Optimal Fleet Assignment Using Ant Colony Algorithm: Rashid Anzoom M. Ahsan Akhtar Hasin
Optimal Fleet Assignment Using Ant Colony Algorithm: Rashid Anzoom M. Ahsan Akhtar Hasin
Optimal Fleet Assignment Using Ant Colony Algorithm: Rashid Anzoom M. Ahsan Akhtar Hasin
Algorithm
Rashid Anzoom M. Ahsan Akhtar Hasin
Department of Industrial and Production Engineering Department of Industrial and Production Engineering
Bangladesh University of Textiles Bangladesh University of Engineering and Technology
Dhaka, Bangladesh Dhaka, Bangladesh
rashidanzoom@butex.edu.bd aahasin@ipe.buet.ac.bd
Abstract— Demand for logistical services is highly dynamic, selection of routes and aircraft. Flight routes are selected
due to its high market growth. In Bangladesh, air logistics is through cautious analysis of customer demand in different
evolving at a fast pace, both in terms of passenger and cargo locations. Similarly, aircraft are selected on the basis of
transportation. As a result, several local private and different performance parameters. Next, the airline is
international airlines are trying to capture market share in this required to optimally fulfill the customer demand with the
promising sector. For survival in such tough competition, available aircraft, which is known as the fleet assignment
optimization in operations is indispensable. This research problem (FAP). It involves properly assigning aircraft of
focuses on the optimal fleet assignment with Ant Colony different capacities to the scheduled flights, based on
algorithm. In this rapidly expanding market, short to mid-term
equipment capabilities, operational costs, and potential
demands of passenger were estimated using regression
analysis. Then, from a database of routes and aircraft capacity,
revenues. The objective of the Fleet Assignment Model
a model was developed to estimate profitability from fleet (FAM) is to maximize the profit from these assignments.
assignment with relative ease. Finally, ant colony algorithm This model can also be useful in directing which types of
was used to find the optimal assignment. Although previous aircraft to acquire, especially when a new airline enters the
researches on fleet assignment were done using Genetic market or an existing airline extends its route.
algorithm, only current-level of demand for logistical service For a better understanding of the fleet assignment
was considered. This research considers it as dynamic and
problem, a popular reference is the basic fleet assignment
projected into the future. The results obtained in this research
model, provided by Hane et al. (1995) [2]. The objective of
show that dynamic demand consideration gives much better
results as well as better utilization of resources. And these
this model is to select the fleet assignment that maximizes
contributed to the reduction of operational cost as well as the profitability, or equivalently, minimizes operating costs less
increase in revenue. As a result, profit was optimized. For mid- total revenue. In this research work, a modified form of
term and long-term projection, demand becomes gradually Hane’s Model has been developed and used for fleet
probabilistic. However, this research considered it as assignment. The objective remains unchanged- that is to
deterministic; otherwise, the problem becomes an NP-hard assign aircraft to scheduled routes with a view to maximizing
problem, which is difficult to solve. This research is expected to the profit. However, the calculation of profit requires the
be of immense help to the air industry of Bangladesh. Because knowledge of revenue and cost from flight operation-that is
of similarity in business nature, this can be marginally adapted where the modification has occurred.
in other countries as well.
Calculation of airline cost can be conducted in two ways-
Keywords— logistics; fleet assignment problem; ant colony Functional and Administrative Approach. [3] In the
algorithm administrative approach, airline cost consists of Salaries,
Materials Cost, Service Cost, Landing Fee etc. Functional
Approach, on the other hand, divides operating cost into
I. INTRODUCTION
three sub-groups: Aircraft operating cost, Ground operating
Airline industry is one of the fastest growing industries in cost and System operating costs. However, majority of the
the world. According to Air Transport Action Group revenue obtained by airlines generate from a single source-
(ATAG), in 2016, the global airline industry consisted of Ticket Sales. As a result, earnings from a flight is heavily
1402 commercial airlines operating 32.8 million flights with dependent upon the pricing of the ticket. Traditionally, there
26065 commercial aircraft and providing service to 3883 are two tiers of seat available- business and economy [4].
