Sunteți pe pagina 1din 8

Renewable Energy 129 (2018) 678e685

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Combustion characteristics of a 16 step grate-firing wood pellet boiler


Joon Ahn a, *, Jun Hwan Jang b
a
School of Mechanical Systems Engineering, Kookmin University, Seoul 136-702, South Korea
b
Korea Institute of Energy Research, Daejeon, 305-343, South Korea

a r t i c l e i n f o a b s t r a c t

Article history: A prototype of a 230 kW-class wood pellet boiler employing a four-step grate, was manufactured and
Received 7 February 2017 tested. The flame extended from the grate to the exit of the combustion chamber, and the combustion
Received in revised form efficiency was found to be unsatisfactory. In order to resolve this problem, the number of fire grates was
29 May 2017
increased to 16, and the combustion chamber was remodeled for performance evaluation. The flame was
Accepted 2 June 2017
Available online 6 June 2017
held on the grate, and as expected, the combustion efficiency improved. However, the last four fire grates
were not used even under an operation load condition of 120% of the design capacity, and the size of the
combustion chamber increased. To find improvements to this, computational fluid dynamics analysis
Keywords:
Wood pellet
using a homogeneous model was performed. The simulation results showed that the increase in the
Boiler aspect ratio of the combustion chamber contributed to complete combustion. However, a recirculation
Combustion region, not contributing to combustion, was observed at the lower end of the grate. Based on the results
Heat transfer of experiments and numerical analysis, a novel furnace design with a lower slope and a smaller number
of grates was proposed to reduce the volume of the combustion chamber while still achieving complete
combustion.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction anticipates that biomass consumption will increase significantly in


the near future [6,7]. Wood pellet boilers have been gradually
As fossil fuels such as oil and coal, are on the verge of depletion, installed over the last few years in Korea with the support of gov-
the use of renewable energy sources has become increasingly ur- ernment subsidies based on the standard regulated in 2011 [3,8].
gent [1]. Competitive renewable energy sources must be widely There are several combustion technologies available for biomass
available and economical. Biomass not only has considerable po- combustion such as grate-firing, fluidized-bed and pulverized
tential as a fuel source but is also cost-effective in comparison to combustion [6]. Grate-firing and fluidized bed are the two most
other renewable energies [2]. Wood pellet is a condensed uni- common types of boilers, because they can fire a wide range of fuels
formly sized form of biomass, designed to have better transport, with varying moisture content, and they require less fuel prepa-
storage and feeding capabilities compared with those of numerous ration [9]. In this study, a grate-firing (Stoker) boiler was developed,
other biomass fuels [3]. as shown in Fig. 1. To establish a well-proven design for the com-
Combustion systems that use wood pellets as a fuel can be bustion chamber, we developed a prototype of a grate-firing
applied to large energy consumption facilities such as boilers, in- combustor by modifying one for refuse-derived fuel [10].
dustrial furnaces, and drying furnaces. From the perspective of Compared with a refuse-derived fuel, wood pellets have a high
widely used distributed energy systems, boilers are considered to volatile matter content, with intermediate heavy hydrocarbons that
be relatively promising for wood pellet combustion. Wood pellet require more time and active mixing [3].
combustion systems have been applied for general public use in Experiments and computational fluid dynamics (CFD) simula-
boilers in Europe, wherein wood pellet fuel has been used previ- tions were conducted to identify the problems of the prototype
ously [4,5]. The U.S. Energy Information Administration also shown in Fig. 1(a) and to propose an improved design [10,11]. As for
the original prototype, the combustion in the gas phase was
delayed compared with the refuse-derived fuel so that the size of
the furnace was insufficient to achieve complete combustion even
* Corresponding author. during partial load operation. To resolve this problem, three
E-mail addresses: jahn@kookmin.ac.kr (J. Ahn), jhjang@kier.re.kr (J.H. Jang).

http://dx.doi.org/10.1016/j.renene.2017.06.015
0960-1481/© 2017 Elsevier Ltd. All rights reserved.
J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685 679

