JOURNAL OF ADVANCEMENT IN

ENGINEERING AND TECHNOLOGY

Journal homepage: www.scienceq.org/JAET.php

Research Article Open Access

Supply Chain Optimization Modeling, a Case Study of a Glass Industry in Nigeria

Alexander Akene 1, Oluwayomi J. Oyejide 2 ,

1,2
Department of Mechanical Engineering, Federal University of Petroleum Resources, Effurun, Delta State, Nigeria.
*Corresponding author: Alexander Akene, E-mail: akene.alexander@fupre.edu.ng
Received: April 15, 2020, Accepted: May 26, 2020, Published: May 26, 2020.
ABSTRACT
The application of processes and tools such as supply chain optimization mathematical programming tool, which ensures cost
optimization or profit maximization of any establishment is of great importance in our ever-changing world. In this study, solution
for supply chain optimization problem for a pre-deterministic demand of supplier to customer requirements in which demand
uncertainties were not considered was carried out using a case study of a glass industry in Nigeria. A mathematical optimization
model with cost minimization as its main objective function given all its sets of constraints was developed. The model was
implemented using Generalized Algebraic Modeling System (GAMS) and results obtained were compared with that of LINGO
solver. Generated data obtained from a glass industry in Nigeria was computed in the proposed models. The outcome of the results
revealed that the objectives function Z (optimal cost) of the GAMS model was about 3% more than that of the LINGO model
system. Besides, the raw material requirement analysis showed a difference of not more than 2% between the two models. From the
GAMS results as compared to that of the test value, it was observed that there was an 8% reduction in the total operating cost of
production when compared to that of the real value obtained. More so, the results of the raw material requirement obtained from
both models showed slight variations as well, validating the results obtained from the model implemented. In addition, an
affectability test with a 5% increase in the transportation cost impacted greatly on the overall cost of operation leading to a higher
cost of production.
Keywords: Optimization, model, supply chain, cost, glass industry

INTRODUCTION thereby allowing it to function in an effective, smooth and timely

In the past, manufacturers produce and distribute goods at their manner [7-8]. Although, researchers has over the years
own pace but in today’s market, customers are the pace setters as continuously formulated methods and techniques to analyze the
manufacturers scramble to meet their everyday demand. Thus, supply chain process for better decision making. However,
the business world has no doubt shifted from manufacturers designing an optimal SC network means the network must be
driven to customers driven. In the light of this, business strategies able to meet the long-term strategic objectives of the company
such as Six Sigma, Total Quality Management, Just-In-Time, [9].
Manufacturing Resource Planning, Quality Circle, etc has all The usefulness of supply chain optimization model is numerous.
been introduced at some point in the past to improve business It includes optimizing business operations to minimize cost or
performance and product delivery time. However, over time maximize profit, efficiency, speed at which finished products are
these strategies lose their shine and in recent years many delivered to customers, etc. In all, Customers value good
businesses and companies are beginning to discover that products and the speed at which such products gets to them is
effective supply chain management (SCM) is the way to go in very important, however, maximizing profitability by the
order to increase market share and profit [14]. The use of manufacturer is equally important. There for, ensuring that
modeling techniques in planning and optimization of supply products are delivered as fast and economical as possible without
chains (SC) activities in various industries is gaining widespread sacrificing quality is of paramount importance for an effective
acceptance [1-3]. Despite the usefulness of such techniques, SCM system [10].
some entrepreneurs still lack the technical knowhow and the Study revealed that multiple periodic (MILP) model provided
gains its acceptance and possible implementation would bring to solutions to factory load allocation problems. The model was
their establishment. SC optimization process is primarily able to determine the production lines, production and finished
concerned with finding ways through which firms can reduce products delivery number and raw material procurement of the
minimize overall operating cost, wastage and make the supply supply chain process [11]. Similarly, [12] evaluated global SCM
chain (SC) process more efficient and faster [4-6]. Therefore, at digital equipment cooperation and in their design approach; a
designing a model that allows firms make better strategic and multi-cycle MILP model was developed for the supply chain
tactical decision rather than relying on personal judgment or process. In other to minimize total cost of the SC, factors such as
experience for profit maximization is of great importance. Supply transportation cost, inventory cost, taxes, production cost, etc
chain optimization is the practice of combining resources in a were considered for the model. A real world case was used to test
supply chain with the purpose of removing delays and any other the model indicating a reduction in the total cost of the SC
problems that may interfere with the supply chain process process. Also, [13] proposed a non-separable, nonlinear, MILP

