IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS

3D heterogeneous bin packing framework for multi-constrained problems

using hybrid genetic approach
To cite this article: S K Rajesh Kanna et al 2018 IOP Conf. Ser.: Mater. Sci. Eng. 402 012203

View the article online for updates and enhancements.

This content was downloaded from IP address 139.81.118.35 on 21/09/2018 at 01:48

2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

3D heterogeneous bin packing framework for multi-

constrained problems using hybrid genetic approach

S K Rajesh Kanna1, *, K C Udaiyakumar2, S Dinesh Kumar3, N Lingaraj4

Professor, Rajalakshmi Institute of Technology, Chennai, India

Vice Principal, SRM Institute of Science and Technology, Ramapuram campus, Chennai
Professor, Rajalakshmi Institute of Technology, Chennai, India
Professor, Rajalakshmi Institute of Technology, Chennai, India

*Corresponding Author:[email protected]

Abstract. This work presents distinct methodologies in using Genetic Algorithm (GA) for
optimizing Three Dimensional (3D) packing of heterogeneous shaped bins with arbitrary sizes
into a prismatic container, by considering the major real time packing constraints such as load
bearing constraint, placement constraint, stability constraint, overlapping constraint, orientation
constraint and weight constraint. The primary aim of this research is focused in optimizing the
packing of heterogeneous prismatic bins of arbitrary sizes into standard rectangular
commercial containers by obeying the above mentioned packaging constraints. Different
genetic approaches adopted to achieve these goals are Binary coded GA, Decimal coded GA
with and without penalty fitness function, Constrained GA with maximization and
minimization fitness function, Heuristic GA and Hybrid GA. GA has been used to minimize
the unused void space in the interior of the container by loading as much heterogeneous bins,
by satisfying the packing constraints. Tweaking Algorithm (TA) is an application
dependent heuristic algorithm applied in this research and has been used to enhance the genetic
output by filling the remaining unused empty space inside the container. TA has also been
enhanced in converting the obtained output into packer readable box packing sequence in
tabular and diagrammatical format. In general, combination of GA and TA are considerably at
par compared with the heuristic techniques for box packing.

1. Introduction

Evolutionary algorithms are based on the principle of biological evolution (Hopper and Turton,
1997). Genetic algorithm is a class of evolutionary algorithm used to optimize a wide range of complex
constrained problems in less time. One of the complex problem faced by the industries and logistic
firms are to pack and distribute their goods in less time and with less dissemination cost to their
customers all around the world. In general, goods are packed into bins and the bins will be loaded into
containers for safe, secure and compact distribution. Freight rate of using the containers are also
included with the product cost, but not adding any additional credits to those products. Thus it became
imperative for the firms to reduce this non-value added cost to the minimum by perfectly utilizing the
available container space. This has been achieved in this research by applying genetic approach.
Aristide et al (2018) developed algorithm with the focus on overlapping of bins in packing process and

