Experimental and numerical simulation study of oxycombustion of fast pyrolysis bio-oil from lignocellulosic biomass

Energy 126 (2017) 854e867

Contents lists available at ScienceDirect

Energy journal homepage: www.elsevier.com/locate/energy

Experimental and numerical simulation study of oxycombustion of fast pyrolysis bio-oil from lignocellulosic biomass S.I. Yang*, M.S. Wu, T.C. Hsu Department of Power Mechanical Engineering, National Formosa University, Yunlin County, 63202, Taiwan

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 October 2016 Received in revised form 16 March 2017 Accepted 17 March 2017

Experimental measurements and numerical simulations were conducted to examine the effects of varying O2 concentrations, oxidant velocity (Vo) levels, and bio-oil proportions on the combustion characteristics of the bio-oil/kerosene mixtures. The results indicated that when the O2 concentration was 30% and the liquid fuel flow rate in the spray combustor was fixed, the flame associated with the spray combustion of pure kerosene decreased in length and increased in luminosity as Vo increased; moreover, the flame temperature increased. When Vo ¼ 5.53 m/s, this phenomenon was more visible when the bio-oil was added to the kerosene. When the bio-oil proportion was 15% and Vo ¼ 3.87 m/s, the flame luminosity increased; however, the flame luminosity decreased when Vo exceeded 3.87 m/s. When the O2 concentration reached 40%, the length, luminosity, and temperature of the flame increased; nevertheless, when Vo ¼ 5.53 m/s, the flame temperature decreased. The effect of the bio-oil proportion was not apparent. Because the bio-oil contained more volatile substances and O2 than did the kerosene, the combustion efficiency of the bio-oil-fossil fuel mixtures varied according to the bio-oil proportion and O2 concentration. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Biomass Fast pyrolysis Kerosene Bio-oil Oxycombustion Spray

1. Introduction Fast pyrolysis bio-oils are completely different from petroleum fuels and other bio-fuels available in the market, as regards both to their physical properties and chemical composition [1]. Bio-oil is produced through the fast pyrolysis of biomass, and its composition varies with the condensation temperature in the fast pyrolysis process [2e7]. Bio-oil can be categorized into two phases, oily and aqueous [8]. Bio-oil contains a higher amount of volatile biomass substances when condensed at a lower temperature. Therefore, all the aforementioned factors affect the efficiency of bio-oil in power machines. Yang et al. [8] produced bio-oil through the fast pyrolysis of various biomasses, and they categorized the phase of the produced bio-oil into two types, namely oily and aqueous phases, according to its condensation temperature; this categorization was based on the varying biomass compositions during the pyrolysis process. Ferdous et al. [9], Wang et al. [10], and Raveendran et al. [11,12] have indicated that the compositions, pyrolysis temperature, and condensation temperature of biomass affect the characteristics

* Corresponding author. Tel.: þ886 5 6315432; fax: þ886 5 6312110. E-mail address: [email protected] (S.I. Yang). http://dx.doi.org/10.1016/j.energy.2017.03.084 0360-5442/© 2017 Elsevier Ltd. All rights reserved.

of bio-oil including its heating value, viscosity, boiling temperature, and pH value. In addition, the stability of bio-oil storage is associated with biomass compositions [13]. Lappas et al. [14] used catalysts to improve the quality of bio-oil relative to that of fossil fuel products. Currently, bio-oil is typically mixed with fossil fuel and subjected to spray combustion for application in power plants. Because bio-oil and fossil fuel differed in their compositions, they must be emulsified to ensure that they are mixed appropriately for spray combustion. Stamatov et al. [15], Nguyen and Honnery [16], and Zheng and Kong [17] have investigated the spray combustion characteristics of bio-oil with different biomass contents when mixed with fossil fuels. Because the combustion of bio-oil is difficult, the bio-oil was added to flammable polar additives or mixed with fossil fuels such as alcohol, kerosene, diesel, and heavy oil. The results revealed that the flames produced by the bio-oil were shorter and brighter because of the flammable polar additives [15]. After an increase in ambient pressure, soot was produced in the outer ring of the spray combustion flame [16]. The exhaust gases emitted by the spray combustion of the bio-oil/fossil fuel mixture contained carbon monoxide (CO), nitrogen oxide (NOx), and sulfur oxide (SOx) [17]. All of these results are related to the characteristics of bio-oil such as its heating value, composition, and moisture.

S.I. Yang et al. / Energy 126 (2017) 854e867

Nomenclature B C CD D E F h J k kt K p Pr r Re S Sct T Y Vo Z

constant constant drag coefficient mass diffusivity energy body force enthalpy of ideal gases diffusion flux of species thermal conductivity turbulent thermal conductivity wave number pressure turbulent Prandtl number diameter of droplet Reynolds number mass source vaporized effective Schmidt number for the turbulent flow temperature mass fraction of species oxidant velocity Ohnesorge number

Mailboom and Tauzia [18] applied bio-oil generated through the fast pyrolysis of wood in a Petter AVB test engine. Yang et al. [19] emulsified bio-oil generated through the fast pyrolysis of cedar with diesel, and they tested its performance in a single-cylinder diesel engine. In a previous study, the O2 concentration in oxidants was increased to change the diffusion and equivalence ratios of reactants, and this increase influenced the equivalence ratios, temperatures, burning velocities, and stability of the flame [20]. Numerous studies have explored the combustion characteristics of coal and biomass through O2-rich combustion in boilers [21e26]. Tan et al. [21] employed a 0.8-MWth pilot-scale oxy-fuel-fired circulating fluidized bed (CFB) to convert an air-fired operation to an oxy-fuel-fired operation, revealing that the CO2 concentration exceeded 90% after the oxy-fuel-fired condition was stabilized. Additionally, the desulfurization rate was associated with the characteristics of the fuel and the temperature of combustion and
n ~ ez et al. [22] thereby affected the mode of combustion. Lupia conducted an experiment to examine the effects of the bed temperature and O2 concentration on SO2 and NOx emissions from the coal oxy-firing process in bubbling fluidized bed combustors. Through numerical simulation and experimental measurement, Dual et al. [23,24] adopted numerical simulation and experiment measurement to conduct a flue gas recycling operation on a 50kWth CFB combuster, testing various types of fuel and examining the conditions for stable oxy-CFB operations in consideration of the unstable heat transfer, mass transfer, and chemical reactions in the gasesolid phase. The results indicated that the combustion process with 22.2%e23.4% O2 concentration yielded higher carbon burnout than did that of normal air combustion, thus improving the desulfurization efficiency. Moreover, the fuel nitrogen conversion ratio in oxy-fuel was much lower than in air combustion. Krzywanski et al. [25,26] established a mathematical model to predict the SO2 emission processes of large and small CFB boilers in air and O2-rich combustion. They revealed that the flame temperature of

