Ex-situ Ion Enhanced Pyrolysis of Biomass
Effects of low power high voltage spark on the pyrolysis products
RIKARD SVANBERG
KTH ROYAL INSTITUTE OF TECHNOLOGY
Abstract
Bio oil from biomass sources can be part of the solution to reduce the net green house gas emissions to the atmosphere. One way of converting biomass resources to bio oil is with the help of pyrolysis that is a term for thermally decompose the feedstock in an oxygen free environment producing gas, liquid and char. However pyrolysis bio oil is heavily oxygenated and can thereby not be a direct substitute for fossil oil. In the quest of removing bounded oxygen form the oil, a subsystem to an fixed bed pyrolysis reactor has been designed, constructed and tested to experimentally evaluate the effects of a low power high voltage spark device with a power consumption of
approximately 2% of the reactor. The different spark scenarios tested are two power levels of direct current and one alternating current. These scenarios were then compared to a baseline scenario without any spark. The pyrolysis products in the states of gas and liquid where analysed with gas chromatography respective gas chromatography-mass spectrometry The effects of the spark is that liquid yield goes down while the gas yield goes up. Formation of hydrogen gas is increased, and sugars are crack to a higher extent. However any conclusions about deoxygenating couldn’t be determined with analysis methods used so further work is recommended.
Sammanfattning
Olja framställt från biomassa kan hjälpa till med att reducera nettoutsläppen av växthusgaser till atmosfären. Genom pyrolys vilket innebär termisk nedbrytning utan tillgängligt syre kan biomassa omvandlas till gas, vätska samt kol. Där bioolja kan utvinnas ut vätskan. Dock innehåller olja från biomasspyrolys hög halt bundet syre vilket medför att den inte kan användas som en direkt ersättning till fossil olja. I strävan att avlägsna det bundna syret i oljan har ett delsystem kopplat till en pyrolysreaktor med fast bädd designats, konstruerats och använts för att
experimentellt utvärdera effekterna av att skicka gnistor/högspänningsurladdningar genom pyrolysprodukterna innan dessa kondenseras. Gnistan motsvara ca 2% av totala reaktoreffekten. De olika scenarierna testade är två där gnistan genereras med likström och en med växelström som jämförs med ytterliggare ett utgångsscenario utan gnista. Pyrolysprodukterna i from av gas och vätska analyserades med
gaskromatografi respektive gaskromatografi följt av masspektrometri. Effekten av
gnistan är att utbytet av vätska gick ned medan mängden gas ökade. Mer vätgas
formades och högre andel socker spjälkades. Emellertid kunde inga slutsatser dras
kring huruvida gnistan påverkade andelen bundet syre i oljefasen, varför fortsatt
arbete föreslås.
Table of content
1 Introduction 1
2 Objective and hypothesis 1
3 Background 2
3.1 Pyrolysis 2
3.2 Gasification 2
3.3 Plasma 3
3.4 Plasma gasification 3
3.5 Biomass 4
3.6 Bio-oil 6
3.7 Gas chromatography 6
4 Methodologies 7
4.1 Pyrolysis reactor 7
4.2 Biomass preparation 8
4.3 Biomass composition 8
4.4 Experimental course of action 9
4.5 Pyrolysis parameters 10
4.6 Experimental scenarios 10
5 Development of ex situ ion enhanced pyrolysis system and repair of
temperature logging system 11
5.1 Spark tube: Design and construction 11
5.1.1 Evaluation of spark tube 12
5.2 DC – Driver: Design and construction 14
5.2.1 Ignition coil 14
5.2.2 Ignition coil driver 14
5.2.3 555-timer circuit 15
5.2.4 Final self built electrical system 16
5.3 AC - Driver 18
5.4 Temperature logging 19
5.4.1 Hardware and driver installation 19
5.4.2 Software - Labview programming 19
5.4.3 Evaluation of temperature logging system 20
5.5 Final system 21
6 Results 22
7 Discussion 27
8 Conclusion 28
9 Future work 29
10 References 30
Appendices 32
Appendix A: Fabricating list of operations 32
Appendix B: Arduino based ignition coil driver circuit 33
Appendix C: Compound categorization 35
List of figures
Figure 1 Schematic view of the Gasplasma® process (Advanced Plasma Power 2017) ... 4
Figure 2 β-(1 4)-D- glucopyranose, building block of cellulose (Kantarelis 2014) ... 4
Figure 3 Building blocks of hemicellulose (Kantarelis 2014) ... 5
Figure 4 Representation of lignin structure and its building blocks (Kärkäs, o.a. 2015) ... 5
Figure 5 Schematic view of a gas chromatograph system (Poole 2012) ... 6
Figure 6 Schematic view of the existing lab setup ... 7
Figure 7 Blender ... 8
Figure 8 Electrically driven sifter ... 8
Figure 9 Drying oven ... 8
Figure 10 Desiccator ... 8
Figure 11 Drawing of designed spark tube/spark plug housing ... 12
Figure 12 Self produced spark plug housing before Swagelok fitting was welded on ... 13
Figure 13 Spark tube housing including Swagelock fitting after multiple experiments ... 13
Figure 14 Schematic view of an ignition coil (The National High Magnetic Field Laboratory 2014) ... 14
Figure 15 Circuit diagram of ignition coil driver ... 15
Figure 16 555-timer in A-stable configuration used to signal ignition coil driver mosfet ... 16
Figure 17 Combined circuit diagram showing the whole DC-driver ... 17
Figure 18 Final signalling circuits based on the 555-timer ... 17
Figure 19 Neon sign driver used to power spark plug in third scenario ... 18
Figure 20 Block diagram of written Labviev program ... 20
Figure 21 Front panel of Labview program ... 20
Figure 22 Schematic view of the final version of the lab setup ... 21
Figure 23 Average yield based on measured results, error bar shows the standard deviation between repetitions with the same parameters ... 22
Figure 24 Average yields where the gas yield is calculated from mass balance. ... 23
Figure 25 Average volumetric gas compositions, error bar shows standard deviation between repetitions within the same scenario. ... 23
Figure 26 Average gas yields in mol normalized to mass of sample ... 24
Figure 27 Part of the compounds shown in figure 26 at different scaling. ... 24
Figure 28 elemental analysis of the gas, standard deviation with regards to differences within scenario ... 25
Figure 29 Ratios based on elemental analysis, standard deviation with regards to differences within scenario ... 25
Figure 30 Average composition of liquid yield, error bar shows standard deviation between repetitions within the same scenario. ... 26
Figure 31 Arduino microcontroller programming code ... 33
Figure 32 Arduino based switching circuit with filtering diodes, driving led as test before driving ignition coil ... 34
List of tables
Table 1 General impact of the relation between the process variables of temperature and residence time (Bridgewater 1994) ... 2Table 2 Proximate and ultimate analysis of biomass copy of table 4.2 in (Kantarelis 2014) ... 9
Table 3 Constant experimental parameters ... 10
Table 4 Experimental scenarios including power consumption of spark device ... 10
Table 5 List of machining operations ... 32
Table 6 GC-MS Peak are percentage of the liquid yield for the different experiments with categorization ... 35
1 Introduction
The nations of the European union have an agreement in order to prevent that global warming exceeds 2 °C compared preindustrial time. The agreement from 2009 involves a reduction of 80-95 % of green house gas emissions by the year 2050 compared to 1990 levels. Since every member state has different circumstances and abilities to reduce emissions, every member state has its own action plan for the years to come. For Sweden this involves for example increase of energy efficiency with at least 20 %, renewable energy should exceed 50 % and renewable energy in the transport sector should be at least 10 % by the year of 2020. The list continues and there are also goals that should be met by the year of 2030 etc. The Swedish
government has also visions in addition to the mentioned agreement that state that Sweden’s net green house gas emissions should be zero to the year of 2050, and by 2030 Sweden should have a vehicle fleet that is independent of fossil fuel.
