Upload
vutram
View
215
Download
0
Embed Size (px)
Citation preview
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
An Effective Genetic Algorithm for Capacitated Vehicle
Routing Problem
Hadeer Awada, Raafat Elshaera†, Adel AbdElmo’ezab and Gamal Nawaraa
a Industrial Engineering Department Faculty of Engineering, Zagazig University
Zagazig, Egypt
[email protected], [email protected], [email protected]
bIndustrial Engineering Department College of Engineering
King Khaled University, KSA
Abstract
The capacitated vehicle routing problem (CVRP) is an NP-hard problem. Therefore, metaheuristics are
often more suitable for practical applications. In this paper, a genetic algorithm (GA) is proposed to solve the problem. The performance of the proposed algorithm is tested on different sets of benchmark instances. The computational results indicate that the algorithm has a satisfactory performance in solving the problem.
Keywords Capacitated Vehicle Routing Problem, Genetic Algorithm
1. Introduction
The capacitated vehicle routing problem (CVRP), introduced by [1], is one of the most attractive topics in operation
research, communications, manufacturing, transportation, distribution, and logistics. It is of paramount importance
to thousands of companies and organizations engaged in the delivery and collection of goods or people. The CVRP
can be formally defined as follows [2]–[4]. it consists of designing a set of routes for a fleet of identical vehicles
with overall minimum route cost which service all the demands such that (1) all vehicles should begin and terminate
at the central depot, (2) each customer should be visited exactly once, by exactly one route, (3) the total demand of
each route does not exceed capacity of vehicle, (4) the traveling distance of each vehicle cannot exceed and the
maximum traveling distance of vehicle, and (5) the split deliveries are not allowed. For details about the variants of
the vehicle routing problem and their solution procedures, we refer the reader to several surveys and taxonomies by
[5], [6] and in many books or book chapters by [7], [8].
It is known that the CVRP is an NP-hard problem in which its real-life applications are considerably large in scale
and finding the optimal solution of an instance is very hard and requires very long computational time. Therefore,
metaheuristics are often more suitable for practical applications and have been applied for CVRP to find a near
optimal solution in a reasonable amount of time, for example: tabu search [9], [10], an adaptive memory
programming method [11], simulated annealing [12], [13], variable neighborhood search [14], [15], large neighborhood search [16], [17], ant colony optimization [16], [18], particle swarm optimization [19]–[21], a genetic
algorithm [22]–[24], and a hybrid genetic algorithm [4], [25]–[27] and other metaheuristics [28]–[30]. For details
about descriptions of these algorithms, we refer the reader to the survey papers [31], [32].
The main purpose of this paper is to present an effective genetic algorithm (GA) for solving the CVRP problem. The
remainder of the paper is organized as follows: The notation and model formulation of CVRP is given in Section 2.
A brief review has been conducted on the different GA algorithms used to solve the problem in Section 3. Section 4
† Corresponding author
374
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
presents the proposed GA algorithm and its default settings. In Section 5, the computational results are presented
and discussed. Finally, we give a brief conclusion in Section 6.
2. Notation and model formulation
The CVRP is defined as an undirected graph 𝐺 = (𝑁, 𝐸), where 𝑁 = {0, 1, … , 𝑛} is the set of nodes, 𝐸 = {(𝑖,𝑗) ∶ 𝑖, 𝑗 ∈ 𝑁, 𝑖 ≠ 𝑗} is the set of edges joining the nodes. Node 0 is the depot and the other nodes represent
the customers having a known demand 𝑑𝑖 for customer 𝑖. The travel distance between node 𝑖 and 𝑗 is defined by
𝑑𝑖𝑗 > 0 and each vehicle 𝑘 has a unique capacity of 𝑄𝑘 . In accordance with these explanations, CVRP can be
formulated [12] as given below, where 𝑋𝑖𝑗𝑘equals to 1 if vehicle 𝑘 travels from node 𝑖 to node 𝑗 and 0 otherwise.
𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 ∑ ∑ ∑ 𝑑𝑖𝑗
𝐾
𝑘=1
𝑋𝑖𝑗𝑘
𝑁
𝑗=0
𝑁
𝑖=0
(1)
Subject to:
∑ ∑ 𝑋𝑖𝑗𝑘 = 1 𝑗 𝜖 {1, … . . , 𝑛} ∶ 𝑖 ≠ 𝑗 (2)
𝑁
𝑖=0
𝐾
𝑘=1
∑ ∑ 𝑋𝑖𝑗𝑘 = 1 𝑖 𝜖 {1, … . . , 𝑛} ∶ 𝑖 ≠ 𝑗 (3)
𝑁
𝑗=0
𝐾
𝑘=1
∑ ∑ 𝑋𝑖𝑗𝑘 𝑑𝑖 ≤ 𝑄𝑘
𝑁
𝑗=0
𝑘 ∈ {1, … . , 𝐾}
𝑁
𝑖=0
(4)
∑ 𝑋𝑖𝑗𝑘
𝑁
𝑗=1
= ∑ 𝑋𝑗𝑖𝑘
𝑁
𝑗=1
≤ 1 𝑓𝑜𝑟 𝑖 = 0 𝑎𝑛𝑑 𝑘 𝜖 {1, … , 𝐾} (5)
∑ ∑ 𝑋𝑖𝑗𝑘
𝑁
𝑗=1
𝐾
𝑘=1
≤ 𝐾 𝑓𝑜𝑟 𝑖 = 0 (6)
Objective function (1) minimizes the total travelling distance. Constraint sets (2) and (3) guarantee that each
customer is served by exactly one vehicle. Constraint set (4) ensures that the total demand of the customers assigned
to a route 𝑘 does not exceed the vehicle capacity. Constraint set (5) indicates that the depot is the start and end node
for the trips of each vehicle. Constraint set (6) guarantees that there are maximum K routes for serving the
customers.
3. Genetic algorithm
Genetic algorithm (GA), proposed by [33], has been widely applied to solve hard combinatorial problems and it is
an effective search and optimization method that simulates the process of natural selection or survival of the fittest.
GA starts with generating random population of chromosomes. The chromosomes evolve through a series of
iterations, called generations. During each generation, the fitness of each chromosome in population is evaluated.
According to their fitness measure, select two parent chromosomes and can be crossed over by exchanging pieces
with each other and/or mutate randomly or be transferred unaltered to the next generation; this process is repeated
until a termination sequence (such as convergence) is reached [34].
