See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303806122

Solving Supply Chain Network Optimization Models Using LINGO

Article in International Journal of Applied Engineering Research · April 2015

CITATIONS READS
28 6,983

3 authors, including:

Anand Jayakumar Arumugham S. Deepa

Suguna College of Engineering Karpagam College of Engineering
34 PUBLICATIONS 267 CITATIONS 182 PUBLICATIONS 1,807 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Anand Jayakumar Arumugham on 05 June 2016.

The user has requested enhancement of the downloaded file.

 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 10, Number 19 (2015)
© Research India Publications ::: http://www.ripublication.com

Solving Supply Chain Network Optimization

Models Using LINGO
Dr.C.Krishnaraj, A.Anand Jayakumar, Dr.S.Deepa Shri

must be selected. For each product-market, a marketing policy

Abstract— A typical SCN design problem sets the setting service and inventory levels, as well as maximum and
configuration of the network and the missions of its locations. minimum sales levels, must also be selected. The objective is
Some facilities may be opened, others closed, while others can be typically to maximize net profits over a given planning
transformed using different capacity options. Several modeling
horizon. Typical costs include fixed location/configuration
approaches can be used to formulate the supply chain network
design problem. The simplest models available are appropriate to costs, fixed vendor and market policy selection costs, as well
solve facility location problems (FLP), which can be either as some variable production, handling, storage, inventory and
capacitated (CFLP) or uncapacitated (UFLP). In this paper we transportation costs
solve a numerical problem using LINGO software.

Index Terms—Facility location problems, LINGO, SCN. II. LITERATURE SURVEY

Several modeling approaches can be used to formulate the
I. INTRODUCTION supply chain network design problem. The simplest models
available are appropriate to solve facility location problems
A manager decides on the location and capacity allocation for
(FLP), which can be either capacitated (CFLP) or
each facility. Besides locating the facilities a manager also
uncapacitated (UFLP). In facility location models, the capacity
decides how markets are allocated to facilities. This allocation
of potential facilities is assumed to be predetermined. As
must account for customer service constraints in terms of
capacity acquisition is a rather fundamental aspect of supply
response time. The demand allocation decision can be altered
chaindesign problems, several authors investigated capacity
on a regular basis as costs change and markets evolve. When
expansion and relocation alternatives.
designing the network, both location and allocation decisions
HasanSelim and IremOzkarahan (1), have developed a
are made jointly.
model to select the optimum numbers, locations and capacity
Successful supply chain management requires many decisions
levels of plants and warehouses to deliver products to retailers
relating to the flow of information, product, and funds. These
at the least cost while satisfying desired service level to
decisions fall into three phases, depending on the frequency of
retailers. A maximal covering approach is used in statement of
each decision and the period over which a decision phase has
the service level.
an impact. The design, planning, and operation of a supply
HadiMohammadiBidhandi et al (2) have proposed a mixed
chain have a strong impact on overall profitability and
integer linear programming model and solution algorithm for
success.
solving supply chain network design problems in
A typical SCN design problem sets the configuration of the
deterministic, multi-commodity, single-period contexts. The
network and the missions of its locations. Some facilities may
strategic level of supply chain planning and tactical level
be opened, others closed, while others can be transformed
planning of supply chain are aggregated to propose an
using different capacity options. Each selected facility is
integrated model. The model integrates location and capacity
assigned one or several production, assembly and/or
choices for suppliers, plants and warehouses selection, product
distribution activities depending on the capacity options
range assignment and production flows.
available at each location. The mission of each facility must
Marc-Andre´ Carle, AlainMartel, NicolasZufferey (3), have
also be specified in terms of product mix and
proposed paper proposes an agent-based metaheuristic to solve
facilities/customers to supply. Key raw-material suppliers
large-scale multi-period supply chain network design
problems. The generic design model formulated covers the
Dr. C Krishnaraj is a Professor of Mechanical Engineering at Karpagam entire supply chain, from vendor selection, to production–
College of Engineering, Coimbatore, Tamil Nadu, India.
(e-mail:[email protected]) distribution sites configuration, transportation options and
Anand Jayakumar A, is an Assistant Professor in the Department of marketing policy choices. The model is based on the mapping
Mechanical Engineering at Kalaivani College of Technology, Coimbatore, of a conceptual supply chain activity graph on potential
Tamil Nadu,India. (email:[email protected])
Dr. S.Deepa shri is a Professor of Civil Engineering at Karpagam College
network locations.
of Engineering, Coimbatore, Tamil Nadu, India. Sanjay Melkote, Mark S. Daskin (4), have presented a mixed
integer programming formulation of the facility location/

