h1 2.1 4.3 2.1 5.1 3.2 2.0 2.7 2.9

Mean riser T C 835 821 831 836 831 827 830 837

Gas inlet T C 360 360 360 397 360 364 360 360
Pgage CFB out hPa (mbar) 79 52 76 n.m. 38 n.m. n.m. n.m.
l kg kg1 0.28 0.26 0.24 0.36 0.35 0.34 0.32 0.38
SB* kg kg1 1.0 1.4 0.85 1.4 1.5 1.2 1.1 1.5
H2O/O2 mol mol1 4.6 6.8 4.4 5.1 5.7 4.5 4.4 5.0
CO2 vol. fraction, dnf % 31.0 30.7 31.7 42.6 45.3 35.6 36.0 35.9
CO vol. fraction, dnf % 33.7 31.7 31.4 12.7 12.7 28.7 30.1 30.7
H2 vol. fraction, dnf % 21.8 21.9 22.8 35.9 32.0 21.2 19.4 21.5
CH4 vol. fraction, dnf % 8.9 8.6 9.5 5.5 5.8 8.0 8.3 8.5
C2H4 vol. fraction, % 2.8 3.3 2.6 1.3 1.4 3.1 3.7 2.6
dnf
C2H2 vol. fraction, % 0.43 0.50 0.34 0.055 0.15 0.48 0.55 0.40
dnf
N2 vol. fraction, dry % 13.8 31.5 12.6 29.8 28.0 22.0 18.7 15.8
H2O vol. fraction, raw % 59 57.1 56.6 49.0 58.0 67.1 53.1 62.8
BTX concentration, dnf g m3 (STP) 38 33 42 6.5 23 n.m. n.m. n.m.
PAH concentration, dnf g m3 (STP) 8.3 10.1 11.4 2.2 3.7 12.6 12.8 11.3
Phenolics g m3 (STP) 2.2 4.4 4.6 0.40 1.2 1.9 1.7 2.0
concentration, dnf

convinced that the benefits of the use of magnesite as bed fresh and pre-treated olivine on tar conversion under real
material justify those drawbacks, they were also aware of the gasification conditions.
existence of other promising bed material candidates. Devi
et al. [15,16] investigated the effect of olivine on the conver- 3.5.1. Pre-treatment
sion of naphthalene in a small bench-scale reactor where The fresh olivine that was supposed to be used as bed material
gasification conditions had been simulated. One of the was pre-treated at two different temperatures: 900 and
conclusions of that research was that a thermal pre-treatment 1200  C. Also an attempt has been made to pre-treat a batch at
(calcination) of the bed material causes the migration of iron 1000  C, but due to the limitations of the oven that was used
molecules towards the surface of the particle. This effect
could contribute to the increase of catalytic activity of olivine
on tar-cracking reactions and has been demonstrated in the
bench-scale reactor mentioned above. The main objective of
Table 8 e Differences in the concentrations of different
Test Case 3 (Olivine case, OC) was to investigate the effect of
hydrocarbons between the gasification experiments with
A-wood and miscanthus.
Case MC-A SD MC-M SD Difference
Table 7 e Average applied SB* and average measured tar (averages) (averages) [%]
concentrations for the two groups of Base Case l 0.35 0.021 0.33 0.042 7.1
experiments.
SB* 1.15 0.23 1.26 0.15 9.7
Average Average Change
Hydrocarbons and tar, dnf [g m3 (STP)]
BC-1 BC-5 [%]
Benzene 14.4 0.794 15.1 5.2
till BC-4 till BC-8
MAH 8.11 1.25 4.96 39
SB* 0.92 1.3 41 PAH 3.54 0.378 2.57 0.393 27
Measured PAH [dnf] 15 8.9 40 Phenols 0.468 0.147 0.308 0.117 34
Measured Phenolics 3.9 3.5 9 Measured 12.1 1.17 7.84 35
[dnf] tar
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S57

Fig. 10 e Overview of all Test Cases: minimum and maximum PAH and Phenolics concentrations and total measured and
combustible gas fraction (top); contribution of each tar class to the total amount of tar measured, toluene and xylenes not
included (bottom).