Fig. 1. Cross-section of the combustion chamber; (a) original prototype with a 4-step grate [10]; (b) novel design with 16-step grate (numbers represent temperature sensors).

measures were suggested. First, the number of grates could be simulation are discussed. In conclusion, further modifications to
increased with higher inclination to distribute wood pellets more the furnace are proposed to achieve complete combustion while
uniformly on the grates. Second, the aspect ratio of the furnace minimizing the volume of the furnace.
could be increased to achieve longer a residence time. Finally, a
baffle could be installed inside the combustion chamber, which
2. Experimental apparatus
would also increase the residence time.
Fig. 1(b) shows a novel combustion chamber reflecting all three
The experimental setup integrated the 16-step grate-firing
design points suggested above. The furnace was manufactured and
wood pellet furnace into the boiler system for the purpose of
built as a 230 kW-class boiler. The performance of the boiler was
measuring combustion characteristics and boiler efficiency. In this
measured at a test facility developed for a wood pellet boiler [8] for
section, modifications of the grate and combustion chamber will be
various operational loads and air flow rates. In this study, the per-
described in comparison with a previous 4-step grate version. The
formance of the novel wood pellet boiler is presented and
boiler was modified to be compliant with the newly developed
compared with that of the original prototype (Fig. 1(a)). Then, the
combustion chamber; its specifications would be introduced.
combustion characteristics inside the furnace are discussed based
Finally, the instruments used and the measured variables will be
on the experimental data at a combustion load of 230 kW.
explained.
A series of numerical simulations were also conducted for the
The cross-section of the combustion chamber in the present
novel combustion chamber. The simulation results are compared
study is shown in Fig. 1(b). The number of grates was increased
with the experimental data to validate the model. Then, the
from 4 to 16 with higher inclination so as to distribute the fuel more
detailed flow, thermal and chemical species fields from the
uniformly on the grates. They should maintain their shape at
680 J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685

temperatures above 1000  C; therefore, they were cast using NieCr Table 1
iron at a high temperature. Proximate composition analysis of the wood pellet
(wt%).
The shape of the combustion chamber was a rectangular
parallelepiped, and cylindrical combustion gas outlets were Moisture 8.3
installed in the upper part. However, the outlet was moved from Volatile Matter 73.7
Fixed Carbon 17.9
the center to the edge to increase the residence time. The residence Ash 0.1
time could also be increased by changing the aspect ratio of the
furnace. The dimensions of the furnace were changed from
1.6 m  1.2 m  1.2 m to 2.5 m  0.9 m  1.6 m. The width was
Table 2
decreased but the length and height were increased to accommo- Ultimate composition analysis of the
date the grates and the secondary combustion chamber made by wood pellet (wt%).
the baffle. Air is supplied at the bottom of the grates to generate an
C 48.8
upward flame. Along with the residence time, temperature is a key H 6.9
factor in achieving good combustion [3]; therefore, the walls were O 43.5
made using an insulating refractory material. N 0.4
The combustion chamber was mounted in a boiler. The boiler
had a vertical flame-tube-type structure in the previous study [10].
However it was changed to a horizontal flame-tube type to be It is designed such that the pellet store is separated from the
compliant with the new furnace (Fig. 1(b)), whose exit was moved combustion zone to prevent the danger of back burn [4]. The ca-
from the center to the edge. To accommodate a capacity of 230 kW pacity is 230 kW, which corresponds to a wood pellet feed rate of
the boiler was equipped with two passes comprising 57 flame tubes 50 kg/h. The feed rate was measured using a balance (Load Cell, AD-
as shown in Fig. 2. 5000) and recorded by data acquisition system (midi logger).
Wood pellet was used as fuel, the composition of which is A small quartz window was installed on one side of the chamber
summarized in Tables 1 and 2. The pellet burner in the present to capture flame images (see Fig. 2), which were recorded by a
study is a top fed type, which is supplied with pellets from the top. video camera (SONY, TRV-30) during the experiments. Thermo-
couples were installed inside the combustion chamber to provide
the temperature distribution, as shown in Fig. 1(b). R-type ther-
mocouples, which can measure up to 1450  C, were installed. Their
tolerance was ±2.5  C, when the temperature was 1000  C.
The flue gas was sampled at the exit of the boiler for composi-
tional analysis. The sampled gas was supplied to a gas analyzer
(TESTO: TESTO 350) through a moisture trap. The gas analyzer
provides the dry-based volume concentrations of O2, CO2, CO and
NOx. The maximum uncertainties for the CO and NOx concentra-
tions are estimated as 5 ppm and 1 ppm respectively, for the
measuring span adopted in the present study. A control system [12]
was installed to monitor the flame and boiler operation and to turn
off the fuel supply when necessary.