14715
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 10, Number 19 (2015)
© Research India Publications ::: http://www.ripublication.com

network design problem. Computational experience with popular management science text. In 1995, The first Windows
problems with upto 40 nodes and 160 candidate links have release of LINGO was shipped. Today, LINDO Systems
been reported. continues to develop faster, more powerful versions.
Phuong NgaThanha, Nathalie Bostelb, Olivier Pe´ton (5),
have proposed a mixed integer linear program (MILP) for the IV. WHAT IS LINGO?
design and planning of a production–distribution system. The LINGO is a simple tool for utilizing the power of linear and
study aims to help strategic and tactical decisions: opening, nonlinear optimization to formulate large problems concisely,
closing or enlargement of facilities, supplier selection, flows solve them, and analyze the solution. Optimization helps you
along the supply chain. find the answer that yields the best result; attains the highest
JukkaKorpela et al (6), have proposed a framework by profit, output, or happiness; or achieves the lowest cost, waste,
which the risks related to a customer supplier relationship, the or discomfort. Often these problems involve making the most
service requirements by the customers and the strategies of the efficient use of your resources—including money, time,
supplier company can be included in production capacity machinery, staff, inventory, and more. Optimization problems
allocation and supply chain design. The framework is are often classified as linear or nonlinear, depending on
demonstrated with a numerical example and it is based on whether the relationships in the problem are linear with
integrating the analytic hierarchy process (AHP) and mixed respect to the variables.
integer programming (MIP). LINGO includes a set of built-in solvers to tackle a wide
variety of problems. Unlike many modeling packages, all of
AtefehBaghalian, ShabnamRezapour, Reza
the LINGO solvers are directly linked to the modeling
ZanjiraniFarahani (7), have developed a stochastic
environment. This seamless integration allows LINGO to pass
mathematical formulation for designing a network of multi-
the problem to the appropriate solver directly in memory
product supply chains comprising several capacitated rather than through more sluggish intermediate files. This
production facilities, distribution centres and retailers in direct link also minimizes compatibility problems between the
markets under uncertainty. modeling language component and the solver components.
RaghebRahmaniani, Mohammad Saidi-Mehrabad, Local search solvers are generally designed to search only
HojjatAshouri (8), have proposed an extension of the until they have identified a local optimum. If the model is non-
capacitated facility location problem under uncertainty, where convex, other local optima may exist that yield significantly
uncertainty may appear in the model’s key parameters such as better solutions. Rather than stopping after the first local
demands and costs. They have developed the mathematical optimum is found, the Global solver will search until the
formulation in order to allow partial satisfaction by global optimum is confirmed. The Global solver converts the
introducing penalty costs for unsatisfied demands. In general, original non-convex, nonlinear problem into several convex,
this model optimizes location for predefined number of linear subproblems. Then, it uses the branch-and-bound
capacitated facilities in such a way that minimizes total technique to exhaustively search over these subproblems for
expected costs of transportation, construction, and penalty the global solution. The Nonlinear and Global license options
costs of uncovered demands, while relative regret in each are required to utilize the global optimization capabilities.
scenario must be no greater than a positive number (p > 0).
V. THE MODELING FRAMEWORK
III. THE LINDO STORY In this section, the provided mathematical formulation by
Since 1979, LINDO Systems software has been a favorite of Sunil Chopra, Peter Meindl and D V Kalra (9) for logistics
business and educational communities alike. LINDO Systems network design problems is considered. Our presented model
has dedicated itself to providing powerful, innovative for deterministic SCND problems is largely inspired from this
optimization tools that are also flexible and easy to use. work.
LINDO Systems has a long history of pioneering powerful The demand allocation problem can be solved using a
optimization software tools. In 1988, LINGO became LINDO demand allocation model. The model requires the following
Systems first product to include a full featured modeling inputs:
language. Users were able to utilize the modeling language to n = number of factory locations
concisely express models using summations and subscripted m = number of markets or demand points
variables. In 1993, LINGO added a large scale nonlinear Dj= annual demand from market j
solver. It was unique in that the user did not have to specify Ki = capacity of factory i
which solver to use. LINGO would analyze the model and Cij = cost of producing and shipping one unit from factory i to
would engage the appropriate linear or nonlinear solver. Also market j (cost includes production, inventory and
unique to the LINGO’s nonlinear solver was the support of transportation)
general and binary integer restrictions. With the addition of xij= quantity shipped from factory i to market j
the nonlinear solver, LINGO essentially replaced GINO as The goal is to allocate the demand from different markets to
LINDO Systems premier product for nonlinear optimization. the various plants to minimize the total cost of facilities,
GINO made its debut in 1984 and was the first ever nonlinear transportation and inventory.
solver available on the PC Platform. In 1994, LINGO became The problem is formulated as the following linear program:
the first modeling language software to be included in a

