Luận án tiến sĩ Kỹ thuật hóa học: Plant-based Biomass Pyrplysis and Reaction Kinetic Models in Different Types of Reactor

As for the fast pyrolysis in tubular reactor, first of all, a comparative study of pyrolysis was carried out using both types of biomass: palm kernel cake major non-cellulosic material a

General Introduction 1.1 Introduction

Biomass source

As commonly known, fossil fuels, namely oil and coal, are limited Oil is estimated to be run out within 40 years and coal within 250 years from now [1]

From this issue, many researchers have focused on finding a new source for energy

Biomass is one of the most interesting alternative sources, which has been paid much more attention to in recent years, owing to its low cost as well as its abundance [2] The term biomass is used to describe all biologically produced substances World production of biomass is estimated at 146 billion metric tons a year Some farm crops and trees can produce up to 20 metric tons of biomass per acre a year Only algae and grasses may produce 50 metric tons per year [3] The source of biomass is infinite and can be replenished through natural processes, hence it is also considered as a renewable source for producing energy For

According to the International Energy Agency, approximately 11% of the energy is derived from biomass throughout the world [4]

Biomass can stem from timber industry, agricultural crops, forestry residues, household wastes and wood [5] Of all these types of biomass, woody biomass makes up the most major amount Therefore, most researches have now focused on this one This material largely contains hemicellulose, cellulose (for wood-based biomass) or cellulose-like compound (for seed-based biomass), and lignin of which the content could change depending on type of biomass Hemicellulose, a branched biopolymer with a random and amorphous structure, is most favorably decomposed

In contrast, lignin is the most difficult one to be degraded due to its cross-linked three-dimensional structure Cellulose and cellulose-like compound are more thermally stable than hemicellulose because of its crystalline and linear-chain structure Also, the structure of cellulose and cellulose-like compound are rather simple compared to a cross-linked three-dimension structure of lignin, leading to its lower thermal stability [5]

A biomass can be cellulosic or non-cellulosic depending on whether its carbohydrate contains cellulose or not Most recent researches have used cellulosic biomass as a feedstock for conversion into energy Apart from cellulosic biomass, the non-cellulosic one is also cropped with a considerable amount every year For instance, only Malaysia can produce an annual quantity of 1.4 million tons of palm kernel cake, a non-cellulosic biomass, as a by-product in the milling of palm kernel oil [6] The major composition of carbohydrate building up this biomass structure is mannan, a cellulose-like compound Although produced with such a large amount, there have been few researches reported about its usage for studying in biomass conversion As can be seen, if the biomass composition changes, it might result in varying the properties and distribution of pyrolytic product Nevertheless, the characteristics of pyrolysis of a non-cellulosic biomass, especially like palm kernel cake, have not been studied yet.

Pyrolysis mechanism

As raw biomass is solid, it is difficult to use in many applications without substantial modification Conversion to gaseous and liquid energy carriers has many advantages in handling and application There is a wide range of processes available for converting biomass and wastes into more valuable fuels, but only two general thermochemical and biological processes are considered as feasible solutions [7] The thermochemical conversion technology tends to be grouped in four distinct categories for fuel production: combustion, pyrolysis, liquefaction, and gasification As for biological technology, this refers to approaches involving

Since bio-oil is now received more interest as an alternative for the depleted fossil-based fuel, much effort has been done to find an effective and economically feasible technology for obtaining as high bio-oil yield as possible A pyrolysis process, or a mild depolymerization of biomass producing pyrolytic oil, can satisfy for such a requirement This is the technique of applying high heat to organic matter (lignocellulosic materials) in the absence of oxygen or in reduced air, typically in the range of 400–650°C The process can produce charcoal, condensable organic liquids (pyrolytic fuel oil), and non-condensable gasses [8]

Also, it can be adjusted to favor charcoal, pyrolytic oil, or gas depending on process conditions [9] This process is considered as a potential approach for biomass conversion into energy Consequently, understanding pyrolysis kinetics is valuable for the in-depthexploration of process mechanisms However, owing to the very complicated nature of reactions occurring in the process, as well as its unpredictable amount of products, the pyrolysis mechanism has not been carefully understood yet As a consequence, it seems to be impossible for specifying the reaction kinetics if each specific compound is considered

Recently, a general and simple kinetic mechanism named lump model has been employed to investigate the pyrolysis [10, 11] For applying this model, all pyrolytic products should be gathered to form various lumps depending on their phases such as gas, liquid or solid This model is just applied effectively if the content of each lump is clearly determined Specifically, these lumps will include remaining biomass, gas, liquid and char In fact, the liquid product and char contents can be easily measured and calculated based on the weight balance before and after pyrolysis The problem derives from how to determine the content of pyrolytic gas product In addition, it can not confirm that the biomass feedstock will be completely decomposed during the process As a result, the content of remaining biomass after pyrolysis is also needed to verify If the pyrolyzing residence time is long enough, it is believed that the biomass will be completely decomposed As a result, the remaining biomass will approach zero and the content of gas product can be easily obtained through the content of liquid and solid by the formula as follows: gas yield = 100% - liquid yield – char yield However, if the pyrolysis is explored in a short time, the biomass is just partially decomposed

Hence, it seems impossible to calculate the gas yield as mentioned above Also, there has been no method for quantifying the biomass content in a mixture of solid phase From this information, it is necessary to find out a feasible method in order to specify the yield of pyrolytic gas and remaining biomass, through which facilitates for applying the lump model in investigating the kinetic mechanism conventional pyrolysis was for decomposition at a low heating rate and long residence time to produce mainly gas fuel or charcoal The other was fast pyrolysis at which the pyrolysis occurred with a very high heating rate in a short residence time, largely for achieving high yield of bio-oil In the pyrolysis process, the type of reactor is very significant to control the bio-oil yield In fact, there have been some researchers trying to apply pyrolysis conducted in several kinds of reactor for biomass conversion into fuel For instance, Li et al [13] or Wei et al [14] utilized different types of biomass for fast pyrolysis in a free fall reactor Lappas et al [15] researched about fast pyrolysis of biomass for liquid fuel in a fluidized-bed reactor

As can be seen for each type of reactor, the reaction might be different, thus leading to changing reaction pathway as well as product yield

Recently, pyrolysis has been considered as a promising technology, largely due to its economical feasibility as well as its operational simplicity There must be an appropriate set of operating parameters for each specific reactor applied in the pyrolysis process For the fast pyrolysis using fluidized bed reactor, these parameters include pyrolysis temperature, residence time, biomass particle size, and biomass feed rate For pyrolysis in a tubing reactor, some factors affecting on the pyrolysis might be pyrolysis temperature, residence time, and biomass particle size In general, there are many factors needed to be investigated in a pyrolysis process However, there have been few researches studying fully and systematically about the effect of these operating parameters on pyrolysis performance In fact, if there are many factors influencing on the process, the operation and technology cost will increase proportionally to its number of controlling variables Therefore, once all these operating parameters can be explored carefully, it is possible to eliminate the insignificant ones out of controlling process, thus resulting in further simplifying its operation

Biomass has been paid attention for energy production because of its abundance and diversity Among the types, the cellulosic biomass is employed as a feedstock for studies the most Much effort is now placed on enhancing pyrolysis performance by exploring the operating parameters [15-17] There are just few works involving the effect of biomass composition on properties and distribution product Besides, how biomass is pyrolyzed when its major carbohydrate is cellulose-like has not been mentioned so far For the compensation to the above lack, this dissertation aims to characterize the pyrolysis of the cellulose-like process can be selectively controlled to accelerate the biomass conversion In fact, there are various researches focusing on pyrolysis mechanism [10-11] Still, it has not been fully understood yet owing to the difficulties in determining product contents for several reasons First, it is impossible to specify the content of too many compounds in bio-oil Second, there is no feasible method for differentiating between char and remaining biomass after pyrolysis process, thus the char yield, or the biomass conversion can not be calculated In order to avoid the first difficulty as previously mentioned, a lumped kinetic model should be applied to simplify the kinetic calculation However, this model can be used only if all product yields are available Consequently, finding out how to determine the product yields is necessary to facilitate the investigation of pyrolysis mechanism

Finally, the most desirable objective of pyrolysis is to upgrade the bio-oil yield This can be obtained by improving the reactor design to work more stably and reliably, which is another target for this research In addition, in spite of pyrolysis’s widespread application, all of its operating factors have not been systematically studied As a result, the dominant factors on pyrolysis performance have not been defined and hence needed to be carried out in this research

There are two major objectives for this research as follows: one is to understand the characteristics of pyrolysis of palm kernel cake (a non-cellulosic feedstock), and the other is to clarify all the effects on both performance and kinetic mechanism of the pyrolysis occurring in various types of reactors

Palm kernel cake, a non-cellulosic biomass, was applied as a feedstock for all experiments Simultaneously, in some experiments, pine wood chip, a cellulosic biomass, was also pyrolyzed similarly to palm kernel cake From this comparative study, the characteristics of palm kernel cake can be clearly understood

In order to carry out pyrolysis, tubing, tubular and fluidized bed reactors were employed for pyrolyzing biomass For each type of reactor, the properties and distribution of products were characterized carefully In addition, appropriate kinetic models are also proposed and then verified From the results, all the rate constants could be investigated clearly, leading to the ability to upgrade the performance of pyrolysis

The detailed objectives are enumerated as follows:

(1) Optimization of operating parameters in pyrolysis of palm kernel cake using a tubing reactor

(2) Kinetic model study of pyrolyzing palm kernel cake using a tubing reactor and a closed tubular reactor

This dissertation consists of seven chapters, wherein the remainder will discuss the following subjects:

In Chapter 2, all the concepts of biomass and pyrolysis were fully expressed to provide a fundamental background of biomass resources, pyrolysis classification, as well as pyrolysis kinetics Also, the theory of the central composite rotatable design, a design of experiment, was also described carefully This design is a useful method for organizing the experiments, through which equations depicting the effect of all operating parameters can be obtained easily

Chapter 3 showed the studies about the pyrolysis of palm kernel cake using a tubing reactor at closed condition The pyrolysis temperature and residence time were designed for experiment following the central composite rotatable design (CCRD) As a result, the model equations describing all the affecting parameters were obtained, leading to approaching the optimal conditions for pyrolysis process

In addition, a kinetic lumped model was also proposed and used for modeling the pyrolysis process in this chapter A non-linear regression method was then applied for experimental data to calculate all the global kinetic parameters Subsequently, a kinetic analysis was performed for better understanding of pyrolysis characteristics and reaction mechanisms during the pyrolysis of palm kernel cake

In Chapter 4, palm kernel cake (non-cellulosic biomass) and pine wood chip (cellulosic biomass) were used for pyrolysis in an open tubular reactor

Different conditions of pyrolysis temperature and sweeping-gas flow rate were also applied for the pyrolysis of both biomass types From this research, the characteristics of pyrolysis of palm kernel cake (a non-cellulosic biomass) were more clarified

Operating parameters

Recently, pyrolysis has been considered as a promising technology, largely due to its economical feasibility as well as its operational simplicity There must be an appropriate set of operating parameters for each specific reactor applied in the pyrolysis process For the fast pyrolysis using fluidized bed reactor, these parameters include pyrolysis temperature, residence time, biomass particle size, and biomass feed rate For pyrolysis in a tubing reactor, some factors affecting on the pyrolysis might be pyrolysis temperature, residence time, and biomass particle size In general, there are many factors needed to be investigated in a pyrolysis process However, there have been few researches studying fully and systematically about the effect of these operating parameters on pyrolysis performance In fact, if there are many factors influencing on the process, the operation and technology cost will increase proportionally to its number of controlling variables Therefore, once all these operating parameters can be explored carefully, it is possible to eliminate the insignificant ones out of controlling process, thus resulting in further simplifying its operation.