(Miljödepartementet 2014) (COM112final 2011) (Regeringskansliet 2010) Pyrolysis oil from biomass could assist these goals and visions. However pyrolysis bio oil from biomass is heavily oxygenated and can thereby not be considered as traditional hydrocarbons (Crocker, 2010).
2 Objective and hypothesis
A spark device is to be designed and constructed to fit downstream an existing pyrolysis reactor and with the help of this device evaluate what will happen with the pyrolysis products when the pyrolysis gases is exposed to an electrical high voltage spark before being condensed. The main goal is to se if this setup can be used to deoxygenate the pyrolysis oil.
The hypothesis is that the spark will excite molecules enough that they might get split in half, or get functional groups broken away. This means that the spark-influenced reactions will favour formation of smaller molecules and should be seen with higher gas yield. With increased spark power bigger changes is expected with regards to yields and composition with regards to gas and liquid, the char is assumed to be non affected since the spark device will be situated downstream the bed. Depending on where the eventual splits might happen deoxygenation of the oil might occur but is not considered as a certainty.
One example of a successful oil deoxygenating reaction could be that a phenol gets its hydroxide split of and turned into benzene and the hydroxide might react with
hydrogen that naturally is formed during biomass pyrolysis to form water.
Another way of seeing proof of deoxygenating will be if more oxygen is present in the
product gas.
3 Background 3.1 Pyrolysis
The word pyrolysis comes from the Greek language where πυρ (pyr) means fire and λύσις (lysis) translates into breakdown or separation. Pyrolysis is considered to be the thermochemical decomposition of organic mater in absence of oxygen generally in the temperature range but not limited to 300-700°C. During the pyrolysis process large macromolecules of biomass decomposes into smaller molecules that can be classified as gas, liquid and solid (Kantarelis 2014). The initial products during pyrolysis are condensable gases and solid char, then depending on residence time at the heated zone the condensable gases may further break down into non-condensable gases (Basu, 2013). In table 1 below it’s shown how the variables of process temperature and residence time affects the pyrolysis yield.
Table 1 General impact of the relation between the process variables of temperature and residence time (Bridgewater 1994)
Process temperature Residence time Favour formation of
Low Long Char products
High Long Gaseous products
Moderate Short Liquid products
The pyrolysis process is also dependent on the heating rate, if the time to heat up the sample to pyrolysis temperature is much longer- or shorter than the characteristic pyrolysis time the pyrolysis process is considered to be slow respectively fast. Fast pyrolysis is used when the goal is to maximize liquid or bio oil yield, with a fast heating rate in order to minimize the decomposition of the sample before target temperature is reached. (Basu 2013)
3.2 Gasification
Closely related to the pyrolysis process is gasification. Unlike pyrolysis, the gasification process needs a gasification agent (usually, steam, air or oxygen) to rearrange the chemical bonds in the organic feedstock to produce smaller molecules.
