Proceedings of IICMA 2013

Research Topic, pp. xx-xx.

A GOAL PROGRAMMING APPROACH TO SOLVE

VEHICLE ROUTING PROBLEM USING LINGO

ATMINI DHORURI1, EMINUGROHO RATNA SARI2, AND DWI

LESTARI3

1Departmentof Mathematics, Yogyakarta State University, [email protected]

2Department
of Mathematics, Yogyakarta State University,
[email protected]
3Department of Mathematics, Yogyakarta State University, [email protected]

Abstract. Vehicle routing problem (VRP) which discusses a set of routes for some
vehicles, starting and ending at a depot, serving a set of customers such that each
customer must be visited once by exactly one vehicle, and having a time constrain is
called vehicle routing problem with time window (VRPTW). This paper presents a
goal programming approach to solve VRP, especially VRPTW. We have considered
an objective function with four main goals: to maximize utilization of vehicle
capacity, to minimize the total waiting time, to minimize the total cost to serve the
customers and to maximize the number of served customers. The proposed model was
implemented and has been solved numerically using LINGO software and the optimal
solution is presented.
Key words and Phrases: goal programming, vehicle routing problem, LINGO.

1. Introduction

Vehicle Routing Problem (VRP) is a set of routes which formed to serve a set
of customers using vehicles, starting and ending at a depot. In VRP, each customer
must be visited once by exactly one vehicle. The route has to be designed such that
the total demands of all customers must not exceed the capacity of the vehicle.
Based on Jolai and Aghdaghi [1], Sousa et al [2], Azi et al [3], if VRP has a
time constrains on the periods of the day in which each customer must be visited, it
is called Vehicle Routing Problem with Time Window (VRPTW). VRPTW is one
of an important problem occurring in distribution systems. So, it has received many
attentions not only on the development of the theory, but also on its application.
For example, postal deliveries, delivery service of food business, Liquefied
Petroleum Gas (LPG) deliveries. Routes which designed should be in a short
duration and must be satisfy time constrain. Larsen [4], Cook and Rich [5],
Cordeau et al [6] proposed an exact method for the VRPTW. As development of
research, there are many other methods to solve VRPTW.
On the other side, because of its wide application to real-life situations,

1
5 A GOAL PROGRAMMING APPROACH…

3. Computational Result
Now we present some tests and results to a real problem of the Liquefied
Petroleum Gas (LPG) agent. The LPG will be distributed to the customers. In this
paper we only take the first 5 customers as an example. We denote each customer
by N1, N2,…, N5. For depot and copy of depot will be denoted by N0 and N6,
respectively. The travel time, travel cost and the demand of the customers are
presented in Table 1, Table 2, and Table 3.

Table 1. Travel time between nodes (minutes)

Depot N1 N2 N3 N4 N5
Depot 0 2 3 3 5 3
N1 2 0 1 4 5 4
N2 3 1 0 5 6 5
N3 5 4 5 0 5 5
N4 5 6 7 3 0 5
N5 3 7 8 6 7 0

Table 2. Travel cost between nodes (x Rp 1000,00)

Depot N1 N2 N3 N4 N5
Depot 0 0.27 0.3 0.41 0.54 0.38
N1 0.25 0 0.06 0.48 0.67 0.48
N2 0.28 0.06 0 0.45 0.70 0.51
N3 0.45 0.41 0.45 0 0.54 0.41
N4 0.54 0.67 0.70 0.35 0 0.51
N5 0.67 0.77 0.80 0.64 0.57 0

Table 3. Demand and service time for each customer

Customer N1 N2 N3 N4 N5
Demand 70 420 80 120 240
(units)
Service 23 140 27 40 80
time
(minutes)

5
6 ATMINI D., EMINUGROHO R.S, AND DWI L.

Complete script of the problem solved in LINGO 11.0 shown bellow

MODEL:
SETS:
N/N0,N1,N2,N3,N4,N5,N6/:q,s,v,h,d2;
R/R1..R2/:B,TR,d4;
E(N,R):Y,W;
D(N,N,R):X;
A(N,N):c,t;
ENDSETS
DATA:
U = 56;
UC = 3;
UT = 6;
q = 0 7 42 8 12 24 0;
s = 0 23 140 27 40 80 0;
!0; !1; !2; !3; !4; !5; !10;
c = !0; 0. 0.27 0.30 0.41 0.54 0.38 0.
!1; 0.25 0. 0.06 0.48 0.67 0.48 0.27
!2; 0.28 0.06 0. 0.45 0.70 0.51 0.30
!3; 0.45 0.41 0.45 0. 0.54 0.41 0.41
!4; 0.54 0.67 0.70 0.35 0. 0.51 0.54
!5; 0.67 0.77 0.80 0.64 0.57 0. 0.38
!10; 0. 0.27 0.30 0.41 0.54 0.38 0. ;

!0; !1; !2; !3; !4; !5; !10;

t = !0; 0 2 3 3 5 3 0
!1; 2 0 1 4 5 4 2
!2; 3 1 0 5 6 5 3
!3; 5 4 5 0 5 5 5
!4; 5 6 7 3 0 5 5
!5; 3 7 8 6 7 0 3
!10; 0 2 3 3 5 3 0 ;
v = 0 1 1 2 1 1 0;
h = 0 1 1 2 1 1 0;

