Sunteți pe pagina 1din 3

Proceedings of the 3rd International Conference on Maritime and Naval Science and Engineering

Three Dimensional Maritime Transportation Models

ELIODOR CONSTANTINESCU Department of Mathematical Sciences Maritime University of Constanta 104, Mircea cel Batran Street ROMANIA mirelcon@yahoo.com, http://www.cmu-edu.eu

Abstract: - There are major differences between standard vehicle transportation problems and ship models. Therefore, the classical transportation problem is not a good model for the maritime transportation. This paper proposes the application of three dimensional transportation problems (3D) in maritime activity. We briefly present the three dimensional maritime transportation models and the main known results obtained in solving such difficult problems.

Key-Words: - maritime, transportation, model, three-dimensional, index, axial, planar

1 Transportation problem

The transportation problem (also called Hitchcock Problem and denoted by TP) is one of the classic problems in operation research, a special type of linear programming problem:

min z

=

m n

∑∑

i =

1

j

=

1

cx

ij

ij

n

j = 1

m

i = 1

x

x

ij

ij

= a i , i = 1,

= b j , j = 1,

,m

,n,

x ij o for all i , j We have to determine the optimal shipments from a given set of origins to a given set of destinations in such a way as to minimize the total transportation costs. The problem is constrained by known upper limits on the supply at the various origins and by the necessity to satisfy the known demand at each destination. The classical transportation model assumes that the per unit cost for each potential origin destination pair is known a priori. The model doesn’t take in consideration the type of ship, the various commodities to be transported, the different characteristics of the vessels and other factors which also can influence the total cost of transportation. Also, it cannot be assumed that carriers will be able to serve every origin-destination pair for which they are the least-cost carrier because of capacity constraints on the various carriers. Consequently, it is impossible to assign, a priori, the appropriate per unit transportation costs necessary to use classic

ISSN: 1792-4707

25

transportation problem in modeling of maritime transportation. There are major differences between classical transportation problems and ship problems. According to [6], we mention:

destination of ships may be changed at sea;

ships are different from each other in their operating characteristics (capacity, speed), as well as their cost structure. Due to frequent fluctuations in the ship market, even two identical ships may have quite different cost structures;

ships

origin;

there are more sources of uncertainty and much longer voyages in maritime transportation. All these arguments denote that the standard transportation model can not apply to simulate ship problems.

2 Maritime Transport Modeling

Obviously, we have to consider more indices to build realistic maritime transportation models. By introducing the third index for the types of vessel used to transport homogeneous commodity, we obtain a three dimensional maritime transportation model. It is very clear that the multidimensional transportation models could be very good representations of real situations, but the computational problems are really very difficult. More indices we introduce, more realistic the maritime transportation models become; in the same time, the problems become more and more difficult

ISBN: 978-960-474-222-6

Proceedings of the 3rd International Conference on Maritime and Naval Science and Engineering

to be solved. Applying the three dimensional maritime transportation model to carry homogeneous goods (containers, oil, chemicals, ore, etc.), we have to analyze the influence of different types of ships.

Therefore, the third index k will define the type of vessel used to transport the commodity. We illustrate below the significance of all indexes and dates:

k the type of vessel used to transport

homogeneous commodity c ijk the unit cost of transportation from i to j

using a k type vessel x ijk the amount of goods loaded in the origin

port i to be transported in the port j with a k type ship, so that the total cost of transportation should be minimum a ij the amount of commodities transported from i to j b jk the amount of commodities demanded in the destination port j and transported with a k type vessel c ik the amount of goods transported from the origin port i with a k type ship the total amount of commodities stored in

the port i

the total amount of commodities demanded

amount of commodities

transportedwith the k type ships. Depending on the known dates and on the types of the constraints adequate to the problem, there are several three dimensional maritime transportation models. Taking into account their main characteristics, we can do a general unit presentation of the 3D case. The objective function become

,

a i

b j

c k

the

total

∑ ∑ ∑

z =

cx

ijk

, I = {1,

,

m}, J = {1,

ijk

 i ∈ I j ∈ J k ∈ K n}, K = {1, ,p} and where independent on the

semnification of indexes i,

j,

{ c ijk }

represents

the

matrix

k

in

each

model,

of

unit

costs

of

transportation and { x ijk } represents the unknown matrix of commodities to be transported. There are two distinct types of constraints for the variables x ijk :

a. Planar Constraints (PC) Fixing two indexes of the matrix { x ijk }, i, j, the summation of x ijk , for all k, should be equal to the

ISSN: 1792-4707

26

elements of a real two indexes sequence a ij :

k K

x

ijk

= a ij ,

i I, j J

The structure of this constraints type imposes the existence of three distinct planar constraints, as maximum, one for each two indexes group {i, j}, {j, k} and {i, k}, denoted by (PC1), (PC2) and (PC3):

k

K

i

I

jJ

x

ijk

=

a ij

x ijk = b jk

x ijk = c ik

 , i ∈ I, j ∈ J, (PC1) , j ∈ J, k ∈ K, (PC2) , i ∈ I, k ∈ K (PC3)

b. Axial Constraints (AC) Fixing one index of the matrix { x ijk }, i.e. i, the double sum of x ijk , for all j and k, should be equal to the elements of a real one index sequence a i :