If the gasification process is compared to combustion, the gasification packs energy into chemical bonds of the product gas by running the system at lower than
stoichiometric conditions. While the combustion process breaks these chemical bonds
further until they are completely oxidized into carbon dioxide and water. During
gasification, hydrogen is commonly used as an additive to strip the organic feedstock
out of its carbon producing a product gas with high hydrogen to carbon ratio. (Basu
2013)
3.3 Plasma
Plasma (from the Greek language πλασµα translates into moldable substance) is the so called forth state of matter, and can be described as the form of a heated gas, heated enough that collisions between the molecules are so intense that electrons are knocked off in the process forming an ionized gas. The problem with producing plasma with heat only is that the gas container most likely will evaporate and become plasma as well du to the excessive heat needed, therefore the gas is usually ionized by actuate an electrical current trough the heated gas, or by shocking the molecules with radio waves. (Goldston 1995)
3.4 Plasma gasification
Plasma gasification or pyrolysis since it does not necessarily need a gasification agent usually heats up an inert gas that passes trough the plasma field before reaching and mixes with the organic feedstock which gets thermally disintegrated into small molecules (e.g. CO and H2) in an complete lack or low oxygen containing
environment. The plasma arc can reach extremely high temperatures (13 000 °C) but in the stage where the inert gas blends with the feedstock the temperature range is 2700 to 5400°C. Plasma reactors is especially appropriated for decomposing municipal solid waste (MSW) since the temperature they reach are enough for tar cracking and is useful for break down dioxin and furan. Inorganic materials such as glass, metals and silicates fuse together at high temperature to form a molten slag that after cooling solidifies as a basaltic slag. Plasma reactors runs on electricity that
makes them insensible to feedstock in comparison to regular gasification methods, which is based on incomplete combustion. This might however make the process more costly than traditional methods due to the electricity used. Other disadvantages are corrosion from the extreme temperatures especially when MSW is used which usually contains highly corrosive elements such as chlorine. (Basu 2013)
Another plasma gasification process worth mentioning is the Gasplasma® system shown in figure 1 which uses a fluidized bed designed for gasification with steam or oxygen as gasification agent that down stream are coupled with a single carbon electrode plasma generator in the cyclone unit. This two-step process greatly reduces the electricity need when the plasma doesn’t stand for all the heating; only 3% of the total energy to fuel the whole process comes from the plasma. The high temperature and reactive environment created by the plasma is used to break down volatile organic compounds but it also melts inorganic materials into a slag (ashes and other
impurities), which is said to become a environmentally stable glass like substance
after cooling which greatly reduces the load on the down stream filters. With this two-
stage system it’s said to have high efficiencies regarding energy, ash recovery and
carbon conversion in its quest of producing clean syngas (Advanced Plasma Power
2017)
Figure 1 Schematic view of the Gasplasma® process (Advanced Plasma Power 2017)
3.5 Biomass
Cellulose, hemicellulose and lignin are the three major components of biomass.
Considering dry biomass 40-45% by weight is cellulose, which gives wood its strength.
The cellulose is structured as an linear polymer chain of of β-(1 4)-D- glucopyranose shown in figure 2.(Sjöström 1981) Thermal decomposition of cellulose occurs around 240-350°C with the main products of anhydrocellulose and levoglukosan (D. Mohan 2006).
Figure 2 β-(1 4)-D- glucopyranose, building block of cellulose (Kantarelis 2014)
Approximately 20-30 %wt of the biomass is hemicellulose, which is a polymer of different monosaccharaides, the main ones are shown in figure 3. A large portion of the acetic acid in pyrolysis oil has its origin from the hemicellulose. (Sjöström 1981) (D. Mohan 2006)
Figure 3 Building blocks of hemicellulose (Kantarelis 2014)
Lignin is the component which 25-35 %wt of dry biomass consist of, but stands for over 40% of the biomass energy content. The structure is highly branched and irregular and built by mainly tree monomers shown in figure 4, and acts as a binder between the linear cellulose polymers and helps with plant water transport and pathogenic defense. (Kärkäs, o.a. 2015) During pyrolysis the lignin decomposes over the range of 200-250°C (Vasile, o.a. 2009).
Figure 4 Representation of lignin structure and its building blocks (Kärkäs, o.a. 2015)
3.6 Bio-oil
Pyrolysis oils have benefits such as being sulphur free, renewable and considered environmentally friendly, and thereby interesting as a replacement for fossil fuels.
However the list of disadvantage is long and limits its use as a replacement fuel resource. Unlike conventional mineral based oils, the bio oil has poor heating value and volatility, the viscosity is high as well as the acidity. It’s also unstable when warm.
The fact that bio oil is heavily oxygenated is to blame for all of these properties.
(Cheng, o.a. 2016) The mixture of molecules in pyrolysis bio oil can roughly be categorized as the following oxygenates: hydroxyaldehydes, hydroxyketones, sugars, dehydrosugars, carboxylic acids and phenolic compounds (Kantarelis 2014). The bio oil form pyrolysis does also react during storage until it reaches thermodynamic equilibrium; aldehydes have been recognized as the compounds that are the most unstable. The carboxylic acids bring down the pH of the bio oil to a value of 2-3, which makes it highly corrosive. (Zhang, o.a. 2007)
3.7 Gas chromatography
Gas chromatography (GC) is an analysis method useful for separation, identification and for qualitative analyse of mixtures. A schematic view of a gas chromatograph is shown in figure 5. The chromatograph uses two phases, one mobile and one
stationary. The mobile phase is an inert gas, which pushes the sample along the column. The stationary phase liquid or solid is a thin film on the column wall. The sample interacts with the stationary phase and depending on the degree of interaction the more a certain compound will be delayed. This makes the different compounds of the mixture to exit the column and reach the detector at different times, called
retention time. The signal form the detector shows as a peak in the chromatogram that is a graphical view of the detector output. When performing gas chromatography on liquid samples, the liquid is first being vaporized before it enters the column.
(Utbildningsstyrelsen, Helsingfors u.d.) (Poole 2012) The detector can also be equipped with a mass spectrometer (MS) that ionizes the compounds leaving the chromatograph. This involves that the ionized compounds gets accelerated trough the mass analyser, where they are separated with regards to mass to charge ratio. The detector inside the mass spectrometer produces a spectrum based on the compounds ionized properties, and this spectrum can be compared to libraries to determine what the specific compounds in the sample are. (Thermo Fisher Scientific u.d.)
Figure 5 Schematic view of a gas chromatograph system (Poole 2012)
4 Methodologies 4.1 Pyrolysis reactor
The existing system shown in figure 6 below is a small fixed bed reactor that uses nitrogen for providing an inert environment. A flow controller controls the nitrogen flow, where the flow rate roughly can be correlated to pyrolysis products retention time since the flow of nitrogen pushes out the product gas out of the reactor. The reactor is heated by electricity where the heater is switched on and off with the help of a temperature controller. The sample temperature is supposed to be logged, but this system was non-functional due to the fact that the computer that logged the
temperature was infected by ransom ware and thus out of order. Downstream the reactor a condensing system consisting of a couple of gas washing bottles submerged in a mix of equal parts of isopropanol and water. Isopropanol was used in order to reduce the freezing point of the liquid. On the gas side after the condenser a gas collection bottle is attached for determining product gas volume before the gas finally is sent to a gas chromatograph. The liquid that condenses in the gas washing bottles is after dilution injected into a unit performing gas chromatography–mass
spectrometry (GC-MS).
.
