Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
A Honey Bee Foraging approach for optimal location of a biomass power plant
David Vera a, Julio Carabias b, Francisco Jurado a,*, Nicols Ruiz-Reyes b
a
Dept. of Electrical Engineering, University of Jan, 23700 EPS Linares, Jan, Spain
b
Dept. of Telecommunication Engineering, University of Jan, 23700 EPS Linares, Jan, Spain
a r t i c l e i n f o a b s t r a c t
Article history: Over eight million hectares of olive trees are cultivated worldwide, especially in Mediterranean countries,
Received 9 November 2009 where more than 97% of the worlds olive oil is produced. The three major olive oil producers worldwide
Received in revised form 18 January 2010 are Spain, Italy, and Greece. Olive tree pruning residues are an autochthonous and important renewable
Accepted 26 January 2010
source that, in most of cases, farmers burn through an uncontrolled manner. Besides, industrial uses have
Available online 13 February 2010
not yet been developed.
The aim of this paper consists of a new calculation tool based on particles swarm (Binary Honey Bee
Keywords:
Foraging, BHBF). Effectively, this approach will make possible to determine the optimal location, biomass
Biomass
Swarm intelligence
supply area and power plant size that offer the best protability for investor. Moreover, it prevents the
Protability index accurate method (not feasible from computational viewpoint). In this work, Protability Index (PI) is
set as the tness function for the BHBF approach. Results are compared with other evolutionary optimi-
zation algorithms such as Binary Particle Swarm Optimization (BPSO), and Genetic Algorithms (GA). All
the experiments have shown that the optimal plant size is 2 MW, PI = 3.3122, the best location corre-
sponds to coordinate: X = 49, Y = 97 and biomass supply area is 161.33 km2. The simulation times have
been reduced to the ninth of time than the greedy (accurate) solution. Matlab is used to run all
simulations.
2010 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2120
2. Optimization problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2121
2.1. Problem description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2121
2.2. Coding of the solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2121
3. Fitness function: Profitability Index (PI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
3.1. Initial investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
3.2. Cash inflows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
3.3. Cash outflows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
3.4. Net present value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
3.5. Profitability index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
4. Metaheuristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
4.1. Genetic Algorithms (GA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
4.2. Binary Particle Swarm Optimization (PSO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2122
4.3. Binary Honey Bee Foraging (BHBF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2123
5. Simulations results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2123
6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2126
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2126
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2126
* Corresponding author.
E-mail address: fjurado@ujaen.es (F. Jurado).
0306-2619/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2010.01.015
2120 D. Vera et al. / Applied Energy 87 (2010) 21192127
Nomenclature
The motivation of using HBF is due to its novelty approach, Therefore, given the total mean efciency of the electric gener-
which is based on the main features of GA and PSO algorithms. ation system (gasiergas turbine), g, the produced electricity, Eg
These algorithms have been widely used to nd exact or approxi- (MWh/yr), is equal to:
mate solutions in optimization and search problems. Moreover,
X
K
the Honey Bees algorithm has been recently used in some state- Eg g Si U i Di LHV i 1
of-the-art approaches related to optimization problems. Pham i1
[1012] is one of the most active authors in this eld proposing
Assuming a plant running time of T (h/yr), the electric power, Pe
several approaches in order to test the reliability of this method
(MW) is:
in optimization problems. Sundareswaran and Sreedevi [13] pro-
posed the use of a colony of honey bees to suit the controller de- Eg
Pe 2
sign of a boost converter. Melin-Batista et al. [14] proposed a T
heuristic approach based on the Honey Bees Mating Optimization The main requirements to build biomass power plants in these loca-
algorithm to solve the Uncapacitated Plant-Cycle Location Prob- tions are listed below. The perspective cost is listed in Table 2:
lem. In [15], Bees Algorithm were used to determine the optimal
allocation of Flexible Ac Transmission Systems (FACTS) devices Microturbines Capstone C200 (e.g. for 2 MW it would be
for maximizing the Available Transfer Capability (ATC) of power installed 10 stack of C200 model). Electrical performance, fuel/
transactions between source and sink areas in the deregulated engine and exhaust characteristics, dimensions and weight and
power system. Nakrani and Tovey [16] employed Honey Bee algo- sound levels are listed in [19].
rithm to dynamically allocate servers to satisfy the unpredictable Gasier and Feedstock Handling System. Other needs for a gas-
HTTP request loads. iermicroturbine system have been described in Table 2.