Motivation

Biomass has been paid attention for energy production because of its abundance and diversity Among the types, the cellulosic biomass is employed as a feedstock for studies the most Much effort is now placed on enhancing pyrolysis performance by exploring the operating parameters [15-17] There are just few works involving the effect of biomass composition on properties and distribution product Besides, how biomass is pyrolyzed when its major carbohydrate is cellulose-like has not been mentioned so far For the compensation to the above lack, this dissertation aims to characterize the pyrolysis of the cellulose-like process can be selectively controlled to accelerate the biomass conversion In fact, there are various researches focusing on pyrolysis mechanism [10-11] Still, it has not been fully understood yet owing to the difficulties in determining product contents for several reasons First, it is impossible to specify the content of too many compounds in bio-oil Second, there is no feasible method for differentiating between char and remaining biomass after pyrolysis process, thus the char yield, or the biomass conversion can not be calculated In order to avoid the first difficulty as previously mentioned, a lumped kinetic model should be applied to simplify the kinetic calculation However, this model can be used only if all product yields are available Consequently, finding out how to determine the product yields is necessary to facilitate the investigation of pyrolysis mechanism

Finally, the most desirable objective of pyrolysis is to upgrade the bio-oil yield This can be obtained by improving the reactor design to work more stably and reliably, which is another target for this research In addition, in spite of pyrolysis’s widespread application, all of its operating factors have not been systematically studied As a result, the dominant factors on pyrolysis performance have not been defined and hence needed to be carried out in this research.

Research objectives

There are two major objectives for this research as follows: one is to understand the characteristics of pyrolysis of palm kernel cake (a non-cellulosic feedstock), and the other is to clarify all the effects on both performance and kinetic mechanism of the pyrolysis occurring in various types of reactors

Palm kernel cake, a non-cellulosic biomass, was applied as a feedstock for all experiments Simultaneously, in some experiments, pine wood chip, a cellulosic biomass, was also pyrolyzed similarly to palm kernel cake From this comparative study, the characteristics of palm kernel cake can be clearly understood

In order to carry out pyrolysis, tubing, tubular and fluidized bed reactors were employed for pyrolyzing biomass For each type of reactor, the properties and distribution of products were characterized carefully In addition, appropriate kinetic models are also proposed and then verified From the results, all the rate constants could be investigated clearly, leading to the ability to upgrade the performance of pyrolysis

The detailed objectives are enumerated as follows:

(1) Optimization of operating parameters in pyrolysis of palm kernel cake using a tubing reactor

(2) Kinetic model study of pyrolyzing palm kernel cake using a tubing reactor and a closed tubular reactor.

Dissertation overview

This dissertation consists of seven chapters, wherein the remainder will discuss the following subjects:

In Chapter 2, all the concepts of biomass and pyrolysis were fully expressed to provide a fundamental background of biomass resources, pyrolysis classification, as well as pyrolysis kinetics Also, the theory of the central composite rotatable design, a design of experiment, was also described carefully This design is a useful method for organizing the experiments, through which equations depicting the effect of all operating parameters can be obtained easily

Chapter 3 showed the studies about the pyrolysis of palm kernel cake using a tubing reactor at closed condition The pyrolysis temperature and residence time were designed for experiment following the central composite rotatable design (CCRD) As a result, the model equations describing all the affecting parameters were obtained, leading to approaching the optimal conditions for pyrolysis process

In addition, a kinetic lumped model was also proposed and used for modeling the pyrolysis process in this chapter A non-linear regression method was then applied for experimental data to calculate all the global kinetic parameters Subsequently, a kinetic analysis was performed for better understanding of pyrolysis characteristics and reaction mechanisms during the pyrolysis of palm kernel cake

In Chapter 4, palm kernel cake (non-cellulosic biomass) and pine wood chip (cellulosic biomass) were used for pyrolysis in an open tubular reactor

Different conditions of pyrolysis temperature and sweeping-gas flow rate were also applied for the pyrolysis of both biomass types From this research, the characteristics of pyrolysis of palm kernel cake (a non-cellulosic biomass) were more clarified

Chapter 5 presented an investigation about the kinetic model of palm kernel cake pyrolysis using a closed tubular reactor A new method for determining the pyrolytic gas content and an innovative procedure for a non-linear regression were described meticulously In this research, the pyrolysis process was modeled based on the kinetic model proposed by Liden [10] From the obtained kinetic constants, the favorable and unfavorable pathway of pyrolysis can be clarified

In Chapter 6, kinetics study in an open condition using thermogravimetric analysis (TGA) was also carried out From TGA data, the activation energy of biomass decomposition can be obtained In order to explore the fast pyrolysis of palm kernel cake, a new improved fluidized bed reactor was designed and then applied The central composite rotatable design was also employed to set up all experiments and to investigate the effects of operating factors on pyrolysis performance such as pyrolysis temperature, residence time, biomass particle size

References

[1] http://www.earthenergy.ws/renewableenergy.htm [2] D.L Klass Biomass for renewable energy, fuels, and chemicals, Academic

[3] D.J Cuff, and W.J Young, U.S Energy atlas, Free Press/McMillan Publishing Co NY, 1980

[4]http://energy.converanet.com/cvn03/cachedhtml?hl=keywords&kw=s%3Acvc%

5C.112ZE&cacheid=ds1- va:p:1001t:8099407855520:7bd79b24c1104e5e:4bdde224&scopeidLink [5] http://www1.eere.energy.gov/biomass/feedstock_glossary.html

[6] http://www.jphpk.gov.my/Agronomi/PKC.htm [7] C.A.C Sequeira, P.S.D Brito, A.F Mota, J.L Carvalho, L.F.F.T.T.G

Rodrigues, D.M.F Santos, D.B Barrio, and D.M Justo, Energy Conversion and

[8] A Demirbas, and D Gullu, Energy Education Science and Technology, 1998, 1, 111–1115

[9] A Demirbas, Energy Education Science and Technology, 1998, 2, 23–28

[10] A.G Liden, F Berruti, and D.S Scott, Chemical Engineering Communications, 1988, 65, 207-221

[11] Y.H Park, J Kim, S.S Kim, and Y.K Park, Bioresource Technology, 2009,

[12] A.V Bridgwater, and G.V.C Peacocke, Renewable and Sustainable Energy Reviews, 2000, 4, 1-73

[13] S Li, S Xu, S Liu, C Yang, and Q Lu, Fuel Processing Technology, 2004, 85, 1201-1211

[14] L Wei, S Xu, L Zhang, H Zhang, C Liu, H Zhu, and S Liu, Fuel Processing Technology, 2006, 87, 863–871

[15] A.A Lappas, M.C Samolada, D.K Iatridis, S.S Voutetakis, and I.A Vasalos,

[16] G Chen, J Andries, Z Luo, and H Spliethoff, Energy Conversion and Management, 2003, 44, 1875–1884

[17] A Demirbas, Journal of Analytical and Applied Pyrolysis, 2004, 71, 803–815.

Literature Review 2.1 Concept of biomass

Biomass definition and classification

Biomass is biological material derived from living, or recently living organisms In the context of biomass for energy this is often used to mean plant based material, but biomass can equally apply to both animal and vegetable derived material [1] Biomass is available in a variety of forms and is generally classified according to its source (animal or plant) or according to its phase (solid, liquid or gaseous).

Chemical composition

Biomass is carbon-based substance which consists of a mixture of organic molecules containing hydrogen, usually including atoms of oxygen, often nitrogen and also small quantities of other atoms, including alkali, alkaline earth and heavy metals As for the plant-derived biomass, it can be classified as wood-based biomass and seed-based biomass Basically, the plant-derived biomass contains carbohydrate, lignin, minerals, and remaining resin (wood)/vegetable oil (seed)

For wood-based biomass, the carbohydrate mainly includes hemicellulose and cellulose This wood-based biomass can be named as cellulosic material A typical composition of a wood-based biomass can be enumerated as following:

• Hemicellulose (20–40% of total feedstock dry matter) is a short, highly branched polymer of five-carbon (C5) and six-carbon (C6) sugars

Specifically, hemicellulose contains xylose and arabinose (C5 sugars) and galactose, glucose, and mannose (C6 sugars) The structure of hemicellulose is shown in Figure 2.1

• Cellulose (30–50% of total feedstock dry matter) is a glucose polymer linked by ò–1,4 glycosidic bonds The basic building block of this linear polymer is cellubiose, a glucose-glucose dimmer The structure of cellulose is shown in Figure 2.2

• Lignin (15–25% of total feedstock dry matter), a polyphenolic structural constituent of plants, is the largest non-carbohydrate fraction of lignocellulose Unlike cellulose and hemicellulose, lignin can not be utilized in fermentation processes The structure of lignin is shown in Figure 2.3

• Other compounds present in plant-derived biomass are known as extractives

These include resins, fats and fatty acids, phenolics, phytosterols, salts, minerals, and other compounds

For seed-based biomass, the carbohydrate largely consists of hemicellulose, and cellulose-like compound This biomass is classified as non-cellulosic material

For instance, the cellulose-like compound existing in palm kernel cake is mannan, with the structure shown in Figure 2-4

Biomass resource

Biomass can be divided into several biomass categories as following: (1) Pulp and paper industry residues, (2) Forest residues, (3) Agricultural or crop residues, (4) Urban wood wastes, and (5) Energy crops Each of these biomass categories comprises different types of biomass, the main ones being products (harvested biomass) and residues (by-products from cultivation, harvesting and processing)

2.1.3.1 Pulp and paper industry residues

The largest source for energy production from wood is the waste from the pulp and paper industry called black liquor [1, 2, 3] Black liquor is generated in the kraft process Usually, it consists of lignin and pulping chemicals used to separate lignin from the cellulosic fraction of wood Wood processing produces sawdust and a collection of bark, branches and leaves/needles

The forest products industry generates large amounts of residual biomass as timber is harvested and manufactured into marketable goods such as lumber and keep transport costs high, and so it is economical to reduce the biomass density in the forest itself

Agriculture crop residues include corn stover (stalks and leaves), wheat straw, rice straw, nut hulls etc Corn stover is a major source for bioenergy applications due to the huge areas dedicated to corn cultivation worldwide

Such waste consists of lawn and tree trimmings, whole tree trunks, wood pallets and any other construction and demolition wastes made from lumber The rejected woody material can be collected after a construction or demolition project and turned into mulch, compost or used to fuel bioenergy plants

Dedicated energy crops are another source of woody biomass for energy

These crops are fast-growing plants, trees or other herbaceous biomass which are harvested specifically for energy production Rapidly-growing, pest-tolerant, site and soil-specific crops have been identified by making use of bioengineering

Herbaceous energy crops are harvested annually after taking two to three years to reach full productivity These include grasses such as switchgrass, elephant grass, bamboo, sweet sorghum, wheatgrass etc

Short rotation woody crops are fast growing hardwood trees harvested within five to eight years after planting These include poplar, willow, silver maple, cottonwood, green ash, black walnut, sweetgum, and sycamore

Industrial crops are grown to produce specific industrial chemicals or materials, e.g kenaf and straws for fiber, and castor for ricinoleic acid Agricultural crops include cornstarch and corn oil, soybean oil and meal, wheat starch, other vegetable oils etc Aquatic resources such as algae, giant kelp, seaweed, and microflora also contribute to bioenergy feedstock.