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

the same considered in this research while calculating the occupied container volume. Ranga et al
(2018) considered the rotation of the bins inside the container for optimal packing and in this research,
instead of free rotation, orthogonal rotation have been considered and achieved by overloading the
mutation operation. Fabrizio and Andrea (2018) carried out packing of two dimensional problems with
the heuristic rules, but this research concentrated on 3D packing and more or less the generalized
algorithm. Martinez et al (2017) developed bin packing algorithm with free rotation of bins with
irregular bins, in this research, arbitrary sizes of bins with four different shapes considered. Xueping
and Kaike (2018) utilized the shop floor algorithm for packing of 2d bins, but this research uses 3D
packing. Célia at al (2018) attempted to identify the solution for the packing problems by considering
the transportation constraints and the same have been considered in this research with penality
function. Gorge and Robinson (1980) introduced the approach for packaging bins into a standard
container using layer-by-layer packing approach. Hopper and Turton (1997) applied the concept of GA
to the box packaging problem and proved that the performance of GA was satisfactory. Bortfeldt and
Gehring (2001) developed by hybridizing GA based on the layer-layer packing concept for container
loading problems with packaging constraint and proved that the feasible solution can be obtained by
satisfying the packaging constraints. Lodi et al (2002) used Tabu search algorithm with a heuristic to
reinforce carton orientation in 3D multiple-container loading problem and showed that the box
orientation also plays a major role in optimal packaging. Martine Labbe et al (2003) framed upper
bound and enumeration algorithm for packaging maximum number of bins into the container in less
computational time. Nihat Kasap and Anurag Agarwal (2004) implemented Augmented artificial
neural network algorithm for solving classical bin packaging datasets which combines priority logic
approach with iterative trial learning approach. Zhoujing Wang and Kevin Li (2006) developed a
tertiary tree algorithm for packaging heterogeneous bins into containers. Wenqi and Kun (2009)
incorporated a new caving algorithm in solving 3D single container loading problem. Leonardo et al
(2010) developed a mathematical model for the problem of loading rectangular bins into containers by
considering cargo stability and load bearing constraint. Rajesh et al (2017) experimented the bin packing
optimization using Firefly algorithm. Allen et al (2011) introduced placement of bin strategy in 3D strip type
packaging problems, which considers packing of standard available set of cuboids into a container of fixed
width, height and unconstrained length. Rajesh et al (2012) developed the genetic algorithm approach to pack
the two dimensional rectangular bin packing inside the larger rectangle. Rajesh and Saravanan (2012)
incorporated genetic approach for the dynamic bin packing problems. Even-though numerous heuristic and
hybrid approaches are existing for 3D bin packing, most of these have simplified the 3D problem into
2D or even to 1D for solving. Also some of the approaches did not consider the major box packing
constraints. As the result of the literature survey, it is clear that there is a necessity to optimize 3D box
packing problem with ‘n’ number of arbitrary sized heterogeneous bins by satisfying packing
constraints.

In this research work, rectangular prismatic container of regular dimension have been
considered for experimenting the packing of ‘n’ number of differently shaped prismatic bins namely
cubical bins, rectangular prismatic bins, cylindrical bins and spherical bins of varying sizes by
satisfying the related packing constraint. The packing constraints considered are load bearing
constraint, placement constraint, stability constraint, overlapping constraint, orientation constraint and
weight constraint. The user defined box data have been encoded and given as input to GA for
optimizing. The developed GA module identifies the best optimal and feasible box packing pattern
which yields maximum container volume utilization without violating the constraints.

Many of the researchers didn’t consider the major packing constraints altogether and it has been
considered in this research to yield feasible optimal solution which can be directly used by the layman
for packing. Also sensitivity analysis carried out to overload the mutation genetic operator, where as
the other researchers didn’t try with mutation overloading. In addition, special tuning heristic
developed to enhance the genetic results in packing. This paper has been organized as follows.

2
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

Background research works carried out by the researchers have been discussed in literature section 2. Section
3 describes the problem formulation followed by the various experimental implementations and the obtained
results. The paper concludes with conclusion and future scope.

2. Bin Packing Problem

Bin packing problem is defined as the problem of identifying the best mix of bins and packing
pattern which utilizes the maximum container volume by meeting the packing constraints. Inputs are
the container specification, number of bins and box specifications along with its related constraints.
The user defined data should be checked for its completeness and sorted in descending order based on
the weightage values. Improper box packing and arbitrary heterogeneous bins lead to some empty
unused space formation in the interior of the container. This void formed inside the container is the
primary problem, which results in box instability, utilizing of additional containers, usage of additional
airbags/dampers/fillers and inturn, increases the cargo/transportation cost and overall retail cost
without adding any additional value to packed components. Hence in this research area, various
frameworks had developed for solving the problem of box packing and the developed modules were
validated with the second order mathematical equations.

3. GENETIC ALGORITHM – BINARY ENCODING

In the Binary encoded genetic approach (BEGA), the user defined box data have been encoded
into binary elements. The encoding format for a box should contain the box dimension, position inside
the container and its constraints. A sample of two parents with four bins each is given in the figure 1.