Nature-inspired methods as HBF are getting very popular in the Processes machines storage area: chipper, packer, shredder and
literature when dealing with optimization problems that requires trucks for biomass transport.
high computation requirements to provide an optimal solution. Biomass storage area.
This method is appropriated for the current problem because Building a transformer and electric line for grid connect mode.
HBF is based on BPSO and GA. Both methods have proven to be
very suitable in similar problems, as exposed in [17]. Therefore,
the idea of using a hybrid method that combines the main features 2.2. Coding of the solution
of PSO and GA should be further exploited. Actually, this is the
main purpose of our approach. Before using the proposed BHBF to determine location of bio-
Finally, the reliability of BHBF has been proved in our concrete mass power plant, the representation of a feasible solution must
problem of determining the optimum location of the biomass be dened. A solution consists of three parts: (1) X component of
power plant, obtaining satisfactory results in terms of protability location plant; (2) Y component of location plant; (3) Size of supply
and efciency. area for the power plant. These components are binary Gray coded
This paper is organised as follows. After introduction, the prob- in order to exploit some useful properties of Gray code related with
lem description and the tness function are presented in Sections 2 the Hamming distance.
and 3, respectively. In Section 4, it will describe the metaheuristics We have considered a rectangular search space with x 2 [1, LX]
techniques considered in this paper (GA, BPSO and BHBF). Simula- and y 2 [1, LY], LX and LY being sizes in X-dimension and Y-dimen-
tion results are shown is Section 5. Finally, conclusions are pre- sion, respectively. Supply area is a square shaped region which has
sented in Section 6. the plant at the centroid. In order to obtain not only the sitting of
the plant but also the sizing of the supply area, a prexed number
2. Optimization problem of supply region sizes have been assumed (i.e. size number 0 cor-
responds to a 1 1 region, size number 1 corresponds to a 3 3
2.1. Problem description region and maximum size number S corresponds to a (2S + 1)
(2S + 1) region. Thus, the total number of bits used to code the
The problem to be solved consists on comparing three optimi- solution is:
zation algorithms (GA, BPSO and BHBF). We are interested in deter-
mining the optimal location, supply area and protability index (or N log2 LX log2 LY log2 S 3
net present value) of the electric generation plant for a biomass
fuelled system.
In this paper, the size of the electric generation system depends
on: Table 2
Capital and installation costs.
Biomass quantity that can be collected from a given region Capital cost (Is) /kWe-net
mainly covered by olive tree by-products.
Gasier and feedstock handling system 400
Technology to produce electricity from biomass, in our case, gas- Gas cleaning system 70
ier with gas turbine. Power generation/interconnect system 650
Emission control system 35
Heat recovery system 70
Location of the biomass-based power plant (parcel p) mainly
depends on the characteristics of the considered region to collect Total facility capital cost 1225
biomass. In this work, K parcels of constant area have been re- Installation cost (INVf)
garded, most of them characterized by a predominant biomass Building works and silencer 45,000
type (olive tree prunings in this paper). These parcels also present Electrical works 23,000
Gas and plumbing works 57,000
other relevant characteristics, such as accessibility [18]. Meters 16,000
The values of the variables involved in the problem are obtained Others 9000
from databases or Geographic Information Systems (GIS), [8,9]. Total installation cost 150,000
These are Si, Ui, Di, LHVi, Lp, dist(p,i), C Cu tu
i and C i .
2122 D. Vera et al. / Applied Energy 87 (2010) 21192127
The objective function takes into consideration costs and bene- The Protability Index (PI) is chosen in this work as objective
ts. Specically, initial investment and collection, transportation, tness function. It is dened as follows:
maintenance and operation costs are considered, together with NPV PV
benets from the sale of electrical energy. Therefore, the protabil- PI 1 8
INV INV
ity index is chosen as the objective function.
In this section some interesting parameters to evaluate the We can also say that an investment is protable when PI > 0.
protability index of the project are reviewed. The initial invest-
ment, the present value of cash inows (benets) and cash out- 4. Metaheuristics
ows (costs) are dened and adapted to the particularities of this
work. In the following, we briey describe the metaheuristic tech-
niques considered in this work. As demonstrated in [17] popula-
tion methods utilized in this work (GA, BPSO and BHBF) are
3.1. Initial investment
more efcient than trajectory algorithms (SA, TB or fuzzy search).