Figure 6 Schematic view of the existing lab setup
4.2 Biomass preparation
The sample of biomass was grounded down with the help of a household blender (Andersson BLR 1.2) shown in figure 7. Then the finer biomass particles were sifted using two strainers shaken by an electrically driven unit shown in figure 8. The biomass particles that ended up between the two strainers of known grid size were collected and put in an oven (figure 9) at 105 °C for a couple of days to reduce the moisture content, and after this drying process the sample were put into a closed container called desiccator filled with a hygroscopic salt to ensure a moisture free environment during storage (figure 10).
Figure 7 Blender Figure 8 Electrically
driven sifter Figure 9 Drying oven Figure 10 Desiccator
4.3 Biomass composition
Proximate and ultimate analysis of the biomass used in the experiments had previously been performed by Efthymios Kantarelis, and presented in his doctoral dissertation 2014: Catalytic Steam Pyrolysis of Biomass for Production of Liquid Feedstock ISBN 978-91-7595-023-5. A copy of his results is shown in table 2 below.
He also concluded that the empirical chemical formula of the biomass could be
written as CH1.43O0.63N0.003.
Table 2 Proximate and ultimate analysis of biomass copy of table 4.2 in (Kantarelis 2014)
Proximate analysis [wt%]
Moisture 9.80
Volatile Matter
db83.00
Ash
db0.31
Fixed carbon
db16.60
Higher heating value [MJ/kg]
a20.46
Ultimate analysis [wt%
db]
C 50.70
H 6.10
O
b42.71
N 0.18
S <120 ppm
Metals [ppm]
Si 49.6 Pb 0.0523
Al 16.2 B 1.84
Ca 760 Cd 0.0556
Fe 23.8 Co 0.0228
K 390 Cu 0.626
Mg 106 Cr 0.14
Mn 95.6 Hg <0.01
Na 12.1 Mo 0.0106
P 31.6 Ni 0.059
Ti 0.659 V 0.0239
As <0.09 Zn 7.49
Ba 9.84
db
-dry basis
a-Calculated
b-by difference
4.4 Experimental course of action
The gas collection bottle was flushed with nitrogen, quenching bath cooler started.
Cooling bottles, tubing and connections weighed and assembled and finally a predefined mass of biomass was inserted in the middle of the reactor tube held in place with a metal mesh. Then the complete system was flushed with nitrogen and after the flush the gas outlet was clamped off for pressure testing the system with the help of the nitrogen flow controller. Leakages in connections were found with the help of a mixture of water and dish soap and tighten further. When the flow controller said less then 1 ml/min at a pressure of 5 bar (pressure controlled with regulator on top of gas bottle) the system was assumed to be tight enough since it runs at close to
ambient pressure when the gas outlet isn’t clamped off.
The reactor was then turned on, when the sample had reached 250°C the spark device
were also turned on, and gas rerouted to the gas collection bottle for the rest of the
duration of the test. Current draw of the electrical spark device was measured a couple
of times during the test using a clamp meter measuring on the ac input side of the
circuit. After the test, the reactor was from the outside cooled with compressed air.
Liquid yield was found by measuring the weight of the cooling bottles and tubes before collecting the oil. The char was extracted and measured as well, residue from the spark tube was cleaned with acetone into a beacon which was left a day for the acetone to evaporate before the final liquid yield could be calculated. The displaced water from the gas collection bottle was scaled to calculate the volume of the collected gas with the help of the density of water.
The gas in the collection bottle was pushed out of the bottle by refilling it with water and thereby injecting the gas into a micro gas chromatograph, which was used to calculate the composition of the gas, from the gas composition elemental composition were also determined.
The liquid yield was diluted with dichloromethane and injected into a GC-MS to identify the molecules in the liquid and get the composition in the form of peak area percentage.
4.5 Pyrolysis parameters
Fixed parameters for the experiments are presented in table 3 below.
Table 3 Constant experimental parameters
Pyrolysis sample Blende of soft woods (pine and spruce) provided by Svenska
Cellulosa Aktiebolaget, SCA Bio Norr
Sample particle size 0,5<0,71 mm
2Sample mass 5 g
Nitrogen flow of 50 ml/min
Reactor set temperature 500 °C
Heat tape set temperature 300 °C
Pyrolysis time: 30 min*
Gas collection: 36 min*
Quenching temperature -15 °C +/-5°C
* From t
sample= 250 °C
4.6 Experimental scenarios
The different spark scenarios tested are presented in table 4 bellow, together with the power consumption of the spark devices.
Table 4 Experimental scenarios including power consumption of spark device
Scenario Driver Spark power
consumption
No spark (baseline) Non -
DC Spark Self built (direct current) 8 W Increased DC spark Self built(direct current) 25W
AC Spark Neon sign driver
(alternating current)
25 W
5 Development of ex situ ion enhanced pyrolysis system and repair of temperature logging system
In order to perform desired experiments some changes to the system had to be done.
Both my supervisor and myself have a mechanical background so we decided without any extensive research to initially base the spark apparatus on car parts (i.e. spark plug and ignition coil). So firstly a spark chamber that could be mounted between the reactor and condensing system had to be built, then an electrical apparatus so the spark plug mounted to the chamber could spark, and finally the broken temperature logging system had to be fixed.
5.1 Spark tube: Design and construction
The spark tube or spark plug housing was design according to the drawing produced in solid edge shown in figure 11 with the criteria’s that the spark would be situated in the middle of the gas stream, and the housing itself mountable in both ends to the existing system between the fixed bed tube reactor and condenser. The spark tube is constructed out of a solid piece of mild steel with a trough hole along the axial
direction, for the gas stream. The spark plug is connected via threads radially situating
the spark end of the plug in the middle of the gas stream. The spark tube is connected
trough corresponding thread to the end of the reactor tube. The reactor tube did set
the limit of thread size to 24,5 mm, which isn’t a standard size for available tap and
die sets, why these where cut by lathe in a process where the rotational speed of the
work piece is matched with the axial feed of the cutting tool on the lath. When cutting
treads in this manner it’s important to end the tread with a bevel (Stamer 2017) so
that the fine tread cutting steel ends up in air at the end of every pass (repetition of
low depth cuts) until decried tread depth is achieved. Without the bevel the risk of
breaking the edge of the cutting tools is immense since it’s practically impossible to
stop the lathe in the exact same position for every pass. In the other end of the spark
tube a threaded pipe fitting from Swagelok was welded on to match a whole range of
equipment in the lab that uses the same type of fittings. In order to tightly screw the
spark tube on to the reactor with a wrench two parallel faces on the outside of the tube
where milled, and to complete the electrical connections a treaded hole were created
to allow for a electrical ground connection. A list of operations for the machining of
the spark plug housing can be seen in appendix A.
