Tar conversion and recombination in steam gasification of biogenic residues: The influence of a countercurrent flow column in pilot-and demonstration-scale

First experiments with biogenic residues and a plastic-rich rejects and woody biomass blend were conducted in an advanced 1 MW dual fluidized bed steam gasification demonstration plant at the Syngas Platform Vienna. Wood chips, bark, forest residues, and the plastic-rich rejects and woody biomass blend were tested and the tar composition was analyzed upstream and downstream of the upper gasification reactor, which is designed as a high-temperature column with countercurrent flow of catalytic material. Each feedstock was gasified with olivine as bed material in demonstration scale and is compared to the gasification of softwood pellets with olivine and limestone in pilot scale. A reduction in tar content was observed after countercurrent column for all feed-stocks. However, a shift in tar species occurred. While styrene, phenol, and 1H-indene were predominant up-stream, naphthalene and polycyclic aromatic hydrocarbons (PAHs) were the prevailing tar species downstream the countercurrent column. Hence, an increase of i


Introduction
The finite nature of fossil resources, dependence on countries with autocratic political governments, and the global climate crisis have emphasized the need for sustainable fuel alternatives.Strategies aimed at achieving the objective of limiting global warming to 1.5 • C compared to preindustrial levels and achieving net zero greenhouse gas emissions were outlined by the IPCC [1].As of 2023, the global temperature has already increased by 1 • C and even with the implementation of current climate pledges, by the end of the century a further increase between 2.4 • C and 2.8 • C is predicted [2].Hence, a major energy system transition is necessary.This transition includes enhanced efforts of utilization and recycling of carbon-based material streams like polymers [3], but also the substitution of fossil resources with non-fossil-based green sources.Efficient utilization of renewable bio-based resources requires, however, advanced technologies reaching far beyond simple energy Abbreviations: aDFB, advanced dual fluidized bed; BTEX, benzene, toluene, ethylbenzene, xylene; CCC, Countercurrent column; CR, Combustion reactor; db, dry basis; DFB, Dual fluidized bed; d sv , Sauter mean particle diameter; GC-MS, chromatography coupled with mass spectrometry; GR, Gasification reactor; HACA, hydrogen abstraction and C 2 H 2 or carbon addition; LHV, lower heating value; LSP, Lower sample point; PAHs, Poly aromatic hydrocarbons; RME, rapeseed methyl ester; SWP, Softwood pellets; USP, upper sample point; wb, wet basis.
* Corresponding author at: BEST -Bioenergy and Sustainable Technologies GmbH, Inffeldgasse 21b, 8010 Graz, Austria.E-mail addresses: miriam.huber@best-research.eu (M.Huber), florian.benedikt@tuwien.ac.at (F.Benedikt), thomas.karel@best-research.eu (T.Karel), matthias.binder@best-research.eu (M.Binder), daniel.hochstoeger@best-research.eu(D.Hochstöger), anna.egger@best-research.eu (A.Egger), katharina.fuersatz@bestresearch.eu (K.Fürsatz), matthias.kuba@best-research.eu (M.Kuba).generation from biomass.Among other thermochemical technologies, biomass gasification is an already proven technology, currently primarily applied in combined heat and power generation due to its high conversion efficiencies [4,5].However, land-use conflicts and price fluctuations arise the need for adapted gasification strategies to optimize the feedstock utilization and flexibility [6].In addition, these advanced technologies should focus on the generation of useful components for follow-up synthesis for various further applications [5].One potential solution is the application of biogenic wastes and non-biogenic residual materials in dual fluidized bed (DFB) steam gasification to produce a product gas of mid-calorific value consisting mainly of hydrogen, carbon monoxide, carbon dioxide, and methane.This product gas offers various pathways for downstream upgrading to renewable gases as i.e. renewable hydrogen or synthetic natural gas [7], transportation fuels [8,9] or chemicals [10] and hence contributes to the goal of global warming mitigation.
However, the gas cleaning steps required to remove impurities from the product gas are usually complex and costly [11].The aim is therefore to suppress the formation of critical species already during gasification to reduce the need for gas cleaning units.A major concern here is the formation of organic impurities, including tar, which is a mixtures of hydrocarbons primarily characterized by their aromatic structure with a higher molecular weight than benzene [12].Tar is undesired in the product gas as it condenses in downstream equipment forming solid or highly viscous deposits that can lead to fouling on heat exchanger surfaces or clogging of the piping [13].Since this unintended condensation of tar in process units is relevant for further applications, special emphasis is placed on tar species with relatively high condensation temperatures, such as polyaromatics.In general, the removal of tar is based on two approaches: primary measures intend to inhibit or transform the formation of tar in the gasification reactor, while secondary measures focus on the downstream removal of tar with chemical or physical treatment e.g.scrubbers [4].In DFB steam gasification, the product gas quality and tar content is strongly dependent on the gas phase interaction with catalytically active compounds [14] and feedstock origin [15].Typically, the bed material serves as catalytically active compound.Olivine, a Mg-Fe silicate, is the most applied bed material in DFB gasification due to its low agglomeration tendency [16] and the catalytic effect of Fe and Mg [17].Additionally, layer formation of feedstock ashes on the particle surface promotes the catalytic activity and improves the gas quality [18]- [20].Further enhancement of the catalytic activity is achieved with additives such as limestone [21,22] and potassium [17,23].Considering these points, one primary measure for tar inhibition is the installation of a countercurrent flow column (CCC) above the bubbling bed of the gasification reactor.This countercurrent flow column, which is operated at higher temperatures than the bubbling bed improves the residence time and the interaction between solids and gas phase.The active bed material hence contacts volatiles from the feedstock and steam from the fluidization at a time.This contact and interaction leads to enhanced gasification and reforming reactions [24,25].Benedikt et al. [15] studied the favorable impact of the CCC on the product gas composition and tar reduction with a broad range of feedstocks in an advanced DFB steam gasification pilot plant.Kolbitsch [26] firstly reported the change of tar species with detailed tar measurements upstream and downstream of the CCC in the same plant.Based on these observations, Mauerhofer et al. [27] investigated the change of product gas composition and tar concentration in each compartment of the CCC.In industrial-scale DFB steam gasification, a parametric study on tar composition without a CCC was conducted with focus on the fate of different tar species, dependent on parameter variation [28].Hence, research on the effect of the CCC has so far been limited to the 100 kW aDFB pilot plant and the transferability of the results to the demonstration-scale is uncertain so far, which is a key research question for the scalability of the aDFB design.In contrast, detailed studies on the underlying mechanisms of the observed recombination of secondary tar to poly-aromatic hydrocarbons (PAHs) is limited to research in lab-scale for the identification of mechanisms [29].
Hence, the aim of the present work is to bridge the gaps between pilot-scale and industrial-scale research and unravel the evolution of tar over the CCC during demonstration-scale experiments.To ensure that the observed behaviors are not feedstock-dependent, four different feedstocks are tested in 1 MW demonstration-scale at the Syngas Platform Vienna, including three lignocellulose residues and one feedstock derived from plastic-rich rejects, a waste stream from the pulp and paper industry.Within this series, all experiments are conducted with olivine as bed material to enhance comparability.To link these results to mechanisms, one key target is the transfer of knowledge gained at labscale for identifying demonstration-scale tar recombination patterns.Due to measurement limitation in demonstration-scale, only observations and connections to mechanisms and patterns are given.The results are further compared to measurements at the 100 kW pilot plant at TU Wien to validate the applicability of pilot-scale observations in demonstration-scale.In addition, the results of the 100 kW pilot plant with olivine are compared to the same 100 kW experimental set-up conducted with limestone, to include the influence of catalytically enhanced bed material.