3. Numerical method

The interior of the combustion chamber of the wood pellet


boiler with a 16-step grate, as shown in Fig. 3, was set as the
computational domain. Because of its symmetrical shape, sym-
metric boundary conditions were applied in the central plane and
only half of the combustion chamber was considered as the
domain. Approximately 280,000 hexahedral grids were used for
the calculation (Fig. 3(a)) to achieve similar or finer resolution
compared with that used in a previous study [10]. The represen-
tative cell size was approximately 10 mm for a side, which was
validated with a detailed temperature distribution via experiments
[10]. The walls in the combustion chamber were insulated with
refractory and insulation materials (Fig. 1(b)), and the adiabatic
condition was applied while performing the analysis. Pressure
boundary conditions were applied to the outlet of the combustion
chamber.
In this study, FLUENT, a commercial software, was used for the
analysis, and the continuity equation, incompressible momentum
equation, energy equation, and species equation were considered
as the governing equations. For the momentum equation, three
turbulence models including the Spalart-Allmaras, standard k-ε,
and realizable k-ε models, were tested; all of them gave same the
results. Hence, the standard k-ε model was used as the turbulence
Fig. 2. Image of the wood pellet boiler used in the present study. model. A simple algorithm was used for the pressure-velocity
J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685 681

CO þ 0.5O2 ¼ CO2 (2)

During the modeling, the atomic balance was determined based


on the results of the ultimate analysis of the wood pellets (Table 2)
used as fuel; the heating value of the fuel (16,873 kJ/kg) was re-
flected in the energy balance.
The boundary conditions (Table 3) at the fuel bed were set by
considering the amounts of pellet input and air. The pellet input
was 50 kg/h for the 100% operation load condition. Operation loads
from 60% to 120% were simulated. The fuel amount at each grate
was estimated based on the amount of ash at the bottom of the
grate; the boundary conditions of the 3rd to the 13th layers of the
grate were set in the ratio 15:15:12:12:12:12:8:5:3:3:3. Almost no
ash was found at the first two and the last three grates. The amount
of air supplied was 547 Nm3/h for the combustion load of 100%. Air
was assumed to be supplied evenly to the 16 grates; therefore, the

Table 3
Boundary conditions.

Inlet Momentum Velocity inlet for air and fuel


Turbulence k ¼ 0.001 J/kg, ε ¼ 0.1 m2/s3
Energy 300 K
Outlet Pressure outlet
Wall Momentum Standard wall function
Energy Adiabatic

Fig. 3. Numerical modeling; (a) computational domain; (b) grid system.

coupling. The density was calculated using an ideal-gas relation,


and other physical properties were calculated using piecewise-
linear interpolations.
In the energy equation, radiative heat transfer was calculated
using the 3  3 discrete ordinate (DO) method. The radiative
properties of gases were evaluated using the weighted sum of the
gray gases model. In the species equation, the chemical reaction
rate was determined by a slow reaction between the finite rate
reaction and the eddy dissipation reaction. For the eddy dissipation
reaction, turbulent mixing of the reactant and the product was
considered [10,13]. For the convection terms of the momentum,
energy and species equations, a second-order upwind scheme was
applied. Convergence criteria were set to 1  106.
The chemical reaction was modeled by assuming that the
thermo-pyrolysis products of the wood pellets were moisture,
volatiles and char. Based on the assumption of a homogeneous
reaction, the rate of creation/destruction of the components was
evaluated using the Arrhenius function. Comparing the Arrhenius
rate and eddy-dissipation, the net reaction rate was taken as the
minimum of these two rates [14,15].
The chemistry of volatile oxidation in turbulent flows is
assumed to follow a two-step global mechanism:

CxHy þ (x/2 þ y/4)O2 ¼ xCO þ (y/2)H2O (1) Fig. 4. Images of the flame: (a) from the side of the fuel bed for the original 4-step
grate furnace [10]; (b) at the end of the grate for the novel 16-step grate furnace.
682 J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685

On comparing the exhaust gas components to determine the


combustion performance (Fig. 5 (b)), it can be confirmed that CO
was greatly reduced due to the redesign of the combustion cham-
ber. After the modification of the combustion chamber, CO
decreased to a lower level than that in boilers using conventional
fossil fuels. The concentration of NOx was observed to increase by
about 30%, although it was within environmental regulations for
the novel design. This may arise from an increase in the tempera-
ture of the secondary combustion chamber wherein thermal NOx is
believed to be generated. In order to prepare for the stricter envi-
ronmental regulations that may arise in the future, measures to
reduce thermal NOx should be studied.
Fig. 6 shows the temperature distribution inside the primary
combustion chamber from the position indicated by 1e8 in
Fig. 1(b), which is measured by conveying each thermocouple in the
spanwise direction. The temperature increases from the first sensor
as it flows downstream and it shows a gentle maximum around the
5th or 6th position. Then, the temperature decreases slightly after
the 6th sensor. The isothermal surface shown in Fig. 6 is almost
two-dimensional, which implies that it was appropriate to impose
the adiabatic boundary condition on the wall when conducting
numerical simulations.

Fig. 5. Performance of the 16-step grate-firing wood pellet boiler compared with the
original 4-step grate version; (a) flue gas temperature; (b) CO and NOx emissions. Fig. 6. Spanwise temperature distribution inside the 1st combustion chamber along
temperature sensors.

velocity boundary conditions were applied. On the grate, the tur-


bulence kinetic energy and dissipation rate were set at 0.01 J/kg and
0.1 m2/s3 respectively by considering the air flow through small
holes on the grate with low turbulence level.

4. Results & discussion

The flame started on the 2nd grate and stretched to the exit of
the chamber in the original 4-step grate furnace as shown in
Fig. 4(a) [10]. The flame from the modified 16-step grate combus-
tion chamber was not stretched in a specific direction but rather
formed to cover the grate (Fig. 4(b)). The intended flame shape was
achieved successfully; however, combustion was completed before
the pellet arrived at the bottom grate at the design load of 230 kW.
The performance of the 16-step grate wood pellet boiler was
compared with the previous 4-step grate boiler and is presented in
Fig. 5. On investigating the temperature of the exhaust gas, as
shown in Fig. 5 (a), it is expected that the thermal efficiency would
increase by 2% because the gas temperature decreases by 35  C or
more depending on the load. It is considered that this was due to an
increase in the combustion efficiency and an increase in the heat
exchange area of the boiler. Fig. 7. Temperature data inside the combustion chamber for various combustion loads.
J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685 683

Fig. 9. Temperature data inside the combustion chamber compared with numerical
simulation.

Fig. 8. Measurement data for 16-step grate firing wood pellet boiler with respect to
excess air ratio; (a) temperature inside the furnace (legend indicates the oxygen
concentration of the flue gas); (b) CO and NOx concentrations of the flue gas.