We

855

Weber number

Greek symbols time scale dynamics viscosity density, kg/m3 surface tension stress L wave length U frequency v droplet velocity

t m r s t

Subscript d ε g i k l m t KH RT

droplet dissipation rate gas phase species turbulence kinetic energy liquid mass turbulent KevineHelmholtz model RayleigheTaylor model

the powdered coal and air decreased significantly when the O2 concentration reached 21% and that the coal was fully combusted; when the O2 concentration exceeded 21%, an increased proportion of CO and NOx was generated in the primary combustion zone, and burnt char particles were produced around the flames. However, liquid biofuel exhibits a high moisture content and low heating values, and it must be mixed with petroleum fuel, which features higher heating values, to improve its combustion characteristics. Because liquid biofuel and fossil fuel are almost incompatible, emulsifiers are typically required to mix these two fuel types. 2. Approaches This study applied experimental measurements [27] and numerical simulations [28] to investigate the combustion characteristics of mixtures of kerosene and a bio-oil, which was produced through the pyrolysis of cedar, at various O2 concentrations. Fig. 1 illustrates the burner employed in this study, indicating that it comprises a fuel inlet, an atomizer having two tangential openings, and a flame holder. The liquid fuel and oxidant coaxially entered the burner through the fuel inlet, and the fuel was released from the burner through the two tangential openings of the atomizer, collided to form droplets, and mixed with the oxidant. The fuel dropleteoxidant mixture was then combusted through the flame holder, preventing flashbacks and stabilizing the flame. 2.1. Experimental methods Fig. 2 depicts the spray combustion test platform adopted in this study for examining the combustion characteristics of kerosene mixed with the cedar bio-oil. This platform comprised the spray-atomized combustor (Fig. 1), fuel supply systems, an image capture system, and temperature and emission analyzers. The

856

S.I. Yang et al. / Energy 126 (2017) 854e867

Burner

Flame holder

Atomizer =0.3mm

O2 /N2

Therefore, this study applied various emulsifier types to the bio-oil and kerosene [19,27]; in addition, liquid fuel samples involving various bio-oil/kerosene ratios were mixed for the experiment. These samples were transferred into the burner by using the oil pump, and they were atomized through the inlet and nozzle. The oxidant was produced in the mixing chamber by using oxygen and nitrogen cylinders with the required concentration levels, and it subsequently entered the burner and mixed with the fuel droplets for combustion. The characteristics of the cold flow field of the atomizer included the droplet flow field, spray angle, and size distribution, and these characteristics were measured using particle image velocimetry, direct imaging, and a laser diffraction particle-size analyzer (LDPA), respectively. The burner was placed inside a combustion chamber comprising four glass sides. Thermocouples were attached to one of the sides to measure the temperature distribution of the flames, and the exhaust gas emissions were assessed at the top of the chamber. The flame images were captured using a high-speed camera, and an optical filter with varying wave bands was used to acquire images of the flame chemiluminescence [27].

2.2. Numerical simulation methods

Bio-oil/Kerosene Fig. 1. Structure of the spray-atomized burner.

bio-oil and kerosene were emulsified for spray combustion. Moreover, the bio-oil used in this study was oily-phase bio-oil produced through the fast pyrolysis of cedar [8]; it exhibited a moisture content of less than 5% and a low heating value of 21.78 MJ/kg. In addition, bio-oil and petroleum fuels differed mainly their hydrophilicity, hydrophobicity, as well as viscosity.

This study integrated Fluent software, a commercially available package for computational fluid dynamics, with a discrete phase model (DPM) and 1D spray model to simulate the spray combustion process. The droplets were considered multicomponent fuel, and two reactions were established to represent the kerosene and bio-oil combustion reactions. To simplify the calculation processes without affecting the physical spray combustion characteristics, the following assumptions were established: (1) The flow in the calculation process was an incompressible flow, and the substances in each controlled volume were evenly distributed; (2) the gravity effect is negligible; (3) the flow field in the calculation was steady; (4) the flow at the boundaries exhibited a nonslip condition; and (5) the droplets were uniformly spherical, and the surrounding pressure was evenly distributed. The spray combustion calculation involved discrete analyses of continuous gas flow fields and droplet sprays; the continuous conservation equations comprised mass, momentum, and energy conservation. The governing equations

Exhaust-gas Analyzer

Computer

Temperature Measure System High-speed camera K-type Thermocouples Power Supply

Burner

+ " #$%&'()*

Oil Tank

! " #$%&'()* Filter

Oil Pump

Flow meter Fig. 2. Spray combustion test platform.

S.I. Yang et al. / Energy 126 (2017) 854e867

and models of this study were established on the basis of these assumptions [28]. 2.2.1. Continuity equation The mass change in each substance in the controlled volume and the subsequent gas-phase mass generated by the droplets in each calculation grid must be balanced.

! ! V $ðr y Þ ¼ Sm

857

equations, which are shown as follows:

   m ! ! ! V $ðrk! y Þ ¼ V $ m þ t V k þ mt S2  rε

(8)

sk

   m ! ! ! y Þ ¼ V $ m þ t V ε þ rC1 Sε  rC2 V $ðrε!

sk

ε2 pffiffiffiffiffi k þ nε

(9)

(1)

where Sm represents the mass source of the atomized discrete phase (droplet). 2.2.2. Momentum equation The momentum equation is expressed as follows:

! ! ! ! !! V $ðr y y Þ ¼  V p þ V $t þ F

(2)

where p represents the static pressure, t represents the stress ! tensor, and F represents the force generated by the interaction between the discrete phase (liquid) and the continuous phase (gas). In a high-momentum spray, gravity can be omitted from the equation.

