EXAMENSARBETE INOM KEMIVETENSKAP, AVANCERAD NIVÅ, 30 HP
STOCKHOLM, SVERIGE 2017
Hydrodeoxygenation of Pyrolysis Oil
Comparing an Iron-based Catalyst with Dolomite
DENNIS FÄLLÉN HOLM
KTH Royal Institute of Technology
School of Chemical Science and Engineering
Master Thesis: 30 Credits
Hydrodeoxygenation of Pyrolysis Oil: Comparing an Iron-based Catalyst with Dolomite
KE2020X
Author:
Dennis F¨ all´ en Holm (dennisfh@kth.se)
Supervisor & Examiner:
PhD Student Pouya H. Moud Professor Klas Engvall
September 24, 2017
Abstract
This thesis evaluates the possibility to use a iron-based catalyst as a pyrolysis vapour conversion catalyst. The iron catalyst was also compared with the mineral dolomite. The experiments were facilitated at Cortus Energy’s demonstration plant in K¨oping, Sweden, by in situ instal- lation of the experimental setup to an outlet of the pyrolyser unit.
The pyrolysis vapour from Cortus Energy was converted for a total of 8 hours by passing it through a packed bed reactor containing the iron-based catalyst while sampling gas and oil from the feed for analysis. The outset for the operation on the dolomite catalyst bed was the same as for the iron catalyst with a resulting collapse of the bed when the pyrolysis vapour was introduced. The permanent gases were analysed on site with a µ-GC unit while oil samples were condensed and analysed with GC-MS, H-NMR and Karl Ficher titration. The carbon laydown and surface area of the catalyst was determined as well as the phase changes of the catalyst surface with XRD.
The results showed clear indications of bio-crude conversion with an eightfold increase of the H2 concentration of the synthetic gas from 3.38 % to 26.69 % on a dry gas basis. The oxygen to carbon (O:C) ratio decreased in the treated pyrolysis oil compared to the untreated oil while the hydrogen to carbon (H:C) ratio showed indications of dehydration of the oil. The gas and water content of the stream increased while 57.2 % of the oil was converted in the process.
Lastly, the iron-based catalyst did not seem to favour the conversion of alkylated phenols.
Key words: Biomass, bio-oil, bio-crude, hydrodeoxygenation, pyrolysis and tar cracking.
Contents
Page
1 Introduction 1
1.1 Background . . . 1
1.2 Aim and Scope . . . 1
2 Biomass 2 2.1 Biomass Products . . . 2
2.2 Biomass Characteristics . . . 3
3 Thermochemical Conversion 4 3.1 Pyrolysis . . . 4
3.2 Cortus WoodRoll®Process . . . 5
3.2.1 Drying . . . 5
3.2.2 Pyrolysis . . . 6
3.2.3 Gasification . . . 6
3.2.4 Combustion . . . 6
4 Hydrodeoxygenation of Pyrolysis Oil 7 4.1 Liquid and Vapour Phase . . . 7
4.2 Catalysts . . . 7
4.2.1 Properties . . . 8
4.2.2 Catalyst types . . . 8
5 Experimental 9 5.1 Experimental Conditions . . . 9
5.2 Catalyst Pre-activation . . . 11
5.2.1 Dolomite . . . 11
5.2.2 Iron-based Catalyst . . . . 12
5.3 Experimental Setup . . . 12
5.3.1 Pipes, Heating and Insulation . . . 13
5.3.2 Filter . . . 14
5.3.3 Reactors and Programmable Terminal . . . 14
5.3.4 Gas Analysis and Oil Sampling Setup . . . 14
5.4 Characterization Methods . . . 15
5.4.1 Catalyst Characterization Methods . . . 15
5.4.2 Permanent Gas Characterization . . . 15
5.4.3 Pyrolysis Oil Characterization . . . 16
6 Theoretical 16 6.1 Mass Balance . . . 16
7 Results and Discussion 17 7.1 Feedstock elemental analysis . . . 17
7.2 Reactor Temperature Profile . . . 17
7.3 Catalyst Analysis . . . 19
7.3.1 Temperature Programmed Oxidation . . . 19
7.3.2 Catalyst X-ray Diffraction Pattern . . . 20
7.3.3 BET Analysis . . . 22
7.3.4 Carbon Laydown . . . 23
7.4 Permanent Gas Analysis . . . 24
7.5 Pyrolysis Oil Analysis . . . 25
7.5.1 Oil Elemental Analysis . . . 25
7.5.2 Condensed Oil . . . 26
7.5.3 Water Content . . . 27
7.5.4 Hydrogen Nuclear Mass Resonance . . . 27
7.5.5 Gas Chromatography-Mass Spectrometry . . . 28
7.6 Mass Balance Analysis . . . 28
8 Conclusion 29 8.1 Challenges and Improvements . . . 29
9 References 30
1 Introduction
1.1 Background
The global energy use is estimated to rise with over 32 % worldwide between 2013-2040, according to the International Energy Outlook from 2015 [1], at the same time as stricter legislative measures forces an increased production of sustainable fuels and energy sources in the European Union (EU) [2] and world wide [3]. As a consequence of this, renewable resources are expected to make up 50 % of the power generation in the European Union by 2040 while becoming the largest global feedstock for power generation by the first years of 2030 [1]. Current trends in the chemical and energy industry focus heavily on the development of sustainable processes that have a satisfactory low environmental impact while still maintaining, or increasing the efficiency and profit, thus eventually leaving the dependence on finite resources such as coal, natural gas and petroleum behind [3].
Biomass, i.e. living or recently deceased organisms, shows a great potential for future energy and chemical production where the rationale of utilization can be attributed to the well-studied green and sustainable nature. Biomass is thus deemed as a promising energy source that can help to reach a sustainable energy sector [1, 4]
Present research and development allows for increasingly interesting possibilities where a broad range of biomass feedstock and techniques have accumulated, obtaining a seemingly mature and broad technology with room for improvement [5]. A furthermore important factor making biomass an influential energy source in the development of future technologies is that a wide variety of inexpensive biomass sources, e.g. waste from the agricultural sector are applicable thus giving developing countries access to sustainable energy sources [6].
In light of the current trends with increased focus on renewable resources that warrant cost effective and proficient technological solutions this thesis studies low temperature treatment of a pyrolysis vapour at atmospheric pressures without addition of hydrogen to the stream.
1.2 Aim and Scope
This master thesis has two aims, firstly the aim is to evaluate the potential performance improve- ment of the pyrolysis vapour due to hydrodeoxygenation of the oxygenated compounds within the pyrolysis oil. The evaluation of the performance improvement will be conducted by targeting a pro- duction of hydrogen gas and a more stable pyrolysis oil as a consequence of the hydrodeoxygenation.
Secondly, the aim is to to evaluate the iron (Fe) based tar-cracking catalyst with undisclosed chem- ical structure by comparing it with the well-studied low-cost alkaline earth metal mineral dolomite (CaMg(CO3)2).
The justification to the first aim of this study is that a successful hydrodeoxygenation of the oxygenated compounds in pyrolysis oil may prove beneficial in a multitude of aspects. It may improve the overall energy balance of the biomass conversion process by lowering the condensation temperature of the oil. The lowered condensation temperature will generate a decreased energy use of the heating of pipes while also lowering the risk of plug formation in the pipes. The hydrodeoxygenation may also improve the efficiency of the burners providing heat to the gasifier by breaking down the complex oxygenates into lighter hydrocarbons.