J. of Advancement in Engineering and Technology Volume 7 / Issue 4 ISSN: 2348-2931 1

for a class of location-allocation problem involving several constraints was developed. The proposed optimization model
echelons for the planning and scheduling of parallel semi- was taken into account (four participants via suppliers,
continuous processes. A successive MILP algorithm where manufacturing plants, distribution centers and customers) in
solution is constrained within trust region was used in solving the trying to optimize or having a near to optimal solution. The
model. [14] Considered a stochastically varying production and model was implemented using Generalized Algebraic Modeling
transportation cost supply chain network optimization model System (GAMS) and results obtained were compared with that of
design. A two-stage supply chain was proposed with the first LINGO solver (a more advanced solver) for results validation.
being a MILP, solved to provide solution for the second model. The assessed information (data) for the model was obtained from
In the study, variables with noise (stochastic model) were a glass company in Nigeria and computed in the proposed
compared to those without noise (deterministic model) using models. The variables of the mathematical model are defined as
Gaussian, Lognormal and Pareto distribution. Research study on shown in Table 1.
linear programming model has been implemented in [15-16]. Table 1. Variables of the Mathematical Model
However, [17-20] demonstrated the use of high-level algorithms Variable Sets and Indices Definitions
in supply chain optimization. [21-22] Developed an integer- r Raw material type, r=1, 2,…………… R
programming model for transportation optimization of oil m Manufacturing plant, m=1, 2, …….., M
products. In his study, distances and cost minimization were set p Product type, p=1, 2,..…………..., P
as the objective function in other to obtain the refinery-to-deport s Supplier, s=1, 2,…………………., S
optimal assignments. In his approach adopted to run the program d Distribution center, d=1, 2,………, D
using I-Log software. Real case study data were selected and the z Customers, z=1, 2,……………….., Z
resulting established are highly feasibly in reality to obtain the
best outcome. [23] Developed a mathematical model of Input parameters
multilevel cost optimization for SC in the form of mixed integer
linear programming (MILP) for optimizing the supply chain from
the prospective of a logistics provider. Production, distribution Tr(sm) Unit transportation cost for raw material, r, from
and transportation cost were adopted as an optimization criterion supplier, s, to manufacturing plant, m.
and the model was implemented in the mathematical modeling
package “LINGO” environment. C(rs) Unit cost of crude material, r, for provider, s, to
Researchers has proposed and implemented various optimization assembling plant, m.
tool and algorithm for solving supply chain management
problems. However, large amount of research work still needs to Fm(m) Fixed expense for assembling plant, m
be done on minimization or maximization of the entire supply
H(pd) Inventory holding cost per unit of item, p, at
chain process (supplier, production, distribution, inventory,
transportation and location) for better optimality of the process. dissemination focus, d.
More so, works still needs to be done in model validation and
Cp(pm) Unit cost of item, p, at assembling plant, m.
raw material requirement with respect to customer’s demand.
Tp(md) Unit transportation cost for item, p, from
MATERIALS AND METHODS
assembling plant, m, to circulation focus, d.
Problem Structure and Model Formulation
As shown in Fig. 1, a four echelon SC process was considered L(pd) Inventory level of item, p, at circulation focus, d.
with a given sets of suppliers, fixed factories and distribution
centers that delivers finished products to its multiple customers. Ud(pz) Demand of item p by client, z.
Sets of logical constraints for each of the SC echelon were Decision variables
developed to ensure coherence in the SC decisions model.
Ur(rsm) Number of units of crude material, r, secured from
provider, s, to assembling plant, m.
Up(pm) Number of units of item, p, delivered at
assembling plant, m.
U(pmd) Number of units f item, p, transported from
assembling plant, m, to circulation focus, d.
Binary variables
Xm(m) 1, if fabricating plant, m, is open; 0, assuming
Fig. 1 Four echelon supply chain network representation
something else.
The problem undertaken is a pre-deterministic demand of
Xd(d) 1, if conveyance focus, d, is open; 0, assuming
supplier to customer requirements in which demand uncertainties
were not considered. Thus, it is a deterministic problem and a something else.
mixed integer linear programming (MILP) model is being
Xc(dz) 1, if dispersion focus serves client, z; 0, assuming
proposed. A mathematical optimization model with cost
minimization as its main objective function given all its sets of something else.
J. of Advancement in Engineering and Technology Volume 7 / Issue 4 ISSN: 2348-2931 2
 Table 2. Describes the basic syntax of GAMS mathematical