P1: 1001001001011011 1110010111011110 0101110111010011 1011100111010010 1110001010111001 1101101111011100 11100101110110

P2: 1011011101111001 0110011010010110 1111011111001001 1110101100111011 1110001010001011 1010111001001011 11100100101111

Figure 1. BEGA Parents

From the figure 1, the first nine binary digits represent the 3D box dimension followed by the
box position inside the container in next three digits. The box position can be left, right, above, in-
front and behind the recently packed bin. The orientation of the box occupies next two digits to denote
vertical or horizontal positions. The last two digits are the auxiliary elements to identify the bin. The
other packing constraints have been checked in the final stage for making the solution feasible. These
binary elements will be feed as input to the developed hybrid genetic module, which in turn identify
the optimal sequence of packing pattern inside the container. The various stages of the GA module are
the initial parent formulation, parent crossover, mutation of child and fitness function calculation. The
fitness function used is to maximize the container volume utilization. The GA output will have the set
of bins to be packed along with its position and orientation. These outputs have been checked for
packing constraints and the pattern which satisfies the packing constraints is considered to be a
feasible solution. Even though the module gives better result, the computational time and memory
usage will be increased with the increase in the problem size. So the developed module had enhanced
to decimal encoding approach.

4. DECIMAL CODED GENETIC ALGORITHM

In the Decimal coded genetic algorithm (DCGA) module, the computational time and memory
usage had reduced using decimal encoding instead of binary encoding. A set of sample decimal coded
parents are given in figure 2.

3
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

P1: 2421344224222242222213233422242213233422123232432422221312432421113232324214223244224233324221211121142321
P2: 4222332114123322214112333232111223334111233221411122332214112223333211123332114112333122214122322133214111
Figure 2. A Sample DCGA Parents

The decimal numbers 1, 2, 3 and 4 represent cube shaped bins, rectangular prismatic shaped
bins, cylindrical shaped bins and spherical shaped bins respectively. In figure 2, starting digits ‘2’, ‘4’,
‘2’, ‘1’, ‘3’ .… represent first box from rectangular prism database, first box from sphere database,
second box from rectangular prism database, first box from cuboid database, first box from cylinder
database respectively and so on. These parents have been allowed to operate by the genetic operators
to identify the best packing pattern. The constraints need to be checked for identifying the feasible box
packing pattern from the set of DCGA outputs. In some cases, the best GA solution was also
eliminated by the constraint checking and thereby the computational time was wasted for identifying
the optimal solution from the infeasible set of patterns. This has been avoided by enhancing the DCGA
module with decimal coded GA with penalty fitness function (DCGAP).

5. DECIMAL CODED GA WITH PENALTY FITNESS FUNCTION

In this DCGAP module, the constraints were checked at the stage of calculating the fitness
function value. The pattern which violates the constraints will have lesser fitness value and will be
eliminated from the generation. The penalty fitness function is given as follows.

Max. fobj(x) = Vol(B) - Pen(B) (1)

n
Vol ( B)   ( Leni  Beri  Hig i ) /(Vol (Ctr )) (2)
i 1
Pen (B) = Plc(X) + Ovr(X) + Stb(X) + Weg(X) + Ori(X) + Lber(X) (3)

whereas, fobj(x) is the objective function for maximizing fitness function which calculates
total volume occupied by the packed bins inside the container, Plc(X) is the Placement (Boundary
crossing) constraint with the penalty value of 0.25. Ovr(X), Stb(X), Weg(X), Ori(X) and Lber(X)
represent the Overlapping of bins within it and the container, Stability aspect constraint, Weight
withstanding capability constraint, Orientation of the bin constraint and maximum Load bearing by the
container respectively with the penalty value of 0.1. Even though the DCGAP generates the optimal
and feasible packing pattern in less computational time and memory usage, the performance may not
satisfied for the lager problem size. This can be solved in this research by modifying the genetic
operators for better performance.

6. HEURISTIC GENETIC ALGORITHM

In Heuristic Genetic Algorithm module (HGA) the crossover and mutation operators had
enhanced to two-point-double-crossover (2PDC) and overloaded mutation operator (OM) respectively.
The implemented 2PDC is explained from the figure 3.