2 Biomass
Biomass is a broad term denoting material originating from organic living matter. It can be used as feedstock in production of power, chemicals, energy-dense fuels in gaseous, liquid and solid form.
Common sources of biomass are, crops, wood and municipal waste. The feedstock may be converted to a useful product by a number of techniques including biochemical-, chemical-, thermochemical- and physical conversion methods [4, 7]. The physical conversion techniques include processes such as distillation and mechanical extraction while chemical conversion techniques contain e.g. hydrolysis and solvent extraction [7]. The thermochemical conversion is carried out adding heat and chemical catalysts that help to promote the conversion of the biomass. The biochemical conversion process is governed by micro-organisms and biological catalysts, i.e. enzymes [4, 5, 7]. Prevalently used thermochemical processes include pyrolysis, gasification and combustion while fermentation is a biologically regulated decomposition process [5]. Biochemical processes are preferable in terms of milder reaction conditions and up-scaling possibilities while thermochemical processes are favoured in terms of residence time and the wider variety of suitable starting materials. Hybrid processes, i.e. combinations of thermochemical- and biochemical processes are currently emerging with the aspiration to mitigate the disadvantages and elevate the advantages of the processes [8].
2.1 Biomass Products
Biofuel products derived from different sources of biomass are generally categorized into one of the four different generations shown in Table (1) below, adapted from Pant and Mohanty [7]. The table gives a good representation of how extensive the range of useful biomass actually is, and the variety of products that can be produced via the different feedstocks.
Table 1: Biofuel generations shown with the common biomass sources, general products and benefits and drawbacks adapted from Pant and Mohanty [7].
Generation 1 Generation 2 Generation 3 Generation 4 Source Sugar, starch Wood, agricultural Micro algae Genetically
crops, vegetable waste, municipal biomass modified crop oil, soybean, solid waste,
animal fat, straw manure
Product Biodiesel, Hydro treating oil, Algae oil Biofuel polyalcohols bio oil, Fischer-
ethanol Tropsch oil
Benefit Environmentally Not competing Ability to add Captures
friendly, with food, nutrients, CO2 readily
economically environmentally leftover algae Carbon neutral
secure friendly is usable
Detriment Limited High oxygen Low growth rate,
feedstock content, viscous, oil extraction is
acidic difficult
A summary of Table (1) shows that first generation fuels are produced using possible food-sources while second generation fuels are made from a plethora of waste products and wood. The third generation is composed of products originating from algae while the fourth generation is derived
reasons, as it may not be deemed as a holistically sustainable approach to use an available food source as an industrial feedstock [9]. Biomass derived chemicals and fuels produced through the biomass conversion techniques, e.g. pyrolysis and gasification, may require upgrading to be suitable for end use.
2.2 Biomass Characteristics
The main main biomass components derived from wood and plants are cellulose, hemicellulose and lignin where the fractions of each component vary depending on the source. Cellulose is a polymer composed of glucose (C6H10O5) covalently bound by 1,4-β-glycosidic bonds where the hydrogen bonding between the cellulose chains promotes stability. Hemicelluloses are a class of sugars com- posed of five or six carbons where D-xylose and L-arbinose make up the former and D-galactose, D-glucose and D-mannose make up the latter group. The chains are highly branched, attributing to the amorphous nature of the sugars. Lignin on the other hand, is not a carbohydrate, instead it is an intricate organic molecule with heavily aromatized structure and a high degree of crosslinking [7]. Wood sources with smaller fractions of lignin are usually favoured for conversion [10].
The biomass composition is usually characterised via an ultimate or proximate analysis. Char- acterization is vital as the properties of the biomass will influence the production and products extensively. The contents of the feedstock can be divided into fixed carbon, moisture, volatile mat- ter and the ashy content in a proximate analysis [7, 11]. The ash is defined as the inorganic, often mineral, matter that remains after thermochemical conversion of the biomass while the volatile matter is the fraction of the feedstock that can be volatilised at elevated temperatures under a low rate of heating. A low ash content is preferential due to the potential of fouling of alkali and phosphorus containing ashes [11]. An atomic distribution is obtained through an ultimate analysis where the carbon (C), hydrogen (H), nitrogen (N), sulphur (S) and oxygen (O) contents are re- ported on the basis of a dry biomass [7, 11].
The higher heating value (HHV) is a good determinant of the energy content of biomass. Equation (1), defined on a dry weight fraction basis and empirically established, has been proposed by Channiwala and Parikh [13].
HHV = 0.3491 C + 1.1783 H − 0.1005 S − 0.0134 O − 0.0151 N − 0.0211 Ash (MJ/kg) (1) The equation shows how the different biomass constituents, determined in an ultimate and proxi- mate analysis contributes to the HHV where it has been shown to give an agreeable estimate of the HHV for gaseous, liquid and solid fuels [13].
3 Thermochemical Conversion
The present section introduces the concept of pyrolysis and gives a general description of the Cortus WoodRoll®Process.
3.1 Pyrolysis
Pyrolysis is a highly adaptable endothermic thermochemical decomposition process where biomass is converted into gas, vapour and solid in an anaerobic environment [12]. The lack of oxygen facilitates the production of the gaseous product of a given composition of carbon monoxide (CO), methane (CH4) and hydrogen gas (H2) as seen in the simplified reaction mechanisms below [7].
CnHmOk−−→ (1−n)CO + (m
2)H2+ C (∆HR= 180 MJ/kmol) (2)
CnHmOk−−→ (1−n)CO + (m − 4
2 )H2+ CH4 (∆HR= 300 MJ/kmol) (3) The other main products from the process are the volatilised moisture, the vaporized pyrolysis oil (bio-crude), the inorganic ash and the solid carbonaceous char, shown as C in Reaction (2) above [7, 11]. The pyrolysis oil is composed of organic molecules of a wide variety of molecular weights where a large quantity is present in an oxygenated form, as shown in Figure (1) below [14].
Figure 1: A summary of some of the general oxygenated compounds present in bio oil [14].
The amount of produced pyrolysis oil differs with the pyrolysis method that is deployed and the size of the particles that are feed to the pyrolyzer, where larger sizes tend to favour a higher char production [7, 15]. The most prevalent methods are categorized as either fast, intermediate and slow pyrolysis [15]. Other methods are flash, vacuum and hydropyrolysis [4]. The temperature range for each of the methods are presented in Table (2) below along with the heating rates that are commonly deployed.
Table 2: Temperature ranges and heating rates for slow, intermediate and fast pyrolysis. Adapted from Tripathi [4, 15].
Slow Intermediate Fast Vacuum Hydro Flash
pyrolysis
Temperature [°C] 550-950 500-650 850-1250 300-600 350-600 900-1200 Heating rate [°C/s] 0.1-1.0 1.0-10 10-200 0.1-1.0 10-300 > 1000
The slow pyrolysis process is mainly used to produce coke/char from biomass, while intermediate pyrolysis gives a more evenly distributed product of gas, oil and char. Fast pyrolysis on the other hand gives a product containing up to 75 % pyrolysis oil. High temperatures favour the production of bio oils as heavy molecules decompose more readily with increasing temperatures [4]. Yang et al. showed that the pyrolysis of hemicellulose occurs at slightly lower temperatures than cellulose followed by the more complex lignin [16].
3.2 Cortus WoodRoll
®Process
The WoodRoll® process from the clean-tech company Cortus Energy provides a novel process design to the well-studied and mature technology of converting biomass to synthetic gas (syngas).