∑∑

jJ k K

x

ijk

= a i ,

i I

Similarly to the planar case, the structure of this constraints type imposes the existence of three distinct axial constraints, as maximum, one for each i, j, k index, denoted by (AC1), (AC2) and (AC3):

∑∑

jJ k K

∑∑

i I k K

∑∑

i I jJ

x

x

x

ijk

ijk

ijk

= a i ,

= b j ,

= c k ,

 i ∈ I, (AC1) j ∈ J, (AC2) k ∈ K, (AC3)

According to the different types of constraints, the model will be called axial, planar or mixed. For every model, the known dates of the problem are distinct, but there are three compatibility relations (CR) between all these dates, which assure the mathematical equilibrium, as a consequence of the economic equilibrium between demand and supply:

j J

i I

jJ

a

a

b

ij

ij

jk

=

=

=

k

K

k

K

i

I

c ik

b jk

c

ik

 = a i , i ∈ I, (CR1) = b j , j ∈ J, (CR2) = c k , k ∈ K, (CR3)

From the above compatibility relations, it follows the well known equilibrium condition for the transportation problem:

i I

a

i

=

jJ

b

j

=

k K

c

k

,

(EC)

Finally it is obvious that the introduction of the third index complicates very much the model structure, although the model still remains a linear programming problem. On the other hand, the three dimensional maritime transportation models allow a better representation of real studied systems.

ISBN: 978-960-474-222-6

Proceedings of the 3rd International Conference on Maritime and Naval Science and Engineering

3

transportation models

Three

dimensional

maritime

Based on the main characteristics of the 3D modeling process applied in maritime transport, we develop five distinguish types of maritime transportation models. The most important three dimensional transportation models are 3A - the three axial problem (the model which contains all three distinct types of axial constraints) and 3P - the three planar problem (contains all three distinct types of planar constraints). Considering only two distinct types of axial or planar constraints, the 3D transportation models are called 2A – two axial model, respectively 2P – two planar model. The AP- axial-planar model takes into consideration an axial restriction and a planar constraint. It’s important to point out that while the 3A and 3P maritime transportation models are well defined, for the other three types of 3D models there are different model types, depending on data considered. All these aspects of three dimensional transport modeling are summarized in table 1:

 a i b j c k a ij b jk c ik 3P - - - * * * 3A * * * - - - 2P - - - * * - 2A * * - - - - AP - - * * - -

K.B. Haley [4] is the author of the algorithm for

solving the three planar transportation problem. His algorithm is an entirely spreading of “modi method”,

a refinement of simplex applied to the classical

transportation problem (G.B. Dantzig [2]). We have

to point out that some problems remain concerning

the computational aspects and also nothing for its solution is described by Haley. W. Junginger [5] made some advances in the above mentioned problems. It is not very difficult to see that the 2A, 2P and AP transportation models are not representative for the three dimensional case; more exactly, these three mentioned models could be solved as classical transportation problems [1]. Taking into account that we also know to reduce the three axial model to a three planar transportation problem, it’s easily to conclude that theoretically, we can solve any three dimensional transportation problem. Practically, it is difficult due to the loose of triangularity of the bases in the multi dimensional

ISSN: 1792-4707

27

case. Introducing a fourth index for the commodities transported, we obtain four dimensional maritime transportation models. These problems are more difficult to solve, but the models are more realistic, especially taking into account the ships which can transport different types of goods. Unfortunately, starting with the four dimensional transportation models, the application of Haley’s algorithm is only theoretical, due to computational aspects. Therefore, another approach will be necessary.

4 Conclusion

When the daily operating costs of ships are thousands of dollars, large profits may be expected from improving its scheduling process. Therefore, the three dimensional maritime transportation models represent a more realistic approach of real problems. It is obviously necessary to develop the multidimensional transportation models applied in maritime transport, in order to obtain significant cost

savings in the operations of shipping companies.

References:

[1] E. Constantinescu, Utilisation de divers modeles

mathematiques en transports maritime, Taol Lagad, Vol. 70, 1995.

[2] G. B. Dantzig, Application of the Simplex

Method to a Transportation Problem, Activity Analysis of Production and Allocation, Koopmans Publishing House, 1951. [3] B. Haley, The Solid Transportation Problem, Operations Research 10, 1962. [4] B. Haley, The Multi-Index Problem, Operations Research 11, 1963. [5] W. Junginger, Zur Losung des dreidimensionalen Transport problems, Operations Research Verfahren, 13, 1972. [6] D. Ronnen, Ship Scheduling, European Journal of Operational Research, 71, 1993

ISBN: 978-960-474-222-6