Biomass applications

An ideal renewable resource is one that can be replenished over a relatively short timescale or is essentially limitless in supply Resources such as coal, natural gas and crude oil come from carbon dioxide ‘fixed’ by nature through photosynthesis many millions of years ago They are of limited supply, can not be source of energy, biomass can be used to produce not only energy, but also chemicals and materials [4, 5] The application of biomass was shown in Figure 2.5

The basic concept then of biomass as a renewable energy resource comprises the capture of solar energy and carbon from ambient CO2 in growing biomass, which is converted to other fuels (biofuels, synfuels) or is used directly as a source of thermal energy or hydrogen One cycle is completed when the biomass or derived fuel is combusted This is equivalent to releasing the captured solar energy and returning the carbon fixed during photosynthesis to the atmosphere as CO2 Hydrocarbons identical to those in petroleum or natural gas can be manufactured from biomass feedstocks This means that essentially all of the products manufactured from petroleum and natural gas can be produced from biomass feedstock Alternatively, biomass feedstock can be converted to organic fuels that are not found in petroleum or natural gas

Figure 2.5 Some typical applications of biomass [4, 5]

Approaches for biomass conversion into energy

In recent year, application of renewable material, especially biomass, for producing energy attracted a great attention to scientists as well as governments [3, 6] Some common approaches were presented in Figure 2.6 Biomass, a natural resource mostly including hemicellulose, cellulose and lignin, was a potential material not only due to its abundance but also its high content of hydrocarbon

Many types of biomass were used as a subject for researches, such as wood chip, rice husk, straw, sawdust, etc However, depending on the composition of cellulose and lignin, there might be the most appropriate method for converting each biomass into fuel There were some popular methods applying for the biomass with the high content of cellulose For instance, fermentation by enzyme or hydrolysis with support of acid or alkaline was applied widely as presented in literatures [7, 8]

As for the high lignin biomass, due to the firm structure of this component, until now, it seems that there was only the thermal decomposition also known as pyrolysis being most effective and most simple This method can be also applied for the biomass with a broad distribution of various compositions Moreover, owing to its reasonable cost and simple operation, pyrolysis has been more accepted as a feasible approach for converting biomass into energy as well as chemical

Figure 2.6 Energy products and classification [3]

Concept of pyrolysis

By definition pyrolysis of a compound containing carbon is incomplete thermal degradation in the absence of oxygen, which results in liquid and gaseous products, and also char [9] Figure 2.7 shows the definition of pyrolysis The liquid product from biomass pyrolysis is known as bio-oil or pyrolysis oil (there are several other names available such as biofuel oil, wood liquid, wood oil, etc) Bio- oils are a mixture of different molecules (alcohols, aldehydes, ketones, esters and phenolic compounds) derived from the fragmentation of lignin, cellulose, hemicellulose and extractives [10] It is much easier to handle and to transport bio- oil than solid biofuel However bio-oil can change during storage if the products in the condensate (bio-oil) have not reached thermodynamic equilibrium during pyrolysis The moisture content of bio-oil is about 15–30 wt% of the original moisture of the feedstock is used in the pyrolysis process [10]

CONDENSATION Catalytic conversion to hydrogen (Optional)

Power Generation or chemical Separation

Pyrolysis of biomass is divided into slow pyrolysis, which is well known to produce charcoal, fast pyrolysis, which produces a high yield of liquid biofuels and other chemicals [11] and flash pyrolysis

• Slow pyrolysis (or carbonisation) requires low temperatures and very long residence time In the carbonisation process the amount of char is maximized

• Fast pyrolysis of biomass occurs usually at 500–700°C and high heating rates (e.g 300°C min -1 ) over a short time Recently, biomass has been converted to bio-oil and then to hydrogen by catalytic steam reforming, however the yield is relatively low Hydrogen can be produced by either gasification of biomass followed by reforming of the syngas or fast pyrolysis followed by reforming (rearrangement) of the carbohydrate fraction of the bio-oil

• Flash pyrolysis is the process in which the heating rate is very high and the reaction time is of only a few seconds Therefore, the particle size of biomass should be fairly small (105–250 μm) for this process [12] Often flash pyrolysis and fast pyrolysis are mentioned as one and the same process in the literature

The gaseous products of fast pyrolysis require rapid cooling or quenching to minimize secondary reactions of the intermediate products (radical components)

These radicals are very reactive and can undergo secondary reactions such as cracking and carbon deposition

There are several parameters that have an effect on the yield and the composition of the volatile fraction of biomass during pyrolysis: the biomass species, the chemical and structural composition of the biomass, the temperature, the particle size, etc [13] Liquid production with high yield from biomass by fast pyrolysis is a promising technique for the replacement of fossil-fuel precursors for different chemicals and fuels with sustainable and renewable energy sources [14]

As biomass is heated, its various components become chemically unstable and thermally degrade or vaporize A number of studies have shown that the main components of most biomass, for example cellulose, hemicellulose and lignin, are chemically active at temperatures as low as 150°C This has recently been indicated by the kinetic parameters determined by Bilbao et al [15-19] Wood is claimed to begin pyrolyzing at 250°C A review of the possible reaction pathways and

The component of wood which has received the most attention is cellulose

Cellulose occurs in most biomass types up to 50 wt% and has a well defined structure which allows its easy purification and separation This has been carried out at two different temperature ranges: up to 300°C and above 300°C A reaction pathway for the pyrolysis of cellulose has been proposed by Shafizadeh [20], Antal [21] and Kilzer et al [22] as shown in Figure 2.8

It is generally considered that primary pyrolysis of pure cellulose occurs by two competing pathways: one involving dehydration and the formation of char, CO2 and water and the second involves fragmentation and depolymerisation resulting in the formation of tarry products consisting mainly of levoglucosan as shown in Figure 2.8 [20-22] At temperatures greater than 300°C, fragmentation or transglycosylation predominates which involve the conversion of cellulose into predominantly a liquid product consisting of levoglucosan and other anhydrosugars b Hemicellulose pyrolysis

Other work has been carried out on the pyrolysis of hemicellulose but this has received less attention due to its lower abundance, variety of constituents, high reactivity and rapid degradation at low temperatures (150-350°C) It is believed that the intermediate levoglucosan is replaced by a furan derivative [23] c Lignin pyrolysis

The complex structure of lignin has led to a lack of understanding of the pyrolysis of this component Lignin is the most thermally stable component but its structure varies according to its source and the method of isolation To date therefore, most detailed work in lignin pyrolysis has been obtained from model compounds Minor decomposition begins at 250°C but most significant lignin pyrolysis occurs at higher temperatures [24]

Low temperature pyrolysis of lignin (< 600°C) has been carried out by a large number of researchers [23, 24] Detailed work using Kraft lignin has also been carried out by Jegers and Klein [25, 26] who identified and quantified 33 products (12 gases, water, methanol, and 19 aromatic compounds such as phenol, cresol and guaiacol) at a range of temperatures from 300 to 500°C Latridis and Gavalas [27] studied the pyrolysis of Kraft lignin at 400 - 700°C using a captive sample reactor, obtaining a total volatiles yield of 60 wt% Nunn et al [28] have also carried out work in this area obtaining a maximum of 53 wt% liquid at 625°C, again in a captive sample reactor

High temperature pyrolysis of lignin (> 600°C) leads to complex cracking, dehydrogenation, condensation, polymerization and cyclization reactions resulting

2.2.3.2 Some typical kinetic models a Global kinetic model

The kinetics of wood degradation and its respective components are obtained by measuring the rate of weight loss of the sample as a function of time and temperature The most common technique for this investigation is thermogravimetric analysis (TGA) [15-19] The global thermal degradation process can be described by a simplistic reaction scheme as shown in Figure 2.9 b Secondary pyrolysis kinetics

Secondary pyrolysis kinetics has been studied by number of researchers to account for the conversion of primary liquids to secondary products such as char, tar and gases Some kinetic models were shown in Figure 2.10 and Figure 2.11

Figure 2.8 Pure cellulose pyrolysis pathways [20-22]: (1): primary pyrolysis; (2): secondary pyrolysis

Figure 2.10 Reaction scheme used by Liden [29], and Diebold [30]

Tar (primarily levoglucosan) Water, char, CO, CO 2

Theory of central composite rotatable design (CCRD)

Experimental design refers to the process of planning, designing and analyzing the experiment so that valid and objective conclusions can be drawn efficiently [32, 33] In practice, the second-order design named central composite rotatable design (CCRD) is mostly employed, especially in chemical engineering [34]

2.3.1 Set up experiment using matrix design

From this design, a matrix of coded variables (X) is initially set up to plan experiment The number of row in this matrix or the total experiment run (n) depended on the number of factor (k) by the expression: n = 2 k + 2k + no Where no is the number of replicated experiment at center point

That means, the number of experiment run includes 2 k experiment at core points (X = ±1), 2k experiment at axial point (X = ± α = ± 2 k/4 ) and n o experiment at center points (X = 0)

Corresponding to each variable set in this matrix, the response value from experiment can be obtained Subsequently, the experimental data were fitted to a polynomial mathematical model of second order which was shown as below:

Xi, Xj is the coded variable from the actual variable xi, and xj Y represents the response value from experiment βo is the value of fitted response at the center point of design; βi, βii, βij are the linear, quadratic and interaction terms, respectively The coded variable was achieved from the actual variable based on the following expressions [35]: o i i i i x x

Where: o x i is the midpoint value of the actual variable x i ( max min

∆xi is the interval value of the actual variable xi ( max min

∆ = ) max min i , i x x is the high and low level of actual variable respectively

All the regression coefficients (βo, βi, βii, βij) were calculated as follows:

Where: a1, a2, a3, a4, a5, a6 and a7 are the constants determined from literature [34]

2.3.3 Verifying the statistical significance of each regression coefficient

The statistical significance of regression coefficients was also estimated A regression coefficient is statistically significant if its absolute value is higher than the confidence interval

2.3.4 Verifying the lack of fit of regression equation

The hypothesis on lack of fit of a regression model is checked by Fisher’s criterion:

Where, S 2 AD is residual variance or lack of fit variance

Residual variance can be calculated by the formula as follows:

Reproducibility can be calculated by the formula below:

Basing on the obtaining value F R , the obtained regression model was then checked for lack of fit

If FR < FT (FT: tabular value of Fisher-criterion), it can be considered that this regression equation is adequate

If FR > FT, there will be a lack of fit (inadequacy) of mathematical model

References

[1] J Twidell, and T Weir Renewable energy resources 2 nd Ed., Taylor and Francis, 2006

[2] D.L Klass Biomass for renewable energy, fuels, and chemicals, Academic

[3] J.H Clark, and F.E.I Deswarte Introduction to chemicals from biomass, John

[4] F.A Farret, and M.G Simoes Integration of alternative sources of energy,

[5] F.R Calle, S.V Bajay, and H Rothman Industrial uses of biomass energy,

[6] A Bridgwater Thermal biomass conversion and utilization - biomass information system, European Commission, 1996

[7] Y Lin, and S Tanaka, Applied microbiology and biotechnology, 2006, 69, 627- 642

[8] Y Sun, and J Cheng, Bioresource Technology, 2002, 83, 1-11

[9] S Czernik, and A.V Bridgwater, Energy Fuels, 2004, 18, 2, 590–598

[10] S Yaman, Energy Conversion and Management, 2004, 45, 651–671

[11] A.V Bridgwater, and G.V.C Peacocke, Renewable and Sustainable Energy Reviews, 2000, 4, 1–73

[12] H.B Goyal, D Seal, and R.C Saxena, Renewable and Sustainable Energy Reviews, 2008, 12, 504–517