Material and methods
In the following section, the design of the pilot and demonstration aDFB plants, as well as sampling and measurement methods are introduced.In addition, the analyses of applied feedstocks and bed materials are presented.

DFB steam gasification
The DFB steam gasification process is based on two interconnected reactors, a gasification reactor (GR) and a combustion reactor (CR), as illustrated in Fig. 1.Steam is used as gasification agent in the GR, operated as bubbling fluidized bed and the feedstock is transported onto the bubbling bed.The heat required for the endothermic gasification reactions is provided by the circulating bed material.In addition, the bed material acts also as catalyst for enhancing the gasification reactions.Residual char from the GR enters the bottom of the CR via a loop seal and is further combusted, enabling a temperature level above 900 • C. Hence, the CR is fluidized with air and operated as fast fluidized Fig. 1.Schematic principle of the DFB steam gasification process [15].

M. Huber et al.
bed [30].DFB steam gasification typically provides a mid-calorific LHV product gas characterized by the neglectable amount of nitrogen.Hence, a typical product gas composition of DFB steam gasification with olivine as bed material is 35 % to 45 % hydrogen, 20 % to 30 % carbon monoxide, 15 % to 25 % carbon dioxide and 8 % to 12 % methane [31].Apart from recent developments to use DFB gasification for solar energy storage [32] or CO 2 [33], the technology is already established in industrial-scale since several DFB plants have been operated in i.e.Güssing [31], Senden [34], Oberwart [35], and Gothenburg [36].The applied feedstock in industrial-scale is limited to woody biomass.
To further enhance the gasification reactions and particle-solid interactions, an advanced GR design was developed by Schmid et al. [37].Above the freeboard section of the bubbling bed, an upper gasification reactor, designed as countercurrent column (CCC), is installed.While the gas phase streams upwards through the upper GR, it is exposed to improved mixing and gas-solid interactions with the downward passing bed material.By this enhanced gas-solid interactions the application of complex feedstocks, as biogenic or industrial waste streams is facilitated [15].This reactor designed is further referred to as advanced dual fluidized bed (aDFB) technology.The first aDFB plant was installed in 2014 at TU Wien, a 100 kW pilot plant [38].In 2022, a 1 MW demonstrationsscale aDFB plant was installed at Syngas Platform Vienna, based on the previous results of the pilot-scale plant.

100 kW aDFB pilot plant at TU Wien
The 100 kW aDFB steam gasification plant is in operation since 2014 and its basic reactor design and the placement of sample points presented in Fig. 2. The surface area of the bubbling bed section is 0.03 m 2 and increases to 0.27 m 2 in the freeboard section, where the lower sample point is located.The total height of the CCC is 3.3 m, with six constrictions to improve the hold-up in the column.To prevent air slip from the CR to the GR, the reactors are interconnected via loop seals.In addition to feedstock flexibility, various bed materials have already been tested [18,22].Hence, both reactors are equipped with gravity separators and cyclones to limit abrasion effects of the bed material.Radiative heat losses are relatively high compared to industrial-scale, since the reactor system is made out of stainless steel and thermally insulated.Thus, auxiliary fuel is added to the CR to maintain a stable temperature level.A detailed plant description and reactor geometries are reported by Schmid et al [39].