Parent 1:
422324442422224222221323342224221323342212323243242222131243242111323232421422324422423332422132132334211342422334221232324214444324411122323134224123
Parent II:
32412322142134112333241342123222114234411111231232333411421312345223341221142312122441444112323143341123311421433212412323221412312324123221
Parent III:
42122221313342324111233121412431423412134242121314231423221422332144113121421413241324342134212332142323321411412232441214123341112313242121
Crossover site: 07, 22, 51 and 68
Temp Offspring:
422324421421341123332423342224221323342212323243242222131243242111323232421422324422423332422132132334211342422334221232324214444324411122323134224123
OffspringI:
422324421421341123332423342224221323342212323243242221422332144113121421231121242242333242213212232334211342422334221232324214444324411122323134224123

Figure 3. 2-Point Double Crossover

4
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

In Figure 3, “Parent-x” are the randomly generated sample parents from a group of randomly
formulated population. For crossover, in this research, crossover sites have also been generated at
random and the sample sites are 07, 22, 51 and 68. So the strings between 58th position to 102nd
position were exchanged between the first and second parents in order to generate temporary offspring
denoted by ‘temp offspring’, which in turn again exchanges the strings from 30 th to 75th position with
the third parent, to give birth to a new child denoted by ‘offspring I’, which is inheriting the properties
from three parents. Thus the solution space increases by generating new combination of parameters.
Inorder to avoid stagnation at local optimal points, the generated offspring’s have been permitted to
operate over mutation genetic operator.

In this research work, as a application specific heuristic, two-point genetic mutation is

overloaded with orientation satisfaction constraint. By overloading, instead of changing the box type,
box orientation has been changed by interchanging the dimensional parameters such as length, breadth
and height of a box at random. Figure 4 explains the mutation operation. Randomly generated
mutation sites are 9 and 65. Decoding the 9th and 65th position gives 4th box in the rectangular prism
database and 12th box in the cylinder database. By applying mutation operation, mutated offspring
structure remains same, but the dimension value of those bins gets interchanged by considering the
user defined orientation constraint and is shown in the figure 4. Thus the stagnation has been avoided
and various orientations also checked by overloading operation. Based on many iterative results, it was
found that the strings in the starting position of the parents in the population remains same, even after
many generations. This can be avoided by using swapping operation.

Before Mutation Operation

Offspring I:42232444242222422222132334222414234121342421213142314232214223323441131214214231212244144411232
3143341211342422334221232324214444324411122323134224123
Mutation Site: 9, 65
Bin at 9th Position: Code : 2 & 4th Bin in Rectangular Prism Database. Dimension: Length = 4; Width = 8; and Height =
15.
Bin at 65th Position: Code : 3 & 12th Bin in the Cylinder Database. Dimension: Length = 15; Width = 8 (Dia.) and Height
= 8.
After Mutation Operation
Mutated New Offspring I:422324442422224222221323342224142341213424212131423142322142233234411312142142312122
441444112323143341211342422334221232324214444324411122323134224123
Bin at 9th Position: Code : 2 & 4th Bin in Rectangular Prism Database. Dim.: Length = 15; Width = 8; and Height = 4.
Bin at 65th Position: Code : 3 & 12th Bin in the Cylinder Database. Dim.: Length = 8 (Dia.); Width = 8; Height = 15.

Figure 4. Mutation Overloading

In this research, single point random swapping was implemented and is explained in the figure
5. Randomly generated swapping site is 82 and then the strings beyond 82nd position were swapped to
the front to form ‘swapped offspring’. Thus the same set of bins will be retained with different packing
sequence.

New Offspring I:422324442422224222221323342224142341213424212131423142322142233234411312142142312122

441444112323143341211342422334221232324214444324411122323134224123
Swapping Site: 12
Swapped offspring: 224222221323342224142341213424212131423142322142233234411312142142312122
441444112323143341211342422334221232324214444324411122323134224123422324442422

Figure 5. Swapping Operation

5
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

The HGA produces better results in par with the benchmark problems. But in real time
situations, logistics firms are receiving higher heterogeneity bins and bins of various sizes. This will
lead to formation of empty spaces near the boundaries and at the same time, the HGA output should be
in packer readable format. So in this research, special heuristic algorithm namely Tweaking Algorithm
(TA) had developed.

7. HYBRID GENETIC-TWEAKING ALGORITHM

Genetic algorithm in combination with Tweaking algorithm namely Hybrid Genetic Tweaking
Algorithm (HGTA) have been used to pack the bins with or without empty space inside the container
based on the concept of First-fit algorithm [4] and also to generate the output in packer readable
tabular and diagrammatical format. A sample box packing output shown in Table 1 denotes the box
type, its dimensions, placement position inside the container along with the volume data.