The process is comprised of separate drying, pyrolysis, combustion and gasification stages where the volatiles from the pyrolysis is combusted to provide indirect heat for the process in an allothermic manner. The process produces an energy dense syngas in the gasification stage where steam is used as the reduction agent [23, 24, 25, 26, 27].
A high degree of control and precision is required to balance the energy balances of the separate stages of the process. A demo plant, located in K¨oping, Sweden, is used to fine tune the process before commercialisation. Figure (2) presents a simplified flowchart of the full WoodRoll® process where each stage is briefly described in the Section (3.2.1-3.2.4).
Figure 2: A schematic representation of the Cortus WoodRoll process [28].
3.2.1 Drying
The biomass, e.g. wood components or crops, is initially dried to decrease the water content of the fuel. The temperature in the drying unit is held above 100 °C at atmospheric pressure thus effectively removing the water from the biomass. High moisture content significantly decreases the energy value of the fuel and it is thus preferable to dry the biomass before the thermochemical conversion commences. The drying is conducted by passing air, that has been heated by heat
exchange with the effluent gas from the combustion stage, through the biomass thus removing moist from the fuel [24].
3.2.2 Pyrolysis
The biomass is fed into the pyrolysis reactor where the pyrolysis occurs in the temperature ranges of 350-450 °C. The solid char is fed to the gasifier while the volatiles, i.e. the bio-oil and gaseous products are fed to the burners for combustion to provide heat for the gasifier.
The hydrodeoxygenation reactor is installed after the pyrolyzer. The pyrolysis oil is rich in oxy- genated compounds at this stage where hydrodeoxygenation of the oxygenates may prove beneficial as it may increases the efficiency of the burners that provide heat for the gasifier. The conversion of the complex hydrocarbons to lighter hydrocarbons and permanent gases lowers the condensation temperature of the oil which decreases the risk of plugging of the pipes due to condensation. It could also lead to cost savings as a lower temperature will be held on the heating tapes on the pipes connecting the reactor unit.
3.2.3 Gasification
The pulverized char product from the pyrolysis stage is fed into the indirectly heated gasification reactor through a Laval nozzle. The transport is facilitated by a process gas that is reduced to syngas in the reactor where steam is used as the reduction agent thereby facilitating a production of a syngas that is rich in H2. The temperature requirements of approximately 1100°C are met via radiation from the combustion reactor.
3.2.4 Combustion
The volatile components from the pyrolyser are combusted in the gasifier burners, thus supplying the required heat for the gasification reactions to take place. Excess energy from the combustion gas is then used to heat the pyrolyser and to dry the biomass in the first stage of the process.
4 Hydrodeoxygenation of Pyrolysis Oil
Hydrogendeoxygenation (HDO) is a hydrolytic cleavage of a carbon and oxygen bond that removes oxygen and as a consequence decreases the sizes of the molecules in the oil. The process can be utilised for tar-cracking, where the decreased amount of oxygen and smaller molecules gives a more stable oil in the process as the use of highly oxygenated pyrolysis oil may incur a number of negative aspects compared to a deoxygenated oil. The problems may both influence the apparatus during the biomass conversion process as well as the product itself. An oxygen rich oil has a higher viscosity and is more corrosive than an oil that has been treated. The pyrolysis oil vapours readily form deposits on solid materials as a result of the carbonaceous coke formation during tar-cracking and the condensation of aerosols in the process. A negative product property is that a bio oil with a large amount of oxygen has a lower heating value than a corresponding bio oil that has been submitted to HDO [30, 31, 32].
Dayton [33] provides a generalized mechanism for the hydrodeoxygenation reaction shown in Re- action (4)-(5) below.
C5H8O4+ 6H2−−→ 6CH2+ H2O (4)
C5H8O4+ 4.5H2−−→ 6CH1.5+ 4 H2O (5) A number of different reactions are attributed to the removal of oxygen from the oil, viz, dehydra- tion, decarboxylation and decarbonylation. Literature covering HDO is largely composed of studies conducted on bio-oil model compounds such as guaiacol or anisole [30, 31] or mixtures of simple aromatic compounds [32] where further knowledge is needed to evaluate how large molecules from pyrolysis of lignin containing biomass influences the process. A number of distinct HDO mecha- nisms based on the type of model compound and catalyst have been suggested in various different literature sources [34]. However, the processing occurs with a pyrolysis oil in a liquid or vapour phase.
4.1 Liquid and Vapour Phase
Liquid phase hydrodeoxygenation has been thoroughly examined. The process is operated with heterogeneous catalysts at temperatures of 200-300°C with pressures of 50 bar H2[30, 37]. Elevated pressures of H2 is required to reach high conversion of oxygenates with a high degree of aromatic ring saturation as an unwanted side reaction [31]. This can have further negative consequences if the bio oil is processed to a fuel as the octane number decreases due to the saturation of the aromatic rings [31]. Compared to liquid phase hydrodeoxygenation, the process conditions for vapour phase HDO are completely different as the process is operated at ambient pressure at temperatures above 400°C to make sure that there is minimum condensation of the bio oil components. The use of a vapour-phase hydrogendeoxygenation-system as the one proposed in this report removes the heat loss in a corresponding liquid phase system attributed to the need to condensate the bio oil prior to the hydrodeoxygenation.
4.2 Catalysts
A number of catalysts have been reported to catalyse the HDO reaction. The modality behind the choice of catalyst is the same in liquid and vapour phase systems. A solid catalyst is preferred as the separation between catalyst and product is more readily achieved [37]. Olcese et al. states that there are a few notable criteria that influence the choice of catalyst, namely, ”low price, easy regeneration and selectivity to stable aromatics even in presence of pyrolysis gases” [30].
4.2.1 Properties
A number of properties of the catalyst i.e. pore size distribution and catalyst life time influences the HDO reaction. It is important that the pores are sufficiently wide to allow for passage of the molecules into the catalyst so that the active sites in the pores are accessible to catalyze the HDO reaction [29]. Furminsky and Massoth further states that a main factor governing the lifetime of the catalyst is the loss of active sites via deactivation. Where the deactivation is dictated by sintering, pore constriction, coking and poisoning [29].
4.2.2 Catalyst types
Several different types of catalysts adhering to different chemical species or classes have been used to promote the HDO reaction. These types of catalysts include carbides, noble metals, reducible metal oxides, nickel molybdenum (Ni-Mo) and cobalt molybdenum (Co-Mo) catalysts as well as low cost base metals [34]. Peters et al. conducted a study where four different hydrogenation catalysts were compared with an inert SiC solid for HDO with anisole and guiacole as model bio-oil compounds [35]. Two of the catalyst were Ni based, one Fe and one commercial zeolite fluidized catalytic cracking catalyst. The results concluded that all of the catalysts favoured higher temperatures, pressures and H2concentration of the feed where the Ni based catalyst outperformed the two other catalyst in terms of conversion and selectivity to the HDO reaction [35].
Wildschut et al. [36] showed that noble metals are active in liquid phase hydrodeoxygenation even though they favour saturation of the rings rather than hydrogendeoxygenation, where the high price is another drawback [36, 37]. Han et al. notes that supported Nickel (Ni) is promising for HDO in commercial use as it is inexpensive and a competent hydrogenation catalyst [37].
Addition of iron (Fe) to Ni further increases HDO instead of the saturation process [31]. Aho et al.
states that iron-based catalysts show a selectivity towards the production of alkyl phenols during hydrodeoxygenation [12].