Mathematical Formulation Nomenclature GAMS Syntax

The expression for the cost optimization of the entire supply Minimum MIN(.,.)
chain (SC) process as formulated based on the objective function Addition +
shown in Equations (1) to Equation (4); Equality in an operation =
Summation SUM(set domain, element)
Multiplication *
(1) Subtraction -
Product PROD(set domain, element)
modeling language
LINGO Model
(2) The LINGO is a high – level optimization modeling tool
for solving mathematical optimization problems. The
model is designed to solve and analyze linear, stochastic,
(3) integer, quadratic and nonlinear optimization problems.
LINGO provides a completely integrated package that
(4) includes a powerful language for expressing optimization
models and a set of fast built-in solvers. In LINGO, MILP
The objective function (Z1) consists of four echelon and problems are solved using the branch and bound technique.
defines the costs of the entire supply chain process. Equation (1) The algorithm for solving MILP in LINGO is as stated in
consists of all related cost associated with getting the raw
Table 3.
material to the manufacturing plant. Equation (2) consists of all
variable cost associated with producing the product and getting it Table 3. The syntax mathematical modeling language as
to the distribution center, while Equation (3) on the other hand represented in LINGO
consists of cost associated with setting up of the distribution Nomenclature LINGO Syntax
center and holding cost, and Equation (4) represent all related Minimum MIN=
cost of getting the products delivered to the customers. Addition +
More so, the cost of the entire supply chain process is subject to Equality in an operation =
observed constraints which include: availability constraint which Summation @SUM(set domain,
ensures that the total produced product do not exceed the plant’s element)
capacity; demand constraint, which specify that the number of Multiplication *
units of products produced in the manufacturing plant equals that Subtraction -
shipped from the plant; constraint which ensures that the total Product @PROD(set domain,
number of units of products shipped from the distribution center element)
satisfies demand; storage capacity constraint which ensures that X ϵ integer (ginX)
the numbers of unit of product do not exceed its storage capacity; X ϵ {0,1} @bin(X)
plant capacity constraint which ensures that the produced j=1,…..,M for each customer(j) in @FOR(customers(j))
products do not exceed the plant capacity; supplier capacity the set of customers
which ensures that the raw materials supplied are within the
supplier’s capacity and the non-negativity constraint. RESULTS AND DISCUSSION
The generated data were inputted separately into the GAMS
Model Implementation and LINGO modeling framework of the MILP problem
Implementation of the model is done with the use of two (2)
different MILP systems; the General Algebraic Modeling System
environment. The outcome yielded the following results as
(GAMS) and the LINGO system, with the former being the main shown in Table 4. As observed from the results in Table 4,
solver while the latter is implemented to compare and validate the objective function Z (optimal cost) of the GAMS model
the main model. was about 3% more than that of the LINGO modeling
framework (See Appendix A1 and A2). The raw material
General Algebraic Modeling System Model requirement when looked at demonstrated a distinction of
The General Algebraic Modeling System (GAMS) is a not more than 2% between the two models. This shows a
mathematical optimization tool designed for modeling and low distinction in the outcome acquired between the two
solving nonlinear, linear and mixed-integer optimization models (GAMS and LINGO) when looked at as a
problems. GAMS systems are adapted for large-scale, complex framework for tackling the MILP studied. There for,
modeling applications. The system uses the branch and cut
validating the results obtained from the main model in
approach for solving MILP problems. The approach is a type of
branch and bound solver that involves a combination of cutting view (GAMS). More so, the LINGO results obtained as
plain and LP relaxation method with its algorithm as shown in compared to the GAMS results indicates that the LINGO
Table 2. mathematical tool is a more efficient and superior tool for
finding solutions to SC optimization as researched [17].

J. of Advancement in Engineering and Technology Volume 7 / Issue 4 ISSN: 2348-2931 3

 Table 4. GAMS and LINGO modeling system results

Model Solver Parameters

Cost (₦) Raw Material Raw Material Raw Material Raw Material
1 2 3 4
GAMS 66,881,645.4545 3,620 2,840 2,230 1,710
LINGO 64,844,564.2586 3,550 2,780 2,195 1,675
Difference 2,037,081 170 60 35 35
% Difference 3.045 1.93334 2.1127 1.5695 1.4619

Table 5. GAMS and Real value result

Cost Raw Material 1 Raw Material 2 Raw Material 3 Raw Material 4
GAMS 66,881,645.4545 3,620 2,840 2,230 1,710
Real Value 72,684,204 3,810 3,000 2,405 1,875
Difference 5,802,559 190 160 175 165
% Difference 7.9837 4.9869 5.3333 7.2765 8.8000