[14] Z Qi, C Jie, W Tiejun, and X Ying, Energy Conversion Management, 2007, 48, 87–92

[15] R Bilbao, J Arauzo and A Millera, Thermochimica Acta,1987, 20, 121-131

[16] R Bilbao, J Arauzo and A Millera, Thermochimica Acta, 1987, 120, 133-141

[17] R Bilbao, A Millera and J Arauzo, Thermochimica Acta, 1989, 143, 137-148

[18] R Bilbao, A Millera and J Arauzo, Thermochimica Acta, 1989, 143, 149-159

[19] R Bilbao, A Millera and J Arauzo, Thermochimica Acta, 1990, 165, 103-112

[21] Jr MJ Antal, Advances in solar energy, 1983, 1, Plenum Press, P 61

[23] R.J Evans and T.A Milne, Energy and Fuels, 1987, 1, P 123

[24] Jr M.J Antal, Advances in solar energy, 1985, 2, Plenum Press, P 175

[25] H.E Jegers, and M.T Klein, Industrial and Engineering Chemistry Process

[28] T.R Nunn, J.B Howard, J.P Longwell and W.A Peters, Industrial and Engineering Chemistry Product Research and Development, 1985, 24, P 844

[29] A.G Liden, A kinetic and heat transfer modelling study of wood pyrolysis in a fluidized bed MASc Thesis, University of Waterloo, Canada, 1985

[30] G Varhegyi, M.J Antal, T Szekely and P Szabo, Energy and Fuels, 1989, 3, 329-335

[31] C.W Gorton and J.A Knight, Biotech and Bioeng Symp., 1984, 14, 14-20

[32] G.E.P Box, and K.B Wilson, Journal of the Royal Statistical Society, 1951, B13 (1), 1-38

[33] D.C Montgomery Design and analysis of experiments, 5 th Edition, John Wiley & Sons Inc, 2001

[34] Z.R Lazíc, Design of experiments in chemical engineering, Wiley – VCH,

[35] N.R Draper, and H Smith Applied regression analysis, 3 rd Edition, John Wiley & Sons Inc, 1998

Effect of Operating Parameters and Kinetics of Pyrolysis in a Tubing Reactor

Palm oil is now produced with the highest capacity all over the world Just Malaysia and Indonesia, the world’s two largest producer and exporter of palm oil, produced about 32 million tons in 2006, amounting to 87% of world production [1]

The palm oil production brought a great economic benefit for the nations but also resulted in many environmental problems owing to its waste biomass Averagely, every ton of palm oil produced generates tons of palm waste biomass, such as empty fruit bunch, fiber, and shell [2] As a result, great focus is now placed on how to employ this abundant material as a renewable energy

Recently, there are various techniques applied for converting this waste biomass into fuel For instance, Yang et al [3] utilized palm waste for pyrolysis in very high temperature to obtain hydrogen-rich gas Mae et al [4] researched on how to promote the gasification of palm shell waste Rozainee et al [5] just used

8] Therefore, if pyrolysis process is not controlled carefully, biomass might not be completely decomposed in such a short residence time Consequently, it is very complicated to differentiate between the char and remained biomass after pyrolysis, thus causing difficulty in determining product yields, especially char yield In a closed reactor, the residence time can be determined accurately based on the time which reactor was heated in furnace and prolonged until biomass being totally degraded As a result, the product yields would be conveniently specified, leading to facilitating pyrolysis kinetic investigation This is the reason why a closed tubing reactor was used for pyrolysis in this study

The objectives of this study are to investigate the effects of operating factors on pyrolysis performance and calculate all rate constants in a closed tubing reactor In order to determine the parameters as well as the optimal operating conditions (if available) for pyrolysis reaction, a method known as central composite rotatable design was applied for organizing the experiment In addition, a lumped kinetic model was proposed for studying pyrolysis kinetics and then employed for calculating the rate constants

The raw palm kernel cake received from Enertech Co (Malaysia) was ground with a knife mill and then exposed in air for 24 h The prepared sample was then sieved to obtain the desirable size, depending on each experiment The moisture and ash contents of the biomass were determined using ASTM (E1756-01, 2001 and E1755-01, 2001) [9] The elemental analysis for C, H, O, N and S were carried out using the automatic elemental analyzer (EA, Flash EA 1112, CE

Instrument) Thermogravimetric analysis of biomass was carried out using TGA (Q50, TA Instrument) with a heating program from 30°C to 800°C at the ramp rate of 20°C/min

3.2.2 Pyrolysis of palm kernel cake

Figure 3.1 shows a schematic diagram of experimental apparatus In all experiments, 5 g of sample was weighed and then fed into the tubing reactor (∅22mm × 110mm) A molten salt bath with the eutectic mixture of KNO3 (59 wt%) and Ca(NO3)2 (41 wt%) was used to heat the tightly-closed reactor following the temperature and residence time as designed in the operational matrix After each run, the tubing reactor was taken out and then cooled to room temperature In order to explore the effect of operating parameters on pyrolysis performance, residence time varied in the range from 4 to 12 min and temperature from 360 -

The reaction products were analyzed by weighing the products of gas, oil and solid After cooling, the tubing reactor was opened to allow gas to release gently Gas yield is defined as (gas weight)/(feed weight) × 100 The remaining solid was filtered with acetone The solid yield was defined as (weight of solid)/(feed weight) × 100, while oil yield is defined as (100 - gas yield - solid yield)

The gas product of pyrolyzing biomass was quantitatively characterized by two GCs which have FID and TCD detectors, respectively The GC-FID (M600D–

Younglin) with HP-Plot/Al2O3 column (50mì0.53mmì15àm) was used to analyze hydrocarbons Meanwhile, GC-TCD (ACME 6000GC–Younglin) with Molecularsieve 5A and Porapak N column (80/100, 6ft×1/8in, HP) was used to detect CO, H2 and CO2 In order to quantify the composition of C1–C6 hydrocarbons in the gas product, the peak areas of these components were calibrated using a standard gas mixture (Supelco–Scotty 14)

The liquid product was dissolved in chloroform solvent and then analyzed by GC-MS (HP-5MS 30mì0.25 mmì0.25 àm) to identify the compositions [10]

The solid residue after pyrolysis reaction was characterized by elemental analysis to determine the contents of C, H, O, and N in the remaining biomass

Figure 3.1 Schematic diagram of experimental apparatus

Results and discussion

Table 3.1 shows the characteristics of palm kernel cake biomass The higher heating value in this table was calculated with the following formula [11]:

HHV (MJ/kg) = {33.5[C]+142.3[H]-15.4[O]-14.5[N]} × 10 -2 (1) where [C], [H], [O], and [N] were the contents (wt%) of carbon, hydrogen, oxygen and nitrogen, respectively

The moisture content of palm kernel cake was 2.0% while the ash content was 5.85% The carbon (C) and oxygen (O) contents of this biomass were 50.1% and 38.2%, respectively The oil extracted from palm kernel cake was also characterized by GC–FID to use as a reference The result shows that this extracted oil was just palm oil with the amount up to 10 wt% of dry basis biomass As shown in Table 3.1, the ash content of biomass was rather low as opposed to carbon content Additionally, in comparison with the higher heating value (HHV) of other conventional fuel such as petroleum (43 MJ/kg), LPG (45.75 MJ/kg), or kerosene (41 MJ/kg) [12], that of palm kernel cake was approximately half of these HHV values Consequently, it would be unfeasible to use this biomass as a source for energy unless the pyrolysis technique was applied

Table 3.1 - Sample characteristic of palm kernel cake

3.3.2 Thermal decomposition analysis of biomass

The conversion (X) and differential rate of conversion (dX/dt) as a function of temperature, which describes TG and DTG graphs respectively, are shown in Figure 3.2 As can be seen from the DTG graph of raw palm kernel cake, there are two distinctive peaks (one peak at 295 o C and the other at 335 o C) and one broad peak (from 300 - 620°C)

As reported by Dusterhoft et al [13], palm kernel cake largely consisted of mannan (a cellulose-like material), cellulose, palm oil, lignin, mineral and small amount of unknown compounds, of which mannan, palm oil and lignin made up the majority Owing to such a composition, palm kernel cake was classified as a non-cellulosic biomass and obviously its data of thermogravimetric analysis would be different with that from cellulosic biomass However, there has no report about the decomposition of mannan using TG analysis published Raveendran et al [14] and Demirbas et al [15] reported that cellulosic biomass commonly consisted of three major components with the decomposition order as follows: hemicellulose (at around 300°C), lignin (at 250 – 550°C), and cellulose (at 300 – 430°C) In comparison with the profile of lignin decomposition in cellulosic biomass as mentioned in literatures [14, 15], it can be inferred that the broad peak locating from 300 - 620°C belonged to the decomposition of lignin

For the second sharp peak at 335°C, when the biomass was washed by acetone to eliminate palm oil remaining in biomass, this second peak disappeared as shown on the DTG graph of the washed biomass Therefore, it was supposed that this second peak involves the degradation of palm oil existing in palm kernel cake In addition, on the TG graph of the raw biomass, the slope of curve abruptly changed in the range at which the second peak in DTG graph appeared The difference of conversion in this temperature range was approximately 10 wt% This result was consistent with the palm oil content (10.7 wt%) determined as mentioned above, strongly confirming the existence of oil decomposition peak on DTG graph Finally, it can be seen that the second sharp peak (at 335°C) and broad peak (300 – 620°C) involve the two major compounds of palm oil and lignin respectively; therefore the first sharp peak at 295°C must be related to the decomposition of the remaining one - mannan

Based on the TG and DTG data as aforementioned, in order to obtain a relative high conversion (around 60 – 70 wt%) in which the experiment would be convenient to characterize the product yield, the reaction temperature for palm kernel cake pyrolysis in tubing reactor should be selected at the temperature higher than 360°C On the other hand, the upper limit of temperature in this eutectic salt solution just approached 480°C and therefore the pyrolysis temperature must be

DTG, dX/dt conversion, X (%) temperature, o C 1

1- biomass washed by acetone 2- biomass

Figure 3.2 TGA and DTG curve for pyrolysis of palm kernel cake using TG method.