The initial investment (INV) for the design, construction and
equipment of the generation plant is expressed as: 4.1. Genetic Algorithms (GA)
X
K The classical version of the PSO algorithm [23,24] operates in a
Cc C cui U i Si Di continuous search space. In order to solve optimization problems
i1
in discrete search spaces, several binary discrete PSO algorithms
The annual cost of biomass transport is have been proposed. In a binary discrete space the position of a
particle is represented by a N-length bit string and the movement
X
K
of the particle consists of ipping some of these bits. Kennedy and
Ct C tui U i Si Di distp; i
i1
Eberhart proposed in [25] the rst binary version of PSO.
In this paper, we have developed and applied a improved ver-
The annual M&O cost is sion of the binary PSO algorithm proposed in [26], which incorpo-
C mo C mof m Eg rates a inertia weight factor, like the classical continuous approach
[23,24]. Now, particle position (xi) and particle velocity (vi) are N-
Finally, the present value of cash outows is: length binary vectors. The algorithm uses the Hamming distance,
and the logical AND (), OR ( + ) and XOR () operators. Particle
K c 1 K Vc u K t 1 K Vt u position is updated by using the XOR operator instead of the
PV OUT C c Ct
1 Kc 1 Kt sum-operator, as in [26].
K mo 1 K Vmou
i;j v i;j ;
xti;j xt1 t
C mo 6 i 1; . . . ; P j 1; . . . ; Z 9
1 K mo
Z represents the number of variables of the function to be optimized
where K c 1r
1d
c
, K t 1r
1d
t
and K mo 1r
1d
mo
.
and P the number of particles in the swarm.In this algorithm, the
velocity vector can be interpreted as a change vector. Thus, if
3.4. Net present value v ti;j = 1, then xti;j xt1 t1 t1
i;j , xi;j being the logical negation of xi;j . How-
ever, if v i;j = 0, then xi;j xi;j (no change happens). The velocity
t t t1
The net present value (NPV) of an investment is dened as
follows: vector (change vector) is updated by applying the following equa-
tion [6]:
NPV PV INV 7
t1 t1
v ti;j xi;j xi;j t1
c1i;j pbesti;j xi;j c2i;j gbestj xt1
i;j
PV PV IN PV OUT being the present value. An investment is prot- 10
able when NPV > 0.
D. Vera et al. / Applied Energy 87 (2010) 21192127 2123
Y
The region under study and the single line diagram illustrating 100
the topology of the test distribution network are shown in Fig. 1. 80 80
Fig. 2 presents the theoretical biomass potential, which is de-
ned from the net density of dry biomass that can be obtained at 60
100
any parcel, Di (t/(km2 yr)), and provides a measure of the primary 40
biomass resource.
20
Fig. 3 shows the available biomass potential. It has been created 120
taking the following parameters into account: Di (t/(km2 yr)), Ui, Si 0
20 40 60 80 100 120
(km2) and LHVi (MWh/t). Multiplying the four variables for all par-
cels that comprise the entire region, it results the available biomass X
potential, expressed in (MWh/yr).
Fig. 2. Theoretical biomass potential (t/(km2 yr)).
Simulation results for BHBF approach (optimal location, supply
area, location coordinates, installed power and protability index)
will be contrasted with BPSO algorithm results. Moreover, compar-
ison between the average protability index obtained in BHBF, 40
BPSO and GA performances for all tests will be attached.
In all tests, BHBF and BPSO have the same simulation parame- 20 35
ters. These parameters are shown in Table 4:
30
40
Population size (P): it is associated to the search space. If this
parameter is too low, the algorithm will probably converge to 25
a local optimum. Otherwise, a high value of this parameter will 60
Y
20
require a high computation cost. For the three algorithms, P = 30
80
15
Table 3 100 10
Standard values for parameters.
5
Parameter Value Ref. Parameter Value Ref.
120
C tu
i (/t km)
0.3 [7] rc 0.06 [6,17] 0
cu (/t) 40 [7] rt 0.08 [6,17] 20 40 60 80 100 120
CL (/km) 3-104 [6] r
mo 0.04 [6,17] X
T (h/yr) 7500 [6,17] Cmof () 1.5 105 [17]
INVf () 1.5-106 [19] m (/MWh) 4.0 [17] Fig. 3. Available biomass potential (MWh/yr)).
ps (/MWh) 128100 [33] g 0.28 [19]
DR 0.08 [6,17] Is, (/MW) 1.225 106 [19]
IR 0.04 [6,17] Vu (yr) 15 Estimated
Table 4
Simulation parameters.