ENDDATA
min = d1 + @sum(N(I)|I#NE#1 #AND# I#NE#7:d2) + d3 + @SUM(R(K):d4);
@FOR(R(K):@SUM(N(I):q(I)*Y(I,K))+d4(K)=U);
@FOR(R(K):(@SUM(A(I,J):t(I,J)*X(I,J,K))+@SUM(N(I):s(I)*Y(I,K)))/60=TR);
@SUM(R(K):TR)<=UT;
@SUM(R(K):TR)-d3=0;
@FOR (R(K):B=@SUM(A(I,J):c(I,J)*X(I,J,K)));
@SUM(R(K):B)<=UC;
@SUM(R(K):B)-d1=0;
@FOR(R(K):@FOR(N(I):@SUM(N(J):X(I,J,K))=Y(I,K)));
@FOR(N(I)|I#NE#1 #AND# I#NE#8:@SUM(R(K):Y(I,K))+d2=1);
@FOR(R(K):@SUM(A(I,J)|J#EQ#7:X(I,J,K))=1);
@FOR(R(K):@SUM(A(I,J)|I#EQ#1:X(I,J,K))=1);
@FOR(R(K):@FOR(N(I):@SUM(N(J)|I#NE#1 #AND# I#NE#7:X(I,J,K))-
@SUM(N(J)|I#NE#7 #AND# I#NE#1:X(J,I,K))=0));
@FOR(R(K):@FOR(N(I):@FOR(N(J):(W(I,K)+S(I)+t(I,J))-100000*(1-
X(I,J,K))<=W(J,K))));
@FOR(N(I):@FOR(R(K): v*Y(I,K) <= W(I,K)));
@FOR(N(I):@FOR(R(K): W(I,K) <= h*Y(I,K)));
@FOR(D(I,J,K):@BIN(X));
@FOR(R(K):@BIN(Y));
@FOR(N(I):@BIN(d2));
END

Fig. 2. Complete Script Using LINGO 11.0

By inputting vehicle capacity 560 units, maximum distribution time 6 hours a

week, and maximum distribution cost Rp 3.000,00 a week, we get two route which
summarize in Table 4.
7 A GOAL PROGRAMMING APPROACH…

Table 4. Summarize output of LINGO

Route Total Distribution Total distribution Total
Cost time (hour) unit LPG
N0-N2-N1-N6 Rp 610,00 2.82 490
N0-N4-N3-N5-N6 Rp 1.970,00 2.77 440
Sum Rp 2.580,00 5.59 930
From Table 4, all customers can be served by the agent. If we decrease the
maximum distribution time to 5 hours a week, then only 4 customers can be served.
It is also logic, if we decrease to 4 hours a week, then only 2 customers can be
served. Decreasing the number of customers which can be served also happened
whenever we decrease maximum distribution cost to Rp 2.000,00. It is only 4
customers can be served.

4. Conclusion
In this paper, we proposed a goal programming approach to solve Vehicle
Routing Problem with Time Windows (VRPTW). Eq. (3) subject to (4) – (17) is
called goal programming model of the VRPTW. The model was implemented and
the results had been obtained using LINGO. From the simulation, by increasing the
value of Tmax , it is followed the increasing customers can be served. As a future
research, we suggest improving our model to solve larger nodes.

References

[1] Jolai, F., & Aghdaghi, M. (2008). A Goal Programming Model for SIngle Vehicle
Routing Problem with Multiple Routes. Journal of Industrial and Systems
Engineering , 154-163.
[2] Sousa, J. C., Biswas, H. A., Brito, R., & Silveira, A. (2011). A Multi Objective
Approach to Solve Capacitated Vehicle Routing Problems with Time Windows Using
Mixed Integer Linear Programming. International Journal of Advanced Science and
Technology , 1-8.
[3] Azi, N., Gendreau, M., & Potvin, J.-Y. (2007). An exact algorithm for a single-vehicle
routing problem. European Journal of Operational Research , 755-766.
[4] Larsen, J. (1999). Parallelization of the Vehicle Routing Problem with Time Windows.
Institute of Mathematical Modelling. Denmark: Technical University of Denmark.
[5] Cook, W., & Rich, J. (1999). A Parallel Cutting-Plane Algorithm for the Vehicle
Routing Problems with Time. Department of Computational and Applied
Mathematics. Houston: Rice University.
[6] Cordeau, J.-F., Gendreau, M., Laporte, G., Potvin, J.-Y., & Semet, F. (2002). A guide
to vehicle routing heuristics. Journal of the Operational Research Society , 512-522.
[7] Hong, S., & Park, Y. (1999). A heuristic for bi-objective vehicle routing with time
window constraints. International Journal of Production Economics , 249-258.
[8] Calvete H.I., G. C. (2007). A Goal Programming Approach to Vehicle Routing
Problems With Soft Time Windows. European Journal of Operational Research ,
1720-1733.
[9] Hashimoto H., I. T. (2006). The Vehicle Routing Problem With Flexible Time
Windows and Travelling Times. Discrete Applied Mathematics , 1364-1383.

7
8 ATMINI D., EMINUGROHO R.S, AND DWI L.