Figure 11 Drawing of designed spark tube/spark plug housing
5.1.1 Evaluation of spark tube
The final version of the spark plug housing worked as desired and became gas tight when used together with thread seal tape (PTFE) on both the treads connecting to the reactor and the treads on the sparkplug. Figure 12 and 13 shows the spark tube before and after all experiments where executed, and it can be seen that the mild steel has gained some surface rust. The inside of the spark tube has no signs of oxidation.
Figure 13 also shows the connecting end of the reactor tube.
Figure 12 Self produced spark plug housing before Swagelok fitting was welded on
Figure 13 Spark tube housing including Swagelock fitting after multiple experiments
5.2 DC – Driver: Design and construction
In this section a short background on how an ignition coil produces its spark, and how I controlled it with the help of a Mosfet that in turn was controlled by a 555-timer integrated circuit. Initially I tried to control the Mosfet with an easily computer programmable microcontroller under the trade name Arduino, but this system didn’t withstand the switching of the ignition coil because of what I believe is high voltage back currents created by the inductive load that the ignition coil is categorised as. A section about the Arduino version together with the program written to control it can be found in appendix B, which can be useful for switching a non-inductive load.
5.2.1 Ignition coil
The ignition coil (illustrated in figure 14) consists of two windings of insulated wires, a primary coil with few turns and secondary coil with many turns both wrapped around an iron core. The primary winding is connected to the battery and the current flow creates a magnetic field. When the current is turned off abruptly, the magnetic field created from the primary winding collapses and an induced voltage is created in the secondary winding that the spark plug is connected to. (Nave 2017) So in order for the spark plug to spark repeatedly, the primary winding has to be turned on and off at a suitable frequency with some sort of ignition coil driver.
Figure 14 Schematic view of an ignition coil (The National High Magnetic Field Laboratory 2014)
5.2.2 Ignition coil driver
To turn the ignition coil primary winding on and off in order to get the spark plug to spark a transistor (N channel mossfet) was used as in the circuit diagram shown in figure 15 below. The mosfet allows current to flow between drain (D) and source (S) when the voltage between gate (G) and source is above the so-called threshold voltage.
The resistor is used to pull down the gate voltage to ground in order for the mosfet to
break the connection between the drain and source, when the signal voltage is turned
of. (AspenCore, Inc)
Figure 15 Circuit diagram of ignition coil driver
5.2.3 555-timer circuit
In order to control the mosfet, a NE555p integrated circuit was used in an A-stable operation mode (see figure 16) to give the mosfet its control signal. The output voltage of the 555-timer in this mode of operation alternates between input voltage and zero.
Choosing different values of resistor R1, R2 and capacitor C1 can alter the
square/rectangular waveform produced by this circuit. Equations for calculating values of the three components are provided below (Eq.1-2) where th is the time that the output signal is provided and tl is the off time. These equations together with a lot more information about the NE555p and different ways to use it can found in the datasheet for this component. (Texas Instruments 2014) Chosen resistor and capacitor values are also provided in the equations with resulting timings
!
!= 0.693 !
!+ !
!!
!= 0.0693 10 + 12 ∗ 10
!∗ 1 ∗ 10
!!= 1,5 [!"] (Eq.1)
!
!= 0.0693 !
!! = 0.0693 12 ∗ 10
!1 ∗ 10
!!= 0.8 [!"] (Eq.2)
Figure 16 555-timer in A-stable configuration used to signal ignition coil driver mosfet
5.2.4 Final self built electrical system
The final electrical DC system shown in figure 17 is a combination of the two circuits
shown in figure 15 and 16 shown where the 555-timers output signal is connected to
the gate pin of the Mosfet. Vcc (positive supply voltage) is connected to 12V and the
combined circuit uses a common ground. To power the complete circuit, a mains AC
to 12V DC power supply was used.
Figure 17 Combined circuit diagram showing the whole DC-driver
5.2.4.1 Evaluation of the self built electronic spark system
The final circuit can be seen in figure 18, the observant reader might also se that a variable resistor in series with R1 is installed to enable some flexibility regarding the charging time of the ignition coil.
Figure 18 Final signalling circuits based on the 555-timer
The 555-timer circuit wasn’t as fragile as the Arduino and still work after many hours of use together with the ignition coil. However a computer-controlled system such as the Arduino had been more ideal for easier laboratory use. Fore example plugging an USB cable in to the computer and update the microcontroller with other on- and off- times had been easier than de- and re-soldering components that is necessarily with this setup since the variable resistor used only operates in a short range, and that is why the power level was altered by changing the spark gap instead.
The power supply maximum output was thought to be a limiting factor why I started to look around for another solution and instead of upgrading the power supply I ran across a Neon sign driver, however the limiting factor was found to be the spark gap i.e. the spark plug housing.
5.3 AC - Driver
The second working version of the electrical system was a neon sign driver HB-C10 (depict in figure 19) said to be able to sustain 30mA at 10kV witch corresponds to up to 120 W of input power from the 220V mains. In order to get a better understanding about how this transformer works the unit was opened only to find a thick layer of black epoxy covering everything. Maybe not that surprising when it has an IP66 rating corresponding to being completely dust proof (IP6x) and also withstand 100kPa of water being sprayed at a flow rate of 100l/min (fire hose with ø 12,5mm nozzle) without any traces of water inside the armature (IPx6) (Glamox).