In the next subsections, the GA and its elements, such as chromosome representation, initial population, GA
operators (such as crossover and mutation) and control parameters are described. Also, the different GA settings
mentioned in the literature concerning the problem under study are presented.
3.1. Chromosome representation
The first and most important step is to determine the chromosome representation (encoding). In CVRP literature,
chromosome representations include the permutation representation (e.g., [24], [26], [35]–[71]), direct
375
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
representation (e.g., [72]–[74]), path representation [27] and binary representation (e.g., [55], [75]). In this study, the
permutation encoding is adopted, so the review of the GA settings in the next sections is focus only on literature that
uses this encoding.
3.2. Initialization of population
There is more than one way to generate initial population. First way is to generate it randomly (e.g., [24], [35],
[37], [39], [41], [42], [44]–[46], [49], [54], [59], [61], [66], [68], [72], [75], [76]). Second way is to use heuristics for
generating the initial population, such as the nearest addition method (NAM), the sweep algorithm (SWA), the
savings algorithm (SA), the Clarke and Wright heuristic (C&W), the Push Forward Insertion Heuristic (PFIH), the
Nearest Neighbor Heuristic (NNH) and Insertion heuristic (IH) , etc, (e.g., [26], [27], [40], [47], [48], [55]–[58], [60],
[62]–[64], [66], [67], [70]). Other way is to a combination of heuristic and randomly (e.g., [36], [38], [43], [50], [51],
[53], [65], [69], [71], [73]).
3.3. Fitness function
The fitness measure for evaluating a chromosome is total travel distance. The objective is to achieve the smallest
value of that measure. The total travel distance is given by Equation 1.
3.4. Selection operators
During the research, four types of selection methods are found: Roulette Wheel Selection (RWS) (e.g., [26], [27],
[35], [39], [42]–[45], [51], [52], [55], [68], [71], [75]), Ranking Selection (RNKS) (e.g., [37], [40], [46], [63], [76],
[77]), Tournament Selection (TTS) (e.g., [24], [35], [36], [38], [39], [41], [43], [47]–[54], [56], [57], [59]–[61], [63],
[65], [70], [72], [73], [76] ) and Uniform Selection (e.g., [43], [51]).
3.5. Crossover operators
In the literature, ten types of crossover operator methods are found: One-Point Crossover (1PX) (e.g., [43], [44],
[46], [59]), Two-Point Crossover (2PX) (e.g., [35], [44], [49]), Order Crossover (OX) (e.g., [26], [36], [37], [43],
[48], [50], [54], [56], [57], [60], [62], [71], [76]), Partially Mapped Crossover (PMX) (e.g., [40]–[43], [45], [46],
[52], [67]), Cyclical Crossover (CX) (e.g., [43], [75]), Route Based Crossover (RBX) (e.g., [61], [70], [72], [76]),
Sequence-Based Crossover (SBX) (e.g., [76]), Single Parent Crossover (SPO) (e.g., [36]), Genetic Vehicle
Representation Crossover (GVR) (e.g., [69]) and Best Cost-Best Route Crossover (BCBRC) (e.g., [38], [53], [64]–
[66]).
3.6. Mutation operators
Six types of mutation operator methods during research are found: Swap Mutation (SWM) (e.g., [24], [26], [35],
[37], [39]–[42], [44], [46], [49], [59], [62], [64], [65], [67]), Inversion Mutation (INVM) (e.g., [24], [45], [53], [64],
[65]), Insertion Mutation (INSM) (e.g., [41], [64], [67]), (Reallocation Mutation (RAM), Exchange Mutation (EXM)
and Reposition Mutation (RPM) (e.g., [61], [72], [73])).
3.7. Termination condition and parameter selection
The control parameters that control the execution of the GA and its operators are population size (𝑝𝑜𝑝𝑆𝑖𝑧 > 10),
crossover rate (𝐶𝑅 ∈ [0,1]), mutation rate (𝑀𝑅 ∈ [0,1]) and stop criteria. Each publication has its values which are
selected based on its own experiment. Many stop criteria can be used to termination the algorithm, but the known
one is the maximum number of generation (𝑚𝑎𝑥𝐺𝑒𝑛). Where 𝑚𝑎𝑥𝐺𝑒𝑛 equals number of explored chromosomes
divided by 𝑝𝑜𝑝𝑠𝑖𝑧.
4. The Proposed GA Characteristics
The good performance of a GA depends on the selection of a good combination of GA operators and parameters.
Based on the GA review in section 3, there are three selection methods (four versions of tournament methods), ten
crossover methods, and six mutation methods as shown in Table 1. For selecting our proposed GA, we design full
376
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
factorial experiment with the GA operators and parameters shown in Table 1 and Table 2 respectively. Where Table
2 shows five, nine and seven levels for population size, crossover probability and mutation probability respectively.
As mention above, the stop criterion, 𝑚𝑎𝑥𝐺𝑒𝑛, is calculated based on the used population size at fixed number of
explored solutions 5000 (i.e. 𝑚𝑎𝑥𝐺𝑒𝑛 = 5000/𝑝𝑜𝑝𝑆𝑖𝑧 =). Therefore, the total number of combinations of GA
operators and parameters equals 113400 (6 × 10 × 6 × 5 × 9 × 7). The implementation model of the proposed has
been coded in C# language and the experiment is applied on the problems (A-n32-k5, A-n44-k6, A-n53-k7 and A-n60-k9) of set A which are mentioned in next section. The statistical analysis conducted in the computational results
shows that the best settings of our proposed GA is as shown in Table 3.
Table 1. Selection, Crossover and Mutation methods
used in CVRP
Table 2. GA parameters
Table 3. Proposed GA settings (Default settings)
5. Computational study
The performance of the proposed algorithm has been validated on groups of benchmark problems. The benchmark
problems consist of six sets as follow: seventy-four instances ( sets A, B and P) from [78], eleven instances (set E)
from [79], three instances (set F)from [80], five instances (set M) from [81]. All the problem instances were
downloaded from the site <http://branchandcut.org>.
The computational results of benchmark problems were obtained by running the coded algorithm on an Intel® Core
(TM) i3-4160 CPU 3.60 GHz personal computer. The results of the problems are as shown in Tables 4-9 for the
problem sets A, B, E, F, M and P respectively. In the first column of the tables the name of each problem is denoted.