Table 5 shows the results obtained from the model

(GAMS) and that of the test values (company’s value).
From the GAMS results as compared to that of the test
value, it can be seen that there is an 8% reduction in the
total operating cost of production when compared to that of
the test value obtained. The 8% decrease achieved which is
about N5.8 million is a very significant variation as this
additional cost can incredibly affect the unit cost of each
products produced, it can also affect the company’s
competitiveness and on the overall smooth running
operations of the company. The reduction in cost observed Fig.2 Raw material sensitivity analysis
was largely due to the reduction in the quantity of raw The affectability examination was likewise done to show
material obtained from the model that should be ordered in how an expansion or a diminishing in the stock level would
relation to customers demand as shown in the Table 5. influence the optimal solution (cost). As appeared in Fig. 3,
Consequently, for the organization to accomplish such a 10% expansion in the stock level brought about a
decrease and limit its working expense according to its comparing increment in the overall cost of production. In
clients’ demand, the amount of raw material as gotten from this manner, the higher the stock level an association keeps
the GAMS model should be implemented. the higher it's operational cost.

An affectability examination on a portion of the results

obtained uncovers that changes was affected adversely or
emphatically on the output data of the model. Through this
the impact of uncertainty on the output of the optimal
solution can be known. A sensitivity analysis test was done
on the quantity of raw material required as to how its
changes would affect the result of the optimal solution
(cost).
Fig.3 Inventory level sensitivity analysis
Figure 2 shows a raw material graph. From it, for as low as Affectability investigation was similarly done on the
a 5% increase in the quantity of raw material ordered, there transportation cost for shipping items to client zones and its
is a corresponding rise in the cost of operation. Indicating corresponding effect on the operational cost is as shown in
that if the proper analyses are not carried out on the right Fig. 4. From the plots of the transportation cost of
quantity of raw material to be ordered with respect to its conveying the items to the client's zone against its
customers demand, the organization will incur more operational cost, a corresponding relationship diagram was
operational cost of production. Therefore, knowing the set up. It was seen from the plot that as low as a 5%
right amount of raw material needed for producing a expansion in the transportation cost affected enormously
certain amount of goods in any organization is of on the general expense of activity prompting a greater cost
paramount importance. of production. This was as an aftereffects of increment in
fuel, mileage and support cost that has huge impact on the
general expense. Then again, if there is a decrease of cost
J. of Advancement in Engineering and Technology Volume 7 / Issue 4 ISSN: 2348-2931 4
of maintenances and fuel there would be a comparing 8. Mengye, C. (2014): “Optimization of A Supply Chain Network For
decrease in the general working expense. Bioenergy Production From Food Waste”, Illinois at Urbana-
Champaign, USA.
9. Park, Y. B. (2007). An integrated approach for production and
distribution planning in supply chain management. International
Journal of Production Research, 43: 1205-1224.
10. Mula, J., Peidro, D., Díaz-Madroñero, M., & Vicens, E. (2010).
Mathematical programming models for supply chain production
and transport planning. European Journal of Operational Research,
204:377–390.
11. Cohen and Moon (1991). Designed of a multiple periodic MILP
model.
12. Arntzen, B., Brown, G.G., Harrison, T., and Trafton, L. (1995).
Fig. 4 Transportation sensitivity analysis Global Supply Chain Management at Digital Equipment
Corporation. Interfaces, 25:69-93
CONCLUSION 13. Mcdonald, C., & Karimi, I. (1997). Planning and scheduling of
Conclusively, the developed model proved to be an effective and parallel semicontinuous processes. Production planning. Industrial
efficient for providing solution to SC optimization problem in the and Engineering Chemistry Research, 36, 2691–2700.
glass industry. The MILP tool adopted (GAMS) for cost 14. Jose, P. (2015) “Supply chain optimization Process and Supply
optimization of the problem undoubtedly facilitated its solution Chain Operations”, Pan American Advance Study institute program
as compared and validated using a more superior mathematical and systems engineering (PASI)
programming package (LINGO). Implementation of both models 15. M.O. Okwu, F. I. Ashiedu, Joachim Gidiagba, Adedoyin Adesuji