Dolomite, a nonmetallic oxide mineral, on the other has been studied in a great degree for its tar-cracking ability [19, 21]. It is a widely distributed naturally occurring rock mineral with the chemical formula CaMg(CO3)2 that has a weight distribution of approximately 30 % CaO, 20%
MgO and 45% CO2 [21, 38]. The major benefits with dolomite is that it has a high activity toward tar-cracking and that there is plentiful native supply of the mineral that makes it a low- cost alternative. A drawback is that it is notably active for tar-cracking at elevated temperatures, usually above 800°C [21].
5 Experimental
The experimental section covers work done prior to, during and after the catalytic hydrogendeoxy- genation tests were conducted at the demo plant in K¨oping. The initial work with the catalyst included activation and structure-determination of the activated dolomite with X-ray diffraction.
The initial stages of work in K¨oping included the installation of the equipment and charging of the reactor with catalyst before the leak testing were done. The composition of the permanent gases in the feed was determined with gas chromatography at two spatial points during the project, upstream and downstream of the reactor. The target time for the catalytic reactor run was set to circa 75 hours thus mimicking the up time of an industrial pyrolyser. The experimental work after the trials in K¨oping consisted of analysis of the catalysts and the bio oil samples with the various methods presented in this section.
An overview of the experimental conditions of the pyrolyser and experimental setup is presented below, prior to a detailed presentation of the experimental work that was conducted during the thesis.
5.1 Experimental Conditions
The feedstock that was used during the experiments contained branches and tree top leftovers from the logging industry, here denoted as Grot, while the pyrolysis temperature was held at 380 °C throughout the period. The provided pyrolysis vapour flow to the reactor unit varied between 2.5-3 Nm3/h.
Table (3) gives the operational conditions for the pyrolysis unit operated on site in K¨oping that provided the pyrolysis vapour for the hydrodeoxygenation reactor unit.
Table 3: The operational conditions for the pyrolysis unit operated on site in K¨oping.
Temperature [°C] Feedstock Outlet flow [Nm3/h]
380 Grot 2.5-3
Table (4) shows the operational log for the testing period. The date of operation is shown with the time that the unit was operated on and at which periods that a stable run with pyrolysis gas through the reactor system was achieved. A number of failures that caused downtime are also presented in Table (4) along with the times when a successful GC-sample was obtained.
Table 4: The operational log for the testing period.
Date of operation Operational hours Stable hours of run Cause of failure Comment
2016-06-01 14:00-24:00 21:20-23:10 Clogged pyrolysis line In-situ catalyst activation
2016-06-02 Aborted Clogged line above reactor
2016-06-07 Aborted Failed heating wire on the flange of the filter
2016-06-15 16:00-20:00 Clogged flanges on pyrolysis line Inlet GC-sample (17:25)
2016-06-16 Aborted Programmable terminal failure
2016-06-28 Aborted Heating wire failure on Cortus pyrolysis line
2016-06-30 Aborted Power failure pyrolysis feeding screw
2016-07-04 Aborted Smoke development from the pyrolyzer
2016-07-05 16:30-24:00 Inlet GC-sample (18:00)
19:25-01:30 Completed activation (20:30)
Outlet GC-sample (20:40) Outlet GC-sample (22:00) Outlet GC-sample (23:10)
2016-07-06 00:00-01:30 Clogged filter inlet Outlet GC-sample (00:15)
15:00-22:00 16:30-18:00 Outlet GC-sample (17:20)
Inlet GC-sample (18:20) 19:00-19:30 Failure of the dolomite bed Dolomite bed
10
Experimental Setup
Table (5) below shows the predetermined operational conditions of the experimental setup. The reactor unit is denoted according to Figure (3) where reactor 1 (R1) remained unused, while essen- tially all of the uptime was spent on reactor 3 (R3).
Table 5: Operational conditions for the experimental setup.
Unit Catalyst Bed volume Temperature Ramping Space Time
[dm3] [°C] rate [°C/min] velocity [h−1] [h]
R1 Iron-based 1.6 400-450 10 0 0
R2 Dolomite 1.0 450 10 2500-3000 0.5
R3 Iron-based 1.0 400-450 10 2500-3000 8
The temperature of the iron-based catalyst was held at 400°C during the activation period whereas the temperature otherwise easily would overreach the cap of 500 °C that may otherwise have caused phase changes of the active surface of the catalyst that favours methanation instead of hydrodeoxygenation. The ramping rate was held at 10 °C in order to mitigate any structural changes of the catalyst surface due to exposure of a rapid temperature increase during the initial stages of heating. The space velocity varied between 2500-3000−1 depending on the flow rate with the upper limit of the total uptime for the iron-based and dolomite catalyst was set to 75 hours. A totalt time of 8 hours on the iron-based catalyst and 0.5 hours on the dolomite was reached.
5.2 Catalyst Pre-activation
The surface of the dolomite was activated by batch-wise calcination while the iron catalyst was activated in situ with H2and CO from the pyrolysis vapour. The activation methods are described in more detail in Section (5.2.1-5.2.2) below.
5.2.1 Dolomite
The dolomite from Sala Mineral AB was cured in order to activate the surface of the catalyst toward the HDO-reaction. The CaCO3 and MgCO3 in the dolomite was converted to CaO and MgO during the curing of the catalyst.
CaMg(CO3)2−−−−→ CaO + MgO + 2 CO900oC 2 (6) (7) The dolomite pellets were calcined batch-wise in batches of 250 ml in a Carbolite CWF 1200 lab- oratory chamber furnace. The batches were prepared by applying a single layer of dolomite on a ceramic tray. This was done in accordance with previous work [?] to counteract the observed in- complete calcination attributed to a build-up of thermal gradients within the sample. The thermal gradients had shown to have a negative effect on the curing, i.e. causing concentration gradients of CO2 within the sample, thus yielding a mineral that was not cured homogeneously.
The first batch was calcined for two hours at a temperature of 900°C with a temperature ramping rate of 5°C/min. Visual inspection indicated that the curing was not complete as white and partially white pellets were observed in the otherwise brownish catalyst sample. A X-ray diffraction (XRD) analysis confirmed this, as signals attributed to calcium carbonate (CaCO3) were observed in the
diffraction pattern, see Section (7.3.2). The batch was therefore calcined for an extra three hours under the same operation conditions as presented above. The remaining batches were calcined for 5 hours at the same temperature and ramping rate making up a total catalyst volume of 2 litre.
XRD analysis of the final calcination product contained CaCO3 albeit in a lesser extent than in the sample that was calcined for two hours. An additional calcination of crushed dolomite powder was conducted in order to fully calcinate the dolomite as a reference which can be seen in Figure (11a-b) in Section (7.3.2)
5.2.2 Iron-Based Catalyst
The Fe-catalyst was initially reduced into its active form by the CO and H2in the pyrolysis vapour.
The reduction reactions that converted the surface of the catalyst to its active form are outlined in Equation (8-11) [40]:
3 Fe2O3+ H2 −−*)−− 2 Fe3O4+ H2O (8)
2 Cr2O3+ 3H2 −−*)−− Cr2O3+ 3H2O (9)
3 Fe2O3+ CO −−*)−− 2 Fe3O4+ CO2 (10)
2 Cr2O3+ 3 CO −−*)−− Cr2O3+ 3 CO2 (11)
5.3 Experimental Setup
A schematic view of the reactor setup is shown in the piping and instrument diagram in Figure (3).
The setup was connected to the main pyrolysis vapour line of the Cortus pyrolyser on site in K¨oping.