3.3.3 Optimization of pyrolysis conditions by CCRD

In pyrolysis experiment using tubing reactor, residence time and pyrolysis temperature are two dominant factors affecting the reaction performance Therefore, the general form for the regression equation can be shown as follows:

Y=β +β +β +β +β +β (2) where Y is the product yield (gas, liquid, and solid yield) and X 1 , X 2 are the coded values corresponding to residence time (t) and pyrolysis temperature (T), respectively The relationship between coded values and actual values can be drawn from the formula as shown below: o i i i i x x

Where: o x i is the midpoint value of the actual variable xi ( max min

∆x i is the interval value of the actual variable x i ( max min

∆ = ) max min i , i x x is the high and low level of actual variable respectively

Regression equation for gas yield:

(4) Regression equation for liquid yield:

Y= −29.68 0.228 t+ × +0.258 T 2.52 10× − × − ×T (5) Regression equation for solid yield:

(6) It can be seen from Figure 3.3, 4 and 5, the residence time and pyrolysis temperature were directly proportional to liquid and gas yield while inversely proportional to solid yield As known, the desirable objective of biomass pyrolysis is to obtain as high bio-oil yield as possible, whereas to limit the solid yield

Consequently, the optimizing objectives for equation (5) and (6) are to achieve the highest liquid yield and lowest solid yield respectively Gaseous product is also an expected product in pyrolysis; thus the highest gas yield is also required for equation (4) In order to approach the global optimum of product yield (if available), a trial and error algorithm was applied for each equation (4) – (6) A repeatable procedure using Matlab software was set up for this algorithm The results show that the yield of gas and liquid product reached the highest value of 17.2 and 38.9 wt% respectively at the condition of 12 min 460°C, while solid yield obtained the lowest one (43.9 wt%)

Table 3.2 Central composite rotatable design 2 2 + 2x2 + 5

Design matrix Operational matrix Response X 1 X 2 x 1 (time) x 2 (temp) Gas Liquid Solid

G as y iel d , % ti m e, m in tempe rature , o C

Figure 3.3 Gas yield in the variation of experiment

L iq u id y ie ld , % ti m e, m in tempe rature , o C

Figure 3.4 Liquid yield in the variation of experiment

So li d y ie ld , % time, min te m pe ra ture o , C

Figure 3.5 Solid yield in the variation of experiment

In order to investigate the pyrolysis in tubing reactor, it was necessary to select a condition for reaction through which the effect of residence time and pyrolysis temperature on the product characteristic could be clarified The optimal conditions were the most appropriate choice for such purpose Therefore, in our study, the condition of 12 min and 460°C was used for experiment to obtain gas, liquid and solid product

The composition of gas product was analyzed using both GC-TCD and GC-FID The result of GC-TCD analysis shows that the content of CO and CO 2 was 4.9 and 20 wt% respectively Meanwhile, the composition of hydrocarbons determined by GC-FID comprised a major content, mainly including CH 4 (23.1wt%), C 2 H 6 (15.8 wt%), C3H8 (14.0 wt%) and C4H10 (5.3 wt%) The GC-FID chromatogram was demonstrated in Figure 3.6 As compared with the results obtained in experiment at open condition reported in literature [3], the content of C 1 – C 4 hydrocarbons in this study were higher, whereas that of CO and CO2 was much lower Therefore, it reveals that closed pyrolyzing condition was preferable to produce more hydrocarbon product than the open one time, min

Figure 3.6 The chromatogram of GC – FID for gas product

Table 3.3 shows the composition analysis data using Elemental analysis method In comparison with analysis data of palm kernel cake in Table 3.1, those of solid product show that the carbon content increased 17.56%, whereas oxygen content decreased 25.42% This might be due to the decomposition of biomass leading to cracking C-O bonds, releasing oxygen atoms and finally forming carbon-rich solid product As a result, the higher heating value of solid product was upgraded to 26 MJ/kg as calculated by the equation (1), compared to just 20.37 MJ/kg in raw biomass

Table 3.4 shows the results obtained from GC-MS analysis of bio-oil

According to shown data, the detectable components of bio-oil were ketones, alkanes, phenolics, aromatics and derivatives of furan Of all products, three major components were 3- penten-2-one, 2-pentanone, 4-hydroxy-4-methyl, and 1- hydroxy, 1-phenyl, propanon-2 with the contents of 14.72, 20.57 and 26.67 % respectively Meanwhile, in the extent of literatures referenced [10, 16-18], the pyrolytic oil at the reaction condition of normal pressure was basically composed of alcohols, aldehydes, ketones, organic acids, esters, levoglucosan, and phenolics

Depending on the pyrolysis temperature, these compounds can change in the content, but seemed to remain the composition regardless of the source of used biomass

Obviously, there were some differences in product distribution between this study and that of other researchers First, there are a considerable amount of alkanes (9%) detected in pyrolytic oil, while some compounds such as esters, aldehydes, organic acids was absent As reported by Ngo et al [19], the vegetable oil was decomposed under thermal condition to produce alkanes, alkenes, alkadiens, or aromatics Therefore, it can be concluded that the presence of alkanes derived from the pyrolysis of palm oil remaining in raw biomass (10.7 wt%) As for the other dissimilarities in bio-oil composition, it can be attributed to changing in pyrolysis kinetics as well as mechanism caused by the high pressure in tubing reactor In general, a factor that increases the number of collisions between particles will increase the reaction rate and a factor that decreases the number of collisions between particles will decrease the chemical reaction rate [20] Based on this theory, it can be seen that the pyrolysis occurred in a closed reactor would facilitate the collision of particles more than in open reactor, causing changing pyrolysis mechanism as well as product distribution Finally, it should be noted that

Table 3.3 Composition of solid product by Elemental analysis

Table 3.4 GC – MS analysis of bio-oil from pyrolysis of palm kernel cake at 460 o C in 12 min

2, 5 - dimethyl Furan 1.94 1-hydroxy, 1-phenyl, propanon-2 26.67

3-pentenone-2 14.72 1,4-dimethyl Pyridinone - 2(1H) 0.95 2-ethyl-5-methyl Furan 0.85 2,4,6-trimethyl Phenol 0.87

As mentioned previously in chapter 2, the biomass mainly included a mixture of organic and mineral compounds Due to the low pyrolysis temperature, the mineral compounds might not participate in the reaction process Therefore, all gas and liquid products would be largely generated just by the pyrolysis of organic compounds (OC) [21 -24] This mechanism was only a particular case when there was no any reaction between gas and liquid product In order to generalize the pathway of the pyrolysis reaction, a formation reaction of gas from liquid and liquid from solid were supplemented and hence a more general mechanism known as a lump model with the assumption that all reactions are first order and irreversible was proposed in this study The proposed model was shown in Figure 3.7.

Based on these assumptions, the rate of reaction corresponding to Solid, Gas, liquid component in term of reaction yields can be drawn as follows:

Where: t is the residence time k1, k2, k3, k4 and k5 are rate constants

Conclusions

The response surface of variation level in pyrolysis of palm kernel cake was set up using the central composite rotatable design The gas and liquid yields were directly proportional to residence time and pyrolysis temperature, whereas solid yield is reversely The highest value of gas yield, liquid yield and the lowest of solid yield were 17.5, 37.2 and 43.6%, respectively at the condition of 12 min, 460°C

The GC analysis for gas product shows that the major components were C 1 - C4 hydrocarbons The pyrolytic bio-oil composition mainly included three major compounds: 9-penten-2-one; 2-pentanone, 4-hydroxy-4-methyl and 1-hydroxy, 1- phenyl, propanon-2 The composition analysis of solid residue after reaction shows that there was a noticeable decrease of oxygen content in biomass due to the cracking of C-O bonds in the biomass molecular structure

The proposed model was found to well describe the experiment results The rate constants obtained from calculation indicated that all the primary reactions were much more dominant than the secondary ones Also, the closed condition in tubing reactor was preferable to generating solid product rather than other products.

Comparative Study of Pyrolysis of Palm Kernel

Experimental

PKC and PWC were ground with a knife mill and then exposed in air for 24 h The size of prepared sample was between 500 and 600 àm after sieving Ash contents of the biomass were determined using ASTM (E1755-01, 2001) [3]

Thermogravimetric analysis of biomass was carried out using TGA (Q50, TA Instrument) with a heating program from 50°C to 800°C at the ramp rate of 20°C/min

The experiment of fast pyrolysis was carried out in an upright tubular reactor (φ2.54 cm × 50 cm) as schematically described in Figure 4.1 For each run, the sample with a weight of 1.5 g was pre-placed in the feed hopper When the furnace temperature achieved the setting values (550, 600, 650, 700 or 750 o C), on/off valve was subsequently switched on to allow both the biomass and sweeping gas (flow rate changed at 20, 200 and 500 mL/min) to enter the reactor The produced liquid was condensed and separated from the aerosol at the bottom of condenser (cooled by coolant at -20°C using chiller), while the gas product was led to, and then trapped at the water column Finally, the flow of N 2 was shut down after ten minutes by switching off the on/off valve, and the pyrolysis was finished

The product yields of fast pyrolysis for each type of biomass were calculated based on the equations as shown below: weight of liquid

Liquid yield (wt%) 100% weight of biomas

Solid yield (wt%) 100% weight of biomas

Gas yield (wt%)0% Liquid yield Solid yield− −The weight of liquid product was determined by the difference in weight of the condenser before and after pyrolysis, and the weight of solid was obtained from balance

Figure 4.1 Schematic diagram of experimental apparatus

Results and discussion

Figure 4.2 shows the thermogravimetric analysis data which was carried out at the heating rate of 20 o C/min Although the kinetics and reaction pathway at this heating rate in TG analyzer were totally different with those of fast pyrolysis conditions, these results give some useful information in respect of understanding the characteristics of biomass

As shown in Figure 4.2, DTG graph of PKC shows two sharp peaks at 297, and 337°C, while that of PWC appears one peak at 380°C For biomass constituted by hemicellulose, cellulose and lignin, Raveendran et al [4] and Demibras et al [5] found that the less stable hemicellulose would decompose first at around 300°C, followed by lignin at 250 - 550°C, and lastly by cellulose at 300 - 430°C Based on the decomposing temperatures as mentioned above and the asymmetry of peak from the DTG graph of PWC, it can be inferred that there arose an overlapping between the peak derived from the decomposition of hemicellulose (25 - 30 wt%) and cellulose (40 - 50 wt%), thus forming only one asymmetric peak at the temperature of 380°C The peak from lignin was almost difficult to observe due to indicates that this peak was resulted from the decomposition of only one major compound Based on all these reasons, it can be confirmed that this peak was due to the decomposition of mannan compound only The second sharp peak at 337°C would disappear if the biomass was washed by acetone to eliminate all the remaining palm oil Therefore, this peak was attributed to the degradation of palm oil remaining in biomass Finally, DTG data suggest that PKC and PWC have completely different carbohydrate compositions: one from major non-cellulosic material and one from cellulosic material

Palm kernel cake Pine wood chip

Figure 4.2 DTG data of TG analysis for PKC and PWC at heating rate of 20 o C/min

4.3.2 Effect of fast pyrolysis conditions on product yield

The product yields of fast pyrolysis using both types of biomass at various temperatures in slow sweeping-gas flow rate (20 mL/min) were shown in Figure 4.3 and 4 (the dotted lines) PKC was pyrolyzed to produce products with the lowest gas yield and highest liquid yield of 15.7 and 58.9 wt% respectively at 550°C PWC was decomposed at the same temperature to form gas and liquid product of which the yield was 35.4 and 43.2 wt% respectively The product distributions of both biomass pyrolysis at 750°C were totally different For instance, the product from PWC reached the highest gas yield of 67.3 wt% and lowest liquid yield of 18.4 wt% As for the product from PKC, gas and liquid yield were 45.0 and 36.1 wt% respectively At slow sweeping-gas flow rate (20 mL/min), char yields just increased slightly with the pyrolysis temperature and varied in the range of 18.9 - 25.4 wt% for PKC and 14.3 - 21.4 wt% for PWC

Generally, if pyrolysis temperature increases, the gas yields would increase, whereas liquid and char yields decrease This tendency can be explained based on the physical chemistry property of cracking reaction As commonly known, cracking is an endothermic reaction Therefore, with increasing the temperature, the cracking reaction is more accelerated, causing easily cleaving heavy molecules to form smaller ones Consequently, the gas yield would increase significantly, while both of liquid and char yields decrease together In addition, as can be seen from these figures, the declination of liquid yields with the increase of temperature in experiment with PKC was smaller than that in experiment with PWC This indicates that the resultant bio-oils from pyrolysis of PKC were more thermally stable than those from PWC

Figure 4.3 and 4 (the dashed and solid lines) also show the product yields of fast pyrolysis at various temperatures in the fast sweeping-gas flow rate (200 and