The next three columns are the most important characteristics, number of nodes (n), capacity of the vehicles (Q), and number of Vehicles (K). The fifth column shows the best-known solution (BKS). Columns 6–9 present respectively
the best, the worst, the average and the standard deviation of 10 runs of the proposed algorithm. The efficiency of
the proposed GA algorithm is measured by the quality of the produced solutions. The quality is given in terms of the
percentage relative deviation from the best-known solution, %𝐷𝑒𝑣 = 100 ∗ (𝐵𝑒𝑠𝑡 − 𝐵𝐾𝑆)/𝐵𝐾𝑆 shown in last
column. The numbers in bold in column (Best) indicates that the solution obtained meets the best known one (i.e.
%𝐷𝑒𝑣 = 0.00). Whereas, a negative %𝐷𝑒𝑣 indicates that the solution obtained by the proposed GA is better than
the best known one and it is marked with (*).
It can be seen from Table 4 (Set A) that the proposed algorithm in seventeen out of the twenty-seven instances has
reached the best-known solution. For the other thirteen instances the quality of the solutions is between 0.17% and
2.33% and the average quality for the twenty-seven instances is 0.26%. For the Table 5 (Set B) the algorithm has
found the best-known solution in fifteen out of the twenty-three instances, for the rest the quality is between 0.08%
and 1.72% and the average quality for the twenty-three instances is 0.18%. From the observation of the Table 6 (Set
E), there are five best known solutions, for the rest the quality is between 0.2% and 3.44% and the average quality
for the 11 instances is 0.75. In Table 7 (Set F), our algorithm obtains one best known solution, two new solutions in
problem (F-n45-k4 and F-n135-k7) and the average quality for the 3 instances is -0.19%. For the Table 8 (Set M) the
Methods Codes
Sele
cti
on
s Roulette Wheel Selection RWS
Ranking Selection RNKS
Tournament Selection (tour size = 2) TTS-2
Tournament Selection (tour size = 3) TTS-3
Tournament Selection (tour size = 4) TTS-4
Tournament Selection (tour size = 5) TTS-5
Cross
over
Order Crossover OX
Partially Mapped Crossover PMX
Cyclical Crossover CX
Route Based Crossover RBX
Sequence-Based Crossover SBX
Single Parent Crossover SPO
Genetic Vehicle Representation Crossover GVR
Best Cost-Best Route Crossover BCBRC
Mu
tati
on
s Swap Mutation SWM
Inversion Mutation INVM
Insertion Mutation INSM
Reallocation Mutation RAM
Exchange Mutation EXM
Reposition Mutation RPM
Parameters Value
Population size, 𝑝𝑜𝑝𝑆𝑖𝑧 10, 20, 40, 50, 100
Crossover probability, 𝐶𝑅 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7,
0.8, 0.9
Mutation Probability, 𝑀𝑅
Number of Generations, 𝑚𝑎𝑥𝐺𝑒𝑛
0.05,0.1, 0.2, 0.3, 0.4, 0.5, 0.6
5000/𝑝𝑜𝑝𝑆𝑖𝑧
Operators / Parameters Type / Value
Selection method Ranking Selection
Crossover operator Best Cost-Best Route Crossover
Mutation operator Exchange Mutation
Crossover probability 0.8
Mutation Probability 0.1
Population size 10
377
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
quality of solutions is between 0.1 and 17.65% and the average quality for the five instances is 6.4%. Of the twenty-
three problem instances in Table 9 (Set P), our proposed algorithm obtains thirteen best known solution, the
percentages above the best-known solutions the quality is between 0.12% and 1.47% and the average quality for the
23 instances is 0.18%.
We summarized the results given in Tables 4-9 in Table 10. The table indicates that the proposed GA algorithm has
a satisfactory performance in solving CVRP problems especially in terms of solution quality. The algorithm is
capable to solve 51 problem instances which reached optimum and finds 2 new best solution problem instances out
of 93 and it is refers to an average success rate (OA%) for both it is 55.4% and 2.2% respectively. Regarding each set, the algorithm solves the problems sets A, B, P, F, E and M with percentage success rates of optimum
achievements 62.9%, 65.2%, 56.5%, 33.3%, 45.5% and 0.00% respectively, and with percentage success rates of
number of new solutions (NNS%) 0.00%, 0.00%, 0.00%, 66.7%, 0.00% and 0.00% respectively. For more
illustration, the performance of the proposed algorithm for each set is shown in Error! Reference source not
found.1.
Table 4. Computational results for the problem set A. Table 5. Computational results for the problem set B.
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
A-n32-k5 32 100 5 784 784 784 784 0.00 0.00
A-n33-k5 33 100 5 661 661 661 661 0.00 0.00
A-n33-k6 33 100 6 742 742 742.9 742.9 0.30 0.00
A-n34-k5 34 100 5 778 778 782 782 4.20 0.00
A-n36-k5 36 100 5 799 799 808.8 808.8 4.80 0.00
A-n37-k5 37 100 5 669 669 673.2 673.2 5.50 0.00
A-n37-k6 37 100 6 949 949 949.7 949.7 1.30 0.00
A-n38-k5 38 100 5 730 730 731.2 731.2 2.20 0.00
A-n39-k5 39 100 5 822 822 823.5 823.5 1.60 0.00
A-n39-k6 39 100 6 831 831 832.5 832.5 2.20 0.00
A-n44-k6 44 100 6 937 937 938.5 938.5 2.20 0.00
A-n45-k6 45 100 6 944 944 949.1 949.1 5.60 0.00
A-n45-k7 45 100 7 1146 1148 1163.5 1163.5 16.8 0.17
A-n46-k7 46 100 7 914 914 920.7 920.7 12.8 0.00
A-n48-k7 48 100 7 1073 1073 1095.4 1095.4 16.3 0.00
A-n53-k7 53 100 7 1010 1010 1017.8 1017.8 5.90 0.00
A-n54-k7 54 100 7 1167 1172 1184.5 1184.5 13.6 0.43
A-n55-k9 55 100 9 1073 1073 1083.6 1083.6 13.9 0.00
A-n60-k9 60 100 9 1354 1354 1373.7 1373.7 14.9 0.00
A-n61-k9 61 100 9 1034 1037 1043.7 1043.7 11.4 0.29
A-n62-k8 62 100 8 1288 1299 1324.7 1324.7 22.3 0.85
A-n63-k9 63 100 9 1616 1621 1633.7 1633.7 8.90 0.31
A-n63-k10 63 100 10 1314 1319 1327.3 1327.3 10.0 0.38
A-n64-k9 64 100 9 1401 1418 1439.1 1439.1 14.3 1.21
A-n65-k9 65 100 9 1174 1177 1182.1 1182.1 6.30 0.26
A-n69-k9 69 100 9 1159 1169 1181 1181 7.80 0.86
A-n80-k10 80 100 10 1763 1804 1833.5 1833.5 33.8 2.33
Table 6. Computational results for the problem set E. Table 7. Computational results for the problem set F.