2.2.6. Discrete phase model The discrete phase model is based on the EulereLagrange method, and it is used to simulate droplet tracks in spray combustion processes. The calculation process in this model involves two phase types: discrete and continuous phases. Therefore, in this study, droplets were placed in the continuous phase for them to interact. Yang et al. [27,28] maintained that under this operating condition, an atomizer typically breaks a liquid thoroughly in the primary breaking area. Hence, the liquid is assumed to be thoroughly atomized upon being released through the outlet orifice. The droplet motion tracks are calculated by summing all forces exerted on the droplets as follows:

2.2.3. Energy equation

  d! yp ¼ FD ! y ! yp dt

3 2 X ! ! ! ! 4 ! ! V $½ y ðrE þ pÞ ¼ V $ ðk þ kt Þ V T  hj Jj þ ðt$ y Þ5

The left side represents the droplet acceleration; moreover, FD represents the resistance encountered by the droplet and is expressed as follows:

(3)

(10)

j

The terms in the square brackets on the right side of this equation denote the conduction, species diffusion, and viscous dissipation, respectively. The energy E is defined as follows:

p ! y $! y E ¼h þ r 2

(4)

where h represents the ideal gas enthalpy value, which is the product of the enthalpy value and the mass fraction of each species.

h¼

X

Yj hj

(5)

FD ¼



rdp up  uj Rep ¼ m CD ¼

! ! ! ! V $ðr y Yi Þ ¼  V $ ji þ Si

(6)

where Si represents the source item generated by the atomization of the droplets. The diffusion-induced species transport in turn ! generates the diffusion flux Ji :

  ! m ! Ji ¼  rDi;m þ t V Yi Sct

(7)

where Di;m represents the diffusion coefficient of the ith species in the mixture and mt represents the turbulence dynamic viscosity coefficient. 2.2.5. Turbulence equations The turbulence field was calculated using the Stander k-ε

(11)

The resistance coefficient CD can be calculated using the dynamic resistance model, which involves the effect of droplet deformation as well as the effect of linear resistance on the droplet. The Reynolds number and CD of the droplet are calculated as follows:

j

2.2.4. Species transport equations Spray combustion involves two types of species: gas phase (oxidant) and liquid phase (fuel). The fuel is atomized into gases and mixed with the oxidant. The transport equation is as follows:

18m CD Re rp d2p 24

  24 1 2=3 1 þ Rep Rep 6

(12)

(13)

2.2.7. Broken mechanism When a droplet passes through the calculated range, the flow field breaks it into smaller droplets that collide with one another. The break-up mechanism occurs when the gas flow field breaks a parent droplet into smaller droplets; this mechanism is divided into two categories. The first category involves the shear stress generated by the velocity difference between a liquid and a gas, and this stress causes a KelvineHelmholtz (KH) instability and breaks the droplets into smaller droplets. The second category involves RayleigheTaylor instability, in which two flows with different densities exist in a flow field; the flow with the lower density pushes the one with the higher density. In this study, both types of break-up mechanism were observed in flow fields. 2.2.7.1. KelvineHelmholtz model. The KH model is based on the KH instability proposed by Reitz [29]. Assume that a cylindrical liquid jet enters a stabilized and incompressible gas flow; the wavelength (LKH) and frequency (UKH) of the most rapidly generated growth

858

S.I. Yang et al. / Energy 126 (2017) 854e867

wave on the liquid surface are expressed as follows:

LKH ¼

UKH

9:02ro

 pffiffiffi  1 þ 0:45 Z ð1 þ 0:4ðTÞÞ0:7  0:6 1 þ 0:8654We1:67 g

Table 1 Experimental conditions.



0:34 þ 0:385We1:5 ¼

 ð1 þ ZÞ 1 þ 1:4T 0:6

Experimental conditions

(14)

sffiffiffiffiffiffiffiffiffi

s rl r2o

(15)

where Weg ¼ rw2r0/s represents the Weber number of the gas phase, s represents the surface tension constant, Z ¼ (Wel)0.5/Rel represents the Ohnesorge number of the liquid phase, Wel ¼ rlw2ro/ s represents the Weber number of the liquid, Rel ¼ rlwro/ml represents the Reynolds number of the liquid phase, and T ¼ Z (Weg)0.5 represents the Taylor number. When a child droplet is formed on the surface of a parent droplet, the child droplet is expressed as follows according to the KH model:

rKH ¼ BO LKH

dr ro  rKH ¼ ; rKH  ro dt tKH

(17)

where tKH represents the KH time scale and is calculated as follows:

3:788B1 ro UKH LKH

(18)

2.2.7.2. RayleigheTaylor model. When the fuel is atomized at a high initial jet speed, the velocity of the droplets is reduced by drag. Therefore, when a liquid film breaks, with the viscosity effect being disregarded in the RayleigheTaylor (RT) model, the frequency (URT) of the most rapidly generated growth wave on the surface of the liquid, number of waves corresponding to the KH model (KRT), and wavelength (LRT) are respectively expressed as follows:

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u h  i3 2 u u 2 a r  r l g t pffiffiffi ¼ rl þ rg 3 3s =

URT

KRT ¼

LRT

3.87, 4.42, 4.97, 5.53 0.038 0/30(B0O30),5/30(B3O30),10/30(B10O30), 15/30(B15O30),0/40(B0O40), 5/40(B3O40), 10/40(B10O40),15/40(B15O40)

bio-oil created in Ref. [8], various bio-oil proportions were added to the kerosene samples to create multicomponent and evenly mixed droplets; the droplets were subsequently subjected to separate, double-component fuel reactions, as shown in the following equations: Kerosene: C12H23 þ 17.75O2 / 12CO2 þ 11.5H2O

(R1)

Bio-oil: C11H132O8.25 þ 39.88O2 / 11CO2 þ 66H2O

(R2)

(16)

The change in the droplet particle diameter over time as the parent droplet breaks into child droplets is expressed as follows:

tKH ¼

Oxidizer Flow Velocity, Vo (m/s) Fuel flow rate , Vf(Liter/min) Bio-oil content in fuel/Oxygen content in oxidant

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi u  ua r  r t l g 3s

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 3s u  ¼ CRT 2pt  a rl  rg

tRT ¼ CT =URT

(19)

(20)

(21)