Figure 19 Neon sign driver used to power spark plug in third scenario
5.3.1.1 Evaluation of neon sign driver
When running the neon sign driver, the power draw was no where near the stated maximum output, so my conclusion is that the unit probably produces a constant current at whatever voltage is needed to push that current trough. This has for obvious safety reasons not been tested, but input power readings went up when the spark gap was increased which is in line with the conclusion about the unit being a constant current power supply. Removing the ground electrode of the spark plug and letting the spark travel between the spark plugs positive electrode to the housing maximized the spark gap and thereby the power output.
5.4 Temperature logging
The existing temperature logging system was out of order (connected computer was infected by ransom ware), thus reinstallation of the computer had to be done together with drivers for the hardware. Finally a Labview program had to be coded as an
interface for the temperature logging system 5.4.1 Hardware and driver installation
To get the thermocouple readout to the computer the following (for the lab already existing) equipment was used. All except the thermocouple and the computer from National instruments. Thermocouple (k-type) ->SCXI 1300 -> SCXI 1102c -> SCXI- 1001 -> PCI-MIO-16XE-50->Computer windows operating system, where “->” shows signal pathway.
The driver used for the computer to communicate with the PCI card was NI MAX 16.0.1.f0, which had to be installed before the PCI card was inserted into the computer. Otherwise Windows self assign some other driver that doesn’t work.
5.4.2 Software - Labview programming
Via the hardware and driver the thermocouple signal was now available in the computer, but to translate and display the signal/temperature in a useful manner a software called Labview (version 16.0-32bit) was used. In figure 20 and 21 bellow a block diagram and font panel of the lab view program created is shown. The short version is that the DAQ assistant reeds in this case the signals and translate them into temperatures of three thermocouples (settings for how often and what type of
thermocouples used can be set in the DAQ Assistant block), and shows the temperatures on three separate graphs on the front panel and a fourth graph
containing all three temperatures. The boxes outside of the gray rectangle in the block diagram, reads previous temperature and then saves current temperature in the end of the for each time the program loops in order to be able to show historical
temperature (diagrams) instead of just momentary readouts. A time function was
used to reed the computers clock to be able to check for example when the data
collection was started.
Figure 20 Block diagram of written Labviev program Figure 21 Front panel of Labview program
5.4.3 Evaluation of temperature logging system
The temperature logging system worked satisfactory, and contained required
functions such as momentary temperature readout, and historical. However, when the
spark device was turned on the readouts became noisy approximately +/- 100°C at the
thermocouple that had its housing connected via the reactor tube to the sparkplugs
ground. But for the experiments need, it worked satisfactory since it was only critical
to know the temperature for when the spark was supposed to be turned on. The
thermocouple user for the temperature controller for the electrical heater was not in
electrical contact with the reactor and was thus not affected by the noise.
5.5 Final system
The final system is presented in figure 22 bellow, and contains the changes made to the system mentioned in previous chapter i.e. functional temperature logging and the spark device together with driver. For description of the rest of the system refer to chapter 4.1.
Figure 22 Schematic view of the final version of the lab setup
6 Results
The pyrolysis yield for the different scenarios is presented in figure 23 where the error bar shows the standard deviation between repetitions of the experiments with the same parameters. Here it shows that different DC sparks only have a slight influence on the yields but in the case of the AC spark less liquid is produced while the gas yield increased.
Figure 23 Average yield based on measured results, error bar shows the standard deviation between repetitions with the same parameters
If the losses are assumed to belong to the gas yield (see discussion on why this might be realistic) the average yields look like figure 24 where the gas yield is calculated from sample mass minus the sum of liquid and char. The trend regarding the AC spark stays the same, but this time the increased DC spark scenario shows a small increase of gas yield and reduced liquid.
0%
10%
20%
30%
40%
50%
60%
70%
Liquid Char Gas
[wt]
Average yield meassured
No Spark DC Spark
Increased DC Spark
AC Spark
Figure 24 Average yields where the gas yield is calculated from mass balance.
The average volumetric composition of the gas for the different scenarios is shown in figure 25, where the error bar shows the standard deviation between repetitions of the same scenario. Here it’s clear that all spark scenarios increase the hydrogen
composition of the gas but for the rest of detected compound the change is
inconclusive or scattered nonlinear with regards to the different power levels of the DC scenarios. Biggest change is however in the AC scenario compared to the baseline scenario i.e. without spark. Were the concentration of carbon monoxide and carbon dioxide went down. A small increase in concentration of ethylene, and acetylene is observed and an increase of propylene is suggested beyond the significant increase of hydrogen gas.
Figure 25 Average volumetric gas compositions, error bar shows standard deviation between repetitions within the same scenario.
0%
10%
20%
30%
40%
50%
60%
70%
Liquid Char Gas
[wt]
Average yield by mass balance
No Spark DC Spark
Increased DC Spark AC Spark
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
[vol]
Average gas composition
No spark DC Spark
Increased DC spark
AC Spark
To get the whole picture the gas composition might not be the best way of presenting the results since the gas yield differed between the scenarios. To see the gas yield in an absolute manner, where the average gas composition is shown on mol basis
normalized to sample mass see figure 26 and 27. Here it’s clear that the DC scenarios increase hydrogen production, and the hydrogen is increased with increased power.
The AC spark increase hydrogen, methane, carbon monoxide, ethylene, acetylene and propylene. When comparing the AC to DC we se that the AC has bigger effect on hydrogen and methane, shows effects on carbon monoxide and propylene, and also produces ethylene and acetylene.
Figure 26 Average gas yields in mol normalized to mass of sample
Figure 27 Part of the compounds shown in figure 26 at different scaling.
Elemental analysis of the gas presented in figure 28 and hydrogen- and oxygen to carbon ratio (figure 29) shows low effect of the DC spark, but the effects seams to be a
0 0,002 0,004 0,006 0,008 0,01 0,012 0,014 0,016
[mol] norm. to sample mass
Average gas composition mol
No Spark DC Spark
Increased DC Spark AC Spark
0 0,0002 0,0004 0,0006 0,0008 0,001 0,0012 0,0014 0,0016
C2H4 C2H6 C2H2 C3H6 C3H8
[mol] norm. to sample mass
Average gas composition mol break out
No Spark DC Spark
Increased DC Spark
AC Spark
function of spark power except for the oxygen to carbon ratio. The AC spark outperforms the DC and shows significant increase of all elements and the ratio between the hydrogen to carbon goes up, while oxygen to carbon goes down.