14716
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 10, Number 19 (2015)
© Research India Publications ::: http://www.ripublication.com

Fig 1: USA Map

VI. NUMERICAL EXAMPLE
VII. LINGO PROGRAM
The following example was taken from the book by Sunil
Chopra, Peter Meindl and D V Kalra (9). MODEL:
We illustrate this with relevant network optimization model SETS:
using the example of two manufacturers of fiber – optic PLANTS1: FIXEDCOST1, CAPACITY1, Y1;
telecommunication equipment. Both TelecomOne and MARKETS1: DEMAND1;
LINKS1(PLANTS1, MARKETS1): TCOST1,
HighOptic are manufacturers of the latest generation of
VOLUME1;
telecommunication equipment. TelecomOne has focused on
PLANTS2: FIXEDCOST2, CAPACITY2, Y2;
the eastern half of the United States. It has manufacturing MARKETS2: DEMAND2;
plants located in Baltimore, Memphis and Wichita and serves LINKS2(PLANTS2, MARKETS2): TCOST2,
markets in Atlanta, Boston and Chicago. HighOptic has VOLUME2;
targeted the western half of the United States and serves ENDSETS
markets in Denver, Omaha and Portland. HighOptic has plants DATA:
located in Cheyenne and Salt Lake City PLANTS1 = BALTIMORE, MEMPHIS, WICHITA;
Plant capacities, market demand, variable production and PLANTS2 = CHEYENNE, SALT_LAKE_CITY;
transportation cost per thousand units shipped and fixed costs FIXEDCOST1 = 7650, 4100, 2200;
per month at each plant are shown in the table. FIXEDCOST2 = 3500, 5000;
Table 1: Input Data – Production and Transportation Cost CAPACITY1 = 18, 22, 31;
Demand City Production and Transportation Cost per CAPACITY2 = 24, 27;
Thousand Units (Thousand $) MARKETS1 = ATLANTA, BOSTON, CHICAGO;
Supply City Atlanta Boston Chicago Denver Omaha Portland
MARKETS2 = DENVER, OMAHA, PORTLAND;
Baltimore 1675 400 685 1630 1160 2800
Cheyenne 1460 1940 970 100 495 1200
Salt Lake 1925 2400 1425 500 950 800 DEMAND1 = 10, 8, 14;
City DEMAND2 = 6, 7, 11;
Memphis 380 1355 543 1045 665 2321
Wichita 922 1646 700 508 311 1797 TCOST1 = 1675 400 685
Monthly 10 8 14 6 7 11
Demand
380 1355 543
922 1646 700 ;
TCOST2 = 100 495 1200
Table 2: Input Data – Capacity and Fixed Cost
Supply City Monthly Monthly
500 950 800;
Capacity Fixed Cost ENDDATA
(Thousand Units) (Thousand $) MIN =
Ki fi @SUM(LINKS1(I,J):TCOST1(I,J)*VOLUME1(I,J)
Baltimore 18 7650 )+
Cheyenne 24 3500
Salt Lake City 27 5000
@SUM(LINKS2(I,J):TCOST2(I,J)*VOLUME2(I,J)
Memphis 22 4100 );
Wichita 31 2200 @FOR(MARKETS1(J):@SUM(PLANTS1(I):VOLUME1(
I,J)) = DEMAND1 (J););
We calculate that TelecomOne has a total production @FOR(MARKETS2(J):@SUM(PLANTS2(I):VOLUME2(
capacity of 71000 units per month and a total demand of I,J)) = DEMAND2 (J););
30000 units per month, whereas HighOptic has a production @FOR (PLANTS1 (I): @SUM (MARKETS1(J) :
capacity of 51000 units per month and a demand of 24000 VOLUME1 (I,J))<= CAPACITY1 (I) * Y1(I););
@FOR (PLANTS2 (I): @SUM (MARKETS2(J) :
units per month. Each year, managers in both companies must
VOLUME2 (I,J))<= CAPACITY2 (I) * Y2(I););
decide how to allocate the demand to their production
@FOR(PLANTS1:@BIN(Y1));
facilities as demand and costs change. @FOR(PLANTS2:@BIN(Y2));