(22)

where a ¼ 3CDrgw2/(8rlr) represents the acceleration of the droplet in its motion direction, and CRT and CT respectively represent the model constant and breaking time controlling the size of the child droplet. 2.2.8. Chemical reactions The atomized liquid was vaporized through heating and mixed with the oxidant for combustion. Regarding the composition of the

3. Results and discussion To examine the spray combustion characteristics of the emulsified bio-oilekerosene mixtures, the oily-phase bio-oil generated through the fast pyrolysis of cedar [8] was emulsified with kerosene, atomized into droplets, and mixed with the oxidant for combustion. Tween80 and span80 were added to the kerosene and bio-oil samples by 1e1.5 vol%; the fuel samples were then proportionally mixed and heated to 30 C-40  C with ultrasound for approximately 2 min [27]. Experimental measurements and numerical simulations were conducted to explore the spray combustion characteristics of bio-oil/kerosene mixtures with a fixed fuel flow velocity and varying oxidant contents (O2eN2 ratios) and flow rates (Table 1). 3.1. Experimental results 3.1.1. Direct imaging of the flames This study estimated that when the O2 concentration was raised to 30%, the flame length decreased as oxidant velocity (Vo) increased (Fig. 3a). In addition, when the bio-oil proportion was increased, the flame luminosity dropped, decreasing as Vo increased; the flame shape also became wrinkled (Fig. 3bed). When Vo increased, the spray angle widened, and the number of droplets in a unit area decreased. Moreover, increasing Vo caused the shear stress of the droplets and oxidant to expand, further enhancing the mixing of the oxidant and droplets. Raising the O2 concentration also enhanced the mixing process and increased the combustion intensity. When the bio-oil proportion was increased, the heating value of the fuel decreased, reducing the spray combustion intensity. When the O2 concentration was raised to 40%, the flame length shortened, and it decreased as Vo and the bio-oil proportion increased; furthermore, the flame intensity decreased, and the flame shape became increasingly wrinkled. The direct imaging results of the spray combustion flame indicated that when the O2 concentration was increased (Fig. 4aed), the turbulence in the flow field enhanced the mixing of the oxidant and the fuel, increased the flame luminosity, and reduced the mixing distance and flame length. Moreover, when Vo increased, the spray angle widened, the number of droplets in a unit area decreased, and the vapor concentration of the fuel decreased. Consequently, the mixing and combustion of the oxidant and fuel vapor were enhanced, and the flame length shortened. In addition,

S.I. Yang et al. / Energy 126 (2017) 854e867

859

Fig. 3. Images of the kerosene flames with an O2 concentration of 30% and various bio-oil proportions: (a) B0O30; (b) B5O30; (c) B10O30; (d) B15O30.

adding the bio-oil lowered the heating value of the fuel. Bio-oil is produced from biomass through fast pyrolysis processes and flash quench; hence, it contains numerous short-chain components and high water content. These components were the first to be vaporized during the heating of the bio-oil droplets, and the component vapor exhibited a low molecular weight; compared with conventional bio-oil vapors [30], the vapor of these components demonstrated a higher mass diffusivity, thereby enabling it to mix and combust more easily with oxygen. This phenomenon became more evident as the O2 concentration increased, and the difference in the mass diffusivity caused the flame to wrinkle. 3.1.2. Flame lengths and liftoffs Fig. 5 illustrates the flame lengths and liftoffs at various bio-oil proportions and O2 concentrations. Each datum point represents the mean value of 1000 images, with the error bars denoting the root-mean-square value of the 1000 images. When the oxidant was composed of 30% O2 þ 70% N2, the flame length decreased as Vo and

the bio-oil proportion increased (Fig. 5a). A similar phenomenon was observed when the O2 concentration reached 40% (Fig. 5b). The liftoff of the turbulent flame was associated with the fuel velocity at the nozzle exit, and the turbulence dissipation rate affected the liftoff height. In this study, the fuel flow velocity was fixed; however, because Vo did not affect the flow field of the atomized droplets, the liftoff height remained unchanged as Vo increased (Fig. 5a and b). When the O2 concentration was 30%, the liftoff height rose after the bio-oil proportion was increased. Nevertheless, when the O2 concentration reached 40%, the liftoff height was not affected by the bio-oil proportion. This is because droplets that exited the nozzle and vaporized in the thermal radiation of the downstream flame were mixed with the oxidant. Therefore, when the bio-oil proportion was increased, the mixing efficiency of the droplet vapor decreased and the liftoff height increased. However, when the O2 concentration was raised to 40%, the intensity of the downstream flame increased, and the thermal radiation encountered by the droplets in the upstream

Fig. 4. Images of the kerosene flames with an O2 concentration of 40% and various bio-oil proportions: (a) B0O40; (b) B5O40; (c) B10O40; (d) B15O40.

860

S.I. Yang et al. / Energy 126 (2017) 854e867

120

300 60

B0O30 B5O30 B10O30 B15O30

150

Temperature (K)

90

Lift off length (mm)

450

Flame Length (mm)

2500

(a) 30%O2/70%N 2

(a) 30%O2/70%N 2

B0O30 B5O30 B10O30 B15O30

2250

2000

1750

30 1500

0 3.5

4

4.5

5

5.5

3

6

3.5

4

4.5

5

5.5

6

Oxidant velocity, V o (m/sec)

Oxidant velocity, V o (m/s) 120

(b) 40%O2/60%N 2 B0O40 B5O40 B10O40 B15O40

300

90

60 150

Temperature (K)

2500

Lift off length (mm)

Flame Length (mm)

450

B0O40 B5O40 B10O40 B15O40

(b) 40%O2/60%N 2

2250

2000

1750

30 1500

0 3.5

4

4.5

5

5.5

6

Oxidant velocity, V o (m/sec) Fig. 5. Length of kerosene flame and liftoff at different O2 concentrations and with various bio-oil proportions.

increased. Because the droplet vapor was thoroughly mixed with oxygen, the liftoff height remained unaffected by bio-oil proportions.

