Fashion Supply Chain Optimization: A Linear Programming

Model

Sara Shoarinejad a , Amin Shoari Nejad b ,

a
Department of Industrial Engineering, K. N. Toosi University of Technology, Tehran, Iran
b
Hamilton Institute, Deparment of Mathematics and Statistcs, Maynooth University, Kildare, Ireland

Abstract

Fashion supply chain management is challenging due to demand uncertainty, short product life cycle, fierce global competition,
increasing expectations for diversity in production, and keeping pace with the current and emerging trends. In this paper, a
brief review of the published works in the field of fashion and clothing supply chain is presented and an integrated linear model
is developed to minimize the costs of a fashion brand, including production and logistics costs, enabling the firm to make better
decisions with regard to its production and replenishment policies and to respond in a reasonable time to customers demands
for various designed items. We show that our model can successfully account for the complex structure of the supply chain
consisting of designers, manufacturers, distributors, and stores. The model is validated using the General Algebraic Modeling
System (GAMS) with the data collected from an Iranian fashion brand.

Key words: Fashion Supply Chain Management; Operations research; Optimization; Mathematical Modelling.

1 Introduction Today fashion industry is bigger than ever and in or-

der to remain in the business it is very important to ef-
ficiently manage the fashion supply chain especially in
The fashion industry as a whole can consist of design, the presence of different players such as suppliers, dis-
procurement of raw materials, production, distribution, tributors and producers that need to collaborate in har-
marketing and sales. Mass production of clothes started mony in order to meet the market demand. Fashion sup-
in mid-19th century when producers started producing ply chain management requires considering the industry
cloths that didn’t need the tailor to take measures for as a complex interconnected system in which businesses
every customer individually. However, fashion did not need to try balancing supply and demand, minimizing
become an industry until the 20th century when the net- costs, maximizing profit, managing risk, and unifying
work of local tailors transformed into bigger production their processes and their associated information flows for
businesses. After world war II textile factories and cloth- optimizing their supply chain and obtaining competitive
ing businesses experienced higher development due to advantage [1,11].
higher demands and cultural and societal changes which
eventually resulted in less regularization and more di- In this paper, we aim at developing a mathematical
versity in the clothing and fashion industry. At the be- model that takes into account demand for multiple fash-
ginning of the 20th century, significant changes in the ion items in different stores and helps businesses to de-
fashion business such as higher attention to creativity cide about their production quantities and distribution
in design, European fashion houses, technological devel- plans while minimizing their costs.
opment such as sewing machines, the emergence of cap-
italism, systematic development of factories and retail- The paper is organized as follows. In section 2, a sum-
ers made the fashion industry bigger and more powerful mary of previous related works is provided. In section
[9,2]. 3, we describe the problem that we try to solve. In sec-
tion 4, we explain our modelling approach. In section 5
we discuss the model validation and results. In section
Email addresses: [email protected] (Sara 6, we discuss the implications of our method and how
Shoarinejad), [email protected] (Amin Shoari our model can help decision makers. We summerize the
Nejad). paper and conclude in 7.

Preprint submitted to engrXiv.org 5 January 2022

2 Literature review

[3] studied a flexible pricing scheme that allows the price

to vary depending on customer. Their goal was to find an
optimal solution using a unified model that considers the
flexibility between the start and the end time of produc-
tion and acceptance/rejection of the orders. They used
a mixed-integer programming and a heuristic method to
solve the problem and showed the significant effect of
unified production on profit considering other decisions
of the supply chain.