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
F-n45-k4 45 2010 4 724 721 725 721.8 1.70 -0.41*
F-n72-k4 72 30000 4 237 237 241 237.8 1.70 0.00
F-n135-k7 135 2210 7 1162 1160 1230 1189.3 20.6 -0.17*
Table 8. Computational results for the problem set M.
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
M-n101-k10 101 200 10 820 842 897 866 16.9 2.68
M-n121-k7 121 200 7 1034 1035 1188 1129 64.2 0.10
M-n151-k12 151 200 12 1053 1095 1175 1137.5 26.0 7.88
M-n200-k16 200 200 16 1274 1319 1394 1348.6 20.1 3.53
M-n200-k17 200 200 17 1373 1500 1665 1603.1 49.2 17.65
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
B-n31-k5 31 100 5 672 672 680 673.5 2.70 0.00
B-n34-k5 34 100 5 788 788 789 788.7 0.50 0.00
B-n35-k5 35 100 5 955 955 955 955 0.00 0.00
B-n38-k6 38 100 6 805 805 805 805 0.00 0.00
B-n39-k5 39 100 5 549 549 550 549.1 0.30 0.00
B-n41-k6 41 100 6 829 829 829 829 0.00 0.00
B-n43-k6 43 100 6 742 742 743 742.1 0.30 0.00
B-n44-k7 44 100 7 909 909 909 909 0.00 0.00
B-n45-k5 45 100 5 751 751 751 751 0.00 0.00
B-n45-k6 45 100 6 678 680 687 683.2 3.40 0.29
B-n50-k7 50 100 7 741 741 744 741.7 1.10 0.00
B-n50-k8 50 100 8 1312 1315 1337 1323.5 6.20 0.23
B-n51-k7 51 100 7 1032 1032 1033 1032.1 0.30 0.00
B-n52-k7 52 100 7 747 724 754 747.9 2.20 0.00
B-n56-k7 56 100 7 707 707 724 712.1 4.40 0.00
B-n57-k7 57 100 7 1153 1153 1153 1153 0.00 0.00
B-n57-k9 57 100 9 1598 1612 1649 1636 10.30 0.88
B-n63-k10 63 100 10 1496 1504 1552 1528.3 17.40 0.53
B-n64-k9 64 100 9 861 861 878 868.2 8.50 0.00
B-n66-k9 66 100 9 1316 1321 1335 1327.2 4.00 0.38
B-n67-k10 67 100 10 1032 1033 1052 1040.8 6.30 0.10
B-n68-k9 68 100 9 1272 1273 1293 1286.6 7.60 0.08
B-n78-k10 78 100 10 1221 1242 1281 1259.9 14.8 1.72
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
E-n22-k4 22 6000 4 375 375 375 375 0.00 0.00
E-n23-k3 23 4500 3 569 569 569 569 0.00 0.00
E-n30-k3 30 4500 3 534 534 537 534.3 0.90 0.00
E-n33-k4 33 8000 4 835 835 835 835 0.00 0.00
E-n51-k5 51 160 5 521 521 539 528.9 4.50 0.00
E-n76-k7 76 220 7 682 691 723 706.4 9.40 1.32
E-n76-k8 76 180 8 735 738 760 749.6 7.10 0.41
E-n76-k10 76 140 10 830 838 855 844.4 6.00 0.96
E-n76-k14 76 100 14 1021 1023 1058 1038.2 10.3 0.20
E-n101-k8 101 200 8 817 843 895 873.9 18.3 3.44
E-n101-k14 101 112 14 1077 1087 1144 1125.4 16.8 1.87
378
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
Table 9. Computational results for the problem set P.
Table 10. Summary of computational study.
OA: Number of optimum achievements; NP: Number of problems;
NNS: Number of new solutions, NDS: Number of deviated solutions
Figure 1. Performance of the proposed GA algorithm for each data set.
Problem n Q K BKS Proposed GA
Best Worst Average SD %Dev
P-n16-k8 16 35 8 450 450 450 450 0.00 0.00
P-n19-k2 19 160 2 212 212 212 212 0.00 0.00
P-n20-k2 20 160 2 216 216 216 216 0.00 0.00
P-n21-k2 21 160 2 211 211 211 211 0.00 0.00
P-n22-k2 22 160 2 216 216 216 216 0.00 0.00
P-n22-k8 22 3000 8 603 603 603 603 0.00 0.00
P-n23-k8 23 40 8 529 529 529 529 0.00 0.00
P-n40-k5 40 140 5 458 458 469 461.7 4.20 0.00
P-n45-k5 45 150 5 510 510 523 516.4 4.90 0.00
P-n50-k7 50 150 7 554 554 582 563.4 8.60 0.00
P-n50-k8 50 120 8 631 632 663 644.3 10.2 0.16
P-n50-k10 50 100 10 696 697 712 703.3 4.90 0.14
P-n51-k10 51 80 10 741 741 776 755.1 10.2 0.00
P-n55-k7 55 170 7 568 570 591 581.9 6.40 0.35
P-n55-k10 55 115 10 694 695 709 704.3 4.40 0.14
P-n55-k15 55 70 15 989 989 989 989 0.00 0.00
P-n60-k10 60 120 10 744 748 768 756.1 6.10 0.54
P-n60-k15 60 80 15 968 974 1001 981 8.60 0.62
P-n65-k10 65 130 10 792 792 820 805.1 6.90 0.00
P-n70-k10 70 135 10 827 828 849 840 6.70 0.12
P-n76-k4 76 350 4 593 595 622 604 8.60 0.34
P-n76-k5 76 280 5 627 629 652 637 7.60 0.32
P-n101-k4 101 400 4 681 691 740 709.6 14.8 1.47
Algorithm Problem
Set OA NNS NDS NP OA% NNS% NDS%
Proposed
GA
A 17 0 10 27 62.9 0.00 37.0
B 15 0 8 23 65.2 0.00 34.8
P 13 0 10 23 56.5 0.00 43.5
F 1 2 0 3 33.3 66.7 0.00
E 5 0 6 11 45.5 0.00 54.5
M 0 0 5 5 0 0.00 100
Overall 51 2 39 92
Average 55.4 2.2 42.4
0%
20%
40%
60%
80%
100%
A B P F E M
Instances
Percentage ofdeviation solution
Percentage successrate of new solutions
percentage of successrate of optimumachievements
379
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
To study the impact of stop criterion (maximum number of generation, 𝑚𝑎𝑥𝐺𝑒𝑛) on the performance of proposed
GA, the following values, 𝑚𝑎𝑥𝐺𝑒𝑛 = {200, 500, 1000, 2000, 5000, 10000, 20000, 50000}, are applied on the
benchmark problems. The average percentage deviation and the standard deviation of ten runs on each set of
problems are computed as mention in Table 11. For more illustration, the 𝑚𝑎𝑥𝐺𝑒𝑛 and its corresponding grand
average of the percentage deviation of all data set are shown in Figure 2. The vertical line represents the standard
error of the average. The figure reveals that the proposed algorithm can keep finding better results as more
generations are explored.