Table 1. Box Packing Output by Tweaking Algorithm

Box
Box Dimension Placement Corner Box Position Volume (X 1000 un its)
Corner
Assigned Box no

Empty Volume
Chromosome

Box Volume
Box Shape

X position

Y position

X position
Y position
Box Heig.
Box Bred.

Container
Box Len.

Column

Volume
Layer
S.No.

Row
1 1 Cube 15 15 15 CB1 0 0 15 15 1 1 1 3375 3375 559125

2 2 Rect. Prism 20 15 10 RP1 15 0 35 15 1 1 2 3000 6375 556125

3 1 Cube 15 15 15 CB2 35 0 50 15 1 1 3 3375 9750 552750

4 2 Rect. Prism 20 15 10 RP2 50 0 70 15 1 1 4 3000 12750 549750

5 1 Cube 15 15 15 CB3 70 0 85 15 1 1 5 3375 16125 546375

6 2 Rect. Prism 20 15 10 RP3 0 15 20 30 1 2 1 3000 19125 543375

7 2 Rect. Prism 20 15 10 RP4 20 15 40 30 1 2 2 3000 22125 540375

8 4 Sphere 20 20 20 SR1 40 15 60 35 1 2 3 4133 30125 536242

9 2 Rect. Prism 20 15 10 RP5 60 15 80 30 1 2 4 3000 33125 533242

10 2 Rect. Prism 20 15 10 RP6 80 15 100 30 1 2 5 3000 36125 530242

This tabular data have been converted to the diagrammatical pictorial format for easy
understanding of the user and packers. The sample diagrammatical format and the top view/plan of the
first bottom layer for the given dataset is shown in the Figure 6. Each box and its numerical value in
the figure 6 represent a bin and its serial number respectively.

6
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

Figure 6. Layer 1 Box Packing Pattern

8. RESULTS
Table 2 compares the utilization percentage of the loaded container for Bischoff and Ratcliff (1995) test
instances by different approaches. The data of HGTA are the best value obtained form continuous ten trials.

Table 2. Container Volume Utilization Comparison in Percentage.

Instance DCGAP HGA HGTA Improvement

BR1 72.56 84.34 92.92 20.36
BR 2 71.23 85.61 93.92 22.69
BR 3 71.89 85.81 93.71 21.82
BR 4 70.26 87.07 93.65 23.39
BR 5 71.74 86.46 93.68 21.94
BR 6 70.56 88.21 93.25 22.69
BR 7 70.24 85.96 93.10 22.86
BR 8 68.23 85.96 92.98 24.75
BR 9 65.84 86.23 92.32 26.48
BR 10 64.25 85.72 92.42 28.17
BR 11 64.32 85.85 91.62 27.30
BR 12 64.10 85.18 91.32 27.22
BR 13 62.36 85.40 91.25 28.89
BR 14 61.38 84.87 91.32 29.94
BR 15 61.24 85.41 91.12 29.88
Average 67.35 85.87 92.57 25.23

Table 2 interpret that, 25.23 % of the left out unpacked boxes were re-packed by special
heuristic TA into the partially loaded container compared to the GA with penalty function. After
applying TA the effectiveness of the box packing increased from 67.35 % to 92.57 %. Thus it is clear
that the effectiveness of the GA is increased by applying TA.

9. CONCLUSION
This research presents a better hybrid genetic approach for tackling 3D packing of different
shaped bins with varying sizes by incorporating the real time packing constraints/limitations. The
experimental results analyzed revels the fact, that heuristic genetic enhancements like double point
double crossover, orientation mutation and swapping operations increases the solution space. Feasible
box packing pattern has been obtained by satisfying the packing constraints. The effectiveness of the
genetic output is further enhanced by TA. Further improvements could also be incorporated by
considering non-prismatic shaped bins.