20
In Spain, electric energy price for a biomass power plant (max- lution, the mean of all tests are near to zero). Consequently, under
imum 2 MW) is 128 /MWh, on the other hand, for plants over 0.5 MW, a biomass power plant would not be protable.
2 MW of generated power the price is 100 /MWh (pg), [33]. In the second simulation, the technical constraints to be consid-
Fig. 4 shows the optimal location and supply area (km2) of the ered are:
biomass power plant for the best found solution (the best PI of
all tests). Output values for the proposed BHBF and BPSO algo- 1. The electric power generated by the plant is limited to 4.5 MW
rithms are displayed in Table 5: 2. The generation system must be located inside the supply area.
As shown in Fig. 5, the proposed BHBF algorithm converges to
accurate solutions in a few iterations. In this gure the average Figs. 6 and 7 present optimal location and PI evolution. Table 6
PI distribution for all tests is represented and compared with BPSO describes simulations results for a 4.5 MW power constraint, it can
and GA results. see the supply area is about 365 km2.
As lower limit it has been established the minimum generated On the other hand, Fig. 6 shows that the maximum generated
power to be 0.5 MW, under this power generation level, the prot- power would be around 44.5 MW. This top limit is not established
ability index will be negative (Fig. 5 shows protability index evo- for the protability index, but it would be a geographic constraint.
PI = 0.0468 PI = 2.5335
Pe = 0.4975 MW 20 Pe = 4.4982 MW
20
Area = 39.4546 km2 Area = 364.9554 km2
40 40
60 60
Y
Y
80 80
100 100
120 120
X X
Fig. 4. Optimal location of biomass power plant for BHBF algorithm (P 6 0.5 MW). Fig. 6. Optimal location of biomass power plant for BHBF algorithm (P 6 4.5 MW).
Table 5 2.6
BHBF vs. BPSO simulation results.
2.3
0.2
PI
2.2
0
0.2 2.1
BHBF
BPSO
0.4
GA 2
0.6
PI
1.9
0.8 0 10 20 30 40 50 60 70 80
Iterations
1
Fig. 7. Average protability index evolution for BHBF, BPSO and GA.
1.2
1.4
Table 6
BHBF vs. BPSO simulation results.
1.6
0 10 20 30 40 50 60 70 80
Algorithm Pe (MW) PI Location coordinates Supply area (km2)
Iterations BHBF 4.4982 2.5335 X = 53, Y = 57 364.9554
BPSO 4.4918 2.5320 X = 53, Y = 64 361.4397
Fig. 5. Average protability index evolution for BHBF, BPSO and GA.
2126 D. Vera et al. / Applied Energy 87 (2010) 21192127
In other words, the supply area for a power plant over 4.5 MW, will Table 8
be greater than beda extension (approximately 400 km2). Mean and standard deviation for BHBF, BPSO and GA.
of olive trees per km2, roads, topographical features, electric lines Method Optimal location PI Simulation time (h)
location, etc.) and determining the economic variables and techno- Greedy X = 49, Y = 97 3.3122 50.47
logical parameters (gasier and microturbine characteristics) in- BHBF X = 49, Y = 97 3.3122 5.60
volved in the mathematical model, the proposed approach allows BPSO X = 50, Y = 95 3.2890 3.84
to get the location, plant size and supply area that offer the best GA X = 49, Y = 98 3.2944 8.12
protability from the investor viewpoint.
All the experiments have shown that the optimal plant size
according to the protability for the investor is 2 MW picts (for a 2 MW power constraint) the mean and standard devia-
(PI = 3.3122). The best location corresponds to coordinate: X = 49, tion values for three algorithms. As can seen, the average
Y = 97. Fig. 8 presents the best location for a 2 MW generated protability index for BHBF algorithm is higher than BPSO and
power (red point). As could be seen in this gure, the highest avail- GA. Otherwise, the lower standard deviation in BHBF algorithm
able biomass potential area coincides with the optimal predicted suggest that our approach converges to the optimal solution with
location of the plant. The rest of best locations are depicted in Table more probability than BPSO and GA.