airports carrying roughly 4.1 billion customers [1]. This However, now-a-days, airlines offer different categories of
inclination towards air transportation can be attributed to the seats based on factors like - date of purchase, available
increasing significance of time and convenience among amenities, and seat location. Pricing the ticket too low might
people. The airline industry is also highly advanced in bereft the company of attainable profit. Contrarily,
technological aspects. With a persistent focus on identifying overpricing of tickets might lead to curtailment of customer
new and emerging technologies, they strive to enhance and market share. Hence, it is essential to select a ticket price
business efficiency and improve customer experiences. in the light of both the profitability of the company as well as
However, to stay conversant with the contemporary the pricing of the competitors. Another important factor to be
technologies, the required expenditure has to be quite lofty- noted is Load Factor- the percentage of total seats sold on a
comprising both the costly acquisition of aircraft as well as route. [5] As the value of load factor differs along flight time,
maintaining high operating cost for regular flights. These route and other factors, its incorporation in fleet assignment
have turned the task of retaining profitability into a grueling model should provide a better and realistic estimation of the
one for the airlines. possible income from a flight.
For ensuring competitiveness and profitability In this paper, the profit from a fleet assignment is
simultaneously, planning in an airline needs to be done with modeled to be a function of aircraft seat capacity and the
utmost vigilance. Two key decisions in this regard are the
Short-Haul Model:
m n
A333 335 Dhaka-Barishal 112
Maximize¦¦ Pijs xij
i =1 j =1
B735 132 Dhaka-Jessore 142
m n
s.t. ¦¦ x
i =1 j =1
ij =m (11) B739 177 Dhaka-Chittagong 211
n
A319 142 Dhaka-Sylhet 200
¦ xij = 1
j =1
(12)
B762 290 Dhaka-Rajshahi 198
xij = 1 or 0 (13)
di ≤ 1000 (14)
TABLE II. INFORMATION FOR LONG HAUL PROBLEM
Long Haul Model:
Seat Flight
m n Aircraft Model Route
Maximize¦¦ Pijl xij
Capacity Distance (km)
i =1 j =1
m n A336 266 Dhaka-Beijing 3025
s.t. ¦¦ x
i =1 j =1
ij =m (15)
MD83 410 Dhaka-Tehran 3960
n
¦x
j =1
ij =1 (16) B737 149 Dhaka-Male 2840
Long-haul Problem:
Aircraft Model
Route
A 336 MD83 B737 A319 B77W
Dhaka-
4161.9 641.28 1421.71 814.71 3381.40
Beijing
Dhaka-
6083.9 1362.72 2409.30 1595.20 5037.30
Tehran
Dhaka-Male 10504.75 3022.01 4680.75 3390.60 8846.00
Dhaka-
9799.99 2547.49 4318.63 3104.40 8238.80
Jakarta
Dhaka-
9671.85 2709.39 4252.80 3052.30 8128.50
Sydney
Fig. 4. Profit vs Iterations for Long Haul Operation As both results match, the solution by ACO is proved to
be authentic and believable.
TABLE IV. SUMMARY OF OPTIMAL FLEET ASSIGNMENT IN
SHORT HAUL OPERATION VI. CONCLUSION
Route Assigned Aircraft High growth and propitious future are drawing an ever-
increasing number of investments in the Aviation industry-
Dhaka-Barishal (DHK-BAR) B735 involving the launch of new airlines or expansion of the
Dhaka-Jessore (DHK-JSR) A319 current ones. However, operational complexities constantly
threaten to jeopardize this trend as maintaining profitability
Dhaka-Chittagong (DHK-CTG) A333
gets tougher day by day. That is why proper fleet assignment
Dhaka-Sylhet (DHK-SYL) B762 holds the key to prosperity for an airline. In this paper, a fleet
Dhaka -Rajshahi (DHK-RAJ) B762 assignment model has been developed to calculate
profitability as a function of seat and distance. And the
Optimum Profit= 25232.96$ application of ACO has assured quick convergence of
optimal solution to the problem. Therefore, this might
TABLE V. SUMMARY OF OPTIMAL FLEET ASSIGNMENT IN facilitate less difficulty in planning for new entrants in the
LONG HAUL OPERATION market with easily accessible information. However, the
modeling had been based on the data from airlines having
Route Assigned Aircraft
operations in Dhaka. It might be modeled with a different
Dhaka-Beijing (DHK-BEI) A336 data set to be applicable in different zones or even be
Dhaka-Tehran (DHK-THR) MD83 modified to a global model if data from different regions are
Dhaka-Male (DHK-ML) B737 accumulated. Besides, more factors can be taken into
consideration while calculating the load factor. There is also
Dhaka-Jakarta (DHK-JKR) A319
a scope of inclusion of uncertainty into the model. All these
Dhaka-Sydney (DHK-SDN) B77W ideas can be considered for inclusion in future research work.