500 mL/min) Of all experiment at this flow rate, the highest liquid yield 63.1 wt% was obtained at 550°C with the feedstock of PKC, whereas the highest gas yield was 56.1 wt% from pyrolysis of PWC at 750°C In comparison with the yields from pyrolysis at slow sweeping-gas flow rate, those at the fast were slightly changed: liquid yields increased, while gas yields decreased Although this change was slightly different in the range of 5.6 - 11.2 wt% for gas yields and 3 - 6.4 wt% for liquid yields, the role of fast sweeping-gas flow rate on retardation of secondary cracking reactions in pyrolysis process could be more clarified Indeed, when the flow rate rose up, the residence time of intermediate vapor compounds would decrease, leading to lowering the cracking reaction and hence improving the bio-oil yields as well as restricting the gas yields temperature, oC

Product yield at condition of 20 mL/min (N 2 ) Product yield at condition of 200 mL/min (N 2 ) Product yield at condition of 500 mL/min (N 2 )

Figure 4.3 Product yield of fast pyrolysis of palm kernel cake temperature, oC

Product yield at condition of 20 mL/min (N 2 ) Product yield at condition of 200 mL/min (N 2 ) Product yield at condition of 500 mL/min (N 2 )

Figure 4.4 Product yield of fast pyrolysis of pine wood chip

Unexpectedly, all char yields from pyrolyzing two types of biomass at the fast sweeping-gas flow rate were also higher than those at the slow This increase, firstly, might be due to the effect of heating transfer on the surface of biomass particles Owing to the fast flow rate, the heat receiving from the wall of the reactor would be easier to be lost, compared to the experiment with the slow flow rate As a result, the core of biomass could not attain sufficient energy for completely decomposing, and hence a small portion of feedstock might remain after pyrolysis

Lastly, the retardation of secondary cracking reactions as mentioned above should be also taken into account for explaining the increase of char yield with sweeping- gas flow rate In fact, this effect might result in somewhat changing in pyrolysis pathway occurring in process As a result, the pyrolysis reaction might be altered preferably following the direction of forming char, thus also causing the increase of char yield

The yields of hydrocarbon-rich gas and mixture of CO, CO2 in gas product obtaining from pyrolyzing PKC and PWC at various temperatures in both slow and fast flow of N2 were shown in Figure 4.5 This figure clearly shows the effect of temperature on fast pyrolysis through the proportional relationship between the

9.88 wt% (based on biomass weight) respectively in the pyrolysis of PWC at the condition of 550°C, 20 mL/min Meanwhile, those of pyrolysis at 750°C, 20 mL/min were 23.3 and 21.7 wt%, respectively Also from GC-TCD analysis, no hydrogen was detected in the gas products of pyrolysis experiment at all the selected conditions Generally, it was known that the forming hydrogen reaction was commonly occurred at high temperature (> 900°C), high pressure (> 1 MPa) with the support of catalysts (transitional metal based catalysts) [6] In comparison with these conditions, that of this study was not sufficiently severe for facilitating such a reaction arising

Finally, also from Figure 4.5, it can be seen that the gas yields from PWC as well as almost the yield of hydrocarbon and mixture CO, CO 2 were always higher than those from PKC at any investigated conditions Regardless of the dissimilarity in structure of major non-cellulosic and cellulosic biomass, the hydrocarbon compositions of both gas products obtained from pyrolysis of PKC and PWC were rather similar A demonstration for the analysis of this hydrocarbon-rich gas using GC-FID was shown in Figure 4.6 The results show that the major components mainly consist of methane, ethane, ethylene, propane and cyclopropane, of which methane always occupies the highest amount

Palm kernel cake Pine wood chip

Figure 4.5 Yield of hydrocarbon and mixture of CO and CO2 in gas products obtaining from fast pyrolysis at different conditions: (1) - 550 o C, 20 mL/min; (2) -

550 o C, 500 mL/min; (3) - 750 o C, 20 mL/min; (4) - 750 o C, 500 mL/min

1- Methane2- Ethane3- Ethylene4- Propane5- cyclopropane6- Propylene

Table 4.1 Bio-oil and water content in liquid product at 550 o C

20 mL/min 500 mL/min 20 mL/min 500 mL/min

Table 4.1 presents the bio-oil and water content of liquid product of pyrolysis at the temperature of 550°C, with sweeping-gas flow rate of 20 and 500 mL/min The data show that depending on biomass type, water, a product of dehydration reaction occurring during pyrolysis [7], would change its content from 15.6 to 25.9 wt% Specially, the water contents obtained from PKC were always higher than those from PWC at all investigated conditions As known, the water content obtained from pyrolysis depended on three factors: oxygen content contained in biomass, degree of dehydration reaction occurring during pyrolysis and oxygen content released to the gas phase in form of CO or CO2 gas It is impossible to determine which factor plays the most dominant role in this pyrolysis process; therefore, there was just a very simple conclusion to be drawn from this result that the biomass composition had strong effect on product distribution

Table 4.2 and 4.3 show the GC-MS analysis for major compounds of bio- oil obtained from pyrolysis of both biomass at the temperature of 550°C and flow rate of 500 mL/min As shown in these tables, the bio-oils from pyrolysis of both biomass have rather similar compositions and mostly consist of phenol derivatives, andehydes, ketones, alcohols, and organic acids, in which the phenol derivatives dodecanoic acid (17 %), tetradecanoic acid (3.4 %) and octadecenoic acid (1.7 %), as opposed to just a minor hexadecanoic acid amount (2.6 %) from PWC As reported by Demirbas [8], oils derived from plants were built up by triglyceride molecules When this molecule was decomposed under thermal conditions, the three branches making the structure of the molecule would be randomly cleaved to generate fatty acids and other smaller molecules [9] Based on this result, it can be concluded that the fatty acids existing in both pyrolyzing bio-oils as mentioned above were derived from the decomposition of the remaining oils in the raw biomass Finally, as can be seen, almost the components of bio-oil products were oxygenated compounds It was the presence of oxygen in compounds that lower the heating value and the stability of bio-oils The upgrading process was therefore recommended for these bio-oils before they can be widely used as fuel

Table 4.2 Major bio-oil product analyzed by GC-MS for fast pyrolysis of palm kernel cake at 550 o C, 500 mL/min

Table 4.3 Major bio-oil product analyzed by GC-MS for fast pyrolysis of Pine wood chip at 550 o C, 500 mL/min

Conclusions

In this study, a comparative study of fast pyrolysis using palm kernel cake (major non-cellulosic material) and pine wood chip (major cellulosic material) was carried out at various conditions The pyrolysis temperature was changed in the range of 550 - 750°C in the sweeping-gas flow rate of 20, 200 and 500 mL/min for determining the distribution as well as properties of products Based on these results, the main conclusions from this present work can be drawn as follows:

• When the pyrolysis temperature was raised, the liquid yields were decreased significantly, while the gas yields were increased The highest yield of liquid product was 58.9 wt%, obtaining from the pyrolyis of PKC at the condition of 550°C, 20 mL/min In contrast, at this condition, PWC was pyrolyzed for the highest gas yield at 67.3 wt% Also, the char yields varied in the range of 6.5 - 7.1 wt%

• For pyrolysis of each biomass at the same pyrolysis temperature, the liquid yields from experiment in the fast sweeping-gas flow rate, were slightly higher (around 3.0 - 6.4 wt%), whereas the gas yields were lower than those from the slow (approximately 5.6 - 11.2 wt%) Meanwhile, the char yields

• Regardless of biomass feedstock, the resultant gases were qualitatively similar, largely including carbon oxide, carbon monoxide, methane, ethane, ethylene, propane and cyclopropane

• The bio-oil obtaining from pyrolysis of both biomass was a mixture of phenol derivatives, aldehydes, acids and alcohols.

References

[2] E.M Dusterhoft, M.A Posthumusb., and A.J.V Voragena, Journal of the

Science of Food and Agriculture, 1992, 59, 151-160

[3] Annual book of ASTM standards, 1997

[4] K Raveendran, A Ganesh, and K.C Khilar, Fuel, 1996, 75 (8), 987-998

[5] A Demirbas, and G Arin, Energy sources, 2002, 24 (5), 471-482

[6] D.A.J Rand, and R.M Dell Hydrogen Energy: Challenges and Prospects, The Royal Society of Chemistry (Great Britain), 2008

[7] S Czernik, and A.V Bridgwater, Energy & Fuels, 2004, 18, 590-598

[8] A Demirbas, Energy Conversion and Management, 2003, 44, 2093–2109

[9] N.H Jayadas, and K.P Nair, Journal of Tribology, 2007, 129 (2), 419 - 423

Kinetic Model of Fast Pyrolysis Using Palm Kernel Cake in a Closed Tubular Reactor

Introduction

Studies focused on the kinetics of fast pyrolysis for biomass conversion into bio-fuel are indispensable for understanding both the mechanism and kinetic parameters There have been various reports referring to mechanisms or kinetic models [1-5] Still, it has not been fully understood yet owing to the difficulties in determining product contents for several reasons First, it is impossible to specify the content of too many compounds in bio-oil Second, there is no feasible method for differentiating between char and remaining biomass after pyrolysis process, thus the char yield, or the biomass conversion can not be calculated In order to avoid the first difficulty as previously mentioned, a lumped kinetic model should be applied to simplify the kinetic calculation However, this model can be used only if all product yields are available Consequently, finding out how to determine the product yields is necessary to facilitate the investigation of pyrolysis mechanism

From a survey of previous reports, it can be deduced that open reactors, such as fluidized-bed or fixed-bed reactors, were the most commonly used for fast pyrolysis of biomass at temperatures around 400 - 600 o C [1, 6-8] There is little research on closed reactors, although under pressurized conditions the mechanism, equilibrium, and kinetics may be completely changed For this reason, in order to investigate the effects of closed conditions on reaction pathways, a closed micro- tubular reactor was therefore designed for fast pyrolysis of biomass and tested in this study

Palm kernel cake is an abundant biomass byproduct of palm oil production in Malaysia and Indonesia This biomass consists of a large amount of noncellulosic carbohydrate, mannan, which is a cellulose-like bio-polymer [9, 10]

The pyrolysis characteristics of noncellulosic biomass might have promising differences from the pyrolysis of cellulose-based biomass, and further the understanding of the role biomass composition plays in the pyrolysis reaction

As a result of aforementioned reasons, fast pyrolysis in a closed-tubular reactor at high temperature (550 - 750 o C) using palm kernel cake was investigated in this study In order to investigate the kinetic pyrolysis of palm kernel cake using a closed tubular reactor, a new method for determining the product yields, especially for gas yield and remained biomass yield, was drawn The pyrolysis process was modeled based on the kinetic model proposed by Liden [1] as shown in Figure 5.1 From the obtained kinetic constants, the favorable and unfavorable

Figure 5.1 Pyrolysis reaction mechanisms proposed by Liden [1]

Figure 5.2 Schematic diagram of experimental apparatus k 3 k 1 k 2

Experimental

Figure 5.2 shows an upright tubular reactor (φ2.54 cm × 50 cm) setup for fast pyrolysis of PKC biomass Before pyrolysis process was carried out, the reactor was purged completely of air by nitrogen sweeping gas Once this step was complete, on/off valves 1 and 2 were closed Subsequently, 1.5 g of the sample was placed in the feed hopper When the setting value of furnace temperature was reached, valve 1 was quickly switched on to allow the biomass sample to enter the reactor, and then closed quickly At each retention time-point (10, 20, 30, 40, 50 and 60 s), the on/off valve 2 was opened, followed immediately by valve 1 to release the product mixture Opening valve 1, allowed nitrogen gas, at a high flow rate of 1 L/min, to sweep all products out of the reactor The liquid produced was condensed and separated from the aerosol at the bottom of the condenser, which was cooled by coolant at -20 o C using chiller, while the gas product was trapped by the water column

5.2.2 Calculating product yield composition from C1 – C6 in gas products, the peak areas of these components were calibrated using a standard gas mixture (Supelco – Scotty 14) A GC–TCD (ACME

6000GC – Younglin Co.), with a packed column of Carbosphere (80/100 mesh, 6ft × 1/8 inch × 0.85 inch, Alltech), was used for determining the content of CH4, CO, CO2 and H2 Pure CH4, CO, CO2 and H2 gas were used as standards for determining the content of gas product

The weight of gaseous product was calculated as follows: n mix i i i 1 x

V mix = V total - V sweeping where Vtotal, Vsweeping were the total volume of gas product collected from the water column, and the volume of sweeping gas at each retention time (10, 20, 30, 40, 50 and 60 s), respectively In order to calculate the gas density, the ideal gas law can be used as shown below: i i i i i i m P

As known, the gas product might include H2, CO, CO2, CH4, C2H6, C2H4, C3H8, C3H6 and C4H10 Therefore, their densities at standard conditions (25°C, 1 atm) were used for calculating the weight of gaseous products above

Finally, the gas yield and bio-oil yield at each reaction time were determined using the following equations: gas feed

= − × (5) where mfeed is the weight of the sample, mliq is the weight of the liquid product after pyrolysis and mwater is the weight of water determined using Orzherovskii’s method [9].