1 MW aDFB demonstration plant at Syngas Platform Vienna
The 1 MW demonstration plant at the Syngas Platform Vienna of BEST was commissioned in early 2022 and first experimental trials with various feedstocks were conducted until July 2022.Fig. 3 shows the 1 MW aDFB steam gasification plant at the Syngas Platform Vienna after commissioning.Similar to the 100 kW pilot plant at TU Wien, the plant is based on the advanced design of the DFB steam gasification technology firstly demonstrated in Güssing, Austria [31].As research facility, the aim of the Syngas Platform Vienna is to investigate feedstock flexibility with particular interest on biogenic and waste materials from municipalities, and the industrial and agricultural sector, in a scale comparable to industrial plants.This has so far been performed successfully in scheduled and planned research campaigns, each with multiple weeks of continuous operation.
The schematic reactor design of the demonstration plant is presented in Fig. 4. In contrast to the pilot plant, the reactor system is refractorylined, and hence, heat losses are decreased.The surface area of the bubbling bed section is 0.31 m 2 and increases to 1.48 m 2 in the freeboard section, where the lower sample point is located.Above the freeboard section, the countercurrent flow column is situated, with a total of five diameter constrictions to generate a particle hold-up in the column.The total height of the CCC is 4.8 m.The CR is equipped with an auxiliary fuel supply to stabilize the temperature profile, as the heat losses surpass  industrial-scale values.Recirculated product gas, heating oil and part of the contaminated RME of the product gas scrubber are used as auxiliary fuel input.A further adaption compared to the 100 kW plant is the lower connection of the two reactors.The bed material and formed char in the GR are transported back to the CR via a chute.The chute hereby facilitates the transport of heterogenous and larger char components to prevent clogging.In order to control the circulation of the bed material and the aimed CR temperature level of 950 • C, the air is supplied at three stages.The primary air stage is beneath the auxiliary fuel supply.The secondary and tertiary air stage are at a higher level and illustrated in the simplified flow sheet of Fig. 4. In order to enhance catalytical reactions during the endothermal gasification process, olivine is used as bed material.Due to layer formation on the olivine particles based on bed-ash interactions, the catalytic activity is even improved during longterm operation [18][19][20]40,41].
The subsequent product gas cleaning line is presented in Fig. 5.As the product gas exits the GR, transported coarse particles of the bed material are separated in the radiation coolers, where a temperature level below 450 • C is reached.Fly char and dust are separated in the downstream cyclone and hot gas filter and are transported back to the combustion reactor (CR).The deposited bed material can also be recirculated onto the bubbling bed of the GR via screw conveyors.Apart from fly char deposition, the hot gas filter separates tar condensed on larger particles, which are periodically led into the CR during filter cleaning.The product gas further passes through a water quench and subsequently a rapeseed methyl ester (RME) scrubbing unit operated at gas outlet temperatures of around 40 • C. Thereby, a major fraction of the tar condenses in the liquid phase of the scrubber.Downstream the scrubbing units, the online continuous gas analyzer for the product gas composition is installed, measuring CO, CO 2 , H 2 , CH 4 , and O 2 .Part of the gas stream can be recirculated into the CR as auxiliary fuel input for temperature regulation.

Tar sampling and analytic procedure
At the 100 kW pilot plant and the 1 MW demonstration plant, tar sampling was performed at two different sample points each to identify the transformations of tar passing through the CCC.The lower sample point is located above the bubbling bed in the freeboard section (see Fig. 2 and Fig. 4: lower sample point) with typical temperature levels between 760 • C and 850 • C for the considered experiments.Due to restriction in the measurement set-up and a varying pressure profile in the freeboard, the isokinetic sampling at this point is limited.After passing the CCC and a radiation cooler, a second sample point is installed at temperatures of above 370 • C (Fig. 2 and Fig. 4: upper sample point).Tar measurement was performed using a methodology based on the Tar Protocol [42] with minor adaption for meeting the needs of the DFB steam gasification process.Hence, toluene was applied as solvent, as its application additionally facilitates the simultaneous measurement of the water content in the product gas.Due to the application of toluene, no values for BTEX species (benzene, toluene, ethylbenzene, and xylene) are available, as toluene is not detectable and the detection of benzene, ethylbenzene and xylene is only restricted and not performed in this study.However, ethylbenzene and xylene typically are not detected even when using isopropanol as solvent.The comparability of the slightly adapted measurement method is shown by Schmid et al. [39].analyzed for their gravimetric tar concentration and the single tar species by gas chromatography coupled with mass spectrometry (GC-MS).A detailed description of the sampling and analytic procedure is given in Wolfesberger et al. [43] und Wolfesberger-Schwabl et al. [44].

Tar classifications and dew point
Tar is formed during the gasification process, being a mixture of hydrocarbons primarily characterized by their aromatic structure, condensing at ambient temperature and thus forming solid or highly viscous deposits [13].These deposits lead to further fouling in downstream units or piping.Several classification and identification system of tar have been proposed in literature.Milne et al. [13] differentiated between primary, secondary, alkyl tertiary, and condensed tertiary tar according to their mechanisms and parameters of formation.Primary tar are cellulose-derived products formed during the devolatilization at relatively low temperature levels between 400 • C and 600 • C. Above 600 • C primary tar decomposes to secondary tar, consisting of phenols Fig. 6.Tar sampling set-up [44].