7. Otherwise, a plant size with an installed power below 0.5 MW Table 9 presents a comparison between our approach and the
(PI = 0.0468) will not be protable for the investment. classical ones has been performed in terms of protability of the
On the other hand, according to [34], the combustion of biomass solution and elapsed time for a 2 MW installed power.
has been considered as a CO2 non-emitting source. Further, accord- As can be seen in the Table 9, BHBF obtain the optimal prot-
ing to Ref. [35], Table 7 shows the CO2 emissions that will be pro- ability with the ninth of time than the greedy (accurate) solution.
duced for a power generation plant based in fossil fuel. Also, the required computation cost is slightly higher than BPSO,
Figs. 5 and 7 show that our algorithm (BHBF) converges in less but lower than GA. However, metaheuristic classical approaches
iterations than other studied algorithms. Furthermore, Table 8 de- did not reach the optimal location. In conclusion, our method has
proven its reliability for this concrete problem, reaching the opti-
mal location for the biomass plant, reducing in a factor of 89
the elapsed time respect to the greedy solution and with similar
PI = 3.3122 computation requirements than the lowest elapsed time method.
20 Pe = 1.9965 MW
Finally, for promoting the olive-based biomass power plants,
Area = 161.3343 km2
investors are nancially supported by the Spanish government
40 (according to the Andalusian Plan of Energetic Sustainability,
20092014 [33]) in some economical issues such as: machinery
for biomass collection and transport, logistic and initial invest-
60
ment. The BHBF approach allows to fast and efciently obtain
Y
References
Table 7
Simulation results and CO2 emissions reduction. [1] Morillo JA, Antizar-Ladislao B, Monteoliva-Snchez M, Ramos-Cormenzana A,
Russell NJ. Bioremediation and biovalorisation of olive-mill wastes. Appl
Pe Biomass supply Best location PI CO2 emissions Microbiol Biotechnol 2009;82:2539.
(MW) area (km2) coordinates reduction (tCO2/yr) [2] Adaptation of renewable energy solutions for the olive oil industry. RESOLIVE;
2009. <http://www.resolive.com>.
0.5 39.4546 (49, 93) 0.0468 2100 [3] Cara C, Ruiz E, Ballesteros M, Manzanares P, Negro MJ, Castro E. Production of
1 78.5186 (49, 89) 1.9104 4200 fuel ethanol from steam-explosion pretreated olive tree pruning. Fuel
2 161.3343 (49, 97) 3.3122 8400 2008;87:692700.
3 247.0798 (79, 70) 2.1710 12,600 [4] Jimnez L, Prez I, Garca JC, Rodrguez A. Inuence of process variables in the
4 325.3055 (53, 50) 2.4403 16,800 ethanol pulping of olive tree trimmings. Bioresour Technol 2001;78:639.
4.5 364.9554 (53, 57) 2.5335 18,900 [5] Jurado F, Ortega M, Cano A, Carpio J. Biomass gasication, gas turbine, and
diesel engine. Energy Sources 2001;23:897905.
D. Vera et al. / Applied Energy 87 (2010) 21192127 2127
[6] Reche Lpez P, Gmez Gonzlez M, Ruiz-Reyes N, Jurado F. Optimization of [21] Goldberg DE. Genetic algorithms in search, optimization and machine
biomass fuelled systems for distributed power generation using particle learning. 1st ed. Addison Wesley; 1989.
swarm optimization. Electr Power Syst Res 2008;78:144855. [22] Michalewicz Z. Genetic algorithms + data structures = evolution programs.
[7] Kinoshita Tsuguki, Inoue Keisuke, Iwao Koki, Kagemoto Hiroshi, Yamagata 2nd ed. Springer Verlag; 1996.
Yoshiki. A spatial evaluation of forest biomass usage using GIS. Appl Energy [23] Kennedy J, Eberhart RC. Particle swarm optimization. Proceedings of IEEE
2009;86(1):18. international conference on neural networks, vol. 4. Piscataway, NJ: IEEE;
[8] Geographic Information Systems of farming plots (SIGPAC); 2009. <http:// 1995. p. 19428.
www.mapa.es/es/sig/pags/sig/intro2.htm> [in Spanish]. [24] Eberhart RC, Kennedy J. A new optimizer using particle swarm theory. In:
[9] Geographic Information Systems of Andalusia (SIGPAC); 2009. <http:// Proceedings of the IEEE sixth international symposium on micro machine and
www.juntadeandalucia.es/agriculturaypesca/sigpacandalucia/> [in Spanish]. human science; 1995. p. 3943.