Results and discussion

When using a micro reactor with a small amount of feedstock, if a low pyrolysis temperature is used, bio-oil formed from pyrolysis might not receive sufficient energy to be broken into smaller-molecule products The bio-oil instead remains and clings to the wall of the reactor, and eventually generates a serious error in the determination of char yield As a result, the fast pyrolysis in this research was carried out at a high temperature range of 550 - 750 o C

CO, CO2, CH4, C2H6, C2H4, C3H8 and C3H6 were 55, 12, 14, 5.5, 8, 1.5 and 4%, respectively Specially, there was no hydrogen detected in the gas products of the pyrolysis experiment under any conditions The compositions of the gas products were qualitatively similar to those described in other reports [10, 11] with the exception of hydrogen As reported by Li et al [10], who carried out fast pyrolysis of legume straw and apricot stone with an open free-fall reactor at 800°C, the hydrogen content was approximately 22% This indicates that either the pyrolysis temperature used in the current study was not adequately high or biomass feedstock was not favorable to promote the hydrogen forming reaction Additionally, the fast pyrolysis reactions were also carried out in a closed reactor and hence, they would not attain any support from the medium for hydrogen formation as opposed to gasification in which the gasifying agent (air, steam, or pure oxygen) is continuously supplied [12]

5.3.2 Pyrolysis kinetics 5.3.2.1 Equations and approaches for calculation

As reported by Liden et al [1], the decomposition reaction, especially fast pyrolysis, mostly occurs by a complex pathway with the formation of numerous products As a consequence, it seems to be impossible to specify the reaction kinetics if each specific compound is considered However, there is a general and simple approach, named the lump model, recently applied by many researchers [1,

5, 13] Of all mechanisms reported, the one proposed by Liden is most commonly used

In order to apply Liden’s mechanism for this study, all products of the fast pyrolysis reaction are gathered to form lumps depending on the product phase

More specifically, the gas product including CO, CO 2 , and light hydrocarbons are collected in a gas lump Meanwhile, the liquid product (after water separation), largely consisting of oxygenate compounds, would establish a lump of bio-oil

Lastly, all solid products would be lumped as char Each lump of products was then used as a complete compound for investigating the reaction kinetics following the mentioned mechanism In order to simplify the calculation, all pyrolysis reactions were assumed to be occurring irreversibly by a first order reaction The reaction order of the biomass pyrolysis was also verified by many researchers and proved that the assumption of the first order in this reaction was consistent with experimental data [14, 15]

Based on these assumptions, the rate of reaction corresponding to biomass (B), bio-oil (O), gas (G) and char (C) in form of product yields could be drawn and shown below:

C 3 B dC k C dt = (9) where k1 is the rate constant of biomass decomposing rate in reaction Biomass (B)  Bio-oil (O); k 2 is the rate constant of bio-oil decomposition rate in reaction Bio-oil (O)  Gas (G), and k3 is the rate constant of biomass decomposition rate in reaction Biomass (B)  Char (C) and Gas (G) The variable t is the retention time C B , C T, and C G were the yields of biomass, bio-oil and gas product at an investigated time respectively

Solving the system of these differential equations led to obtaining all equations expressing the relationship between product yields and retention time

The results were shown as follows:

C is the initial yield of volatile organic in biomass

The agreement of the product distribution obtained from both experiments (550, 650, and 750°C) and the applied pathway was verified by using a nonlinear least-square regression based on the system of these above model equations The prerequisite of this comparison is that all product yields must be determined, such as yields of biomass, gas, bio-oil and char component at a specified retention time

However, it is impossible to determine the biomass yield at a certain retention time Methods for differentiating between biomass remaining after the reaction and the char obtained from pyrolysis are lacking Instead of this, an iterative approach was performed for all experiment data and is described later

Initially, the biomass decomposition was assumed to occur via a first order reaction

As a result, the relationship between biomass yield and retention time complies with the exponential function as shown in equation (10) or described generally in the form of C B =C Bo exp(-kt) On the other hand, based on the experimental data, the gas yields remained almost constant at a reaction time of 60 s for all experiments from 550 to 750°C This meant that the biomass was completely decomposed at 60 o values from 0.001 to 0.5 However, as can be seen in Figure 5.3, and based on the boundary condition (C B = 0 at t = 60 s), the range of k values that were from 0.001 to 0.05 can be ignored because at this range, biomass remained after pyrolysis For instance, the biomass yield at k = 0.05 was around 4.5 wt% at 60 s Also, the pyrolysis reaction seemed to occur immediately at k > 0.5, resulting in the fact that the biomass yield was almost zero at the retention time of 10 s, which was not in accord with experimental data Therefore, k values from 0.05 to 0.5 were selected for the iteration used to calculate all rate constants of the reaction pathways proposed by Liden that are described in Figure 5.1

As can be seen in Figure 5.1, the number of rate constants was three, while there were five equations (10, 11, 12, 13 and 14) which can be used for regression

Of all these equations, only two equations (11 and 12) were applied for calculation, respectively based on the experimentally-determined contents of bio-oil and gas

The other two equations (13 and 14), owing to their dependence on the impossibly- determined biomass and char yield, were employed as two constraints for the whole regression process As for equation (10), it was used for iteration by substituting various values of k constant as mentioned above In order to carry out the regression calculation for obtaining all rate constants, an iterative program was set up First, the k value was pre-fixed at 0.05, allowing the determination of a value of biomass yield, C B, at a specific retention time and describing the relationship between k and k1, k3 (k=k1+k3, equation (15)) Subsequently, equation

(15) was combined with equations (11 and 12) to create a system of three functions for regression calculation of bio-oil and gas yield from the experiment after various retention times As a result of this step, the values of k1, k2, and k3, together with the coefficients of determination R 2 , were established The set of k1, k2, and k3 was then substituted into equations (10, 13, and 14) for verifying the agreement If this condition was met and the R 2 value was in the range of 0.95 to 1, the loop for this sequence would be stopped, and all rate constants were achieved Otherwise, the initial value k which was prefixed at 0.05 would be increased at an increment of 0.001 and the iteration would be repeated until the all conditions were satisfied

Figure 5.3 Relationship between biomass yield and reaction time described by equation (10)

5.3.2.2 Obtaining rate constants from regression calculations

All rate constants were obtained from a fitting calculation and are tabulated in Table 5.1 At the same temperature, k1 had the highest value, followed by k2 and lastly by k3 A large difference between k3 and k1 (or k2) indicates that the second reaction pathway forming gas (G) and char (C) from biomass was not preferable as compared with the first one (forming bio-oil (O) from biomass) The order of these rate constants is in good agreement with the results reported by Liden’s et al [1], albeit with different biomass and at a lower temperature Specially, with the smallest value of k3, in a few reports the decomposition reaction of biomass to generate gas (G) and char (C) was neglected or considered a secondary reaction [1]

Regarding the value of each the rate constants, it is difficult to compare these with values from past studies due to the differences in reaction conditions For instance,

Liden et al [1] carried out pyrolysis of cellulosic biomass at the open condition at temperatures lower than 550°C Meanwhile, in this current study, fast pyrolysis was performed in a closed reactor at very high temperatures (550 - 750°) using noncellulosic biomass Eventually, it was observed that once the reaction conditions, specifically pressure, were changed, the reaction pathway would be

Figure 5.4 Gas yields were directly proportional to both the retention and pyrolysis temperatures, but not to bio-oil yields With conditions of 750°C and 60 s, the highest yield was obtained at 79.9% while the lowest yield of bio-oil was 0.1% As for char yields from calculations in experimental data at 550°C, there was just a very slight increase following the retention time, varying around the value of 10.77% When the pyrolysis temperature was increased to 750°C, this char yields tended to decrease to approximately 9.25% For the remained-biomass yields achieved from equation (10), the data show that biomass was almost completely decomposed after 20 s for experiments at all investigated pyrolysis temperatures

The calculated data fitted well with experimental results, indicating that the pyrolysis of biomass could occur via the proposed model

Product yield from experiment at temperature of 550 o C Product yield from experiment at temperature of 650 o C Product yield from experiment at temperature of 750 o C Product yield from calculation

Figure 5.4 Product yields of fast pyrolysis obtained from experimental data and calculations at various conditions

Conclusions

In this study, fast pyrolysis of PKC was carried out under various conditions The pyrolyzing gas product consisted largely of carbon monoxide mixed with a smaller fraction of carbon dioxide and light hydrocarbon gases

Notably, no hydrogen was detected in the product gas The yields of gas, bio-oil and char during fast pyrolysis were in the range of 32-80.8, 0.1-33, and 8.4-10.7 wt%, respectively

The kinetic lump model of Liden was employed to calculate the rate constants for the fast pyrolysis reaction The results showed that the experimental data was well described with the applied mechanism Also, based on the difference between k1 and other rate constants (k2 and k3), it was found that the pyrolysis was more favorable via the reaction forming bio-oil followed by the decomposition of bio-oil to produce gas, rather than reaction forming the mixture product of char and gas.