Table 1
Selected indicator tar species sorted by their condensation point.[45,46].In addition, the Energy Research Centre of the Netherlands (ECN) [42] established a tar classification system based on physical properties being their condensing and water solubility behavior to identify restrictions in downstream product gas units.Class I tar include GCundetectable and high polarity tar, condensing at high temperatures and corresponding to the gravimetrically measured tar.Class II (e.g.phenol, Table 1) is characterized by heterocyclic compounds with high polarity and thus showing high water solubility.Mono-aromatic compounds are part of Class III (e.g.benzene and toluene) and light polyaromatic hydrocarbons with 2 to 3 rings of Class IV (e.g.naphthalene, Table 1).These light polyaromatics are characterized by condensation at intermediate temperatures and high concentrations.Class V (e.g.pyrene, Table 1) consists of polyaromatics with 4 to 7 rings condensing at high temperatures with low concentrations.This classification scheme is closely linked to the tar dew point.Based on the individually occurring tar species and their concentrations, an overall dew point can be determined.If the temperature of the product gas undercuts this dew point, condensation may happen.Thus, the tar dew point represents an indicator for potential fouling and process restrictions.For tar dew point calculations, the ECN provides an online tool based on the measured GC-MS tar species and concentrations [47].This application is also used to determine the tar dew point in the demonstration trials.
To facilitate a comparison, eight indicator tar species were chosen based on their measured concentrations and presented in Table 1.A full list of measured tar components to determine total mass of GC-MS measured tar is presented by Schmid et al. [39].For species from tar classes II, III, and IV with typically decreased concentrations after the CCC, one species of each tar group is selected.Since tertiary tar from Class V are particularly emphasized due to their different behavior four tar species are selected.This allows a comparison within the class to verify whether all species are affected.Naphthalene acts as a transition tar species, since it is the lightest molecule to increase over the CCC.

Feedstocks, bed materials, and operational parameters
Four gasification trials on the 1 MW demonstration plant and two trials on the 100 kW pilot plant with each different feedstock and bed material combinations are presented in this paper.In 1 MW demonstration-scale, wood chips, forest residues, bark, and plastic-rich rejects and woody biomass blends were applied.Softwood pellets were applied in 100 kW pilot-scale.The proximate and ultimate analyses of each feedstock is illustrated in Table 2. Wood chips are used as benchmark feedstock, based on the homogenous quality and composition.Forest residues and bark are lignocellulose residues with a relatively high share of nitrogen, and plastic-rich rejects are characterized by their high amount of chlorine and volatiles as well as an inhomogeneous composition.However, bark and plastic-rich rejects are both waste streams from the pulp and paper industry and thus, bear potential for closing circular economy gaps when applied in gasification processes.While bark originates from on-sight debarking, the plastic-rich rejects are derived from waste paper recycling.Plastic-rich rejects are a mixture of all materials disposed of in the waste paper bin, with the majority consisting of different varieties of plastic.Hence, they contain biogenic and non-biogenic residues from waste paper collection that are not applicable for waste paper recycling.Both, bark [15,48,49], and plastic-rich rejects [21] were already investigated at the 100 kW aDFB pilot plant with rigorous feedstock pre-treatment such as drying and pelletization, to meet the demands of pilot-scale.In contrast, the feedstock preparation for industrial DFB gasification is limited to feedstock drying and chipping [50].The drying in industrial-scale is performed by the integration of excess process heat at low temperature [35].For this 1 MW campaign, however, no drying was performed in advance, and it was shown, that even with up to 50 wt% moisture content, an application is feasible.Nevertheless, limiting factors are given by the provision of heat and maintenance of the temperature level in the GR and drying is benevolent for the overall process efficiency.To further meet the demands of demonstration-scale, all applied feedstocks were shredded to G30 size, corresponding to 60 wt% of all particles having an average area of 30 mm 2 .However, even after shredding the plastic-rich rejects posed some challenging properties for the application in demonstrationscale: the initial content of metals, the pourability, and the moisture content.To prevent clogging in smaller pipes, metals were additionally separated.As the experience with plastic-rich rejects at high moisture content is yet limited and the pourability a crucial factor for application in demonstration-scale, both feedstock properties were improved by creating a blend with wood chips.Hence, the pourability was tested with a wood chips share ranging from 0 wt% to 70 wt% on wet basis.The blend of 50 wt% plastic-rich rejects and 50 wt% wood chips offered appropriate pourability properties to allow continuous operation.In addition, the 50 wt% plastic-rich rejects and 50 wt% wood chips blend set the water content to 37.1 %, and thus, below a water content successfully tested in the demonstration plant.
The main differences of the applied feedstocks occur with ash content, amount of volatile matter, and the lower heating values (LHV) on dry basis (db), as presented in Table 2.The ash content is increasing from wood chips (0.9 wt.% db ) to plastic-rich rejects and woody biomass blends (12.5 wt.% db ), influencing the ash handling during operation.The amount of volatile matter determines the devolatilization step during gasification and is in the same order of magnitude for wood chips and plastic-rich rejects and below 80 wt.% daf for bark and forest residues.Similarly, the LHV db is at 18 MJ/kg for all lignocellulose material and at 26 MJ/kg for plastic-rich rejects.However, the main influencing parameter for gasification is the high moisture content for all materials.The presented moisture content values were measured during analysis in the lab, meaning that they may differ marginally during the actual test run.For forest residues, the highest moisture content was measured, leading to a LHV wb of below 10 MJ/kg.
As the experiments at the 100 kW plant focused on the impact of catalytically enhanced bed material, olivine and limestones were each used as bed material.The total bed material inventory amounted 80 kg for each trial.In demonstration-scale, olivine was the sole bed material used, since the feedstock varied.However, the experiments in demonstration-scale were conducted consecutively in one campaign, and thus, the bed material was not changed.As described above, the long-term use of olivine in DFB steam gasification leads to formation of active ash layers positively influencing the product gas quality.The experiments were conducted in the following sequence: wood chips, forest residues, bark, and the plastic-rich rejects and wood chips blend.This effect should be kept in mind when further discussing the results of demonstration-scale.Table 3 presents the initial properties of the applied bed materials for both plants.In the demonstration plant the total bed material inventory amounted 1200 kg.
Table 4 presents the main operational parameters of the compared experiments.In the pilot plant, a feedstock feeding rate around 20 kg/h, corresponding to about 95 kW thermal feedstock input, was applied.Due to the low moisture content of the softwood pellets, the steam to fuel ratio is below 1 and for the experiment with limestone as bed material even at 0.69.The feedstock feeding rate in demonstration-scale is ranging between 200 kg/h and 260 kg/h dependent on the feedstock.For wood chips, this corresponds to i.e. 1112 kW thermal feedstock input and 750 kW for the plastic-rich rejects and woody biomass blend.As the steam input was kept constant to guarantee fluidization, the steam to fuel ratio is influenced by the feedstock input and moisture content of the feedstock.As a result, the plastic-rich rejects and woody biomass blend shows a rather high steam to fuel ration.The moisture content analogously affected the temperature level in the GR and CCC.Therefore, temperature levels were higher during the pilot plant experiments with olivine.