Table 11. Effect of number of explored solutions on the %Dev
Problem
set
Number of generations, 𝑚𝑎𝑥𝐺𝑒𝑛
Size 200
500
1000
2000
5000
10000
20000
50000
Avg SD
Avg SD
Avg SD
Avg SD
Avg SD
Avg SD
Avg SD
Avg SD
A 27 3.17 3.71 1.53 1.89 0.89 1.21 0.64 1.06 0.57 1.01 0.39 0.69 0.34 0.63 0.26 0.52
B 23 1.94 1.89 1.08 1.47 0.78 1.23 0.48 0.79 0.37 0.68 0.31 0.57 0.23 0.46 0.18 0.40
P 23 2.95 3.02 1.69 2.14 0.96 1.48 0.72 1.41 0.49 0.61 0.51 0.86 0.22 0.35 0.18 0.34
F 3 1.44 2.51 1.28 2.09 0.92 1.97 0.81 1.77 0.58 1.38 0.06 0.51 0.04 0.47 -0.19 0.21
E 11 6.78 7.68 4.34 6.04 3.35 4.78 2.33 2.83 1.59 2.27 1.44 2.17 1.02 1.50 0.75 1.09
M 5 37.7 15.3 27.1 14.6 21.5 15.8 16.5 13.6 13.5 9.14 8.88 9.16 10.6 8.53 6.37 6.90
Overall 92 5.05 7.99 3.17 5.84 2.29 4.69 1.68 3.61 1.33 2.96 0.98 1.94 0.91 2.35 0.59 1.41
Figure 2. Impact of number of generation, 𝑚𝑎𝑥𝐺𝑒𝑛 on the performance of proposed GA.
6. Conclusions and future research
In this paper, the capacitated vehicle routing problem (CVRP) is addressed using a genetic algorithm (GA). Due the
quality of the performance of any genetic algorithm is depends on the good selection of its operators and parameters,
based on a review on CVRP literature solved using GA, three selection methods, ten crossover methods, and six
mutation methods are found. For selecting our proposed GA, a full factorial experiment with the GA operators and
parameters is designed. The performance of the proposed algorithm is tested on different sets of benchmark
instances. The computational results indicate that the algorithm is able to solve CVRP instances with a satisfactory
performance and proved to be very stable and efficient based on the comparisons performed with the best-known
solutions. Future researches will be focus on implementing the proposed GA on stochastic routing problems, like Vehicle Routing Problem with Stochastic Demand (VRPSD).
References [1] Dantzig, G. B. and J. H. Ramser, “The Truck Dispatching Problem,” Manage. Sci., vol. 6, no. 1, pp. 80–91,
Oct. 1959.
[2] Cordeau, J.-F., M. Gendreau, G. Laporte, J.-Y. Potvin, and F. Semet, “A guide to vehicle routing heuristics,”
J. Oper. Res. Soc., vol. 53, no. 5, pp. 512–522, May 2002. [3] Lysgaard, J., A. N. Letchford, and R. W. Eglese, “A new branch-and-cut algorithm for the capacitated
vehicle routing problem,” Math. Program., vol. 100, no. 2, pp. 423–445, Jun. 2004.
[4] Prins, C., “A simple and effective evolutionary algorithm for the vehicle routing problem,” Comput. Oper.
Res., vol. 31, no. 12, pp. 1985–2002, Oct. 2004.
-1
1
3
5
7
9
200 500 1000 2000 5000 10000 20000 50000
% D
ev
Number of generation, maxGen
380
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
[5] Eksioglu, B., A. V. Vural, and A. Reisman, “The vehicle routing problem: A taxonomic review,” Comput.
Ind. Eng., vol. 57, no. 4, pp. 1472–1483, 2009.
[6] Braekers, K., K. Ramaekers, and I. Van Nieuwenhuyse, “The vehicle routing problem: State of the art
classification and review,” Comput. Ind. Eng., vol. 99, pp. 300–313, Sep. 2016.
[7] Cordeau, J.-F., G. Laporte, M. W. P. Savelsbergh, and D. Vigo, “Chapter 6 Vehicle Routing,” in
Transportation, vol. 14, no. 06, 2007, pp. 367–428.
[8] Golden, B., S. Raghavan, and E. Wasil, The Vehicle Routing Problem: Latest Advances and New
Challenges, vol. 43. Boston, MA: Springer US, 2008. [9] Jin, J., T. Gabriel, and A. Løkketangen, “Computers & Operations Research A cooperative parallel
metaheuristic for the capacitated vehicle routing problem,” Comput. Oper. Res., vol. 44, pp. 33–41, 2014.
[10] Cordeau, J.-F. and M. Maischberger, “A parallel iterated tabu search heuristic for vehicle routing problems,”
Comput. Oper. Res., vol. 39, no. 9, pp. 2033–2050, 2012.
[11] Tarantilis, C. D., “Solving the vehicle routing problem with adaptive memory programming methodology,”
Comput. Oper. Res., vol. 32, no. 9, pp. 2309–2327, 2005.
[12] Lin, S. W., Z. J. Lee, K. C. Ying, and C. Y. Lee, “Applying hybrid meta-heuristics for capacitated vehicle
routing problem,” Expert Syst. Appl., vol. 36, no. 2 PART 1, pp. 1505–1512, 2009.
[13] Fleming, C. L., S. E. Griffis, and J. E. Bell, “The effects of triangle inequality on the vehicle routing
problem,” Eur. J. Oper. Res., vol. 224, no. 1, pp. 1–7, 2013.