initially, the maximum temperature reached was 980  C. That 3.5.2. Results
led to the division of applied olivine into three categories: The results presented in this work consider the experiments
carried out with two different fuels, namely B-wood and
 fresh (untreated) olivine; miscanthus. Fig. 9 shows the main results in terms of gasifi-
 olivine pre-treated at temperature between 900 and 1000  C; cation parameters, gas composition and tar concentration.
and Immediately it is clear that the gas composition is similar to
 olivine pre-treated at 1200  C. the Base Case and Test Case 1. This means that the kind of
olivine used here (fresh or pre-treated) does not have any
In all cases the duration of the treatment was 8 h (hold time noticeable influence on, e.g., wateregas shift equilibrium, as
at setpoint) plus the time necessary to heat up and cool-down was observed in Test Case 2. The measured amount of PAH
the batch (ca. 200  C h1). and Phenolics is comparable to their amounts measured in
S58 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

SC-B and SC-M. During the analysis of the SPA samples lead to increased conversion of tar (and, therefore,
collected during the 2009-2 gasification campaign the GC increased hydrogen yield), although that effect will
analysis method was changed slightly, which allowed the diminish when SB* will increase further beyond these
detection and quantification of a number of PAH heavier than values. This has also been discussed earlier;
pyrene. As those components were not considered in the  in terms of the effect on hydrogen production the perfor-
earlier results, they have been shown separately from other mance of olivine as bed material (OC) is similar to that
PAH in Fig. 9, in order not to bias the comparison of the observed in the BC. The presence of magnesite (MC) leads to
measured tar data. Unfortunately, the effect of the pre- a significant increase in hydrogen concentration at the
treatment of olivine cannot be assessed unambiguously expense of carbon monoxide. This was already made clear
based on the obtained results. What is clear and remarkable is during one of the experiments in SC with A-wood where
that the total percentage of measured combustibles increases magnesite was solely used as an additive; the measured
with increasing l (case OC-M). In the cases presented before volume fraction of hydrogen was approximately 50% higher
exactly the opposite could be seen. Also the sum of the volume than in the BC [10]. As stated earlier, simultaneous decrease
fractions of the measured main gas components is observed of the volume fraction of CO points towards the effect on
to increase with the olivine pre-treatment temperature, both homogenous wateregas shift reaction. Furthermore, the
for OC-B as OC-M, from 96.8 to 103.3% and from 94.8 to 99.6%, effect of magnesite is not limited to the wateregas shift
respectively. Obviously the value exceeding 100% is a conse- reaction. Also the of methane, C2-compounds and tar are
quence of measurement inaccuracies, nonetheless the trend reduced;
is clear. The increase in the total percentage of the measured  the highest volume fraction of hydrogen is achieved in MC,
main gas components is an indication of the fact that less during miscanthus gasification. Values close to 40% on
unknown or non-measurable species are present in the gas. a dry, nitrogen-free basis were measured during five
Such species could be tar. However, the results of the SPA experiments. It should be mentioned that hydrogen
analysis do not give a clear confirmation of that hypothesis. It concentrations measured during miscanthus gasification
