Genetic Algorithm For Integrated Models of Continuous Berth Allocation Problem and Quay Crane Scheduling With Non Crossing Constraint
Genetic Algorithm For Integrated Models of Continuous Berth Allocation Problem and Quay Crane Scheduling With Non Crossing Constraint
Genetic Algorithm For Integrated Models of Continuous Berth Allocation Problem and Quay Crane Scheduling With Non Crossing Constraint
Indian Journal of Science and Technology, Vol 8(32), DOI: 10.17485/ijst/2015/v8i32/92140, November 2015 ISSN (Online) : 0974-5645
Abstract
Background/Objectives: This paper focused on integrated Continuous Berth Allocation Problem (BAPC) and Quay
Crane Scheduling Problem (QCSP) by considering non-crossing constraint to make it more realistic. Methods/Statistical
analysis: Genetic Algorithm (GA) is a metaheuristic method that has been used extensively in Berth Allocation Problem
(BAP). Crossover and mutation are selected as operators in this paper. Findings: The integrated model is formulated as a
Mix Integer Problem (MIP) with the objective to minimize the sum of the processing times. A vessel's processing time is
measured between arrival and departure includingwaiting time to be berthed and servicing time.The new algorithm of GA
arecompatible with the integrated model and useful for finding near-optimal solutions. Three phase new algorithms of GA
are proposed and provide a wider search to the solution space. Application/Improvements: Three phase of GA is another
significant and promising variant of genetic algorithms in BAPC and QCSP. The probabilities of crossover and mutation
determine the degree of solution accuracy and the convergence speed that GA canobtain. By using fixed values of crossover
and mutation, the algorithm utilize the population information in each generation and adaptively adjust the crossover and
mutation. So, the population diversity and sustain the convergence capacity is maintained.
Keywords: Continuous Berth Allocation, Genetic Algorithm, Non Crossing,Quay Crane Scheduling
2 Vol 9 (32) | November 2015 | www.indjst.org Indian Journal of Science and Technology
Azyanzuhaila Hasan Basri, Zaitul Marlizawati Zainuddin and Nurhidayu Idris
• Each vessel is divided along its length into holds 3 or xlk+xkl ≤ 1 ∀k ,l ∈V and k<l (3)
4 container rows. ylk+yk ≤ 1 ∀k ,l ∈V and k<l (4)
• Work needs to continue on a hold until completion tl ≥ ck+(xkl-1)M ∀k ,l ∈V and k<l (5)
when it’s started. bl ≥ bk+hk+(ykl-1)M ∀k ,l ∈V and k<l (6)
• A vessel can leave the port only after the proses of tk ≥ ak ∀k ∈V ∀k ∈V (7)
t t
loading and unloading container is completed on tl ≥ tz ki+(1- z ki)T ∀k ∈V∀i ∈{ 1,…..,hk}∀t ∈{1,…,T}
every hold. (8)
t i
• Only one QC can work on a hold at a given time ck ≥ tz ki+p k ∀k ∈V∀i ∈{ 1,…..,hk }∀t
period. ∈{1,…,T} (9)
t
ck ≥ tz ki+p max
k
∀k ∈V (10)
• QCs are on the same tracks and cannot cross each t
∑Tt=1z ki=1 ∀k ∈V ∀i ∈{ 1,…..,hk }
other.