Results and discussion
In this section, the tar concentrations upstream (freeboard section) and downstream (after radiation cooler) the CCC are presented and discussed.For the demonstration-scale experiments, four different feedstocks are applied to confirm that the observed trends are reactor design-dependent rather than feedstock-dependent.The experiments with feedstock variation were all conducted with olivine as bed material.To further assess the tar behavior over the CCC, experiments with catalytically enhanced bed material from the pilot plant are compared.
As the catalytic effect of limestone is well established, experiments with 100 % limestone are compared to experiments with 100 % olivine [22].Softwood pellets were selected as benchmark feedstock to minimize the individual impact of the feedstock.Since the experiments with bed material variations are conducted at the 100 kW pilot plant with the moisture content of the feedstock significantly different from the feedstock applied in demonstration-scale, these results are rather comparable in respect of the magnitude of reduction with those of the demonstration-scale rather than in respect of absolute values.

Feedstock variation in demonstration-scale
To obtain an overview of the tar behavior in the CCC, the tar concentration is measured in the freeboard (lower sample point) and after the radiation cooler (upper sample point), thus after the CCC.The temperature after the radiation cooler is at a level of 370 • C to 405 • C for each experiment and hence, not affecting tar condensation [42].For gravimetric tar, the minimum concentration was measured with the most-established feedstock, wood chips, though bark only marginally surpassed this value (Table 5).At the upper sample point, a reduction in gravimetric tar compared to the lower sample point occurred for all feedstocks.While the gravimetric tar concentration of biogenic feedstocks was reduced to 3 g/Nm 3 to 6 g/Nm 3 , the product gas of plasticrich rejects and woody biomass blends still contained 8.9 g/Nm 3 before further gas conditioning.A similar trend is visible with GC-MS tar (BTEX not included).After the CCC, the GC-MS tar concentration amounted to 50 % (for Forest Residues) to 65 % (for Bark and the Plastic Rejects & Wood Chips Blend) of the initial concentration at the lower sample point.Lignocellulose residues thus all show a consistent tar level before and after the CCC.Higher concentrations were measured for the plastic-rich rejects and woody biomass blends, this can be attributed to the lower gasification temperatures in the bubbling bed and the CCC.
In addition to the total GC-MS tar concentration, the amount of each tar species was measured in the freeboard section and after the CCC.Two opposing trends were identified in every experiment.The first trend observed applied to Class II and Class III tar, where an apparent reduction occurred over the CCC.In all conducted experiments, theses tar species were either significantly reduced or even limited to a nondetectable amount (Fig. 7).After the CCC, dibenzofuran was the only detected Class II species, amounting to less than 2.5 % of its initial concentration.For the Plastic Rejects Blend no Class II species was detected after the CCC.
Similar to Class II tar, the concentration of Class III is significantly reduced for all feedstocks.Class III tar are typically mono-aromatic compounds as styrene, toluene, and xylene.However, as BTEX components are not included, the major species is styrene.Over the CCC, the total Class III amount is reduced to 28 % for Plastic Rejects Blend and to 12 % for Forest Residues of the initial concentration in the freeboard.
The second behavior represents tar species forming over the CCC.As presented in Fig. 7, this phenomenon includes Class V tar during all trials and Class IV tar in trials with biogenic feedstock.Condensed Class IV tar like naphthalene and anthracene followed the second behavior type, hence an increase in concentration after the CCC was measured.In contrast, cyclopentane compounds and uncondensed species as 1Hindene or biphenyl are significantly reduced after the CCC.As Class IV contains tar species behaving after both observed phenomena, hence, increasing and decreasing after the CCC, no significant increase of Class IV tar occurred.However, according to the classification system established by Milne et al. [13], the decreased species are categorized as secondary tar, like Class II and Class III tar.Thus, the increase over the CCC was observed for tertiary tar only.
The indicator species are presented in Fig. 8, as mentioned above, the Class II tar phenol was not detected after the CCC.Styrene (Class III) and 1-H indene (Class IV) both decreased significantly over the CCC.Naphthalene, a two-ring PAH, was the lightest tar species with increased concentration after the CCC.The relative increase of naphthalene was rather low for the Plastic Rejects Blend, as the absolute values were already at a rather high level (more than 5 g/Nm 3 ).In contrast, a significant increase in concentration after the CCC was observed with all lignocellulose feedstocks.This increase occurred analogously with anthracene, where the initial and the concentration after the CCC were in the same order of magnitude for all experiments (Fig. 8).Class V tar, characterized by heavier PAH compounds with four to seven rings, follow the same behavior as condensed Class IV species.In all experiments, an increase of two to three times the initial concentration was measured, with the selected indicator species being fluoranthene and pyrene.Benzo[a]pyrene (Fig. 8), as even higher PAH, was measured only with Wood Chips and the Plastic Rejects Blend and Bark after the CCC.
When discussing the altered tar concentration over the CCC, the total product gas amount has to be considered as well.By enhancing the gasification reactions, the amount of product gas increases over the CCC leading to a dilution of tar concentration when comparing the amount before and after the CCC.As product gas measurement is limited in the freeboard, a simulation approach may serve but has not yet been tested in demonstration-scale, hence this topic is further discussed in section 3.3.Considering this dilution of tar concentration, the decrease of secondary tar species is mitigated and the increase of tertiary tar is further promoted.
With the determination of the individual tar species, the tar dew point was also determined.Fig. 9 presents the tar dew points in the freeboard and after the CCC.Since the dew point is primarily affected by the concentration of large hydrocarbons [42,51], its increase after the CCC corresponds to the increase of PAHs.As an increase of dew point comes with easier condensable tar and fouling, the subsequent coarse gas cleaning unit has to be considered as well.Though the gas cleaning unit surpasses the scope of this work, the results have to be regarded in the context of further processing.For the demonstration plant a hot gas filter, water quench and RME scrubber are installed.As the hot gas filtration is conducted above 350 • C, no problems with condensation occurred.Typically, heavier tar condensate in the water quench, where RME is added to counteract potential clogging of the viscous tar.Hereby, the separation efficiency for Class II to uncondensed Class IV tar is significantly lower than for PAHs [19,36].Saturated RME is afterwards applied to the combustion reactor as auxiliary fuel.Thus, the tar is thermally recycled and decomposed in the CR.The recombination to heavier tar leads to an overall facilitated product gas cleaning scheme in the demonstration plant.