Figure 28 elemental analysis of the gas, standard deviation with regards to differences within scenario
Figure 29 Ratios based on elemental analysis, standard deviation with regards to differences within scenario
0 0,005 0,01 0,015 0,02 0,025 0,03 0,035 0,04
H C O
[mol] norm. to sample mass
Gas - Elemental analysis
No Spark DC Spark
Increased DC Spark AC Spark
0 0,2 0,4 0,6 0,8 1 1,2 1,4
H/C O/C
Atom ratio
Hydrogen- and oxygen to carbon ratio in produced gas
No Spark DC Spark
Increased DC Spark
AC Spark
The results from the liquid analysis are shown in figure 30. The diagram shows the average area percentage from the GC-MS output, since calibration numbers are not available these area percentages cannot be converted into concentrations but instead can only give a hint on changes Only compounds that in at least one experiment had a concentration of minimum 1% is included in this analyze. The unknown represents compounds that the GC/MS system didn’t recognize to a fulfilling certainty (50%), or gave so many different suggestions that it was impossible to draw any conclusions about categorising them. The categorisation together with GC/MS residence time is shown in appendix C. All changes compared to baseline scenario when it comes to the DC sparks are within the standard deviation between repetitions and no change can be confirmed. For the AC scenario acids, alcohols, furans, ketones and phenols went up, while sugars goes down.
Figure 30 Average composition of liquid yield, error bar shows standard deviation between repetitions within the same scenario.
0%
10%
20%
30%
40%
50%
60%
GC/MS peak area percentage
No spark DC spark
Increased DC spark
AC spark
7 Discussion
The spark device was turned on and the gas flow was redirected into the collection bottle when the internal sample temperature reached a specific temperature. The reason for this was because the volume of the gas collection bottle was to small for the entire time of the experiment. This temperature was chosen because it was a little bit lower than what visual inspection could see that products exited the reactor. First of all, all gases has not a color to it, many of them are transparent so choosing starting point based on visual inspection is not to recommend. Secondly inside the sample, there is a temperature gradient, and the part of the sample that is closest to the reactor wall will have a higher temperature then the middle part where the
thermocouple was situated. Thirdly, in literature there was found that biomass starts to decompose at lower temperatures than the one chosen as starting point of the spark device and gas collection. So there was probably pyrolysis product produced before the gas was collected. However to the rescue is that the collection bottle was situated approximately 6 minutes after the reactor with predefined inert nitrogen flow, but some transparent gases probably already escaped before the collection was started.
Because of this, the method fore deciding the yields should be to assume the gas based on mass balance (results shown in figure 24) instead of calculation based on the
collected gas (figure 23). The gas collection calculations has also a flaw, the calculations are based on an assumption that the gas is stored under ambient
pressure, however the gas is stored under a small water column making it a little bit compressed.
Since the GC-MS used isn’t calibrated with regards to oxygenated compounds the result form the analysis (figure 30) only show peak area percentage. If the GC-MS where calibrated, those peaks could have been translated to concentrations and those concentrations could been used to evaluate if the bio-oil had been deoxygenated together with a test of water content in the liquid, since the GC-MS does not detect water. As of now I can only say that the liquid is slightly deoxygenated because of the higher amount of elemental oxygen in the gas phase as shown in figure 28, but if this is because of an change in water content or if the oxygen comes from the oil part of the liquid yield is hard to say, and that is why water content is proposed as further work.
Also if the water content together with analysis of the char is performed, the elemental oil composition could be calculated since the input (biomass) and other outputs
already is known, leaving only the oil part of the liquid part unknown. Which would
be easy to calculate with the simple equation that input equals the sum of outputs.
8 Conclusion
The effect of the small DC spark is inconclusive except for gain in hydrogen production.
The effect of the increased DC spark suggest that the yields is a function of spark power where increased spark power increases gas yield unlike the liquid yield that goes down. The gas compositions changes with higher hydrogen concentrations with higher spark power.
The effect of the AC spark is that it increases gas yield while liquid goes down.
Regarding the gas yield, more hydrogen, methane, carbon monoxide and propylene is produced. Also ethylene and acetylene is formed which was non detected without spark. When it comes to the liquid yield sugars goes down, while acids, alcohols, furans, ketones and phenols goes up.
Comparing AC to DC it’s clear that the AC has bigger effects regarding yield. The same goes for gas composition where the AC also shows growth where the DC did not. The AC spark also produced acetylene and ethylene, which nether of the DC- or baseline scenario did. For the liquid fraction only the AC scenario showed significant changes.
Proof of eventual oil deoxygenation via methods and measures stated in hypothesis
cannot be given. Phenolic compounds went up, and no sign of benzene in the GC-MS
results. The gas for the AC scenario was proved to contain more oxygen but it can only
conclude that the liquid yield is deoxygenated, not that the oil contained in the liquid
yield is deoxygenated, assuming that the in situ char fractions is constant between the
ex situ experiments.
9 Future work
Since proof of deoxygenating couldn’t be given further analyze of the liquid yield is recommended. With the help of Karl-Fisher titration, the water content and total acid number can be determined. Depending on those results decision about further
investigation about oxygen content can be made.
Redo experiments but collect gas during the whole experiment.
Online measurements on gas yield might give information about when the hydrogen formation is high, and from the timing the momentary sample temperature can be checked and with this information it might be possible to track sources of compounds that got hydrogen split of by the spark, especially if there is some peaks before target temperature is reached.
Redo experiments, but with higher spark power, however this needs another spark tube with longer spark gap if either one of the drivers used during these experiments will be used.
Perform elemental analysis of the char in order to by difference calculate elemental
composition of oil, when water content from the Karl-Fisher titration also is done.
10 References
Advanced Plasma Power. Advanced Plasma Power - The energy from waste solution.
2017. http://advancedplasmapower.com (accessed 8 17, 2017).
AspenCore, Inc. Electronics tutorials. http://www.electronics- tutorials.ws/transistor/tran_7.html (accessed 02 2017).