The temperature distributions inside the combustion chamber


were measured to be similar for combustion loads from 60% to
120% (Fig. 7). The temperature of the combustion chamber was
below 500  C at measuring points 1 and 2. This indicates that the
combustion was completed before the last four steps at the end of
the grate. For 60% load operation, the maximum temperature was
observed at measuring point 5, whereas it was at point 6 for loads
of 80% or more. The temperature was lower than the neighboring
measurement points at the 8th point nearest to the pellet input
position. Here, the pellet was supplied over the local equivalence
ratio, wherein the combustion was delayed and the temperature
was lowered due to lack of oxygen (see Fig. 8).
Temperature data at the center of the combustion chamber
obtained from the numerical analysis and the experiment results
are compared in Fig. 9. The points at which data were extracted are
shown in Fig. 2(b). The numerical data from the 1st combustion
chamber (sensor positions from 1 to 8) follow the trend of
measured temperatures well, which validates the modeling. The
temperature drop in the 2nd combustion chamber (measuring
Fig. 10. Velocity vectors from numerical simulation: (a) 60% load; (b) 120% load.
points 9 and 10) was also predicted by the numerical simulation.
684 J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685

combustion chamber, which are expected to be the locus of the


flame. For all combustion loads considered, high-temperature re-
gions were observed around the flow turning between the 1st and
the 2nd combustion chamber. From the temperature distribution,
the flame was expected to be stretched into the 2nd combustion
chamber. The temperature distribution suggested maintaining the
length of the combustion chamber to ensure sufficient residence
time for combustion.
Figs. 13 and 14 show the mass fraction of fuel and oxygen,
respectively. Oxygen was lean around the turn, and fuel was
abundant around the downstream grates. This implies that com-
bustion was active from the 3rd to the 8th grate, and it was delayed
after the 13th grate. As a result, it is possible to explain the

Fig. 12. Temperature distributions: (a) 60% load; (b) 80% load; (c) 100% load; (d) 120%
load.

Fig. 11. Isosurface of 1.0 m/s velocity: (a) 60% load; (b) 120% load.

The discrepancy between the experimental and numerical data is


considered to be caused by the heat loss through the combustion
chamber wall.
Fig. 10 shows the velocity vectors for the minimum and
maximum combustion loads considered. The flow fields from CFD
showed similar patterns for all combustion loads considered. The
gas supplied at the grate flowed to the exit and a long recirculation
flow was made in the secondary combustion chamber above the
guide vane because of the flow separation. The recirculating flow
accelerates the combustion gas flowing out of the furnace. Another
recirculation flow was found over the last 3 grates in the 1st
combustion chamber. The maximum velocity increases to 15.7 m/s
for 120% load operation, and it is 13.2 m/s for 60% load.
As mentioned in section 2, the pellet fuel is supplied from the
top, so its trajectory can be affected by gas flows over 1 m/s. Fig. 11
shows the iso-surface of the velocity of 1 m/s. The area of the
surface increases by up to 15% as the combustion load increases
from 60% to 120%.
The temperature distributions from the numerical simulation
are shown in Fig. 12. High-temperature regions were found from
the exit of the 1st combustion chamber to the entrance of the 2nd
Fig. 13. Mass fraction of fuel: (a) 60% load; (b) 80% load; (c) 100% load; (d) 120% load.
J. Ahn, J.H. Jang / Renewable Energy 129 (2018) 678e685 685

chamber was significantly improved. However, the last four grates


were not used for combustion, so there is some room for further
modification of the present design.

5. Conclusions

In the present study, we modified a Stoker-type boiler for wood


pellets, and then conducted a performance test. We also conducted
a numerical simulation of the combustion chamber of the boiler.
The results can be summarized as follows:

(1) In our modified furnace, the re-sizing (the change of the


aspect ratio and increase of the volume) and the baffle to
make a secondary combustion chamber worked together to
achieve complete combustion.
(2) The flame locus and flow pattern do not change significantly
even for the overloaded operation; this suggests that further
modification of the grates is required to use the volume of
the furnace more effectively.
(3) The chemical species distribution from the CFD analysis
showed that combustion did not actively occur after 10th
grate. This suggests that a decrease in the number of grates or
Fig. 14. Mass fraction of O2: (a) 60% load; (b) 80% load; (c) 100% load; (d) 120% load. the angle of the slope could be helpful to decrease the vol-
ume of the furnace while maintaining the combustion
efficiency.