Tar recombination patterns
Tar formation is to a great extent dependent on temperature and exposure to catalysts [46,52], hence the temperature profile in the GR affects the formation of tar species.During the demonstration-scale experiments, the temperature of the bubbling bed was between 765 • C (for Forest Residues) and 840 • C (for Bark) and thus complied with design values and previous trials in the pilot plant [39].While Class II, Class III, and Class IV tar are predominantly formed at temperatures below 800 • C, the principal dehydrogenation, condensation and aromatization, deoxygenation and dealkylation reactions leading to the formation of PAHs occur above 800 • C [53].A parametric study for tar formation, conducted in an industrial DFB steam gasifier [50] identified the gasification temperature as major influencing parameter, since tar species with a molecular mass lower than naphthalene (e.g.benzofuran, styrene, and 1H-indene) increased with decreasing gasification temperature.Thus, the temperature profile of the gasification reactor in Fig. 10 corresponds to the detected tar species at the respective sample points.
Oxygen-containing species of Class II were predominantly undetected after the CCC, as phenolic compounds are converted between 700 • C and 850 • C [54].In contrast, rather low concentrations of PAHs were measured in the freeboard section, due to the lower temperature level.The elevated temperatures in the CCC promote the formation of heavier aromatic ring systems.Potential pathways for PAH formation from Class II and Class III tar are introduced by Dufour et al. [55] and presented in Fig. 11.The conversion of phenol and indene to benzene and naphthalene at temperatures between 800 • C and 900 • C is studied.Hence, PAH formation is enhanced by indenyl and cyclopentadienyl radicals and further recombination to phenanthrene and acenaphthylene is suggested [55][56][57][58][59].These recombination patterns from literature correspond to the measured tar species in the freeboard section and after the CCC.Though a broad variety of PAH formation and growth mechanisms has been proposed in the last decades, an identification of the exact mechanism for demonstration-scale experiments is challenging due to the heterogenous feedstock, non-uniform reactions conditions and the numerous participating radicals and molecules, implying competing reactions [29].One of the most established mechanisms of PAH growth is the hydrogen abstraction and C 2 H 2 or carbon addition (HACA) [60,61].Based on the HACA mechanism, Shukla and Koshi [62]   described the growth of PAHs from naphthalene to acenaphthylene and further to fluoranthene and for phenanthrene to pyrene.HACA reactions predominantly occur at two different fusion sites.Triple fusing sites were identified as efficient growth routes and correspond to the observed increase of pyrene after the CCC, formed at the triple fusing site of phenanthrene.In contrast, production of pentacyclo-fused aromatics is enhanced by HACA at double fusing sites.Corresponding observation over the CCC occurred with the formation of acenaphthylene from naphthalene.
Overall, the increased temperature in the CCC enhances the decomposition of secondary tar (Class II, Class III, non-condensed Class IV) leading to the recombination to PAHs.Simultaneously, the increased residence time and thus, contact time promotes tar reaction to growth.Hence, in addition to established decomposition reaction of light tar to light gases [63], selected tar species recombine to higher tar molecules.However, as the total tar content is decreasing, presented in Section 3.1, the tar cracking reactions still dominate the process.