3.1.3. Effects of the O2 concentrations and bio-oil proportions on the flame temperature Increasing the flame temperature reduced the flame length of the bio-oil/kerosene mixture. The maximal temperature obtained during the determination of the flame temperature distribution was set as the flame temperature. When Vo increased, the flame temperature also increased; when Vo ¼ 4.97 m/s, the flame temperature peaked. Moreover, when Vo was increased up to 5.53 m/s, the flame temperature dropped to approximately 1690e1780 K. When the O2 concentration reached 40% (Fig. 6b), the flame temperature increased. Similarly, adding bio-oil caused the flame temperature to rise. When Vo ¼ 5.53 m/s, the flame temperature reached approximately 1680e1770 K. According to these results, increasing the flame temperature promoted the combustion intensity, increasing Vo shortened the distance required to mix the oxidant and the fuel, and increasing the O2 concentration expanded the flame length and increased the combustion intensity. Because of production and raw material factors, the bio-oil featured a different boiling point. In this study, cedar served as the source for fast pyrolysis. In addition to emulsification-induced

3

3.5

4

4.5

5

5.5

6

Oxidant velocity, V o (m/s) Fig. 6. Temperature of the kerosene flame at various O2 concentrations and with various bio-oil proportions.

microexplosions, volatile substances in the droplets were released earlier because of the environmental temperature effect, causing them to mix with the oxidant and combust [27,28,30]. Therefore, the liftoff length increased after the bio-oil proportion was increased. Nevertheless, because increasing the O2 concentration intensified the combustion, the thermal radiation released by the downstream flames affected the vaporization of the upstream droplets. Increasing the O2 concentration also shortened the liftoff length at the nozzle. When Vo ¼ 5.53 m/s, the oxidant flow field shortened or quenched the flame, reducing the measured temperatures. 3.1.4. Effects of the O2 concentrations and bio-oil proportions on exhaust gas emissions Increasing the O2 concentration changed the proportions of the emitted exhaust gas. Increasing the O2 concentration while keeping Vo constant increased the dropleteoxygen mixture and reduced the equivalence ratio. Fig. 7 illustrates the exhaust gas emissions of kerosene with different O2 concentrations and bio-oil proportions. To compare the effects of the O2 concentrations and Vo on the gas emissions, the amount of emissions was adjusted to a level corresponding to a 6% O2 concentration. When the O2 concentration reached 30%, the CO emitted by the spray combustion process decreased as Vo increased. Moreover, when Vo ¼ 4.97 m/s, the

S.I. Yang et al. / Energy 126 (2017) 854e867 B0O30 B5O30 B10O30 B15O30

Emission (ppm) corrected to 6% O2

Emission (%) corrected to 6% O2

2.5 2 1.5 1

(i) CO

25 20 15

(ii) CO2

125 100 75 50

(iii) HC

(a) 30%O2/70%N 2

200 150 100

(iv) NO

3.5

4

4.5 5 Oxidant velocity, V o (m/sec)

5.5

Emission (ppm) corrected to 6% O2

Emission (%) corrected to 6% O2

2.5

6

B0O40 B5O40 B10O40 B15O40

2 1.5 1

(a) CO

25 20 15 10

(b) CO2

125 100 75 50

(c) HC

200

(b) 40%O2/60%N 2

150 100

3.5

(d) NO 4

4.5 5 Oxidant velocity, V o (m/sec)

5.5

6

Fig. 7. Exhaust gas emissions of the kerosene at different O2 concentrations and containing various bio-oil proportions.

minimal CO concentration was 1.01%; when Vo ¼ 5.53 m/s, the CO concentration was 1.48%, and the amount of CO emissions slightly increased with the bio-oil proportion (Fig. 7a). Conversely, the CO2 emitted by the spray combustion process increased with Vo; when Vo ¼ 4.97 m/s, the CO2 concentration was 22.10% (B0O30). When Vo ¼ 5.53 m/s, the CO2 concentration dropped to 15.62% (B0O30), and the amount of CO2 emissions increased with the bio-oil proportion (Fig. 7a). The trend of hydrocarbon (HC) emissions was consistent with that of CO emissions, and the trend of NO emissions was identical to that of CO2 emissions. When the O2 concentration was increased to 40%, the changes in the emission rates remained the same (Fig. 7b); however, the emissions of CO, CO2, HC, and NO decreased slightly. These results reveal that increasing the O2 concentration in the oxidant reduced the equivalence ratios in

861

certain spray combustion areas. Consequently, the flame shortened and widened, and its luminosity and temperature increased; raising Vo increased the spray combustion intensity. Therefore, when Vo was increased, the CO and HC concentrations decreased; by contrast, the CO2 and NO concentrations increased. When Vo ¼ 5.53 m/s, the oxidant flow velocity increased, lowering the flame length and luminosity (Figs. 4 and 5); hence, the CO and HC concentrations increased, and the CO2 and NO concentrations decreased. Increasing the O2 concentration in the oxidant thus raised the flame temperature in the middle of the spray combustion flow and facilitated achieving a more thorough combustion. The CO and HC concentrations also increased with Vo. Furthermore, the bio-oil composition was complex and related to its production process. Because the bio-oil contained numerous oxygenated organic compounds and water molecules in addition to exhibiting a low boiling point, increasing its proportion enhanced droplet vaporization in the spray combustion process. The microexplosions as well as reduced O2 requirement shortened the flame lengths and liftoff and increased the flame luminosity. These results indicate that adding bio-oil to the fuel increased its combustion efficiency in addition to increasing the CO, HC, CO2 and NO concentrations. Because the fuel mixture contained O2 molecules, the effect of the O2 concentration in the oxidant on the combustion efficiency was not apparent. Increasing the bio-oil proportion lowered the heating value of the fuel mixture. Although the microexplosions promoted the combustion, the moisture of the bio-oil reduced the heating value of the fuel and lowered the combustion efficiency. Moreover, at a high Vo, the length and combustion efficiency of the flame were reduced by the effect of turbulence stress, consequently increasing the HC and CO concentrations and reducing the CO2 and NO concentrations. 3.1.5. Effects of the O2 concentrations and bio-oil proportions on the chemiluminescence of the combustion flow field A high-speed camera (phantom v7.3, frame rate: 2000 frames/ sec) and an optical filter with various wave bands were employed to record the C*2 (532 ± 5 nm) and CH* (430 ± 5 nm) chemiluminescence of the kerosene combustion attained using various O2 concentrations and bio-oil proportions. The images were then digitalized into distribution charts (Figs. 8 and 9). Fig. 8 shows the chemiluminescence distributions of C*2 (left) and CH* (right) at a 30% O2 concentration and various bio-oil proportions. When no bio-oil was added to the kerosene, the flame length decreased as Vo increased (Fig. 8a). The chemiluminescence distribution observed when Vo ¼ 4.42 m/s was shorter and wider than that observed when Vo ¼ 3.87 m/s, and the C*2 signal became particularly strong at the flame center. When Vo ¼ 4.97 m/s, the chemiluminescence distribution shortened and narrowed, and when Vo ¼ 5.53 m/s, the distribution weakened, shortened, and narrowed further. Moreover, when 5% bio-oil was added to the kerosene (Fig. 8b) and Vo ¼ 3.87 m/s, a strong but discontinuous C*2 signal appeared at the flame center, and a discontinuous CH* signal was observed. When Vo ¼ 4.42 m/s, C*2 became more intense at the flame center, and CH* became narrower in range but more intense. Furthermore, when Vo ¼ 4.97 m/s, C*2 and CH* became narrower in range, and their intensity levels demonstrated the lowest change; when Vo ¼ 5.53 m/s, C*2 and CH* narrowed further and became discontinuous, weakening or quenching the flame. When 10% biooil was added to the kerosene and Vo ¼ 3.87 m/s (Fig. 8c), C*2 and CH* shortened and narrowed; however, C*2 became relatively strong at the flame center, whereas CH* became weak and discontinuous. When 15% bio-oil was added to the kerosene and Vo ¼ 3.87 m/s (Fig. 8d), C*2 became strong only at the center, and CH* became weak. The C*2 and CH* signals decreased as Vo increased, and when Vo ¼ 5.53 m/s, the signals weakened to as low as zero, indicating