The pipes, filter and reactors were fitted with heaters and thermocouples (T1-T10) to monitor and maintain the temperature throughout the process. Valves (V1-V19) were used to direct and control the flow while the raw- and hydrodeoxygenated gas was analysed with gas chromatography in situ.
The pyrolysis gas was initially fed from the main pyrolysis line through line a where larger partic- ulates were removed in the ceramic filter. The filter was periodically purged with N2 to prevent a build up of particulates on the filter. The pyrolysis vapour flowed through line b where it was fed through one of the three parallel-coupled reactor units (R1-R3) where the hydrodeoxygenation re- action was catalysed. The fourth column of the reactor setup is the bypass used when feed-through of pyrolysis vapour or inert gas was required. The hydrodeoxygenated gas was fed back to the pyrolysis vapour line through line c where it was fed through to the burner. Sections (5.3.1-5.3.4) provides a detailed description of the various sections of the setup.
Figure 3: Catalytic reactor setup: T-i denotes the heating tape-thermocouple pairs, V-i the valves while R-i are the reactors.
5.3.1 Pipes, Heating and Insulation
The connective piping system was constructed using 12 mm o.d. stainless steel pipes that were temperature resistant up to 800°C. The total length of pipes was approximately 16 m. Insulation and heating tapes from BriskHeat were used to mitigate the build up of axial thermal gradients in the pipes thus counteracting plugging due to condensation of tars in the vapour stream. The heating tapes were controlled with 10 temperature controllers that were connected to thermocou- ples fitted between the heating tapes and pipes thus enabling a precise control of the temperatures of the pipes. Two pairs of heating tapes with equal power and electric potential were connected in parallel to decrease the number of temperature regulators. The temperature controllers were in term connected to circuit breakers and solid state relays that switched the current according to the defined set point given on the temperature regulators.
The temperature of the heating tapes were maintained at 450 °C during the testing period to mitigate the chance of condensation and plug formation during operation. The thermocouple tem- perature controller pairs were installed as shown in Figure (3) above.
5.3.2 Filter
The filter had been constructed to remove larger particulates from the gas stream before it was introduced into the reactor. The temperature of the filter was maintained at 450 °C to prevent condensation of the tars while still removing large particles that had been transported with the fluid from the pyrolysis stage. The temperature of the filter was monitored with three thermocouples coupled to temperature controllers. The filter vessel was periodically purged with nitrogen (N2) every hour throughout the experiment thus mitigating carbonaceous cake formation on the ceramic filter preventing pressure drop and eventual plugging. The filter setup was powered with a 16 Ampere three phase cable to allow for the energy requirements of the heating process.
5.3.3 Reactors and Programmable Terminal
The reactor setup, as schematically shown in Figure (3), was constructed to separately operate one of the parallel-coupled reactors for the hydrodeoxygenation reaction. Each cylindrical reactor had a maximum catalyst loading volume of 2.6 litre. Inert Duranit® 1/8” balls were added to the bottom and top of the reactor in order to increasing the thermal conductivity of the otherwise empty reactor space and allow for a precise placement of the catalytic bed so that the top and bottom temperatures as well as the gas inlet temperatures could be monitored with the three pre- welded thermocouples that monitored the inside temperature of the vessels. The reactor vessels and connecting piping were heated by 19 heating tapes with thermocouples monitoring the temperature.
The reactor setup was heavily insulated to diminish the heat loss from the system.
Figure 4: Reactor setup shown with temperature measuring points controlled by the programmable terminal.
5.3.4 Gas Analysis and Oil Sampling Setup
A pyrolysis vapour sampling system as seen in Figure (5) was fitted to the system to allow for a continuous sampling of vapour upstream and downstream of the reactors thus allowing for the determination of the composition of the permanent gases and lighter hydrocarbons at any given point of time with a micro gas chromatography (µ-GC) unit. The sampling was operated during stable conditions in the reactor by switching the appropriate valve thereby feeding the sample through the cascade of 5 impinger bottles and a pentoxide column where moist was removed before the sample gas was introduced to the sensor. The first two and the last impinger bottles contained
in the series were placed in a cooling water bath that helped to condense the bio oil and prevented the iso-propanol from boiling.
Figure 5: The gas and oil sampling setup: The bio oil is collected in the two first isopropanol impinger bottles in the cascade while theµ-GC collects the permanent gas samples. The three first bottles were placed in a cooling bath.
The purpose of the bottle sequence was to condense the volatiles into the iso-propanol for further study and to cool the permanent gases before the gas analysis. The condensed oil was stored in appropriate flasks in a fridge awaiting analysis. The flasks were weighed before and after the condensation of the oil to allow for a gravimetric study of the pyrolysis oil.
5.4 Characterization Methods
The catalysts and pyrolysis products were analysed with a variety of different characterization methods elaborated on in Section (5.4.1-5.4.3) below. Firstly, the catalyst characterization methods are presented followed by the permanent gas and pyrolysis oil characterization methods.
5.4.1 Catalyst Characterization Methods
The catalysts were characterized prior to and after the hydrodeoxygenation. The initial character- ization determined the active phase of the dolomite with XRD while the post-trial characterization was implemented to ascertain any changes of the active surface of the dolomite catalyst. The surface area of the iron catalyst was determined with nitrogen adsorption using the Brunauer- Emmett-Teller (BET) theory while a LECO CS230 and ELTRA CS-2000 instrument was used to determine the carbon and sulphur laydown on the iron-based catalyst.
The calcined dolomite mineral was characterized using a Siemens D5000 X-ray diffractometer in order to determine the chemical structure of the catalyst after curing. Scattering angles (2Θ) from 10°to 90°with a step length of 0.020°and a step time of 2 s was implemented. Results are shown and discussed in Section (7.3.2). Furthermore, temperature-programmed oxidation was used to to explore the carbon depositions on the iron-based catalyst by analysing the catalyst with TPO before and after the HDO.
5.4.2 Permanent Gas Characterization
The permanent gas composition was as previously described determined on site with theµ-GC unit that allowed for real time sampling and analysis of the permanent gases in the system. A number of
volatile lighter hydrocarbons, namely Ethane, Ethene and 1-Butene that remained in the gas phase after the stream had passed through the cascade of impinger bottles and the pentoxide column was also detected with the chromatography unit.
5.4.3 Pyrolysis Oil Characterization
The condensed pyrolysis oil was subjected to a number of different analysis methods including gas chromatography-mass spectrometry (GC-MS), hydrogen nuclear mass resonance (H-NMR) and Karl Fischer titration and an elemental analysis. GC-MS was used in order to analyse the bio oil components of the pyrolysis vapour. H-NMR was used to determine the changes of the functional groups of the pyrolysis oil while the water content of the pyrolysis oil was determined with ISO 6296 standard Karl Fischer titration. ASTM 5291 standard elemental analysis determined the C, H and O ratio of the oil. The experimental setup did however not allow for an effective determination of the condensation temperature of the pyrolysis oil.