Bed material variation in pilot-scale and plant comparability
At the 100 kW pilot plant similar tar reduction patterns occurred over the CCC with olivine as bed material and softwood pellets as neutral benchmark feedstock.For the pilot plant trial, a temperature in the bubbling bed of 845 • C and 951 • C in the CCC was measured.The total gravimetric tar amount is reduced from 19.4 g/Nm 3 to 6.7 g/Nm 3 , hence to 34 % of the initial concentration in the freeboard (Table 5).GC-MS tar species were decreased to 44 % of the initial concentration in the freeboard and thus analogous behavior to the demonstration plant is observed.This also applies to the classification of the tar species, displayed in Fig. 12 (top).No Class II tar were detected, and Class III were reduced to 10 % of the freeboard concentration.The reduction of Class IV tar was predominant for non-condensed species as 1H-indene, whereas condensed Class IV tar as naphthalene increased.Following this trend, an increase of Class V tar to 2.9 times of initial concentration was observed.Overall, the recombination patterns of the demonstrationscale analogously occurred in pilot-scale for experiments with olivine as bed material.When comparing the absolute values of tar concentration in pilot-and demonstration scale, the tar concentration of pilot-scale softwood pellets in the freeboard is significantly higher than tar concentrations in the freeboard of the demonstration plant.Three main influencing factors are determined for this deviation: as mentioned in section 2.4, the isokinetic sampling in the freeboard is limited, hence, plant-specific fluid dynamics may lead to individual sampling errors.Secondly, olivine in the demonstration plant was applied for all presented experiments and thus, exposed to several days of feedstock ash.Since these ash layers positively influence the product gas quality, lower tar concentrations than with fresh olivine, as applied in pilot-scale, were expected.This catalytic impact of ash formation even surpasses the effect of temperature [52] and is thus identified as a main driving force for the lower tar concentrations in demonstration-scale.The third factor targets the feedstock properties.While softwood pellets and wood chips show a substantially similar composition, the moisture content of wood chips is significantly higher (17.8 %) than that of softwood pellets (7.2 %) as presented in Table 2.As the moisture content has a major impact on the tar content, and lowest tar emissions where found to occur around 20 wt% water, experiments with softwood pellets were again expected to increase tar content.In particular with low moisture content below 10 wt% tar peaks occurred in prior studies at an industrial DFB steam gasification plant [64].Nevertheless, the comparability of pilot-and demonstration-scale is given, as the recombination patterns and simultaneous decrease of total tar concentration occurred with both plants.Despite the higher initial tar concentration in the freeboard for softwood pellets, the relative reduction of tar over the CCC is slightly increased in the pilot plant when compared to wood chips in demonstration-scale.This is attributable to higher temperatures in the CCC in the pilot plant, promoting tar elimination reactions.
However, with limestone as bed material a different trend was observed, as the initial tar concentration was significantly lower than with olivine.For gravimetric tar, a freeboard concentration of 4.2 g/ Nm 3 was measured and reduced to 0.7 g/Nm 3 after the CCC.Class V tar concentrations were limited to below 0.1 g/Nm 3 in the freeboard and after the CCC.Fig. 12 (bottom) presents similar concentrations occurring for Class II and Class III tar after the CCC.The only tar class significantly measured at the upper sample point is Class IV.Hence, light polyaromatic compounds represent the main tar class remaining.From these tar species, naphthalene remains at similar level as measured in the freeboard.This behavior is in accordance with Jordan and Akay [65] and Yongbin [66] were no Class V tar were detected with the application of CaO.CaO hence affects the hydrocarbon formation reactions benevolently as primary tar is reacting with CaO [67,68].Mauerhofer et al. [22] more specifically discussed the influence of limestone as bed material in aDFB steam gasification and concluded that the catalytic activity of CaO further enhances tar reduction, compared to a predominance of thermal effects on tar reduction with fresh olivine.
Due to the enhanced endothermic gasification and reforming reactions with limestone as bed material, the temperature in the bubbling bed was decreased when compared to experiments with olivine [21].Hence, 787 • C are measured in the bubbling bed.With the improved gasification and reforming reactions, an increased amount of product gas volume flow is provoked, leading to the dilution of measured tar concentrations after the CCC compared to the freeboard tar concentration, as already mentioned in section 3.1.This increase of product gas amount is not limited to limestone, but also affects olivine experiments,  as the CCC enhances gas-solid interactions and with this the gasification reactions.Thus, when measuring a decrease of tar concentration after the CCC the amount of increased product gas has to be considered.
The determination of product gas amount in the freeboard is based on simulation tools and for the pilot plant calculated for selected experiments by Benedikt [21].Thereby, an increase of product gas in the range of 23 % to 41 % is observed after the CCC for differing bed material shares of olivine and limestone.This corresponds to a product gas yield of around 1.4 Nm 3 db/kg feedstock daf .While this range is not applicable for the demonstration plant operated with differing feedstocks and tar concentrations, it serves as an initial attribution of the order of magnitude.When considering this increase of product gas, major emphasis is on the fate of naphthalene.Naphthalene showed an absolute increase with every experiment with olivine as bed material.In contrast, a decrease was observed with limestone in the pilot plant.However, considering the additional formation of product gas, further formation of naphthalene must have taken place even with limestone to maintain the observed concentration level after the CCC.This thesis is also applicable to demonstration-scale and verification by a simulation tool bears potential for further research.