[7] focused on fashion products under deterministic de-

mand, economic order quantity (EOQ) and limited re- Fig. 1. Schematic of the fashion supply chain.
plenishment. Fashion retailers usually use dynamic pric-
ing (e.g. offering discount) to deplete their stocks. They high operations flexibility to respond to demand in a
found the optimal order quantity and order point in or- timely manner. [4] developed an optimization model for
der to maximize profit and showed the effect of them on a supply chain of a fashion retailer for a fast response to
demand by considering the customer saving price (the demand and having an optimized replenishment policy.
minimum price that a customer is willing to pay). They Their model is cable of calculating the optimal quantity
utilized discrete and dynamic pricing schemes and alter- that needs to be transferred between stores in order to
nated between them multiple times in order to satisfy meet the demand in different locations. They suggested
customers demand in different periods and minimized that their model can help international retailers to make
the inventory cost. better decisions for replenishing their stocks according
to feedback from the market and sales data.
[10] introduced a system dynamics model to analyze
the fashion supply chain. They evaluated the effect of
work duration and transportation delays on the perfor- 3 Problem definition
mance of a fast fashion supply chain including three lev-
els of producer, distributor and retailer. They showed In this section we outline the problem. The goal is to de-
how each level is affected by the delays. velop a model for a multilevel fashion supply chain en-
abling fast response to customers demands. We suggest
[8] developed a causal loop diagram for fashion supply a model considering a direct supply chain and we aim
chain. They divided the supply process into two parts: at developing a decision support system for production
pre sale season and the sale season. The former includes and replenishment quantities. We use data provided by
collection planning, demand forecasting, purchase, logis- Queen Manto, a fashion brand based in Iran. Our sug-
tics and replenishment and the latter includes balancing gested model could be an extension to the [4] by con-
supply and demand. sidering four levels of fashion supply chain instead of
one. The supply chain of our problem includes designing
[5] focused on the problem of determining initial order center, manufacturing centers, distribution centres and
quantity and replenishment of a retail product to min- stores. The functionality of each level are as follows:
imize the opportunity cost, cost of returned products
and expired inventory. They framed the problem into a (1) Designing center: Determining the kind and
two-step stochastic dynamical programming model and quantities of different designs to be produced by
suggested a novel heuristic method to find the optimal the manufacturing centers.
solution. They showed that their solution could double (2) Manufacturing centers: Producing the placed
the profit. orders and sending them to the distribution centers.
(3) Distribution centers: Products are stored in
[6] in collaboration with ZARA, designed and tested a these centers before being sent to stores.
support system model for replenishment and distribu- (4) Stores: Products received from the distribution
tion. Their model is data-driven and is designed to up- centers are sold to customers in stores.
date demand forecasting and allocate inventory through
time using dynamical optimization methods.
4 Modelling procedure
Fast fashion is a competitive area in the fashion industry
and new and diverse trends, highly uncertain demand, In this section we explain the details of our modelling
and high inventory are some of its important characteris- approach for the supply chain schematically shown in
tics. Therefore, in this area, decision makers should have figure 1. We consider the total cost, including fixed and

2
 Table 1. Nomenclature used in the model.

Indices Description
i : i ∈ {1, 2, . . . , I} designs
j : j ∈ {1, 2, . . . , J} manufacturers
k : k ∈ {1, 2, . . . , K} distribution centers
e : e ∈ {1, 2, . . . , E} stores
Decision variables Description
transijk number of item i to be sent from manufacturer j to distribution center k
transike number of item i to be sent from distribution center k to store e
xij number of item i to be produced by manufacturer j
Model parameters Description
cij cost of producing item i by manufacturer j
ctfjk fixed cost of sending item i from manufacturer j to distribution center k
ctvjk variable cost of sending item i from manufacturer j to distribution center k
ctfke fixed cost of sending item i from distribution center k to store e
ctvke variable cost of sending item i from distribution center k to store e
capi capacity of manufacturer j
capk capacity of manufacturer k
cape capacity of store e
demie demand for item i in store e
distjk distance from manufacturer j to distribution center k
distke distance from distribution center k to store e

varying logistics costs for transferring the products be- their capacity:
tween the entities of the supply chain and also produc-
tion costs, as the objective function that needs to be X
minimised. Table 1 describes the nomenclature used in xij ≤ capj ∀j (2)
the model. i

We develop a linear programming model as follows:

• The amount of transferred goods from the producer to
• The cost Z is defined as: the distribution center must be less than the available
XX quantity of that product in the producer’s warehouse:
Min z = xij cij +
i j
X
!!! xij − transijk ≥ 0 ∀i∀j (3)
X X X k
ctfj + ctvj distjk transijk +
j k i
!!!
• The amount of transferred goods to the distribution
X X X
ctfk + ctvk distke transike
k e i centers must be less than their capacity:
(1)
XX
transijk ≤ capk ∀k (4)
In this function the first sum accounts for the cost of i j
producing the designed cloths by the producers and
the 2nd and the 3rd sums are accounting for the cost
of transportation of the goods.
• The amount of goods transferred to the stores must
• The quantity of the produced goods must be less than be less than the amount of goods transferred from the