862

S.I. Yang et al. / Energy 126 (2017) 854e867

Fig. 8. C*2 and CH* chemiluminescence distributions of the kerosene at a 30% O2 concentration and containing various bio-oil proportions.

that the flame was almost fully quenched. When the O2 concentration was raised to 40% and Vo ¼ 3.87 m/s (Fig. 9a), the C*2 and CH* signals at the center of a pure kerosene flame were shorter and more concentrated compared with those observed when the O2 concentration was 30%. In addition, when the O2 concentration was maintained at 40% while Vo increased, these signals decreased more significantly compared with those observed when the O2 concentration was 30%. Adding 5% bio-oil to the kerosene (Fig. 9b) enhanced the C*2 and CH* signals, and the intensity of these signals increased with Vo; furthermore, when Vo ¼ 4.97 or 5.53 m/s, the intensity levels of the signals decreased. The C*2 and CH* signals also weakened further when the bio-oil proportion was 10% or 15% (Fig. 9c and d). These results indicate that raising the O2 concentration strengthened the C*2 and CH* signals of the pure kerosene flame;

moreover, the combustion intensified, the upstream droplets became preheated because of the increased flame temperature, and the vaporization of the droplets accelerated and increased the C*2 and CH* signals. Therefore, when Vo ¼ 3.87 m/s, increasing the O2 concentration promoted the combustion efficiency of the pure kerosene. However, when Vo ¼ 4.42, 4.97, or 5.53 m/s, increasing the O2 concentration reduced the C*2 and CH* signals. When Vo exceeded 4.97 m/s, the burning velocity increased, the flame length shortened, and the combustion reaction zone shifted toward the upstream direction. Regarding the bio-oil/kerosene mixture, increasing the bio-oil proportion and O2 concentration shifted the intense C*2 and CH* signals toward the upstream direction, signifying that the combustion reaction zone moved upstream. Because the bio-oil contained numerous O2 molecules, when the O2 concentration in the oxidant was fixed, appropriately adding bio-oil to

Fig. 9. C*2 and CH* chemiluminescence distributions of the kerosene at a 40% O2 concentration and containing various bio-oil proportions.

S.I. Yang et al. / Energy 126 (2017) 854e867

863

Fig. 10. Spray combustion flow fields at a 30% O2 concentration and with various bio-oil proportions.

the kerosene reduced the equivalence ratio and improved the combustion efficiency; increasing the O2 concentration rendered this phenomenon more apparent. When Vo ¼ 5.53 m/s, the C*2 and CH* reaction zones became scattered because the turbulence in the flow field engendered a tensile expansion of the flame, thus quenching it. Furthermore, the lower heating value of the bio-oil rendered this phenomenon more visible. However, increasing the O2 concentration had a nonsignificant effect on this phenomenon.

3.2. Numerical simulation results 3.2.1. Effects of the O2 concentrations and bio-oil proportions on the combustion flow fields The effects of various O2 concentrations, bio-oil proportions, and oxidant velocities on the spray combustion characteristics were simulated numerically. Fig. 10 illustrates distribution diagrams demonstrating the effects of Vo on the temperature, R1 reaction, and heat of reaction associated with the spray combustion of pure kerosene involving a 30% O2 concentration. On these diagrams, the

circles represent the droplet distribution. When Vo increased, the high-temperature zone of the spray combustion developed toward the upstream direction, and the R1 reaction intensified (Fig. 10a, b, c). However, when Vo ¼ 4.42 m/s, the intensity of the R1 reaction decreased; when Vo ¼ 5.53 m/s, the R1 reaction intensity reached its nadir (Fig. 10d). As illustrated in the distribution diagrams of the heat of reaction, the intensity of the heat release decreased as Vo increased. These results reveal that increasing Vo significantly affected the kerosene combustion reaction. The effect of Vo on the chemical reactions of the kerosene with 5% bio-oil was numerically simulated. R1 and R2 represent the combustion reactions of the kerosene and bio-oil, respectively. When the bio-oil proportion was 5% and Vo ¼ 3.87 m/s, the R1 reaction slightly decreased in range but increased in intensity (Fig. 11a). When Vo increased, the range of the R1 reaction for the bio-oil decreased further, but the intensity of this reaction increased, becoming higher than that of the R1 reaction for the pure kerosene (Fig. 10). When the bio-oil proportion was increased to 15%, the R1 reaction decreased in range but increased in intensity.

864

S.I. Yang et al. / Energy 126 (2017) 854e867

Fig. 11. Spray combustion reaction flow fields at a 30% O2 concentration and with various bio-oil proportions.