Conclusions and outlook
First experiments using biogenic residues and plastic-rich rejects and woody biomass blends in a 1 MW demonstration-scale DFB steam gasification plant were conducted.The focus in this study is on the tar conversion and recombination in the GR, which is equipped with a CCC as a primary measure for tar mitigation.It was observed that for every feedstock applied the content of gravimetric tar and sum of GC-MS tar declined significantly after the CCC.When emphasizing on the tar concentrations of the individual feedstocks, it is apparent that the tar concentrations were consistently at the same level for all lignocellulose feedstocks at the same sample point.With the plastic-rich rejects and woody biomass blend, rather higher concentrations were measured.Nevertheless, the relative reduction over the CCC corresponds to the experiments with lignocellulose materials.The tar reduction over the CCC thus appears independent of the feedstock and the absolute concentrations rather dependent on the feedstock.
While the total tar content significantly decreases over the CCC, the individual tar species show different behavior.The concentration of oxygen containing Class II tar and mono-aromatic Class III is analogously declined or undetectable for selected species like phenol.Class IV tar can be subdivided into condensed light-aromatics and uncondensed aromatics based on their behavior in the CCC.Thus, the uncondensed 1H-indene declined in each experiment, while an increase of naphthalene was measured.This increase was also observed for all occurring Class V PAH species.To confirm that the observation is not plantspecific, the results were compared to experiments on the 100 kW pilot plant, which registered the same phenomenon.Hence, the increase of tertiary PAH tar with a simultaneous decrease of lighter secondary tar species was systematically observed over a CCC in DFB steam gasification, revealing recombination patterns from secondary tar to tertiary tar.To some extent, these patterns are coherent with mechanisms discussed in literature, though the comparability of demonstration-scale experiments and measurement methods is limited.Even though the coarse gas cleaning system of the plant is not target of this study, it requires consideration as the peaking of Class V tar affects the downstream tar removal strategies.In fact, uncondensed Class IV and Class V tar are effectively removed with condensation in a quench and an adjacent RME scrubber, when compared to the lower separation efficiencies of e.g.1Hindene [19,36].Thus, recombination to tertiary tar can result in an improved overall tar separation.
In addition, the introduction of limestone as catalytically enhanced bed material indicates a dependence of the tar recombination on the bed material and the initial tar concentration in the freeboard.In contrast to experiments with olivine, no absolute increase of tar species was detected, yet the increased product gas formation and thus dilution of tar concentration were not considered, suggesting further research.The quantitative comparability of test runs with pilot-and demo-scale using olivine as bed material is shown within this study.Due to the convincing results using pure limestone as bed material, its use is suggested in the demonstration plant.The low attrition resistance of limestone did not lead to operational problems during pilot plant operation but has to be investigated in long-term test runs at demonstration-scale in more detail.
The application of residues derived from various origins as feedstock for gasification implies new challenges for ensuring the product gas quality.It could be demonstrated, however, that the total tar content is significantly reduced over the CCC and the resulting change in tar composition is beneficial for the further product gas cleaning system.Since this phenomenon was observed with various waste materials and plant sizes, advances were achieved not only for the identification of tar behavior in the GR, but also holistically for the application of residues in DFB steam gasification for further upscaling.Hence, the CCC is validated as a primary method for tar mitigation, when upscaled from pilot-to demonstration-scale and downstream gas cleaning requirements are thus reduced.In demonstration-scale, the tar concentrations are

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. Schematic reactor design of the 100 kW aDFB steam gasification pilot plant and location of the sample points from Benedikt et al. [15]; ILS = internal loop seal, LLS = lower loop seal, ULS = upper loop seal.

Fig. 6 Fig. 4 .
Fig. 4. Schematic reactor design of the 1 MW MW aDFB steam gasification demonstration plant and location of the sample points.

Fig. 5 .
Fig. 5. Integrated product gas cleaning of the demonstration plant.

M
. Huber et al.

Fig. 7 .
Fig. 7. Concentration of Class II, Class III, Class IV, and Class V tars in the freeboard and after the CCC for the experiments in demonstration-scale.

Fig. 8 .
Fig. 8. Concentration of indicator tar species in the freeboard and after the CCC for Wood Chips, Plastic Rejects Blend, Bark, and Forest Residues.

Fig. 9 .
Fig. 9. Tar dew points in the freeboard and after the CCC.

Fig. 10 .
Fig. 10.Temperature profile of the GR for the demonstration-scale trials.
Both tar classes occur mostly during the drying and pyrolysis step.With further temperature increase and additional air supply, alkyl and condensed tertiary tar are formed.Methyl derivates of aromatics such as toluene, indene, and methylnaphthalene are typical alkyl tertiary tar, whereas condensed tertiary tar without substitutes consist of polycyclic aromatic hydrocarbons (PAHs) forming at over 800 • C such as naphthalene and benzene [13,42]ntFormula Structure Tar group based on ECN and Milne[13,42]Condensation point

Table 2
Proximate and ultimate analyses of applied feedstocks, plastic-rich rejects were not applied alone, a Mixed blend of 50 wt.%wb /50 wt.% wb -calculated values, b Measured on site.

Table 3
Properties of the used bed materials olivine and limestone.

Table 4
Overview of operational parameters.

Table 5
Gravimetric and GC-MS tar concentration in the freeboard and after the CCC in the pilot and demonstration plant with olivine and limestone, LSP lower sample point, USP upper sample point.