(11)
• QC can be shifted from hold to hold both within t i
∑Tt=1 ∑hki=1 ∑ft=f-pi +1 z ki ≤ Q∀t ̅∈{p k..., T} (12)
vessels and between vessels, as long as QCs are not bk≤B - hk+1
k
∀k ∈V (13)
cross one another. bk≥1 ∀k ∈V (14)
i t
L t ≤ L i+1- 1 ∀k ∈V (15)
4. Model Formulation
t
1≤ L i ≤ B ∀k ∈V (16)
xlk∈{0,1},ylk∈{0,1}, ∀k,l ∈V and k≠l (17)
This section proposed mathematical formulation for ztki ∈ {0,1} ∀k ∈V∀i ∈{ 1,…..,hk }∀t ∈{1,…,T}
IBAPCQSP. The study concentrated on the dynamic (18)
arrival where a set V of vessels with known arrival times, Constraints (2) through (4) guarantee that no vessel
where n = |V |. For each vessel k ∈ V, the study defines: rectangles overlap. Constraints (5) and (6) ensure that
the selected berthing times and berthing positions are
B: Set of berths equal size sections. consistent with the definitions of xlk and ylk, where M is
Q: A set of identical quay cranes operating on a single set a large positive scalar. Constraint (7) forces berthing to
of rails. make sure no earlier than arrival time, and Constraint (8)
T: Time period of vessels. forces hold processing to start after berthing. Constraint
v: A set of vessels with known arrival time. (9) ensures that vessels depart only after all holds are
M: A large positive scalar. processed, and Constraint (10) is a valid inequality that
: Location of crane i at time period t. provides a lower bound on ck given tk. Constraint (11)
hk: Number of holds of vessel k. ensures that work starts on each hold of each vessel and
: Processing time of hold iof vessel k Constraint (12) ensures that no more than Q quay cranes
: Maximum hold processing time for vessel k (pkmax : maxi are used at any time period. Constraints (13) and (14)
pi }). guarantee that all vessels fit on the berth. Constraint (15)
k
ak: Arrival time of vessel k. and (16) ensure that cranes cannot cross over each other.
bk: Berthing position of vessel k. The objective ∑nk=1(ck - ak) + ∑nk=1fk(ck - dk) is to minimize
tk: Berthing time of vessel k. the sum of processing time, where a vessel’s processing
ck: The earliest time that vessel k can depart. time is measured between arrival and departure time
tk: Berthing time of vessel k. including waiting times to be berthed and servicing time.
xlk = { 0
1 if vessel k berth after vesselldeparts
otherwise:
5. Genetic Algorithm
if vessel k berth completely above vessel l onthe time -space diagram
Genetic algorithms are based on population mechanisms.
y lk = { 01 otherwise ;
Every two parent solutions mate produced two child
{ 1 if vessel k berth after vfessel l departs solutions which transferred a new combination of genes
t
z ki = 0 otherwise ; to form of new chromosomes13. To create the populations,
chromosomes are generated randomly and proceeds for
next generations.
xlk+xkl+ylk+ykl ≥ 1 ∀k, l ∈V and k<l (2) Firstly, initial solution (L0, B0) is generated randomly
Vol 9 (32) | November 2015 | www.indjst.org Indian Journal of Science and Technology 3
Genetic Algorithm for Integrated Models of Continuous Berth Allocation Problem and Quay Crane Scheduling with Non
Crossing Constraint
4 Vol 9 (32) | November 2015 | www.indjst.org Indian Journal of Science and Technology
Azyanzuhaila Hasan Basri, Zaitul Marlizawati Zainuddin and Nurhidayu Idris
6. Numerical Example
The small problem in Table 1 is used for the model, where
B = 7 and Q = 4. Commercial software, LINGO 14.0 was
Figure 4. Initial solution of IBAPCQSP on time space
adopted in this research for validation process. In this
diagram.
problem, each vessel is divided into equal size sections that
we call holds. The length of a vessel can be represented by Time space diagram in Figure 4 represented processing
the number of holds it has. Furthermore, the length of a time for vertical axis and berth section for horizontal
hold is also set equal to the length of a berth section, and axis. The location of cranes in any holds illustrated using
it is assumed that the berth is B holds in total length. Each a solid and empty square. Solid square means crane in
hold consists of 3 or 4 container rows. Multiple vessels can processing activities (loading or unloading) and empty
moor at the berth and receive service simultaneously but square means cranes in idle situation.
only one QC can work on a hold at a given time period. Table 2 shows the optimal solution for this problem.
Each hold requires a known processing time to allocate The ordered list is L = (2,4,1,3,5) and B = (1,1,5,4,1).
QC for every hold. A vessel can leave the port only after The objective function values of the optimal solution is
loading and unloading process is completed on every 26 hours. Figure 5 shows time space diagram for optimal
hold. solution of the problem.