7
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

REFERENCES
[1] Allen, S.D., Burke, E.K. and Kendall, G. 2011 A hybrid placement strategy for the three-
dimensional strip packing problem, European Journal of Operational Research. 209,219-227.
[2] Bischoff, E.E. and Ratcliff 1995 Issues in the development of approaches to container loading,
OMEGA. 23,377-390.
[3] Bortfeldt, A. and Gehring, H. 2001 A hybrid genetic algorithm for container loading problem,
European Journal of Operation Research.131,143-161.
[4] George, J.A. and Robinson, D.F 1980 A heuristic for packing bins into a container, Journal of
Computers and Operations Research. 7,147-156.
[5] Rajesh Kanna S.K. and Udaiyakumar K.C. 2017 A complete framework for multi-constrained
3D bin packing optimization using firefly algorithm, International Journal of Pure and Applied
Mathematics,114,267-282.
[6] Hopper, E. and Turton, B 1997 Application of genetic algorithm to packing problems – a review,
Springer Verlag, London. 279 -288.
[7] Leonardo Junqueira, Reinaldo Morabito, and Denise Sato Yamashita 2012 Three-dimensional
container loading models with cargo stability and load bearing constraints, Journal of
Computers & Operations Research,39, 74-85.
[8] Lodi, Silvano Martello, and Daniele Vigo.2002 Recent advances on two-dimensional box
packing problems, Journal of Discrete Applied Mathematics,123,379 – 396.
[9] Martine Labbe, Gilbert Laporte, and Silvano Martello.2003 Upper bounds and algorithms for the
maximum cardinality box packing problem, European Journal of Operational Research, 149, 490–
498.
[10] Nihat Kasap, and Anurag Agarwal 2004 Augmented-Neural-Networks approach for the bin-packing
problem, Proceedings of 4th International Symposium on Intelligent Manufacturing Systems, Sakarya
University, Department of Industrial Engineering, 348-358.
[11] Rajesh Kanna S K and Saravana Manigandan 2012 3D Arbitrary Sized Bin Packing
Optimization Using 2PDC Heuristic: An Adaptive Genetic Approach. ARPN Journal of Science
and Technology.2,26-31.
[12] Wenqi Huang, and Kun He 2009 A caving degree approach for the single container loading
problem, European Journal of Operational Research, 196,93–101.
[13] Zhoujing Wang and Kevin W. Li 2006 A heuristic algorithm for the container loading problem
with heterogeneous boxes,IEEE International Conference on Systems, Man and Cybernetics.
[14] Bremermann, H. J., Rogson M., and Salaff S.1966 Global properties of Evolution Processes in
natural Automata and useful Simulation, Spartan Books.
[15] Rajesh Kanna S.K and Malliga. P. 2012 Multi-Constrained Optimization of Rectangular Bin
Packing Problem using Binary coded Evolutionary Algorithm, International Journal of Materials
Manufacturing and Optimization,1,27-35.
[16] Aristide Grange, Imed Kacem, Sébastien Martin 2018 Algorithms for the bin packing problem
with overlapping items, Computers & Industrial Engineering, 115, 331-341
[17] Ranga P. Abeysooriya, Julia A. Bennell 2018 Antonio Martinez-Sykora, Jostle heuristics for the
2D-irregular shapes bin packing problems with free rotation, International Journal of
Production Economics.195, 12-26
[18] Fabrizio Marinelli, Andrea Pizzuti,2018 A Sequential Value Correction heuristic for a bi-
objective two-dimensional bin-packing, Electronic Notes in Discrete Mathematics, 64, 25-34.
[19] A. Martinez-Sykora, R. Alvarez-Valdes, J.A. Bennell, R. Ruiz, J.M. Tamarit, 2017
Matheuristics for the irregular bin packing problem with free rotations, European Journal of
Operational Research, 258,440-455.
[20] Xueping Li, Kaike Zhang, 2018 Single batch processing machine scheduling with two-
dimensional bin packing constraints, International Journal of Production Economics,196,113-
121

8
2nd International conference on Advances in Mechanical Engineering (ICAME 2018) IOP Publishing
IOP Conf. Series: Materials Science and Engineering 402 (2018) 012203 doi:10.1088/1757-899X/402/1/012203
1234567890‘’“”

[21] Célia Paquay, Sabine Limbourg, Michaël Schyns 2018 A tailored two-phase constructive
heuristic for the three-dimensional Multiple Bin Size Bin Packing Problem with transportation
constraints, European Journal of Operational Research, 267, 52-64.