6 Theoretical
6.1 Mass Balance
A mass balance was developed in order to account for the carbon, hydrogen and oxygen in the process by evaluating the inlet (index i) and outlet (index o) streams on the basis of the carbon content. The accumulated carbon on the catalyst surface as well as the carbon in the condensed tar that was collected in the impinger bottles in the cleaning and cooling cascade prior to theµ-GC is accounted for. The accumulation of carbon on the catalyst surface was assumed to be linear in this model to easily be able to account for the deposited carbon by discounting the mode of carbon build up. The flow to theµ-GC that bypassed the reactor, thus retaining the inlet conditions was determined as follows:
˙nc,i= yc˙ni+ xc Mw
G (I)
Where ˙nc,i, is the molar flow rate of carbon through the inlet, yc the molar fraction of carbon in the gas and ˙ni is the total molar flow of gas and xc is the molar fraction of carbon in the oil. Mw
is the molar weight of carbon and G is the mass condensation rate of the oil. The reactor outlet stream has the additional accumulation on the catalyst surface and is thus evaluated as:
˙nc,o= yc˙no+ mc Mwtc
+ xc Mw
G (II)
Where the accumulation on the catalyst surface is determined by the total mass of the accumulated carbon mc and the total run time through the reactor system tc as described above.
7 Results and Discussion
The results are presented and discussed in this section. The outline follows that off the experimental section where the obtained data associated with the catalyst, permanent gas and pyrolysis oil are presented in separate sections. The permanent gas and pyrolysis oil data that are presented are from the untreated pyrolysis vapour stream and the stream that was treated with the iron catalyst as no reliable oil or gas samples were obtained from the tests on the dolomite catalyst. The data that was collected for the dolomite catalyst was under unstable conditions with a low flow and large pressure drop due to a collapse of the dolomite bed shortly after the pyrolysis vapour stream was introduced.
7.1 Feedstock elemental analysis
The elemental analysis of the Grot feedstock that was pyrolysed by Cortus AB during the testing period is shown in Table (6) below where the composition of the components are within the range of what is to be expected of a wood derived biomass.
Table 6: Elemental analysis of the grot biomass feedstock.
C [wt%] H [wt%] N [wt%] S [wt%] O [wt%] Ash [wt%] Cl [wt%]
50.4 6.0 0.43 0.030 41.6 1.6 < 0.02
7.2 Reactor Temperature Profile
The temperature on the inside of the reactor was monitored in real time to prevent that the catalyst was exposed to temperatures over 500°C as well as giving an indication of when the catalyst was fully activated.
The thermocouples were placed so that the temperatures of the inlet gas, the top section of the catalytic bed and the bottom section of the catalytic bed were monitored during the run as shown in Figure (6).
Figure 6: The temperature measuring point placement in the reactor for the reactor temperature profile.)
Figure (7) shows the start up period of reactor 3 (R3) containing 1 litre of the iron-based catalyst.
The abscissa shows the date and time while the ordinate displays the temperatures on the inside of the reactor.
Figure 7: The temperature profile during start-up of reactor 3 (R3)
A sharp increase of the bottom temperature of the reactor, as can be noted by the blue curve in Figure (7), is apparent during the activation of the catalyst. The nature of the temperature profile may be due to an overall exothermic activation of surface thus causing a sharply defined temperature gradient with elevated temperatures in the bottom of the reactor vessel during the initial stages of operation. The direction of flow would thus have transferred the exothermically released heat with the fluid through the reactor vessel giving the profile that is seen above. To counteract the increasing temperature the gas flow rate was decreased thus preventing the temperature to pass the threshold temperature of 500 °C as can be seen in the initial decreasing trend of the blue bottom temperature curve. The gas flow rate was then increased as the initial temperature spike settled to continue with the activation of the catalyst surface. The fluid flow rate through the system was controlled by reducing or increasing the suction of the ejector at the burner section of the setup.
The highest temperature that the bed reached was 489°C before the decreased flow rate mitigated the sum total of released heat due to the assumed sum total exothermic activation reactions. The temperature dip in the top of the reactor vessel can be attributed to endothermic reactions occurring on the surface of the solid material that may be the latter stages of the activation phase of the surface.
Downtime on the pyrolysis reactor led to a continued activation of the catalyst at a latter stage as can be seen in Figure (8) below. The final activation is assumed to have occurred at the bottom temperature increase at circa 20:20 and corresponding decrease in the top temperature of the bed.
Figure 8: The temperature profile during the continued start-up and initial GC-sampling of reactor 3 (R3)
The testing period commenced after the activation of the catalyst and was conducted at stable conditions with regards to the temperatures of the inside of the system as can be seen from the temperature profiles. Figure (9) below shows the final segment of the testing period on the iron-
°C increase of the gas inlet can be noted which stems from a 10 °C
increase of the heating tapes covering the reactor from 440°C to 450 °C.
Figure 9: The temperature profile during the final day of run of reactor 3 (R3).
7.3 Catalyst Analysis
A number of physical aspects of the iron-based catalyst is presented in this report. This includes the temperature programmed oxidation, a general XRD conclusion, a surface area and pore volume analysis as well as the carbon laydown on the catalyst. The failure of the dolomite bed on the other hand led to the inference that only the diffraction pattern of the fresh and used catalyst and a visual presentation of the catalyst was of interest. Earlier studies [39] suggested that nitrogen adsorption was not applicable on the dolomite catalyst and was thus not implemented on the catalyst to determine the pore size distribution of the catalyst.
7.3.1 Temperature Programmed Oxidation
TPO was implemented in order to quantify the carbonaceous depositions on the iron catalyst.
Figure (10) below shows the TPO results for the fresh catalyst (yellow curve) and the used catalyst samples from the top (grey curve), middle (blue curve) and the bottom (orange curve) of the catalytic bed. The red curve shows the linear increase of the temperature during the oxidation process. The temperature range spanned from a starting temperature of circa 30°C up to a final temperature of 800°C where the fresh catalyst shows carbon peaks at just under 100 °C, at circa 250°C and one with a peak at 750 °C as the carbon is oxidised by O2.
The results on all of the used catalyst samples give a clear indication that there are carbon de- positions on the catalyst surface as a sharp signal with the peak just below 450°C or at a time of approximately 70 minutes is apparent from Figure (10) below. The fact that the peak was not present in the results from the fresh catalyst shows that the carbon species attributed to this peak is not originating from the catalyst itself, hence it is a result of coke deposition on the catalyst surface.
Figure 10: Temperature programmed oxidation curve for the fresh iron-based aswell as the bottom, middle and top samples of spent iron-based catalyst.
7.3.2 Catalyst X-ray Diffraction Pattern
As introduced above, the diffraction patterns of the fresh and used dolomite was the main results for the failed dolomite catalyst while only a summarized result from the XRD analysis of the iron-based catalyst is presented below.
Fresh Dolomite
Figure (11a) shows the diffraction pattern after five hours of curing at 900 °C with a ramping rate of 5 °C/min. The crystal structures of calcium oxide (CaO) are denoted with a (4) while magnesium oxide (MgO) peaks are shown with a (). Peaks corresponding to uncalcined dolomite are given by (?). The presence of CaCO3 in the final product is attributed to the fact that the dolomite mineral was calcined in pebbles with particle diameters of approximately 1-2 mm causing non-uniform curing of the catalyst.
In reference, Figure (11b) shows the diffraction pattern for dolomite that had been crushed in a mortar to much finer sizes than the ones given in Figure (11a) prior to calcination thus facilitating a more uniform mass transport of CO2 from the catalyst. The curing conditions were identical to the ones in Figure (11a) in every other aspect.
(a) Dolomite pebbles calcined at 900°C for 5 hours with a 5 °C/min ramping rate.
(b) Dolomite powder calcined at 900°C for 2 hours with a 5 °C/min ramping rate.
Figure 11: X-ray diffraction pattern for calcined dolomite. CaCO3 is shown with ?, CaO with 4 and MgO with .