[14] Bouzid, M. C., H. Aït Haddadene, and S. Salhi, “An integration of Lagrangian split and VNS: The case of
the capacitated vehicle routing problem,” Comput. Oper. Res., vol. 78, pp. 513–525, Feb. 2017.
[15] Chen, P., H. kuan Huang, and X. Y. Dong, “Iterated variable neighborhood descent algorithm for the
capacitated vehicle routing problem,” Expert Syst. Appl., vol. 37, no. 2, pp. 1620–1627, 2010.
[16] Akpinar, S., “Hybrid large neighbourhood search algorithm for capacitated vehicle routing problem,” Expert
Syst. Appl., vol. 61, pp. 28–38, Nov. 2016.
[17] Sze, J., S. Salhi, and N. Wassan, “A hybridisation of adaptive variable neighbourhood search and large neighbourhood search: Application to the vehicle routing problem,” Expert Syst. Appl., vol. 65, pp. 383–
397, 2016.
[18] Yu, B., Z. Z. Yang, and B. Yao, “An improved ant colony optimization for vehicle routing problem,” Eur. J.
Oper. Res., vol. 196, no. 1, pp. 171–176, 2009.
[19] Ai, T. J. and V. Kachitvichyanukul, “Particle swarm optimization and two solution representations for
solving the capacitated vehicle routing problem,” Comput. Ind. Eng., vol. 56, no. 1, pp. 380–387, Feb. 2009.
[20] Marinakis, Y., M. Marinaki, and G. Dounias, “A hybrid particle swarm optimization algorithm for the
vehicle routing problem,” Eng. Appl. Artif. Intell., vol. 23, no. 4, pp. 463–472, Jun. 2010.
[21] Chen, A., G. Yang, and Z. Wu, “Hybrid discrete particle swarm optimization algorithm for capacitated
vehicle routing problem,” J. Zhejiang Univ. Sci. A, vol. 7, no. 4, pp. 607–614, 2006.
[22] Baker, B. M. and M. A. Ayechew, “A genetic algorithm for the vehicle routing problem,” Comput. Oper.
Res., vol. 30, no. 5, pp. 787–800, Apr. 2003.
[23] Marinakis, Y., A. Migdalas, and P. M. Pardalos, “A new bilevel formulation for the vehicle routing problem
and a solution method using a genetic algorithm,” J. Glob. Optim., vol. 38, no. 4, pp. 555–580, Jun. 2007.
[24] Nazif, H. and L. S. Lee, “Optimised crossover genetic algorithm for capacitated vehicle routing problem,”
Appl. Math. Model., vol. 36, no. 5, pp. 2110–2117, 2012.
[25] Berger, J. and M. Barkaoui, “A new hybrid genetic algorithm for the capacitated vehicle routing problem,” J. Oper. Res. Soc., vol. 54, no. 12, pp. 1254–1262, Dec. 2003.
[26] Wang, C.-H. and J.-Z. Lu, “A hybrid genetic algorithm that optimizes capacitated vehicle routing
problems,” Expert Syst. Appl., vol. 36, no. 2, pp. 2921–2936, 2009.
[27] Marinakis, Y. and M. Marinaki, “A hybrid genetic – Particle Swarm Optimization Algorithm for the vehicle
routing problem,” Expert Syst. Appl., vol. 37, no. 2, pp. 1446–1455, Mar. 2010.
[28] Teymourian, E., V. Kayvanfar, G. M. Komaki, and M. Zandieh, “Enhanced intelligent water drops and
cuckoo search algorithms for solving the capacitated vehicle routing problem,” Inf. Sci. (Ny)., vol. 334–335,
pp. 354–378, Mar. 2016.
[29] Alinaghian, M. and M. Naderipour, “A novel comprehensive macroscopic model for time-dependent vehicle
routing problem with multi-alternative graph to reduce fuel consumption: A case study,” Comput. Ind. Eng.,
vol. 99, pp. 210–222, Sep. 2016.
[30] Szeto, W. Y., Y. Wu, and S. C. Ho, “An artificial bee colony algorithm for the capacitated vehicle routing
problem,” Eur. J. Oper. Res., vol. 215, no. 1, pp. 126–135, 2011.
[31] Boussaïd, I., J. Lepagnot, and P. Siarry, “A survey on optimization metaheuristics,” Inf. Sci. (Ny)., vol. 237,
381
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
pp. 82–117, 2013.
[32] Blum, C. and A. Roli, “Metaheuristics in Combinatorial Optimization: Overview and Conceptual
Comparison,” ACM Comput. Surv., vol. 35, no. 3, pp. 268–308, Sep. 2003.
[33] Holland, J. H., Adaptation in natural and artificial systems. University of Michigan Press, Ann Arbor.,
1975.
[34] Ho, W., G. T. S. Ho, P. Ji, and H. C. W. Lau, “A hybrid genetic algorithm for the multi-depot vehicle
routing problem,” Eng. Appl. Artif. Intell., vol. 21, no. 4, pp. 548–557, 2008.
[35] Cheng, C.-B. and K.-P. Wang, “Solving a vehicle routing problem with time windows by a decomposition technique and a genetic algorithm,” Expert Syst. Appl., vol. 36, no. 4, pp. 7758–7763, 2009.
[36] Liu, S., W. Huang, and H. Ma, “An effective genetic algorithm for the fleet size and mix vehicle routing
problems,” Transp. Res. Part E Logist. Transp. Rev., vol. 45, no. 3, pp. 434–445, 2009.
[37] Karlaftis, M. G., K. Kepaptsoglou, and E. Sambracos, “Containership routing with time deadlines and
simultaneous deliveries and pick-ups,” Transp. Res. Part E Logist. Transp. Rev., vol. 45, no. 1, pp. 210–221,
2009.
[38] Ghoseiri, K. and S. F. Ghannadpour, “Multi-objective vehicle routing problem with time windows using
goal programming and genetic algorithm,” Appl. Soft Comput., vol. 30, no. 2, pp. 286–296, 2010.
[39] Yu, S., C. Ding, and K. Zhu, “A hybrid GA-TS algorithm for open vehicle routing optimization of coal
mines material,” Expert Syst. Appl., vol. 38, no. 8, pp. 10568–10573, 2011.
[40] Shanmugam, G., P. Ganesan, and P. T. Vanathi, “Meta Heuristic Algorithms for Vehicle Routing Problem
with Stochastic Demands,” J. Comput. Sci., vol. 7, no. 4, pp. 533–542, 2011.