allowed for obvious variation of the overall cost of production, (2017). Application of Linear Programming Model and Sensitivity
raw materials requirement, and number of products produced as Analysis in Multi-product Multi-destination System. Canadian
International Journal of Science and Technology ISSN 2356-9085,
compared to the real values obtained from the glass industry. 168-180
Based on the experimental results and analysis, the study 16. Opara Ignatius. (2012) Optimization Modelling for Multi-Objective
revealed a significant reduction in the overall cost of operation of Supply Chains, A Case Study of the Oil and Gas Sector.
about 8% which is about N5.8 million when compared to that of International Journal of Science and Technology, Bahir Dar,
the real value obtained. The model implemented also showed an Ethiopia Vol 1 (3), 54-65
appropriate and appreciable measure of performance when 17. Okwu M.O. and Lagouge K Tartibu (2020) Sustainable supplier
affectability analysis was performed on some of its input selection in the retail industry: A TOPSIS- and ANFIS-based
parameters. Therefore, a well formulated and implemented evaluating methodology. International Journal of Engineering
mathematical modeling tool can assist an organization in make Business management. Volume 12, 1-14. DOI:
10.1177/1847979019899542
useful decisions in supply chain management. 18. Olufemi Adetunji and M.O. Okwu (2018) A comparative study of
REFERENCES artificial neural network (ANN) and adaptive neuro-fuzzy inference
1. Zhang, Q., Shah, N., Wassick, J., Helling, R., & Egerschot, P. V. system (ANFIS) models in distribution system with
(2014). Sustainable supply chain optimisation: An industrial case nondeterministic inputs. International Journal of Engineering
study. Computers & Industrial Engineering, 74:68–83. Business management. Volume 10, 1-17 DOI:
2. Torabi, S., & Hassini, E. (2008). An interactive possibilistic 10.1177/1847979018768421
programming approach for multiple objective supply chain master 19. Chukwu V.U., Oguoma O. (2019) Application of Artificial
planning. Fuzzy Sets and Systems, 159(2): 193–214. Neural Network Model for Cost Optimization in a Single-
3. Stacey, J., Natarajarathinam, M., & Sox, C. (2007). The storage Source, Multi-destination System with Non-deterministic
constrained, inbound inventory routing problem. International Inputs. In: Rojas I., Joya G., Catala A. (eds) Advances in
Journal of Physical Distribution & Logistics Management, 37: 484 Computational Intelligence. IWANN 2019. Lecture Notes in
– 500 Computer Science, vol 11507. Springer, Cham
4. Songsong, L., & Papageorgiou, L. G. (2013). Multiobjective DOIhttps://doi.org/10.1007/978-3-030-20518-8_45.
optimisation of production, distribution and capacity planning of 20. M. O. Okwu, Benjamin U. Oreko, Stanley Okiy, Austin C. Uzorh
global supply chains in the process industry. Omega, 41(2): 369– & Onyewuchi Oguoma,(2018) Artificial neural network model for
382 cost optimization in a dual-source multi-destination outbound
5. Shi, J., Zhang, G., & Sha, J. (2012). A Lagrangian based solution system, Cogent Engineering (2018), 5: 1447774.
algorithm for a build-to-order supply chain network design problem 21. Satish, V. B and Atul, B. B. (2017) “Domain Of Supply Chain
. Advances in Engineering Software, 49:21–28 Management – A State Of Art”, Journal of Technology
6. Selim, H., Am, C., & Ozkarahan, I. (2008). Collaborative Management and Innovation”, Vol. 2, Issue 4.
production–distribution planning in supply chain: a fuzzy goal 22. Waleed K. A. (2013) “Transportation optimization model of oil
programming approach. Transportation Research Part E: Logistics products” Scientific Research and Essays, Academic Journals, Vol.
and Transportation Review, 44(3): 396–419. 8(5), pp. 211 -219, 4 February, 2013
7. Safaei, A. S., S.M., M. H., Z., F. R., F., J., & Ghodsypoura, S. 23. Paweł, S. and Jarosław, W. (2018) “Mathematical Programming
(2010). Integrated multi-site production-distribution planning in Model of Cost Optimization For Supply Chain from Perspective of
supply chain by hybrid modelling. International Journal of Logistics Provider”, Management and Production Engineering
Production Research, 48(14): 4043-4069. Review, Vol. 3. No 2. pp. 49–61.

Citation: Alexander Akene et al. (2020). Supply Chain Optimization Modeling, a Case Study of a Glass Industry in Nigeria. J. of
Advancement in Engineering and Technology, V7I4.07. DOI: 10.5281/zenodo. 3855528.
Copyright: © 2020: Alexander Akene. This is an open-access article distributed under the terms of the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and
source are credited.

J. of Advancement in Engineering and Technology Volume 7 / Issue 4 ISSN: 2348-2931 5