4 2 4 1 2
- 2 1 0 2
- - - 1 4
Vol 9 (32) | November 2015 | www.indjst.org Indian Journal of Science and Technology 5
Genetic Algorithm for Integrated Models of Continuous Berth Allocation Problem and Quay Crane Scheduling with Non
Crossing Constraint
handled by one QC. In this problem, the travel time of a CIAM) for their financial funding through FRGS vote
QC not considered. number R.J130000.7809.4F470, the UTM. This support is
greatly acknowledged.
9. References
1. Zhang C, Liu J, Wan Y, Murty KG, Linn RJ. Storage space
allocation in container terminals. Transportation Research.
2003 Dec; 37(10):883–903.
2. Han X, Lu Z, Xi L. A proactive approach for simultaneous
berth and quay crane scheduling problem with stochastic
arrival and handling time. European Journal Operation Re-
search. 2010 Dec; 207(3):1327–40.
3. Imai A, Chen HC, Nishimura E, Papadimitriou S. The si-
Figure 6. Space time diagram for quay crane in optimal multaneous berth and quay crane allocation problem.
solution of IBAPCQSP. Transportation Research Part E. 2008 Sep; 44(5):900–20.
4. Zhou P, Kang H. Study on berth and quay-crane allocation
under stochastic environments in container terminal. Syst
This integrated model showed that the BAPC and Eng - Theory Pract. Systems Engineering Society of China.
QCSP can be solved simultaneously and become more 2008 Jan; 28(1):161–9.
practical. The efficient utilization of this technical 5. Song L, Cherrett T, Guan W. Study on berth planning prob-
equipment will minimize vessel handling times. Hence, lem in a container seaport: Using an integrated program-
well planned QC operations are important for terminal ming approach. Computer Industrial Engineering. 2012;
Feb; 62(1):119–28.
efficiency. Solving these two problems separately may 6. Ponomarev A, Dullaert W. A two-phase genetic algorithm
cause the terminal operators need to solve BAPC first and for the berth and quay crane allocation and scheduling
then solve QCSP later. Two different problems need to problem. Available from: https://www.researchgate.net/
model for solving the BAPC and QCSP and it may cause publication/254449729
inefficiency situation. 7. Frojan P, Correcher JF, Alvarez-Valdes R, Koulouris G,
Tamarit JM. The continuous berth allocation problem in a
container terminal with multiple quays. Expert Syst Appl.
7. Conclusion .2015 Nov; 42(21):7356–66.
8. Hu Z. Multi-objective genetic algorithm for berth alloca-
tion problem considering daytime preference. Computers
In this paper, the mathematical formulation is presented
and Industrial Engineering. 2015; 89:2–14.
for IBAPCQCSP. This integrated model developed by 9. Imai A, Nishimura E, Papadimitriou S. The dynamic berth
considering non crossing constrain to make it more allocation problem for a container port. Transp Res Part B
realistic. The new approach of GA is able to maintains a Methodol. 2001 May; 35(4):401–17.
rapid convergence and obtain near-optimal solutions in 10. Shan T. Genetic algorithm for dynamic berth allocation
lower computational times. For future work, we propose problem with discrete layout. Prooceding of the 2nd In-
ternational Conference Computer Application and System
to focus on realistic assumptions to the formulation, such
Modelling; China. 2012. p. 261–4.
as safety distance between vessels and the traveling time 11. Li N, Jin Z, Massami E. A new genetic algorithm for contin-
of cranes during scheduling process. Sensitivity Analysis uous berth and quay crane allocation problem. Application
for mathematical model also something important to Mechanical Material. 2011; 98:619–22.
explore. 12. Glover F, Kelly P, Laguna M. Genetic algorithm and tabu
search: Hybrids for optimization. Computer Operational
Research. 1995; 22(1):111–34.
8. Acknowledgment 13. Ganji SRS, Babazadeh A, Arabshahi N. Analysis of the con-
tinuous berth allocation problem in container ports using
The authors would like to thanks the Malaysian Ministry a genetic algorithm. Journal of Marine Science Technology.
2010 Jun; 15(4):408–16.
of Higher Education and Universiti Teknologi Malaysia
Centre for Industrial and Applied Mathematics (UTM-
6 Vol 9 (32) | November 2015 | www.indjst.org Indian Journal of Science and Technology