[41] Ursani, Z., D. Essam, D. Cornforth, and R. Stocker, “Localized genetic algorithm for vehicle routing
problem with time windows,” Appl. Soft Comput. J., vol. 11, no. 8, pp. 5375–5390, 2011.
[42] Tasan, A. S. and M. Gen, “A genetic algorithm based approach to vehicle routing problem with
simultaneous pick-up and deliveries,” Comput. Ind. Eng., vol. 62, no. 3, pp. 755–761, 2012.
[43] Lu, C.-C. and V. F. Yu, “Data envelopment analysis for evaluating the efficiency of genetic algorithms on solving the vehicle routing problem with soft time windows,” Comput. Ind. Eng., vol. 63, no. 2, pp. 520–
529, 2012.
[44] Anbuudayasankar, S. P., K. Ganesh, S. C. Lenny Koh, and Y. Ducq, “Modified savings heuristics and
genetic algorithm for bi-objective vehicle routing problem with forced backhauls,” Expert Syst. Appl., vol.
39, no. 3, pp. 2296–2305, 2012.
[45] Zhang, T., W. A. Chaovalitwongse, and Y. Zhang, “Scatter search for the stochastic travel-time vehicle
routing problem with simultaneous pick-ups and deliveries,” Comput. Oper. Res., vol. 39, no. 10, pp. 2277–
2290, 2012.
[46] Moon, I., J. H. Lee, and J. Seong, “Vehicle routing problem with time windows considering overtime and
outsourcing vehicles,” Expert Syst. Appl., vol. 39, no. 18, pp. 13202–13213, 2012.
[47] Vidal, T., T. Crainic, M. Gendreau, N. Lahnrichi, and W. Rei, “A hybrid genetic algorithm for multi-depot
and periodic vehicle routing problems,” Oper. Res., vol. 60, no. 3, pp. 611–624, 2012.
[48] Vidal, T., T. G. Crainic, M. Gendreau, and C. Prins, “A hybrid genetic algorithm with adaptive diversity
management for a large class of vehicle routing problems with time-windows,” Comput. Oper. Res., vol. 40,
no. 1, pp. 475–489, Jan. 2013.
[49] Pop, P. C., O. Matei, and C. P. Sitar, “An improved hybrid algorithm for solving the generalized vehicle
routing problem,” Neurocomputing, vol. 109, pp. 76–83, 2013. [50] Liu, R., X. Xie, V. Augusto, and C. Rodriguez, “Heuristic algorithms for a vehicle routing problem with
simultaneous delivery and pickup and time windows in home health care,” Eur. J. Oper. Res., vol. 230, no.
3, pp. 475–486, 2013.
[51] Barkaoui, M. and M. Gendreau, “An adaptive evolutionary approach for real-time vehicle routing and
dispatching,” Comput. Oper. Res., vol. 40, no. 7, pp. 1766–1776, 2013.
[52] Kergosien, Y., C. Lenté, J.-C. Billaut, and S. Perrin, “Metaheuristic algorithms for solving two
interconnected vehicle routing problems in a hospital complex,” Comput. Oper. Res., vol. 40, no. 10, pp.
2508–2518, Oct. 2013.
[53] Amorim, P. and B. Almada-Lobo, “The impact of food perishability issues in the vehicle routing problem,”
Comput. Ind. Eng., vol. 67, no. 1, pp. 223–233, 2014.
[54] Vidal, T., T. G. Crainic, M. Gendreau, and C. Prins, “Implicit depot assignments and rotations in vehicle
routing heuristics,” Eur. J. Oper. Res., vol. 237, no. 1, pp. 15–28, 2014.
[55] Nguyen, P. K., T. G. Crainic, and M. Toulouse, “A hybrid generational genetic algorithm for the periodic
vehicle routing problem with time windows,” J. Heuristics, vol. 20, no. 4, pp. 383–416, 2014.
382
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
[56] Liu, R., Z. Jiang, and N. Geng, “A hybrid genetic algorithm for the multi-depot open vehicle routing
problem,” OR Spectr., vol. 36, no. 2, pp. 401–421, 2014.
[57] Koç, Ç., T. Bektaş, O. Jabali, and G. Laporte, “A hybrid evolutionary algorithm for heterogeneous fleet
vehicle routing problems with time windows,” Comput. Oper. Res., vol. 64, pp. 11–27, Dec. 2015.
[58] Barkaoui, M., J. Berger, and A. Boukhtouta, “Customer satisfaction in dynamic vehicle routing problem
with time windows,” Appl. Soft Comput. J., vol. 35, pp. 423–432, 2015.
[59] Beheshti, A. K., S. R. Hejazi, and M. Alinaghian, “The vehicle routing problem with multiple prioritized
time windows: A case study,” Comput. Ind. Eng., vol. 90, pp. 402–413, Dec. 2015. [60] Vidal, T., M. Battarra, A. Subramanian, and G. Erdogˇan, “Hybrid metaheuristics for the Clustered Vehicle
Routing Problem,” Comput. Oper. Res., vol. 58, pp. 87–99, Jun. 2015.
[61] García-Nájera, A., J. A. Bullinaria, and M. A. Gutiérrez-Andrade, “An evolutionary approach for multi-
objective vehicle routing problems with backhauls,” Comput. Ind. Eng., vol. 81, pp. 90–108, 2015.
[62] Bae, H. and I. Moon, “Multi-depot vehicle routing problem with time windows considering delivery and
installation vehicles,” Appl. Math. Model., vol. 40, no. 13–14, pp. 6536–6549, Jul. 2016.
[63] Pierre, D. M. and N. Zakaria, “Stochastic partially optimized cyclic shift crossover for multi-objective
genetic algorithms for the vehicle routing problem with time-windows,” Appl. Soft Comput., Sep. 2016.
[64] Shi, Y., T. Boudouh, and O. Grunder, “A hybrid genetic algorithm for a home health care routing problem
with time window and fuzzy demand,” Expert Syst. Appl., vol. 72, pp. 160–176, 2017.
[65] AbdAllah, A. M. F. M., D. L. Essam, and R. A. Sarker, “On solving periodic re-optimization dynamic
vehicle routing problems,” Appl. Soft Comput., vol. 55, pp. 1–12, Jun. 2017.