Pyrolysis kinetics and parametric effects on fast

Results and discussion

6.3.1 Kinetic parameters of pyrolysis using TGA

The differential rate of conversions (dX/dt) versus temperature for the sample of raw PKC and washed PKC, described in form of DTG graphs, were shown in Figure 6.2 Thermogravimetric analysis is an effective method for calculating all kinetic parameters of biomass pyrolysis, as reported elsewhere in the literature [10-12] All calculations were based on the equation shown below:

X = conversion of biomass decomposition in TG analyzer A = pre-exponential factor, min -1

R = gas constant (8.314 J.mol -1 K -1 ) t = time, min

T = absolute temperature, K In order to calculate the value of activation energy (E), it must be assumed that the values of A and E in the equation (2) were constants independent of temperature The reaction order of biomass decomposition as reported by Michael [13] and Zhang et al [14] varied around the unit value Therefore, the value of n in equation (2) was substituted by 1 In addition, it can be seen that equation (2) would present a linear relationship between ln(dX/dt) and 1/T if the value of ln[A(1-X)] was constant at a given conversion X For each heating rate, there would be a pair of ln(dX/dt) and 1/T value, and hence with four heating rates, a data set of four pairs of ln(dX/dt) and 1/T value would be obtained and a linear regression could then be applied to the data This process was repeated over a conversion range of 10 to 90% over which pyrolysis occurred from 200 to 565 o C

The linear relationship between ln(dX/dt) and 1/T at each iso-conversion point was shown in Figure 6.3

From the slopes of these regression lines (slope = -E/RT), the activation conversion of 40% In addition, the high activation energies were inclined to distribute concentratively at the conversion from 20 - 60% which corresponded with the strongly decomposed region of biomass (287 - 353 °C) Hence, it can be concluded that the activation energy for biomass decomposition would be in the range of 205 - 306 kJ/mol This activation energy was relatively similar to that of the pyrolysis of cellulose (195 - 213 kJ/mol) as reported in previous work [15]

5 o C/min 10 o C/min 15 o C/min 20 o C/min PKC washed by acetone

Figure 6.2 DTG data from thermogravimetric analysis of palm kernel cake

Figure 6.3 The relationship between ln(dX/dt) and 1/T at different iso-conversion points

Figure 6.4 Calculated activation energy of pyrolysis as a function of conversion

6.3.2 Fast pyrolysis using fluidized bed reactor

Table 6.1 shows the actual and coded levels of the process parameters The coded values were designated by -1 (minimum), 0 (center), +1 (maximum), -α and +α Selection of levels for each factor was based on some literature reports on the pyrolysis of other biomass using fluidized bed reactor The investigated region of temperature was 400 - 500°C since the biomass effectively occur at these temperatures with the decomposition of hemicellulose, cellulose and lignin as reported by Demirbas [5] Moreover, based on the TGA data, it was observed that palm kernel cake was almost decomposed in the range of 400 - 500°C The level of residence time, 0.6 - 0.9 second, was limited by the fact that the pyrolysis was just effective for obtaining bio-oil if the residence time is below one second [4] The biomass particle size was selected between 300 - 600 àm in order to make the screw feeder operate conveniently Because palm kernel cake remains palm oil (10.7 wt%), it will be sticky if the particle size is so small As for the feed rate of feedstock, depending on the size of reactor, the feed rate was just appropriate in the range of 160 - 300 g/hr Table 6.2 presents the conducted experiments based on the design matrix

Factors Unit Variation levels Variation interval

Residence time, x3 second 0.45 0.6 0.75 0.9 1.05 0.15 Particle size, x 4 àm 150 300 450 600 750 150

Table 6.2 Experimental design matrix and response value

Run X 1 X 2 X 3 X 4 Liquid yield (%) Experiment at core point Experiment at axial point

Table 6.3 shows the regression coefficients calculated based on the equations (3) - (6) in chapter 2, section 2.3.2 and the F-test of the obtained models

Both the F-test value of regression model for liquid yield was smaller than the F- value from the standard F-distribution with the confidence of 99% This means that the obtained regression equation was adequate Figure 6.5 shows the relationship between actual and predicted value of liquid yield This figure indicates that the model was successful in capturing the correlation between the process parameters to the liquid yield with a correlation coefficient R 2 =0.99

After checking the adequacy of the regression equations, the significance of each regression coefficient was also carried out The t-test value was obtained by dividing each coefficient by its standard error One coefficient was just considered as significant if the magnitude of its t-test value is larger than the standard t- distribution at a certain confidence, for example 95% A large t value implies that the coefficient is much greater than its standard error In addition, it should be noted the meaning of the t-test sign For instance with β i , the response value (liquid yield) will increase or decrease together with the variable Xi, depending on the t- test sign being positive or negative, respectively As can be seen in the column of t- test for liquid yield in Table 6.3, most of the coefficients were significant except for the coefficient β14 and β33 Therefore, this coefficient can be eliminated from the regression equation for liquid yield The final predictive response equation containing all the significant coefficients was shown as below:

Table 6.3 Statistical significance of regression coefficients coefficient t-test significance βo 45.710 238 + β1 -1.697 -16.3 + β2 6.142 59.06 + β 3 3.182 30.6 + β 4 0.446 4.3 + β12 2.906 22.9 + β13 1.794 14.1 + β14 -0.269 -2.1 - β23 -4.069 -32 + β24 1.044 8.2 + β34 -0.894 -7 + β 11 -2.277 -24.2 + β22 -5.285 -56.2 + β33 0.082 0.87 - β44 -0.330 -3.5 +

Significant: +; insignificant: -, degree of freedom f = 6; t-Student (0.05) (6) = 2.45 FT (a)

6.3.2.2 Effect of operating parameters on performance of fast pyrolysis

From the regression equation, it can be seen that there are three types of effects of variable on the response value, including: main, squared and interaction effects corresponding to coefficient βi, βii and βij As for the main effects, the magnitude of t-test was obtained with the order as follows: β2 (59.06) > β3 (30.6) > β1 (-16.3) > β4 (4.3) It was also known that the higher the t-test value of a coefficient, the more significant effect of that coefficient would be Therefore, it can be concluded that the variable X2 (pyrolysis temperature) and X3 (residence time) were the two important factors having the strongest effects on the response value in comparison with the others The order of significant effect on response value can be arranged as following: X2 > X3 > X1 > X4 Since, there are four factors affected the pyrolysis process, it is impossible to present all their effects on the same 3D graph As a result, two factors would be fixed at constant, while the other varied, leading to obtaining the 3D graphs as shown in Figure 6.6, 7 and 8

Indeed, pyrolysis is a cracking reaction for breaking the high molecular hydrocarbon chain to form the smaller molecular compound It is always accompanied with the endothermic effect Consequently, when the pyrolysis until a limit decomposing temperature was achieved, and then decrease with increasing temperature

From experiment as shown in Figure 6.6, at low temperature like 400°C, the receiving energy might be not enough for completely decomposing biomass If the flow rate of fluidizing gas was increased or residence time was decreased, it would draw more and more the incompletely decomposed biomass out of reactor, thus leading to reducing the obtained product yield At high temperature like 500°C, the biomass can be decomposed completely; however, bio-oil might be partially degraded Hence, if the flow rate of fluidizing gas was increased, it can push the product out of reactor as quick as possible, therefore resulting in increasing the liquid yield

Analysis of response surface from the mathematic model of liquid yield indicated the stationary point (X1=0.149; X2=0.597; X3=0.356; and X4=1.139) at which the derivative of equation is zero to be located outside the surveyed region

The Eigen values of the characteristic equation are of different signs implying that the stationary point is a saddle point (λ11 = 1; λ22 = -6.63; λ33 = -0.529; λ44 = -1.729) Therefore, the optimal response values might be determined at one of the boundary of the investigated variable ranges [9]

In order to specify the optimal response values, a trial and error algorithm was applied for the equation of liquid yield in coded units A procedure using Matlab was also programmed to set up a loop for searching the condition in the surveyed range at which the response function achieved the highest value

Once obtained, the optimal condition (X 1 = -0.1, X 2 = 1; X 3 = -1; and X 4 = 1) was then converted into the respective uncoded (actual) units using the formulae described in equation (4) Finally, the optimal values of feed rate, pyrolysis temperature, residence time, and particle size would be 225 g/hr, 500°C, 0.6 s, and 600àm respectively, corresponding to the highest liquid yield of 49.5 wt%

P redi ct ed v al ue

Figure 6.5 Relationship between actual and predicted value of liquid product yield

0.60 0.65 0.70 0.75 0.80 0.85 Li qui d y iel d, w t% tem per at ur e, o C

Figure 6.6 Liquid yield at the condition of feed rate = 160g/hr, particle size 300àm

Pa rti cl e si ze , à m

Figure 6.7 Liquid yield at the condition of pyrolysis temperature = 400 o C, residence time = 0.9 sec

6.3.3 Characteristic of pyrolyzing liquid product

Table 6.4 shows GC-MS analysis data of bio-oil obtained from pyrolysis experiment at two temperatures 400 and 500°C at the same condition of feed rate (300 g/hr), particle size (300 àm) and residence time (0.6 sec) According to shown data, the detectable components of bio-oil were acid acetic, ketones, derivatives of furan, phenolics, esters, β-D-allose and fatty acids As reported by Yaman [16], the bio-oil from pyrolysis of lignocellulosic biomass was largely composed of alcohols, aldehydes, ketones, organic aicds, ester, phenolics, and levoglucosan

Obviously, it can be realized that there existed some differences in product distribution between this study and that of other researchers The most noticeable point was that the presence of β-D-allose (a C3 epimer of glucose [17]) with the amount up to around 20% while some compounds such as alcohols, aldehydes and phenolics were in a smaller quantity It can be explained based on the difference in biomass composition Indeed, the major composition of palm kernel cake is mannan - a biopolymer formed by a great number of sugar mannose (C 6 H 12 O 6 ) units (C2 epimer of glucose [17]) Basically, when a polymer was thermally decomposed, its chain would be cracked to release lower molecular compounds like monomer, dimer, trimer, tetramer, oligomer, etc and some other compounds depending on the reaction condition During this cracking process, various types of reaction might occur, such as transposition of C, H or cleavage of the C-C linkage [18] For a monomer, there might be a change in molecular structure as compared with its initial structure unit; however, the molecular formula might not change

Conclusions

Kinetic tests on pyrolysis of palm kernel cake were carried out using a thermogravimetric technique (TGA) at heating rate of 5, 10, 15 and 20°C/min The activation energy increased with the conversion in the range of 10 - 40%, and decreased in the range of 40 - 90% with the value in the range of 205 - 306 kJ/mol

The fast pyrolysis of palm kernel cake directed by the central composite rotatable design has been carried out in a fluidized bed reactor in a surveyed region of variables as follows: feed rate of 160 - 300 g/hr, pyrolysis temperature of 400 - 500 °C, residence time of 0.6 - 0.9 sec, and particle size of 300 - 600 àm The results showed that the experimental data was rather fitted with the second-order regression equation Therefore, the effect of four operating parameters on the performance of fast pyrolysis was elucidated Of the four operating parameters, the order of significance for each factor was arranged as follows: pyrolysis temperature

> residence time > feed rate > particle size

In the surveyed region of variables, the liquid yield achieved the highest value of 49.5 wt% at the condition of 225 g/hr, 500°C, 0.6 s, and 600àm The pyrolysis conditions carried out at two temperatures 400 and 500°C were used for performing the pyrolysis reaction by which all obtained products would be subsequently analyzed in order to determine the characteristic of reaction The pyrolyzing oil largely contained 1, 2-benzenedicarboxylic acid - bis (2-ethylhexyl) ester, β-D-allose, and fatty acids.

Conclusions and further researches 7.1 Conclusions

Further researches

• Since the product distribution (gas, liquid and char) as well as product properties are strongly affected by the composition of biomass, various types of biomass with different biomass composition should be investigated more As mentioned previously, the main purpose of pyrolysis is to obtain bio-oil Therefore, finding out which biomass can result in the highest bio-oil yield is the most priority option

• In addition, the major components in bio-oil are oxygenated compounds, which was undesirable for using as a fuel owing to its reducing-heating value effect The upgrading of bio-oil by decreasing as much as oxygenated compound as possible is also preferable for further researches In order to upgrade the bio-oil quality, the catalytic pyrolysis using zeolite-based catalyst mainly ZSM-5 has been recently interested

• Of all types of reactor, the fluidized bed results in the highest bio-oil yield The studies on how to increase the bio-oil yield and optimize the operating parameter should be carried out Finally, the kinetic pyrolysis of fast pyrolysis in fluidized bed reactor is also needed to be investigated through which the mechanism and kinetic parameters can be obtained.