The R2 reaction of the kerosene with bio-oil became increasingly intensified and concentrated following the increase in Vo. When Vo ¼ 4.97 m/s, the intensity of the R2 reaction increased, and when Vo ¼ 5.53 m/s, the intensity levels of both the R1 and R2 reactions became extremely low (Fig. 11a). The numerical calculations revealed that when the bio-oil proportion was 15% and Vo ¼ 3.87 m/s, the R1 and R2 reactions intensified and became concentrated in range; however, when Vo exceeded 3.87 m/s, the intensity levels of the R1 and R2 reactions decreased (Fig. 11b). These results are consistent with those of the direct imaging and chemiluminescence analyses. Increasing Vo widened the spray angle and enhanced the mixing of the oxidant and droplets. Moreover, raising the O2 concentration to 30% reduced the equivalence ratio; this phenomenon became increasingly apparent as Vo increased. Because bio-oil is generated through the fast pyrolysis and condensation of biomass, the composition and condensation temperature of the biomass influence the composition of the bio-oil. In particular, the condensation temperature is closely related to the properties of volatile substances in

the bio-oil [8], which affect bio-oil combustion processes [19]. The results of the current study reveal that adding the bio-oil to the kerosene shortened the flame length and increased its intensity. The volatile substances in the bio-oil vaporized earlier than did kerosene, thus activating the reaction of the kerosene, shortening the flame length, and intensifying the flame. However, because the bio-oil contained numerous O2 molecules, the equivalence ratio decreased when the O2 concentration and Vo were increased, thus lowering the flame intensity. The temperature of the pure kerosene flame observed when the O2 concentration reached 40% and Vo ¼ 3.87 m/s was higher than that observed when the O2 concentration was 30%. In addition, the R1 reaction became shorter and more concentrated in range and higher in intensity (Fig. 12 a,b). When Vo increased, the ranges of the R1 reaction and heat of reaction decreased, but their intensity levels increased. When Vo ¼ 4.97 m/s, the flame temperature, R1 reaction intensity, and reaction heat intensity peaked (Fig. 12c), but they decreased when Vo ¼ 5.53 m/s (Fig. 12d). This is because raising the O2 concentration increased the flame temperature;

S.I. Yang et al. / Energy 126 (2017) 854e867

865

Fig. 12. Spray combustion flow fields at a 40% O2 concentration and with various bio-oil proportions.

however, increasing Vo expanded the spray angle and enhanced the mixing of the oxidant and the spray combustion vapor, altering the equivalence ratio and constraining the flame intensity. According to the results, the increase in Vo increased the spray angle and caused oxidant to mix with fuel at the edge of the spray cone, thereby enhancing combustion efficiency. When Vo reached 5.53 m/s, oxidant mixed with fuels around the nozzle and therefore the flame length became shorter; the increase in turbulence intensity stretched the flame, thereby resulting in incomplete combustion and reducing combustion efficiency. In addition, the increase in O2 concentration shortened the flame length, and the effect of turbulence partially extinguished the flame and reduced combustion efficiency.

Fig. 13 (a, b) depict the changes in the R1 and R2 distributions of the kerosene containing 5% and 15% bio-oil when Vo increased. When the O2 concentration was 40%, adding bio-oil to the kerosene reduced the R1 and R2 reaction intensity levels (Figs. 11 and 13); moreover, when Vo increased, the R1 and R2 reaction zones quickly shifted upstream, shortening the flame length. These results are consistent with the experimental results. Furthermore, these results indicate that the volatile substances in the bio-oil provided an ignition mechanism required for the kerosene spray combustion. However, when the O2 concentration increased, the O2 molecules in the bio-oil shortened the flame length. The low nitrogen concentration in the oxidant reduced the dilution phenomenon and increased the flame temperature. When Vo ¼ 5.53 m/s, an increased

866

S.I. Yang et al. / Energy 126 (2017) 854e867

Fig. 13. Spray combustion reaction flow fields at a 40% O2 concentration and with various bio-oil proportions.

spray angle enhanced the evenness of the oxidantedroplet mixture and reduced the equivalence ratio to a level lower than that required for flammability, thus quenching the flame. Through the aforementioned experimental measurements and numerical simulations, this study clarified the effects of bio-oil proportions, oxidant concentrations, and Vo levels on the kerosene spray combustion characteristics. The spray combustion characteristics of the bio-oil, including its production process, O2 concentration, and Vo, affected the spray combustion characteristics of the kerosene. Hence, this study clarified the effects of bio-oil and oxidants on kerosene spray combustion and provides a reference for future bio-oil production processes and nozzle and boiler operations. 4. Conclusion This study investigated the combustion characteristics of kerosene at varying O2 concentrations, bio-oil proportions, and Vo levels. The kerosene was mixed with various bio-oil proportions to create different liquid fuel samples. The fuel was atomized into droplets through a burner at a fixed liquid fuel flow rate and was

mixed with an oxidant for combustion; the combustion characteristics of the fuel were measured and examined. This study derived the following findings: 1. Increasing the bio-oil proportion, O2 concentration, and Vo shortened or quenched the flame. Appropriately increasing the bio-oil proportion, O2 concentration, and Vo improved the combustion efficiency of the kerosene spray. 2. Increasing Vo enhanced the mixing of the fuel and oxidant and increased the turbulence intensity of the flow field, thereby improving the combustion efficiency. When Vo reached 5.53 m/ s, the flame underwent dilution and tensile stretching because of the large amount of oxidant, consequently reducing the combustion efficiency. 3. Adding an appropriate amount of bio-oil to the kerosene effectively improved its combustion efficiency. In the emulsification of the bio-oil and kerosene, the secondary atomization induced by the microexplosions and the bio-oil substances improved the fuel spray combustion characteristics. However, adding an excessive amount of bio-oil reduced the spray

S.I. Yang et al. / Energy 126 (2017) 854e867

combustion efficiency because of the moisture and low heating value of the bio-oil. 4. The numerical simulation results revealed that the volatile substances in the bio-oil were vaporized and mixed with the oxidant for combustion earlier than did the kerosene. Therefore, the R2 reaction, which represented the bio-oil combustion reaction, triggered the R1 reaction, resulting in a kerosene spray with improved combustion efficiency. Acknowledgments The authors thank the Ministry of Science and Technology, Taiwan, for financially supporting this research under MOST 1042221-E-150-037 and MOST 104-3113-E-008-001, and express their appreciation to Yu Hsuan Yeh and Ying Yi Lin for their assistance in conducting the experiments.