Spent Dolomite
The spent dolomite was analysed with the same setup as the freshly produced dolomite as described in Section (5.4.1). The dolomite particles in the top section of the catalytic bed were most visually altered compared to the freshly cured catalyst where they had obtained a black colour and were hence used for the XRD analysis. The diffraction pattern of the catalyst are shown in Figure (12)
Figure 12: X-ray diffraction pattern for spent dolomite. Calcite (CaCO3) is shown with ?, while the MgO is shown with .
The diffraction pattern from the spent dolomite sample taken from the top section of the catalytic bed shows deactivation of the CaO to CaCO3 as can be seen by comparing Figure (12) with Figure (11a-11b). The diffraction pattern does however indicate that the magnesium is still in its calcinated MgO form after the failed run.
Iron-based catalyst
The XRD analysis on the iron-based catalyst was conducted on samples that were gathered from the top, middle and bottom of the catalyst. The phases of the catalyst were fully active after the testing campaign and the XRD crystallites were as expected for a spent catalyst of this type.
7.3.3 BET Analysis Iron-based catalyst
The surface area of the iron-based catalyst was determined with BET as introduced in Section (5.4.1). The surface area of the fresh catalyst was presented as 71.9 m2/g in the previous thesis [39] which is compared to the values of the spent catalyst presented in Table (7) below.
Table 7: Surface area [m2/g] for the iron catalyst. Samples from the top, middle and bottom of the reactor bed.
Unit Top section Middle section Bottom section
Surface area [m2/g] 39 35.2 34.4
The decrease of the surface area can be attributed to ageing of the catalyst where the thermal effects causes a sintering of the bed with gradually collapsing pellets and pore closure as a result.
The trend of decreasing surface area can thus be explained with the results from Figure (7) in mind where it is shown that the catalyst pellets in the bottom of the bed are exposed to higher temperatures than the top sections of the bed. This is shown in it great effect as there is a well defined axial temperature gradient in the bed that is in agreement with the results shown in Table (7) above.
There is no strong indication that the pore size distribution changes significantly between the top, middle and bottom of the catalytic bed as can be seen in Figure (13) below where the incremental pore volume is plotted against the average diameter of the pores.
Figure 13: Pore size distribution of the top, middle and bottom section of the spent iron-based catalyst.
7.3.4 Carbon Laydown Dolomite
Figure (14) shows the light microscope view of the fresh and used dolomite catalyst where Figure (14a) shows the morphology of the surface of the fresh dolomite catalyst while Figure (14b) shows the used dolomite catalyst. The spent sample which was retrieved from the top of the catalytic bed shows a notable change in color in comparison to the fresh sample which stems from the carbon deposits on the catalyst surface.
The light microscope pictures in Figure (14) gives a good visual indication of the carbonaceous deposition of failed dolomite catalyst but does not give a quantifiable deposition of the carbon content on the catalyst surface as can be seen for the Fe catalyst that follows in the section below.
(a) (b)
Figure 14: Light microscopy of fresh dolomite shown to the left (a) and of the spent dolomite catalyst to the right (b) .
Iron-based catalyst
The carbon laydown is shown in Table (8) for the top, middle and bottom of the reactor which was determined using the LECO CS230 and ELTRA CS-2000 instruments. It can be noted that the top section of the catalytic bed shows a larger laydown of carbon compared to the bottom section with the middle of the bed found in between the aforementioned sections.
Table 8: The carbon laydown in the top, middle and bottom of the iron catalyst.
Top [wt%] Middle [wt%] Bottom [wt%]
∆C 10.8 10.2 9.7
The carbon laydown was used as basis of calculation of the accumulated carbon on the catalyst surface used to derive the mass balance in Section (7.6).
7.4 Permanent Gas Analysis
The average concentrations of the components of the treated and untreated pyrolysis gas on a dry and nitrogen free basis are shown in Figure (15a) while Figure (15b) shows the inlet and outlet concentrations of lighter non-condensed hydrocarbons that were detectable with theµ-GC equipment.
An increase of the H2 fraction of the permanent gas is apparent in Figure (15a) while the concen- tration of the rest of the permanents gases CO2, CH4and CO decreases from the inlet to the outlet samples.
(a) (b)
Figure 15: Average gas chromatography results for the inlet and outlet gases on a dry and nitrogen free basis. The blue bars (left) shows the inlet concentration while the orange bars (right) shows the average outlet concentrations.
The increase of H2 is a sign of treatment of the gas stream indicating that the catalyst successfully promotes the cracking of the tars in pyrolysis vapour stream. The increased H2 concentration is however not a general indicative of hydrodeoxygenation as a consumption of H2, in agreement with Reaction (4)-(5), would be expected for the HDO to occur. The results does however agree with the findings of Dayton [33] that showed an increased H2concentreation while using iron-based catalysts for HDO and also with previous studies conducted on the iron catalyst [39]. The water gas shift (WGS) reaction (CO + H2O −−*)−− H2+ CO2) may promote the production of extra hydrogen at the expense of CO but it is in that case not clear why the CO2 concentration decreases instead of increases as seen in Figure (15a) and Table (9).
Table 9: Molar component distribution in the inlet and outlet of the detectable gases on a dry and nitrogen free basis.
Sample CO2 CH4 CO H2 C2H4 C2H6 C4H8 Inlet [mol%] 56.07 5.73 33.06 3.38 0.78 0.71 0.27 Outlet [mol%] 54.56 3.76 13.33 26.69 0.78 0.70 0.18
7.5 Pyrolysis Oil Analysis
The pyrolysis oil was characterised with a number of different analysis methods described in the sections that follows. Firstly, an elemental analysis of the oils and the mass flow rates of oil are shown followed by the tar and water content. The results from the GC-MS and H-NMR are shown thereafter with the disclaimer that the analysis was conducted on a iso-propanol free oil.
7.5.1 Oil Elemental Analysis
Table (10) shows the elemental weight distribution of carbon, hydrogen and oxygen in the inlet and outlet oil streams. A noticeable decrease of the oxygen content of the oil is apparent from the results below with a resulting increase of carbon and hydrogen in the treated oil.
Table 10: The elemental analysis of the untreated (inlet) oil and treated (outlet) oil in terms of C, H and O.
Sample point C [wt%] H [wt%] O [wt%]
Inlet oil 50.3 4.9 44.8
Outlet oil 59.3 5.0 35.7
The results shows that the deoxygenation of the pyrolysis oil stream is in effect. Figure (16) below shows the van Krevelen diagram for the Grot biomass used in the experiment shown in Table (6) and the untreated (raw) and treated (outlet) oil. The hydrogen to carbon (H:C) ratio is plotted against the oxygen to carbon (O:C) ratio.
Figure 16: The van Krevelen diagram for the grot biomass feedstock and the inlet and outlet oil.
The atomic ratio of H:C plotted against O:C.
It can be seen that the heating value of the biomass feedstock is greater than the raw oil as the oxygen to carbon ratio is lower while the hydrogen to carbon ratio is higher. The outlet oil on the other hand shows a lower oxygen to carbon ratio and a lower hydrogen to carbon ratio compared to the raw oil and the feedstock. This is an indication that deoxygenation and dehydrogenation is occurring simultaneously during the tar cracking process.
7.5.2 Condensed Oil
The amount of condensed oil per minute of flow for the outlet and inlet streams are presented in Table (11) below where it is noticeable that the amount of oil per minute flow decreased from the inlet to the outlet.
Table 11: The number of grams of condensed oil per minute for the untreated (inlet) oil and treated (outlet) oil.