[10] Pham DT, Castellani M. The Bees algorithm: modelling foraging behaviour to [25] Kennedy J, Eberhart RC. A discrete binary version of the particle swarm
solve continuous optimization problems. Proc Inst Mech Eng, C: J Mech Eng Sci algorithm. In: Proceedings of IEEE conference on systems, man and
2009;223(12):291938. cybernetics; 1997. p. 410408.
[11] Pham DT, Kalyoncu M. Optimisation of a fuzzy logic controller for a exible [26] Afshinmanesh F, Marandi A, Rahimi-Kian A. A novel binary particle swarm
single-link robot arm using the Bees algorithm. In: IEEE international conference optimization method using articial immune system. In: Proceedings of IEEE
on industrial informatics (INDIN); 2009. p. 47580 [art. no. 5195850]. international conference on computer as a tool (EUROCON 2005); 2005. p.
[12] Pham DT, Ghanbarzadeh A, Otri S, Ko E. Optimal design of mechanical 21720.
components using the Bees algorithm. Proc Inst Mech Eng, C: J Mech Eng Sci [27] Baig AR, Rashid M. Foraging for tness: a honey bee behavior based
2009;223(5):10516. algorithm for function optimization. Technical report, NUCES, Pakistan;
[13] Sundareswaran K, Sreedevi VT. Design and development of feed-back November 2006.
controller for a boost converter using a colony of foraging bees. Electr Power [28] Li-Pei Wong, Malcolm Yoke Hean Low, Chin Soon Chong. A bee colony
Componen Syst 2009;37:46577. optimization algorithm for traveling salesman problem. In: IEEE international
[14] Melin-Batista B, Moreno-Vega JM, Vaswani N, Yumar R. A nature inspired conference on modelling and simulation; 2008.
approach for the uncapacitated plant cycle location problem. Stud Comput [29] Blackwell T, Branke J. Multiswarms, exclusion, and anti-convergence in
Intell 2009;236:4960. dynamic environments. IEEE Trans Evol Comput 2006;10 (4).
[15] Mohamad Idris R, Khairuddin A, Mustafa MW. Optimal allocation of FACTS [30] Boone G, Chiang HD. Optimal capacitor placement in distribution systems by
devices for ATC enhancement using Bees algorithm. World Acad Sci, Eng genetic algorithm. Electr Power Energy Syst 1993;15:15562.
Technol 2009;54:2070372. [31] Eberhart RC, Shi Y. Particle swarm optimization: developments, applications
[16] Nakrani S, Tovey C. On honey bees and dynamic server allocation in internet and recourses. Proc Evol Comput Congr 2001;1:816.
hosting centers. Adapt Behav 2004;12(34):22340. [32] Naka S, Genji T, Yura T, Fukuyama Y. A hybrid particle swarm optimization for
[17] Reche Lpez P, Ruiz-Reyes N, Garca Galn S, Jurado F. Comparison of distribution state estimation. IEEE Trans Power Syst 2003;18:608.
metaheuristic techniques to determine optimal placement of biomass power [33] Agencia Andaluza de la Energa. Andalusian plan of energetic sustainability,
plants. 1st ed. Nova Publishers; 2008. 20092014; 2009. <http://www.juntadeandalucia.es/agenciadelaenergia/nav/
[18] Freppaz D, Minciardi R, Robba R, Rovatti M, Sacile M, Taramasso A. Optimizing com/contenido.jsp?pag=/contenidos/incentivos/incentivos_09> [in Spanish].
forest biomass exploitation for energy supply at a regional level. Biomass [34] International Energy Agency. IEA statistics: CO2 emissions from fuel
Bioenergy 2004;26:1526. combustion; 2009.
[19] Capstone Turbine Corporation. CR200 MicroTurbine Renewable Fuels; 2009. [35] Johannes Schmidt, Sylvain Leduc, Erik Dotzauer, Georg Kindermann, Erwin
<http://www.capstoneturbine.com/_docs/CR200%20Renewable.pdf>. Schmid. Cost-effective CO2 emission reduction through heat, power and
[20] Holland JH. Adaptation in natural and articial systems. 1st ed. Ann Arbor: The biofuel production from woody biomass: a spatially explicit comparison of
University of Michigan Press; 1975 [London: The MIT Press; 1992]. conversion technologies. Appl Energy 2009.