can be seen that the Phenolic species decrease sharply with are clearly higher than the values measured during the
increasing treatment temperature, however, on the other gasification of other fuels under similar process conditions.
hand PAH are increasing. The sum of CxHy hydrocarbons is This could be a confirmation of the special properties of
rather constant; the remaining possibility is that benzene or 1- miscanthus ash, which was already suspected based on
ring aromatic tar compounds are being converted at a higher previous experience with this fuel at Delft University of
rate if the bed material has been treated under the more Technology.
severe conditions. Unfortunately, due to the problems with
the adsorption of the light hydrocarbons in the sampling line
(on the silica gel beads used as a desiccant), no reliable values 3.6.2. Light hydrocarbons and tar
for BTX have been recorded to confirm the above. The formation of tar is nearly inevitable in fluidized bed
gasification, as due to moderate reactor temperature and
3.6. Overall comparison relatively short gas residence time long hydrocarbon chains
cannot be fully converted to short(er) molecules. However,
3.6.1. H2 and other combustibles different steps can be taken to reduce the amount of
One of the main objectives of the CHRISGAS project was to produced tar. The work within this project was focussed on
investigate the possibilities for the production of clean and the primary measures: the measures taken in the gasifier
hydrogen-rich synthesis gas. Therefore, the volume fraction itself.
of H2 and other combustibles in the product gas received The effect of l, SB* and the choice of the bed material has
considerable attention during the experiments. Process already been globally discussed in the earlier subsections.
parameters l and SB*, and also the choice of the bed material Most of the attention has concentrated on the polyaromatic
have an important influence on the gas composition and and phenolic fraction of tar, as these compounds not only
hydrogen yield. The following observations can be made cause the most problems in downstream applications, but
based on the results of the experiments carried out within this also contain a significant fraction of the total heating value of
work: the product gas stream. ECN [17] proposed a classification of
tar that is based on the composition, detectability, water
 the lowest hydrogen volume fractions are obtained during solubility and condensation behavior of the individual
the BC, SC and OC; all with sand or sand-like bed materials. compounds:
The H2 volume fractions on dnf basis do not exceed 25%;
 for the BC and OC the hydrogen concentration (initially)  Class 1 e GC undetectable tar.
seems to increase slightly with increasing lambda value.  Class 2 e heterocyclic compounds: tar, highly water soluble,
This could be explained by the increased tar conversion at e.g., phenol, cresols.
higher lambda, and also seems to be confirmed by the tar  Class 3 e aromatic compounds: single ring, do not pose
analysis data. However, in the BC the tests with the highest condensation and solubility problems, e.g., toluene, xylene,
lambda value had also been performed with the highest SB*, styrene.
so the effect of enhanced availability of steam should also  Class 4 e light polyaromatic hydrocarbons: 2 and 3-ring
be taken into account. From the literature it is known that compounds condense at relatively high concentrations at
the increase of SB* from ca. 0.6 to up to ca. 1.2e1.5 will still moderate temperatures, e.g., naphthalene, fluorine.
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S59