Basu, Prabir. Biomass Gasification, Pyrolysis and Torrefaction. Vol. 2, 147-176, 199- 248, Pages 249-313. 2013.
Bridgewater, A.V. "Catalysis in thermal biomass conversion." Applied catalysis A:
general 116 (1994): 5-47.
Cheng, Tingting, Han Yehua, Zhang Yanfen, and Xu Chunming. "Molecular
composition of oxygenated compounds in fast pyrolysis bio-oil and its supercritical fluid extracts." Fuel 172 (2016): 49-57.
COM112final. 2011.
Crocker, Mark. " Thermochemical Conversion of Biomass to Liquid Fuels and Chemicals." In Royal Society of Chemistry, 289. 2010.
D. Mohan, C.U. Pittman Jr, and P.H. Steele. "Pyrolysis of Wood/Biomass for Bio-oil:
A Critical Review." Energy & Fuels 20 (2006): 848-889.
EETech Media, LLC. Introduction to Diodes And Rectifiers.
https://www.allaboutcircuits.com/textbook/semiconductors/chpt-3/introduction-to- diodes-and-rectifiers/ (accessed 05 2017).
Glamox. IP klasser. http://glamox.com/se/ip-klasser (accessed 07 2017).
Goldston, R.J. Introduction to Plasma Physics. London: Institute of Physics Publishing, 1995.
Kärkäs, Markus D, Bryan S Matsuura, Timothy M. Monos, Gabriel Magallanes, and Corey R. J Stephenson. "Transition-metal catalyzed valorization of lignin: the key to a sustainable carbon-neutral future." Department of Chemistry, University of
Michigan, Ann Arbor, Michigan 48109, USA, 12 2015.
Kantarelis, Efthymios. Catalytic steam pyrolysis of biomass for production of liquid feedstock. Stockholm: KTH – Royal institute of technology – School of industrial engineering and management, 2014.
Miljödepartementet. "klimatfardplan-2050---strategi-for-hur-visionen_H2B153."
Kommittédirektiv 2014:53, Stockholm, 2014.
Nave, Carl Rod. 2017. http://hyperphysics.phy-
astr.gsu.edu/hbase/magnetic/igcoil.html (accessed 08 01, 2017).
Poole, Colin. Gas Chromatography Gas Chromatography. Elsevier Inc., 2012.
Regeringskansliet. "The Swedish National Action Plan for the promotion of the use of renewable energy in accordance with Directive 2009/28/EC and the Commission Decision of 30.06.2009." 2010.
Sjöström, E. " Wood Chemistry - Fundamnetals and Applications." Espoo: Academic Press, 1981.
Stamer, Jan. Stockholm, (02 2017).
Texas Instruments. "xx555 Precision Timers." 09 2014.
The National High Magnetic Field Laboratory. Ignition Coil - Maglab. 12 10, 2014.
https://nationalmaglab.org/images/education/magnet_academy/tutorial/ignitioncoi l.jpg (accessed 02 2017).
Thermo Fisher Scientific. Gas Chromatography Mass Spectrometry (GC-MS) Information. https://www.thermofisher.com/us/en/home/industrial/mass- spectrometry/mass-spectrometry-learning-center/gas-chromatography-mass- spectrometry-gc-ms-information.html (accessed 09 25, 2017).
Utbildningsstyrelsen, Helsingfors. Laboratorieanalyser-Gaskromatografi.
http://www.edu.fi/laboratorieanalyser/analysmetoder/2_5_gaskromatografi (accessed 09 25, 2017).
Vasile, Mihai, Brebu, and Cornelia. "THERMAL DEGRADATION OF LIGNIN – A REVIEW." “Petru Poni” Institute of Macromolecular Chemistry, 41A, Gr. Ghica Voda Alley, 700487 Iasi, Romania, 2009.
Zhang, Q., J. Chang, J. Wang, and Y. Xu. "Review of biomass pyrolysis oil properties
and upgrading research." Energy Conversion and Management 48 (2007): 87-92.
Appendices
Appendix A: Fabricating list of operations
A list of operations for guidance during construction of the spark tube is presented in table 5.
Table 5 List of machining operations
Op.
nr Operation Machine Tool
1 Facing Lathe Facing blade
2 Drill trough hole HSS Ø6 mm
3 Turn profile Roughing and
finishing blade
4 Cut off to length Parting blade
5 Turn work piece around 180° and reattach in lathe
6 Turn outside thread Lathe Fine thread cutter
7 Smoothen sharp edges File
8 Cut outer faces Milling
machine
End mill 9 Drill blind hole
10 Counter bore sink 90° cutter or end
mill
11 Thread for sparkplug By hand Tap
12 Drill hole on outside face for ground
screw Drill press M5
13 Thread hole By hand
14 Turn inside thread corresponding to
spark tube on reactor tube Lathe Fine thread cutter
Appendix B: Arduino based ignition coil driver circuit
The first design to switch the power to the ignition coil on and off in order for the spark plug to spark, was based on a computer programmable micro controller board sold under the trade name Arduino, that is programmed to produced a pulsed signal voltage to the transistor (n-channel logic level mosfet) which in turn acted as the on and off switching device for the ignition coils primary winding. In figure 28 the computer code produced for the Arduino is show, which tells the microcontroller to give a 5V signal for a certain amount of time, turn of that signal for another amount of time and repeat. This circuit worked fine for approximately one minute before the Arduino got destroyed. The next step was to try filtering back currents created by the switching of the inductive load/ignition coil with the help of diodes that works as one way check valves for usage of hydraulic analogy, i.e. only letting current flow in one direction (EETech Media, LLC u.d.). Both this circuit and the one previously
described worked fine for driving a couple of light emitting diodes (LED) shown in figure 29, but when the LED was changed to the ignition coil, it only took a short time before the Arduino gave up. To go further into filtering which is out of my electrical knowledge I did put this Arduino based circuit to rest and had success with a 555- timer integrated circuit instead.
Figure 31 Arduino microcontroller programming code
Figure 32 Arduino based switching circuit with filtering diodes, driving led as test before driving ignition coil