[66] Yousefi, H., R. Tavakkoli-Moghaddam, M. Taheri Bavil Oliaei, M. Mohammadi, and A. Mozaffari,
“Solving a bi-objective vehicle routing problem under uncertainty by a revised multichoice goal
programming approach,” Int. J. Ind. Eng. Comput., vol. 8, no. 3, pp. 283–302, 2017.
[67] Xiao, Y. and A. Konak, “A genetic algorithm with exact dynamic programming for the green vehicle routing
& scheduling problem,” J. Clean. Prod., vol. 167, pp. 1450–1463, Nov. 2017. [68] Mohammed, M. A., M. K. Abd Ghani, R. I. Hamed, S. A. Mostafa, M. S. Ahmad, and D. A. Ibrahim,
“Solving vehicle routing problem by using improved genetic algorithm for optimal solution,” J. Comput.
Sci., vol. 21, pp. 255–262, 2017.
[69] Mendoza, J. E., A. L. Medaglia, and N. Velasco, “An evolutionary-based decision support system for
vehicle routing: The case of a public utility,” Decis. Support Syst., vol. 46, no. 3, pp. 730–742, 2009.
[70] Baños, R., J. Ortega, C. Gil, A. L. Márquez, and F. De Toro, “A hybrid meta-heuristic for multi-objective
Vehicle Routing Problems with Time Windows,” Comput. Ind. Eng., vol. 65, no. 2, pp. 286–296, 2013.
[71] Liu, S., “A hybrid population heuristic for the heterogeneous vehicle routing problems,” Transp. Res. Part E
Logist. Transp. Rev., vol. 54, pp. 67–78, 2013.
[72] Garcia-Najera, A. and J. A. Bullinaria, “An improved multi-objective evolutionary algorithm for the vehicle
routing problem with time windows,” Comput. Oper. Res., vol. 38, no. 1, pp. 287–300, 2011.
[73] Chiang, T.-C. and W.-H. Hsu, “A knowledge-based evolutionary algorithm for the multiobjective vehicle
routing problem with time windows,” Comput. Oper. Res., vol. 45, pp. 25–37, 2014.
[74] Hsu, W. H. and T. C. Chiang, “A multiobjective evolutionary algorithm with enhanced reproduction
operators for the vehicle routing problem with time windows,” IEEE World Congr. Comput. Intell., pp. 10–
15, 2012.
[75] Yücenur, G. N. and N. Ç. Demirel, “A new geometric shape-based genetic clustering algorithm for the multi-depot vehicle routing problem,” Expert Syst. Appl., vol. 38, no. 9, pp. 11859–11865, Sep. 2011.
[76] Jozefowiez, N., F. Semet, and E.-G. Talbi, “An evolutionary algorithm for the vehicle routing problem with
route balancing,” Eur. J. Oper. Res., vol. 195, no. 3, pp. 761–769, Jun. 2009.
[77] Bae, H. and I. Moon, “Multi-depot vehicle routing problem with time windows considering delivery and
installation vehicles,” Appl. Math. Model., vol. 40, no. 13–14, pp. 6536–6549, Jul. 2016.
[78] Augerat, P., Belenguer, J.M., Benavent, E., Corbern, A., Naddef, D., & Rinaldi, G., “Computational results
with a branch-and-cut code for the capacitated vehicle routing problem,” Res. Rep. 949-M, Univ. Joseph
Fourier, Grenoble, Fr., 1995.
[79] Christofides, N. and S. Eilon, “An Algorithm for the Vehicle-dispatching Problem,” J. Oper. Res. Soc., vol.
20, no. 3, pp. 309–318, Sep. 1969.
[80] Fisher, M. L., “Optimal Solution of Vehicle Routing Problems Using Minimum K-Trees,” Oper. Res., vol.
42, no. 4, pp. 626–642, 1994.
[81] Christofides, N., Mingozzi, A., & Toth, P., “The vehicle routing problem. In N. Christofides, A. Mingozzi,
P. Toth, & C. Sandi (Eds.), Combinatorial optimization. Chichester: Wiley.,” pp. 315–338, 1979.
383
Proceedings of the International Conference on Industrial Engineering and Operations Management
Bandung, Indonesia, March 6-8, 2018
© IEOM Society International
Biographies
Gamal Nawara is an emeritus professor at the Industrial Engineering Department, Faculty of Engineering, Zagazig
University; Zagazig, Sharkia, Egypt. He received his B.Sc., from University of Ain Shams, Egypt 1963 in
Mechanical Engineering and Ph.D., from University of Leipzig, Germany 1969 in Industrial Engineering. Prof.
Nawara has several distinguished activities in the field of industrial engineering. He is planner, evaluator and
coaching projects, project manager, and trainer in several projects. In the last decade he has focused on
Development of Small and Medium Enterprises. He is also a member in number of Supreme council of Egyptian
universities. Prof. Nawara has more than 50 articles in different industrial engineering topics.
Raafat Elshaer is an Associate professor in Industrial Engineering Department, Faculty of Engineering, Zagazig
University; Zagazig, Sharkia, Egypt. He received his B.S. degree in Production Engineering from Faculty of Engineering, Helwan University in 1996, M.Sc. degree in Industrial Engineering from Faculty of Engineering,
Zagazig University in 2004, and Ph.D. in Industrial Engineering from Faculty of Engineering, Zagazig University in
2009 as a joint program between Zagazig University and Rutgers University, USA. He has published journal and
conference papers. His research interests include optimization, scheduling, project management, earned value
management and others.
Adel AbdElmoez is an Assistant professor in Industrial Engineering Department, Faculty of Engineering, Zagazig
University; Zagazig, Sharkia, Egypt. Now, He is an Assistant professor in Industrial Engineering Department,
College of Engineering, King Khaled University, Abha, Kingdom of Saudia Arabia. He received his B.Sc. degree in
Construction Engineering from Faculty of Engineering, Zagazig University in 1985, M.Sc. degree in Systems
Engineering from Faculty of Engineering, Zagazig University in 1992, and Ph.D. in Industrial Engineering from
Faculty of Engineering, Zagazig University in 1998. He has published journal and conference papers. His field of
interests include optimization, scheduling, project management, safety engineering, and human factors engineering.
Hadeer Awad is a teaching assistant in Industrial Engineering Department, Faculty of Engineering, Zagazig
University; Zagazig, Sharkia, Egypt. She received her B.S. degree in Industrial Engineering from Faculty of
Engineering, Zagazig University in 2014. She is currently working in designing and developing genetic algorithms for solving different variants of Vehicle Routing Problems.
384