 Class 5 e heavy polyaromatic hydrocarbons: 4e7 rings, higher though (0.6e0.7). The same is valid for the OC
condense at high temperatures at low concentrations. (0.7e0.8). The tests with A-wood (SC-A) with magnesite as
Examples: fluoranthene, pyrene, chrysene. additive, where the ratio has increased to 1.2e1.4 depending
on the amount of additive fed, are an exception. This has
Fig. 10 shows the overall results of tar measurement been presented elsewhere [10] in detail;
grouped and averaged per case and fuel. The only exception is  MC shows obvious increase in the H2:CO ratio. The average
the olivine case, where in this graph the grouping has been value is about 2.4, the minimum 1.7 and maximum 2.7; the
done by the applied treatment instead of fuel. The upper highest values have been measured during MC-M
graph shows the sum of the concentrations of the analyzed experiments.
and combustible gaseous compounds and the minimal and
maximal concentrations of Class 4 and Class 5 tar (summed According to the results described here, only the operation
up) and the minimal and maximal concentrations of Class 2 with magnesite as bed material (MC) yields high-quality
tar. It was decided to sum up Class 4 and Class 5 tar for the product gas. The gas produced in other Test Cases will
sake of clarity of Fig. 10. The bottom graph shows the contri- require more extensive upgrading especially in terms of tar
bution of each Class (except Class 1), to the total tar measured removal and wateregas shift. This result is very promising for
in each case. It has to be stressed here, that in the calculation the future implementation of the process on industrial scale
of the fractions, to obtain the best comparison between cases and also for the Varnamo plant. In the 1990s a lot of experi-
the contribution of the BTX compounds and the Class 5 PAH ence was obtained with the operation of the gasifier with
heavier than pyrene have not been taken into account. The magnesite as bed material. The extension of that experience
reason for this is, that these species have not been quantified with the results of this project could lead to the simplification
(reliably) in each of the cases presented here, and, therefore, of the process flow diagram and, therefore, to the successful
would lead to biased comparisons. Fig. 10 (top) largely illus- demonstration of this powerful biomass-to-syngas tech-
trates what already has been concluded in previous sections, nology on industrial scale.
but it gives an overall comparison of the results related to tar.
In Fig. 10 (bottom) two interesting things can be observed. 3.7. Effect of the choice of fuel and bed material
First, the averaged results of MC-S show a clear shift from combination on the operability of the gasifier
Class 4 to Class 2 tar. From the data presented in earlier
sections it can be seen that the average SB* applied in MC-S is From Table 2 it is clear that all fuels used within this work,
1.5, while the average of other cases is 1.2. Additionally, MC-S except A-wood, contain an increased amount of alkali
is the only case where magnesite has been used together with elements, and especially potassium. As stated earlier, these
kaolin. It is difficult to conclude here what was exactly the elements will interact with silica present in the reactor (either
cause of that shift between Class 2 and Class 4, but these two from the bed material or the ash), and at temperatures above
variables could be taken into consideration. Second, a shift ca. 750  C sticky eutectics will be formed that will deposit on
from Class 2 to Class 4 tar (i.e., the opposite as in MC-S) can be the bed material particles leading to agglomeration and
observed in OC, as the bed material pre-treatment tempera- defluidization phenomena. This process is accelerated by the
ture increases. This would imply an addition of for instance presence of chlorine, as this element acts as a shuttle for alkali
two phenol molecules with the simultaneous dehydrogena- metals in their vaporization. Based on the composition of the
tion and dehydration reactions leading to the formation of fuels it was expected that A-wood would not cause any
acenaphthylene. This would be confirmed by the increasing problems with fluidization, while miscanthus and especially
concentration of that compound, which has been actually straw would very probably lead to agglomeration of a sand bed
observed here. However, the proof of such a mechanism and if no countermeasures are taken. B-wood has a composition
other detailed considerations of the fate of hydrocarbons and potential behavior in between the two cases above, but
under different process conditions require further in-depth from experience it was rather expected to cause agglomera-
investigations and, therefore, fall outside the scope of this tion [18,19]. The results of the gasification tests related to ash
overview paper. effects are summarized in Table 9. As expected, only A-wood
can be gasified using a sand bed without problems. B-wood
3.6.3. H2:CO ratio and miscanthus led to agglomerate formation if no counter-
The main application of the clean and hydrogen-rich syngas measures were taken, while in view of the poor outlook no
produced during the CHRISGAS project was intended to be the attempt was made to gasify straw using a sand bed. As
synthesis of liquid transportation fuels, like FT-diesel or DME. explained in Section 2.3.2 magnesite had been shown to be
For these processes the molar (volumetric) ratio between H2 a very effective bed material in terms of counteracting (post-
and CO should be close to 2 in order to obtain proper synthesis poning) the agglomeration. This observation has been
conditions. Basing the experimental data presented in confirmed during the experiments performed in this work, as
previous sections the following remarks can be made about both B-wood and miscanthus could be gasified without any
this important ratio: additional measures. These measures involved in this case
the use of an additive, to act as an alkali getter and to avoid the
 the BC results show the H2:CO ratio to be typically between formation of the sticky deposits. Based on the previous
0.50 and 0.60 and a maximum at 0.8; experience kaolin was chosen as an additive (see Section
 SC shows the values of H2:CO ratio of the same order of 2.3.3). Its use gave very satisfactory results; allowing reliable
magnitude as the BC. The values are generally slightly operation with B-wood and miscanthus on the sand bed as
S60 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