Optimum Profit= 577266.74$. Nevertheless, this research is expected to assist airlines in
planning better to maintain profitability in their operation.
For verification of the results, Hungarian Algorithm was
applied to the problems as well. And the results came out to REFERENCES
be the same as ACO for both short-haul and long-haul
[1] Air Transport Action Group, Aviation Benefits Beyond Borders, July
operation. The results are shown in Table VI and Table VII, 2016.
where the cells written in bold italic arrangement represent [2] Hane et al., “The fleet assignment problem: solving a large-scale
the optimal assignment. integer program,” Math. Prog., vol. 70(1-3), October 1995, pp.211-
232.
TABLE VI. RESULT OF SHORT HAUL OPTIMIZATION USING [3] P. Belobaba, A. Odoni and C. Barnhart eds., The Global Airline
HUNGARIAN ALGORITHM Industry, 2nd ed., John Wiley & Sons, 2015.
[4] Q. Zhang, H. Yang & A. Zhang, “Market power and its determinants
Aircraft Model in the chinese airline industry”, Trans. Res. Pt. A: Pol. & Prac., vol.
64, 2014, pp.1-13.
Route
A333 B735 B739 A319 B762 [5] F. C. Y. Chen & C. Chen, “The effects of strategic alliances and risk
Dhaka- pooling on the load factors of international airline operations”, Trans.
4161.90 641.28 1421.71 814.71 3381.40
Barishal Res. Pt. E: Log. & Trans. Rev., vol. 39, no. 1, 2003, pp.19-34.
Dhaka- [6] N. Fumo & M. R. Biswas, “Regression analysis for prediction of
6083.90 1362.72 2409.30 1595.20 5037.30
Jessore residential energy consumption”, Ren. & Sus. Ener. Rev., vol. 47,
Dhaka- 2015, pp.332-343.
10504.75 3022.01 4680.75 3390.60 8846.00
Chittagong
Dhaka- [7] M. Dorigo et al., Ant Colony Optimization and Swarm Intelligenece,
9799.99 2547.49 4318.63 3104.40 8238.80 Springer-Verlag Berlin Heidelberg, 2004.
Sylhet
Dhaka- [8] J. Abara, “Applying integer linear programming to the fleet
9671.85 2709.39 4252.80 3052.30 8128.50 assignment problem”, Interfaces, vol. 19, no. 4,1989, pp.20-28.
Rajshahi
[9] A. Abdelghany and K. Abdelghany, Modeling Applications in the
Airline Industry, Routledge, April 2015.
[10] W. M. Swan and N. Adler, “Aircraft trip cost parameters: a function [14] M. Dorigo, G. D. Caro and L. M. Gambadella, “Ant algorithms for
of stage length and seat capacity”, Trans. Re. Pt. E: Log. And Trans. discrete optimization”, Art. Lif., vol. 5, no. 2, 1999, pp.137-172.
Rev., vol. 42, no. 2, 2006, pp.105-115. [15] J. E. Bell and PR McMullen, “Ant colony optimization techniques for
[11] Y. Park and M. E. O’Kelly, “Examination of cost-efficient aircraft the vehicle routing problem”, Adv. Eng. Info., vol. 18, no. 1, January,
fleets using empirical operation data in US aviation markets”, Jour. 2018, pp.41-48.
Of Air Trans. Man., vol. 69, 2018, pp.224-234. [16] H. R. Lourenco & D. Serra, “Adaptive search heuristics for the
[12] J. Zuidberg, “An econometric analysis of the factors affecting aircraft generalized assignment problem”, Math. & Soft. Comp, vol. 9, ,no. 2-
operating costs”, Jour. Of Air Trans. Man., vol. 40, 2014, pp.86-95. 3, 2002.
[13] P. Kevin and N. Piersma, Airline Revenue Management: An
Overview of OR techniques 1982-2001, ERIM Report Series, 2002.