Sample Condensed oil [g/min]
Inlet oil 1.2
Outlet oil 0.6
The oil content per minute of flow showed a 50 % reduction between the inlet and outlet samples where an increased gas production balances out the loss of mass flow of condensed oil. The results does thus once again indicate of treatment of the pyrolysis oil.
7.5.3 Water Content
The water content of the pyrolysis vapour is presented in Table (12) below. The increase of con- densed water vapour from the inlet to outlet samples follow that of the proposed mechanisms by Dayton, Reaction (4)&(5) in Section (4). The dehydration of the oil will also work to increase the water content of the stream. The water gas shift reaction may however be levering the water production in the process.
Table 12: The water content of the pyrolysis vapour before and after HDO with iron catalyst.
Based on the mass flow of the sampling stream.
Sample type Water content [g/min]
Inlet gas 0.3
Outlet gas 0.49
7.5.4 Hydrogen Nuclear Mass Resonance
Table (13) below shows the type of functional groups from the H-NMR analysis. The results is not presented in a quantitative manner as it only shows the changes in concentration between the groups of untreated (inlet) and treated (outlet) oil as it is not taking in account the decreased amount of bio oil flow between the inlet and outlet samples. As can be seen, there is a notable increase in the concentration of phenolic and aliphatic protons while the integral values of the rest of the protons in the functional groups are decreasing from the inlet to outlet samples.
Table 13: The H-NMR results for the untreated (inlet) and treated (outlet) pyrolysis oil. The chemical shift region, type of protons as well as the area of the signals.
Chemical shift [ppm] Type of protons Inlet area [%] Outlet area [%]
8.0-10 Aldehyde 0.11 0.05
6.4-8.0 Aromatic 0.56 0.37
4.2-6.4 Phenolic 18.18 31.95
3.0-4.2 Methoxy 48.34 24.35
2.3-3.0 1,54 1.79
1.8-2.3 Aliphatic carbons 2.79 3.46
and alcohols
0.2-1.8 Aliphatic (alkanes) 28.49 38.03
An increase in the aliphatic protons may be due to saturation of aromatic structures in the pyrolysis oil seen as an unwanted effect of the HDO reaction. At the same time, the decrease of methoxy groups (ROCH3) can be described as an indication of tar cracking in effect as the oxygen bridging the alkyl-groups is removed by the hydrodeoxygenation, which in turn will cause a cracking of the oligomers in the oil. The increase of the integral of the phenolic protons is on the other hand noteworthy and may be explained by that the iron-based catalyst may not be activated towards the conversion of phenolic compounds.
7.5.5 Gas Chromatography-Mass Spectrometry
The GC-MS results for the main functional groups in the bio oil are shown below in Figure (17 below where a decrease of all of the functional groups are notable. This can be linked to a conversion of the oxygenated compounds within the bio oil. It can be seen that the GC-MS values for the phenolic compounds contradicts the H-NMR results as the phenolic compounds seem to be converted by the iron-based catalyst in the analysis of the GC-MS results where the H-NMR results showed an increase in the phenolic signal.
Figure 17: GC-MS results for the pyrolysis oil.
However, the increase of alkyl phenols at the cost of decreasing methoxy phenols is in agreement with the findings of Dayton [33], and Aho et al. [12] that observed the same behaviour while using an iron-doped catalyst. The results show that there is some saturation of the aromatic rings structures by analysing the results for the BTX (Benzene, Toluene, Xylene) compounds in the bio oil which can be expected while treating aromatic compounds with hydrogen. On the other hand, as previously stated, the catalyst does not seem to favour the conversion of alkylated phenols that only show an 8 % conversion while the catalyst seem to readily convert the methoxy phenols (such as the frequently used model compounds guaicol) with a notable 60 % conversion.
7.6 Mass Balance Analysis
The carbon of the process is accounted for by utilising Equation (I) & (II) developed in Section (6.1). It was found that 95.6% of the carbon in the process can be accounted for by analysing the carbon given in Table (9), Table (8) and Table (10) while all of the hydrogen and oxygen was accounted for thus closing out the mass balance for those species. The rest of the carbon is assumed to been have vented out through a side valve that was open when the reactor was cooled down with N2 according to the designated cool down procedure.
The mass balance for the three major components of the stream, i.e. gas, oil and water is shown in Table (14) below It can be noted that the gas and water content increases between the inlet and outlet sample while the oil content decreases.
Table 14: Mass balance of the permanent gas, oil and water in the pyrolysis vapour stream.
Sample Gas [wt%] Oil [wt%] Water [wt%]
Inlet 44.5 44.4 11.1
The increase in water content is attributed to hydrodeoxygenation and dehydration of the bio oil while the resulting decreased oil content with a 57.2 % conversion of the oil as a result. Furthermore, the increase of permanent gas with a high increase of the H2 concentration of the gas is a further indication of the tar-cracking phenomena.
8 Conclusion
Under the prevailing conditions it can be concluded that the iron catalyst was superior to the dolomite catalyst in terms of hydrodeoxygenation of the gas stream with a noticeable production of a synthetic gas rich in H2and a decrease of oxygen in the oil without any significant decrease in the hydrogen to carbon ratio. Furthermore it can be noted that the iron-based catalyst exhibits a decreased surface area and carbonaceous depositions, however the results do not indicate deactiva- tion of the catalyst after 8 hours of run. A longer testing period would be preferential to ascertain when the deactivation of the catalyst becomes evident. The dolomite catalyst on the other hand did not successfully hydrodeoxygenate the pyrolysis vapour as the bed collapsed 30 minutes after the vapour stream was introduced with clear indications of deactivation of the active surface of the dolomite catalyst.
Lastly, it can be concluded that the primary aim of the project was reached with a noticeable production of hydrogen gas and a more stable pyrolysis oil with a lower oxygen content. The secondary goal, i.e. a valid comparison between the iron-based and dolomite catalyst was partially fulfilled. It was noted that the iron-based catalyst was superior to the dolomite. However, the failure of the dolomite bed does question the comparison where a set of reliable data points would have been preferable to the reached outcome. The results from this study are promising for future applications both within the Cortus biomass conversion technique as well as other biomass conversion techniques.
A more complex and cohesive study where the change in condensation temperature and the burner improvement is determined will however also be required in regards to determining the benefits of this type of hydrodeoxygenation setup in a biomass conversion plant.
8.1 Challenges and Improvements
A number of challenges and problems occurred during the thesis work that ended up pushing the testing campaign closer to the deadline of the project than originally planned for. Which as a result, forced an extension of the project to allow for a successful testing and analysis period. The issues were of an assorted nature with the common denominator of causing major downtime. Furthermore, a sizeable time period was spent on maintenance of malfunctioning equipment, both on the reactor setup and its immediate vicinity. The extension of the project did in the end allow for a total of 8 hours run on the iron catalyst bed and a short duration on the dolomite catalyst before the bed collapsed.
The initial plan was to activate the iron-based catalyst by feeding a mixture of methanol and water through the catalytic bed prior to the introduction of the pyrolysis oil feed, thereby saturating the surface with H2. A T-connection, as can been seen in Figure (3) was fitted upstream of the catalytic reactor to facilitate a controlled introduction of methanol and water through the reactor which would decrease the consumption of H2in the pyrolysis oil during the hydrodeoxygenation reaction by pre-saturating it with H2. A strained time plan coupled with other maintenance priorities did however prevent this from being realised. A number of extra samples of both the inlet and outlet gas would be favourable to confirm theµ-GC results.