Table 9 e Overview of the tested bed material/additive/biomass combinations and the problems observed.
Sand Sand & kaolin Olivine Olivine & kaolin Magnesite Magnesite & kaolin

A-wood Stable e Stable e Stable e


B-wood S(7 h) A Stable S(5 h) A Stable Stable e
Miscanthus A(<1.5 h) Stable e Stable Stable e
Straw e e e e Ae(<0.5 h) S(1.5 h) D

Symbols used:
Stable: no agglomeration occurred, stable operation possible during multiple hours.
S(7 h) A: stable operation for x hours, but agglomerates found in the bed after the experiment.
A(<1.5 h): agglomeration and defluidization occurred within x hours.
S(1.5 h) D: stable operation of the gasifier for x hours, but problems related to the deposition of KCl from vapor phase occurred downstream the
HT filter, see Fig. 11.

well as olivine bed. Gasification of straw proved to be chal- calculating the total flow rate was applied in the first instance.
lenging, even when magnesite was used as bed material. In It is based on the assumption that nitrogen fed into the reactor
fact, agglomerates were formed also during those tests, but does not undergo any chemical reactions and thus the
most likely their origin was in the straw ash itself, i.e., the amount leaving the reactor should be the same as the input
interaction between alkali and silica from the ash. Finally the flow. The amount (volume fraction) of nitrogen in the product
decision had been taken to use kaolin also during straw gas is measured on dry basis as indicated in Table 1. Applying
gasification on magnesite bed. This seemed to solve the the law of mass conservation (no accumulation of mass takes
problem with defluidization, but unfortunately caused new place in the reactor) and with the assumption of nitrogen
problems (fouling) to the colder parts of the process located remaining inert, the molar flow of dry product gas can be
down the line. This is shown in Fig. 11, which presents estimated using the following simple equation:
a picture of the inner part of the V-cone flowmeter located
F N2 ;in h i
$
downstream the process (see Fig. 2) after short period of $
1
F tot;dry gas kmol,h (8)
operation with straw. The gap between the wall and the bluff yN2 ;dry
body is completely blocked by an alkali deposit. That deposit
was analyzed by XRF, and the results confirmed that it The contribution of the carbon in the solids (char, ashes) to the
consists mainly of an alkali salt, namely KCl. total carbon flow is calculated from the elemental C balance e
it is assumed that all the solids leaving the reactor (fly ash)
3.8. Mass balance and process benchmarks consist of carbon. More detailed information about the
material balance introduced above can be found in [10]. There
In order to verify the consistency of the measurements and to also two mass balances from two different cases (BC-16 and
gain more insight into the fate of the main elements (C, H, N, MC-A01) are presented in detail. For all the cases presented in
O) a material balance (mole balance) was set up for each case earlier sections a mass balance discrepancy has been deter-
studied. Most of the data necessary to perform this operation mined. This discrepancy is the result of the measurement
came directly from the steady state data. One crucial figure errors or incomplete measurements and in practice can never
that is not listed among that data is the total flow rate of the be eliminated completely. It is defined as follows:
gas leaving the gasifier. Although Fig. 2 indicates the presence $ $
of a flowmeter in the outlet line, that measurement has only MBDabsolute F out  F in (9)
been used in the first eight Base Case experiments (BC-01 till
BC-08). For later experiments a different approach for $ $
F out  F in
MBDrelative $ ,100% (10)
F in

Fig. 12 shows the average relative MBD for all discussed


cases, grouped in the same way as in Fig. 10. The error bars
show the minimum and maximum relative MBD for each
group. For most of the cases the discrepancy is negative,
meaning that outgoing mass (or mole) flow is lower than the
incoming flow. This situation is more justifiable, as even with
very accurate determination of the input it is clear that not all
the constituents of the product gas are quantified. This is
especially true in case of gasification where the composition
of the product gas is very complex including a wide range of
Fig. 11 e The space between V-cone flowmeter wall and hydrocarbons. Furthermore the method of the calculation of
its bluff body entirely blocked by alkali deposits (white the total flow and the gravimetric water measurement are
ring) consisting mainly of KCl. considered to be the primary sources of material balance
b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2 S61

MB deviation Carbon Conversion Cold Gas Efficiency C in ash ufl utr-1

100 3.3

80 2.9

Average Mass Balance Deviation [%]


Average Cold Gas Efficiency [%]
Average Carbon Conversion [%]

Average solid C in ash [kg h-1]


60 2.5

40 2.1

20 1.7

0 1.3

-M

-M
-B
d

-A

-S
e

-B

r.

00

0
ille

as

20
nt
C

C
SC

r9
C
SC

-U
C

r1
M

M
m

-T
-20 0.9

-T
se
e-

C
O
as

C
Ba

O
C
es
Ba

-40 0.5
Case ID

Fig. 12 e Overview of all Test Cases: mass balance discrepancy, carbon conversion, cold gas efficiency and solid carbon (fly
ash) flow rate.