References [1] Lehto J, Oasmaa A, Solantausta Y, Kyto M, Chiaramonti D. Review of fuel oil quality and combustion of fast pyrolysis bio-oils from lignocellulosic biomass. Appl Energy 2014;116:178e90. [2] Ly HV, Kim SS, Woo HC, Choi JH, Suh DJ, Kim J. Fast pyrolysis of macroalgae Saccharin japonica in a bubbling fluidized-bed reactor for bio-oil production. Energy 2015;93(2):1436e46. € [3] Ozbay N, Apaydın-Varol E, Uzun BB, Pütün AE. Characterization of bio-oil obtained from fruit pulp pyrolysis. Energy 2008;33(8):1233e40. [4] Aysu T, Küçük MM. Biomass pyrolysis in a fixed-bed reactor: effects of pyrolysis parameters on product yields and characterization of products. Energy 2014;64:1002e25. [5] Ngo TA, Kim J, Kim SS. Characteristics of palm bark pyrolysis experiment oriented by central composite rotatable design. Energy 2014;66(1):7e12. [6] Pattiya A, Sukkasi S, Goodwin V. Fast pyrolysis of sugarcane and cassava residues in a free-fall reactor. Energy 2012;44(1):1067e77. [7] Nam H, Capareda SC, Ashwath N, Kongkasawan J. Experimental investigation of pyrolysis of rice straw using bench-scale auger, batch and fluidized bed reactors. Energy 2015;93(2):2384e94. [8] Yang SI, Wu MS, Wu CY. Application of biomass fast pyrolysis Part I: pyrolysis characteristics and products. Energy 2014;66:162e71. [9] Ferdous D, Dalai AK, Bej SK, Thring RW. Pyrolysis of lignins: experimental and kinetics studies. Energy Fuels 2002;16:1405e12. [10] Wang G, Li W, Li B, Chen H. TG study on pyrolysis of biomass and its three components under syngas. Fuel 2008;87:552e8. [11] Raveendran K, Ganesh A, Khilar KC. Pyrolysis characteristics of biomass and biomass components. Fuel 1996;75:987e98. [12] Raveendran K, Ganash A. Heating value of biomass and biomass pyrolysis

867

products. Fuel 1996;75:1715e20. [13] Fahmi R, Bridgewater AV, Donnison I, Yates N, Jones JM. The effect of lignin and inorganic species in biomass on pyrolysis oil yields, quality and stability. Fuel 2008;87:1230e40. [14] Lappas AA, Samolada MC, Iatridis DK, Voutetakis SS, Vasalos IA. Biomass pyrolysis in a circulating fluid bed reactor for the production of fuels and chemicals. Fuel 2002;81:2087e95. [15] Stamatov V, Honnery D, Soria J. Combustion properties of slow pyrolysis bio-oil produced from indigenous Australian species. Renew Energy 2006;31:2108e21. [16] Nguyen D, Honnery D. Combustion of bio-oil ethanol blends at elevated pressure. Fuel 2008;87:232e43. [17] Zheng JL, Kong YP. Spray combustion properties of fast pyrolysis bio-oil produced from rice husk. Energy Convers Manag 2010;51:182e8. [18] Mailboom A, Tauzia X. NOx and PM emission reduction on an automotive HSDI Diesel engine with water-in-diesel emulsion and EGR: an experimental study. Fuel 2011;90:3179e92. [19] Yang SI, Hsu TC, Wu MS, Chen KH, Hsu YL, Li YH. Application of biomass fast pyrolysis part II: the effects that bio-oil has on the performance of diesel engines. Energy 2014;66:172e80. [20] Hjartstam S, Andersson K, Johnsson F, Leckner B. Combustion characteristics of lignite-fired oxy-fuel flames. Fuel 2009;88:2216e24. [21] Tan Y, Jia L, Wu Y, Anthony EJ. Experiences and results on a 0.8 MWth oxy-fuel operation pilot-scale circulating fluidized bed. Appl Energy 2012;2012(92): 343e7. ~ ez C, Guedea I, Bolea I, Díez LI, Romeo LM. Experimental study of SO 2 [22] Lupi
an and NO x emissions in fluidized bed oxy-fuel combustion. Fuel Process Technol 2013;106:587e94. [23] Zhou W, Zhao C, Duan L, Chen X, Liu D. A simulation study of coal combustion under O 2/CO 2 and O 2/RFG atmospheres in circulating fluidized bed. Chem Eng J 2013;223:816e23. [24] Duan L, Sun H, Zhao C, Zhou W, Chen X. Coal combustion characteristics on an oxy-fuel circulating fluidized bed combustor with warm flue gas recycle. Fuel 2014;127:47e51. [25] Krzywanski J, Czakiert T, Blaszczuk A, Rajczyk R, Muskala W, Nowak W. A generalized model of SO 2 emissions from large-and small-scale CFB boilers by artificial neural network approach: Part 1. The mathematical model of SO2 emissions in air-firing, oxygen-enriched and oxycombustion CFB conditions. Fuel Process Technol 2015;2015(137):66e74. [26] Krzywanski J, Czakiert T, Blaszczuk A, Rajczyk R, Muskala W, Nowak W. A generalized model of SO 2 emissions from large-and small-scale CFB boilers by artificial neural network approach Part 2. SO 2 emissions from large-and pilot-scale CFB boilers in O 2/N 2, O 2/CO 2 and O 2/RFG combustion atmospheres. Fuel Process Technol 2015;139:73e85. [27] Yang SI, Wu MS, Hsu TC. Spray combustion characteristics of kerosene/bio-oil part I: experimental study. Energy 2017;119:26e36. [28] Yang SI, Hsu TC, Wu MS. Spray combustion characteristics of kerosene/bio-oil part II: numerical study. Energy 2016;115:458e67. [29] Reitz RD. Modeling atomization processes in high-pressure vaporizing sprays. At Spray Technol 1987;3:309e37. [30] Wu MS, Yang SI. Combustion characteristics of multi-component Cedar Biooil/kerosene droplet. Energy 2016;113:788e95.