References

[1] K. Christopher, T. Spachos, N. Moussiopoulos, Exergy analysis of renewable


energy sources, Renew. Energy 28 (2003) 295e310.
[2] A. Demidras, Potential application of renewable energy sources, biomass
combustion problems in boiler power systems and combustion related envi-
ronmental issues, Progr. Energy Combust. Sci. 31 (2005) 171e192.
[3] Y.W. Lee, C. Ryu, W.J. Lee, Y.K. Park, Assessment of wood pellet combustion in
a domestic stove, J. Mater. Cycles Waste Manag. 13 (2011) 165e172.
[4] F. Fiedler, The state of the art of small-scale pellet-based heating systems and
relevant regulations in Sweden, Austria and Germany, Renew. Sustain. Energy
Rev. 8 (2004) 3624e3629.
[5] S.K. Kaer, Numerical modeling of a straw-fired grate boiler, Fuel 83 (2004)
1183e1190.
[6] R. Saidur, E.A. Abdelaziz, A. Demidras, M.S. Hossain, S. Mekhilef, A review on
biomass as a fuel for boilers, Renew. Sustain. Energy Rev. 15 (2011)
2262e2289.
[7] C.D. Ray, L. Ma, T. Wilson, D. Wilson, L. McCreery, J.K. Wiedenbeck, Biomass
boiler conversion potential in the eastern United States, Renew. Energy 62
(2014) 439e453.
[8] S.B. Kang, J.J. Kim, K.S. Choi, B.S. Sim, H.Y. Oh, Development of a test facility to
evaluate performance of a domestic wood pellet boiler, Renew. Energy 54
(2013) 2e7.
[9] C. Yin, L.A. Rosendahl, S.K. Kaer, Grate-firing of biomass for heat and power
production, Progr. Energy Combust. Sci. 34 (2008) 725e754.
[10] J. Ahn, J.J. Kim, Combustion and heat transfer characteristics inside the com-
bustion chamber of a wood pellet boiler, J. Mech. Sci. Tech. 28 (2014)
Fig. 15. Mass fraction of CO: (a) 60% load; (b) 80% load; (c) 100% load; (d) 120% load. 789e795.
[11] J. Ahn, J.H. Jang, CFD (Computational Fluid Dynamics) study on partial-load
combustion characteristics of a 4-step-grate wood pellet boiler, Trans.
Korean Soc. Mech. Eng. B 38 (2014) 365e371.
experimental results with lower temperatures at the 7th and 8th [12] S.B. Kang, J.J. Kim, K.S. Choi, B.S. Sim, H.Y. Oh, Development of an air fuel
measurement points. As for the combustion load, the oxygen-lean control system for a domestic wood pellet boiler, J. Mech. Sci. Tech. 27 (2013)
regions increased and a 2nd oxygen-lean region was found near 1701e1706.
[13] D. Djurovic, S. Nemoda, D. Dakic, M. Adzic, B. Repic, Furnace for biomass
the exit for higher loads.
combustion - comparison of model with experimental data, Int. J. Heat. Mass
As for carbon monoxide (Fig. 15), CO was generated mainly at Transf. 55 (2012) 4312e4317.
the turn where oxygen concentration was low. Compared with the [14] R. Vijeu, R.L. Gerun, M. Tazerout, C. Castelain, J. Bellettre, Dimensional
previous combustor [10], the CO level was decreased at the exit due modeling of wood pyrolysis using a nodal approach, Fuel 87 (2008)
3292e3303.
to the increase of the residence time. Compared with the previous [15] V. Novozhilov, B. Moghtaderi, D.F. Fletcher, J.H. Kent, Computational fluid
prototype with a 4-step grate, the combustion efficiency within the dynamics modeling of wood combustion, Fire Saf. J. 27 (1996) 69e84.

S-ar putea să vă placă și