discrepancies (listed in order of importance). Fig. 12 also


presents two important process benchmarks: 4. Conclusions and recommendations
!
The experimental research presented in the preceding
$
F C;residue
CC 1 $ ,100% (11) sections resulted in the important further steps taken in
F C;feed
pushing the state-of-the-art in fluidized bed gasification
technology forward. The results are not only a large amount of
P$
LHVgas MW F i ,LHVi recorded process data that will be used for the validation of
CGE $ (12)
LHVfuel MW F ,LHV mathematical models of the process. Also operational expe-
fuel fuel
rience and novel insights have been obtained, which may
As already indicated before, the highest carbon conversion result in an improved and simplified process flow diagram of
was achieved with milled fuel particles in the Base Case. a clean syngas production plant with Circulating Fluidized Bed
Olivine (fresh and pre-treated) showed the relatively lowest as the first hot conversion step. The main conclusions of this
carbon conversion, which obviously is linked with the highest work can be summarized as follows:
(calculated) solid carbon flow rate out of the system. The
initial results of the quantification of the mass of the collected  the optimal bed material and fuel combination depends
filter ashes from various experiments show too high disper- strongly on the application of the product gas. In particular
sion to confirm or deny this calculated result. The main frac- advanced applications, like the production of platform
tion of the unconverted (fine) carbon particles is expected to chemicals, synthetic transportation fuels or hydrogen will
be found downstream of the gasifier as the accumulation of benefit from the use of magnesite as bed material;
carbon in the reactor is found to be negligible compared to the  when sand is used as bed material the use of additive to
total carbon throughput during an average test run. The suppress the agglomeration phenomena is necessary,
reason for the decreased carbon conversion in OC has been unless clean wood (A-wood) will be used as the fuel.
primarily sought in different fluidization behavior of olivine. When straw is being used as the fuel the use of an
Afterwards it had been observed that especially in OC(U) and additive to counteract agglomeration is necessary irre-
OC(1200) the ratio between the applied fluidization velocity spective of the bed material used. This is caused by the
(ufl) and the transport velocity of the particular bed material fact that the alkali present in the fuel ash will form low-
(utr) was too low (see Fig. 12). However, this does not fully melting eutectics with silica, that is also present in the
explain the reduced carbon conversion, as ufl:utr applied in ash. From the agglomeration point of view olivine can be
OC(900) is above 1 and still the carbon conversion is lower designated as fractionally better than sand e this has
than in SC and MC. Another possible reason could be the use been proven during two 13-h test runs with that bed
of B-wood as fuel (large particles, low conversion) that would material and miscanthus as fuel;
be in line with the reduced carbon conversion in SC-B, or the  gasification of any fuel tested in this project on a sand or
use of an additive (in line with MC-S, but contradicting to SC- olivine bed will yield similar results in terms of hydrogen
M). However, from the available data it is difficult to isolate the concentration in the gas, tar and light hydrocarbon
individual effect of those variables on carbon conversion. concentration and H2:CO ratio;
S62 b i o m a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) S 4 0 eS 6 2