14717
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 10, Number 19 (2015)
© Research India Publications ::: http://www.ripublication.com

END X. CONCLUSION
Thus the global optimal solution was found. Thus in this paper
VIII. COMPUTATIONAL EFFICIENCY we have used LINGO program to solve the SCND problem.
An intel CORE i5 processor with 4GB RAM was used to
process the model. Branch and Bound solver was used.
REFERENCES
A. Numerical Problem size
[1] HasanSelimand IremOzkarahan,―A supply chain distribution network
Total variables: 20 design model: An interactive fuzzy goal programming-based solution
Nonlinear variables: 0 approach‖, International Journal of Advanced Manufacturing
Integer variables: 5 Technology (2008) 36:401–418.
[2] HadiMohammadiBidhandi, RosnahMohd. Yusuff,
Total constraints: 12 MegatMohamadHamdanMegat Ahmad,MohdRizam Abu Bakar,
Nonlinear constraints: 0 ―Development of a new approach for deterministic supply chain network
Total nonzeros: 50 design‖, European Journal of Operational Research 198 (2009) 121–128.
[3] Marc-Andre´ Carle, AlainMartel, NicolasZufferey, ―The CAT
B. Run Time metaheuristic for the solution of multi-period activity-based supply
chain network design problems‖, International Journal of Production
The problem was solved in less than 1 second. Economics 139 (2012) 664–677.
[4] Sanjay Melkote, Mark S. Daskin, ―Capacitated facility location/network
design problems‖, European Journal of Operational Research 129 (2001)
481-495.
[5] Phuong NgaThanh, Nathalie Bostel, Olivier Pe´ton, ―A dynamic model
for facility location in the design of complex supply chains‖,
International Journal of Production Economics 113 (2008) 678–693.
[6] JukkaKorpela, KaleviKylaKheiko, AnttiLehmusvaara,
MarkkuTuominen, ―An analytic approach to production capacity
allocation and supply chain design‖, Int. J. Production Economics 78
(2002) 187-195.
[7] AtefehBaghalian, ShabnamRezapour, Reza ZanjiraniFarahani, ―Robust
supply chain network design with service level against disruptions and
demand uncertainties: A real-life case‖, European Journal of Operational
Research 227 (2013) 199–215.
[8] RaghebRahmaniani, Mohammad Saidi-Mehrabad, HojjatAshouri,
―Robust Capacitated Facility Location Problem: Optimization Model
and Solution Algorithms‖, Journal of Uncertain Systems, Vol.7, No.1,
pp.22-35, 2013.
[9] Sunil Chopra, Peter Meindl and D V Kalra, ―Supply Chain Management
Strategy, Planning and Operation‖, 5th Edition, Pearson.
[10] LINDO Systems Inc, ―LINGO The Modeling Language and Optimizer‖.
[11] LINDO Systems Inc,―Optimization Modeling with LINGO‖, 6th
Edition.
[12] C Krishnaraj, KM Mohanasundram, SR Devadasan, NM Sivaram,
Total failure mode and effect analysis: a powerful technique for
overcoming failures, International Journal of Productivity and Quality
Management 10 (2), 131-147, 2012.
Fig 2: Result [13] KM Mohanasundaram, NGR Ebenezer, C Krishnaraj, Forward
Kinematics Analysis of SCORBOT ER V Plus using LabVIEW,
IX. RESULT European Journal of Scientific Research 72 (4), 549-557, 2012.
[14] C Krishnaraj, KM Mohanasundram, S Navaneethasanthakumar
The optimal demand allocation is presented below. Observe Implementation Study Analysis Of Ftfmea Model In Indian Foundry
that it is optimal for TelecomOne not to produce anything in Industry,Journal of Applied Sciences Research 8 (2), 2012.
the Wichita facility even though the facility is operational and [15] C Krishnaraj, KM Mohanasundram, Design and Implementation Study
of Knowledge Based Foundry Total Failure Mode Effects Analysis
the fixed cost is incurred. Technique, European Journal of Scientific Research 71 (2), 298-311,
Table 3: Result - TelecomOne 2012.
Atlanta Boston Chicago
Baltimore 0 8 2
Memphis 10 0 12
Wichita 0 0 0

Table 4: Result – HighOptic

Denver Omaha Portland
Salt Lake 0 0 11
Cheyenne 6 7 0

14718

View publication stats