 increasing the steam-to-biomass ratio from ca. 0.7 to ca. 1.3 ceramic filter elements during biomass gasification. Biomass
will contribute to tar reduction and increased H2 yield; Bioenerg; 2010. this edition.
 the use of magnesite either as an additive or as a bed [5] Brage C, Yu Q, Chen G, Sjostrom K. Use of amino phase
adsorbent for biomass tar sampling and separation. Fuel
material leads to significant increase in hydrogen volume
1997;76:137e42.
fraction in the product gas. The maximum volume fraction [6] Ergudenler A, Ghaly A. Agglomeration of silica sand in
of hydrogen measured was close to 40% during miscanthus a fluidized bed gasifier operating on wheat straw. Biomass
gasification on magnesite bed. Gasification of other fuels at Bioenerg 1993;4(2):135e47.
similar conditions gave lower H2 volume fractions, which [7] Zevenhoven-Onderwater M, Backman R, Skrifvars BJ,
could be an implication of special (catalytic) properties of Hupa M. The ash chemistry in fluidised bed gasification of
miscanthus ash; and biomass fuels. part I: predicting the chemistry of melting
ashes and ash-bed material interaction. Fuel 2001;80(10):
 also in terms of tar reduction and the boosting of H2:CO ratio
1489e502.
magnesite has shown excellent results. The sum of [8] Bartels M, Lin W, Nijenhuis J, Kapteijn F, van Ommen JR.
measured polyaromatic and phenolic compounds could be Agglomeration in fluidized beds at high temperatures:
reduced below 2 g m3 (STP) on raw gas basis, and the H2:CO mechanisms, detection and prevention. Prog Energ Combust
ratio increased to values between 2 and 3. 2008;34(5):633e66.
[9] Visser H. The influence of fuel composition on agglomeration
behaviour in fluidised-bed combustion. Tech. Rep. ECN-
According to the results of this project, only the operation
Ce04e054. Petten, the Netherlands: ECN; 2004.
with magnesite as bed material yields high-quality product
[10] Siedlecki M, Nieuwstraten R, Simeone E, de Jong W,
gas. The gas produced in other Test Cases will require more Verkooijen AHM. Effect of magnesite as bed material in
extensive upgrading especially in terms of tar removal and a 100 kwth steameoxygen blown circulating fluidized-bed
wateregas shift. This result is very promising for the future biomass gasifier on gas composition and tar formation.
implementation of the process on industrial scale. The Energy Fuels 2009;23(11):5643e54.
extension of the experience from the past with the results of [11] Gil J, Aznar MP, Caballero MA, Frances E, Corella J. Biomass
gasification in fluidized bed at pilot scale with steameoxygen
this project could lead to the simplification of the process flow
mixtures. Product distribution for very different operating
diagram and, therefore, to the successful demonstration of this conditions. Energy Fuels 1997;11:1109e18.
powerful biomass-to-syngas technology on industrial scale. [12] Narvaez I, Orio A, Aznar M, Corella J. Biomass gasification
with air in an atmospheric bubbling fluidized bed. Effect of
six operational variables on the quality of the produced raw
gas. Ind Eng Chem Res 1996;35(7):2110e20.
Acknowledgements [13] Turn S, Kinoshita C, Zhang Z, Ishimura D, Zhou J. An
experimental investigation of hydrogen production from
biomass gasification. Int J Hydrogen Energy 1998;23(8):641e8.
The financial support provided by the European Commission
[14] Herguido J, Corella J, Gonzalez-Saiz J. Steam gasification of
(EC) 6th Framework Programme (CHRISGAS Project contract
lignocellulosic residues in a fluidized bed at a small pilot
number SES6-CT-2004-502587) is gratefully acknowledged. scale. Effect of the type of feedstock. Ind Eng Chem Res 1992;
The authors also thank Claes Brage from KTH for performing 31:1274e82.
the analysis of the tar samples taken during the gasification [15] Devi L, Ptasinski KJ, Janssen FJ, van Paasen SV, Bergman PC,
experiments. Kiel JH. Catalytic decomposition of biomass tars: use of
dolomite and untreated olivine. Renew Energ 2005a;30.
565e557.
[16] Devi L, Ptasinski KJ, Janssen FJ. Pretreated olivine as tar
references removal catalyst for biomass gasifiers: investigation using
naphthalene as model biomass tar. Fuel Process Technol
2005b;86:707e30.
[1] Glazer MP, Khan NA, de Jong W, Spliethoff H, Schurmann H, [17] van Paasen S. Tar formation in a fluidised-bed gasifier. Tech.
Monkhouse P. Alkali metals in circulating fluidized bed Rep. ECN-Ce04e013. Petten, the Netherlands: ECN; 2004.
combustion of biomass and coal: measurements and [18] Khan A. Combustion and co-combustion of biomass in
chemical equilibrium analysis. Energy Fuels 2005;19(5): a bubbling fluidized bed boiler. PhD thesis, Delft University of
1889e97. Technology; 2007.
[2] Gupta R, editor. Hydrogen fueldproduction, transport and [19] Bizon K, De Jong W, Siedlecki M, Chirone R. Investigation of
storage. Boca Raton, FL, USA: CRC Press; 2009. agglomeration phenomena during fluidized bed combustion
[3] Milne T, Evans R, Abatzoglou N. Biomass gasifier tars: their of biomass in a 1 mwth shallow bed boiler. In: Ragucci R,
nature, formation, and conversion. Tech. Rep. NREL/TP-570- editor. 32nd Meeting of the Italian section of the combustion
25357. Golden, CO, USA: NREL; 1998. institute e combustion colloquia. Napoli, Italy: University of
[4] Simeone E, Nacken M, Haag W, Heidenreich S, de Jong W. Naples Federico II, The Italian Section of The Combustion
Filtration performance at high temperatures and analysis of Institute; April 26e28, 2009, p. V-8.