3
 manufacturer to the distribution center.
X X
transijk − transike ≥ 0 ∀i∀k (5)
j e

• The amount of products transferred to the stores must

be less than their capacity:
XX
transike ≤ cape ∀e (6)
i k

• The amount of products transferred to the stores must

must satisfy the demand: Fig. 2. The optimal quantity of item i to be produced by
X manufacturer j.
transike ≥ demie ∀i∀e (7)
k

• Necessary boundaries:

transike , transijk , xij ≥ 0 (8)

5 Model validation

To validate the suggested model, historical data associ-

ated with costs and demands of the Queen Manto brand
are being used. The direct supply chain of our problem
consists of ten different designed products, three manu-
facturer, four distribution centers and three stores.

We used GAMS to solve the problem. The solver suc-

cessfully converged under two seconds and found the op-
timal value of the objective function to be Z = 935209. Fig. 3. The optimal quantity of item i produced by manufac-
The decision variables values for the solution are given turer j (i-j is coded on the y-axis) to be sent to distributor k.
in figures 2-4.
such as logistics between different centers, and make de-
cisions that result in more efficient use of resources. For
6 Discussion instance, if the managers come to realize that the amount
of goods transferred between two centers are frequently
Supply chain managers in the fashion industry have to high in different periods, they might want to use more
make reasonable and fast decisions due to high demand suitable and efficient transportation vehicles for those
and its complex dynamics. Our novel model can help the routes to further minimize their costs.
managers to make better decisions about their produc-
tion, replenishment, and logistics policies in a reasonable In summary, our model consisting of multiple levels from
time. As opposed to previous works such as [4] that was the designer to the customers can give valuable insights
limited to only manufacturers and stores, our suggested about the different parts of the supply chain and enable
model considers the decisions of designers, manufactur- the decision makers to optimize their operations encom-
ers, distribution centers and stores in a more compre- passing production quantities, logistics, inventory lev-
hensive way. els, and sales. However, our model has some limitations
which we believe can be addressed by adding some ex-
We believe managers in the fashion industry equipped tensions to it and making it more comprehensive and
with our model and with an analytical point of view, can more flexible. Accordingly, future works could possibly
have better control over their supply chain components, include these extensions: considering dynamic pricing,

4
 responsiveness model. In 2018 4th International Conference
on Optimization and Applications (ICOA), pages 1–5. IEEE,
2018.
[5] Marshall Fisher, Kumar Rajaram, and Ananth Raman.
Optimizing inventory replenishment of retail fashion
products. Manufacturing & service operations management,
3(3):230–241, 2001.
[6] Jérémie Gallien, Adam J Mersereau, Andres Garro,
Alberte Dapena Mora, and Martı́n Nóvoa Vidal. Initial
shipment decisions for new products at zara. Operations
Research, 63(2):269–286, 2015.
[7] Yeu-Shiang Huang, Chia-Shuo Hsu, and Jyh-Wen Ho.
Dynamic pricing for fashion goods with partial backlogging.
International Journal of Production Research, 52(14):4299–
4314, 2014.
[8] Raffaele Iannone, Giada Martino, Salvatore Miranda,
and Stefano Riemma. Modeling fashion retail supply
chain through causal loop diagram. IFAC-PapersOnLine,
48(3):1290–1295, 2015.
[9] Bob Lowson, Russell King, and Alan Hunter. Quick
Response: Managing the supply chain to meet consumer
demand. Wiley, 1999.
[10] Marzieh Mehrjoo and Zbigniew J Pasek. Risk assessment for
Fig. 4. The optimal quantity of item i that distributor k (i-k the supply chain of fast fashion apparel industry: a system
is coded on the y-axis) has to send to store e. dynamics framework. International Journal of Production
Research, 54(1):28–48, 2016.
including promotions effects, and stochastic demand. [11] N Salehi Sadghiani, SA Torabi, and N Sahebjamnia.
Retail supply chain network design under operational and
disruption risks. Transportation Research Part E: Logistics
7 Conclusion and Transportation Review, 75:95–114, 2015.

We developed a model to minimize production and lo-

gistics costs for fast response to customer demands and
efficient replenishment in fashion supply chain with four
major levels including designer,manufacturers, distribu-
tion centers, and stores. We validated our model using
data from a fashion brand and solved for the optimal
solution using GAMS. We reported the results of our
model including the optimal quantity for production of
the designers’ designed items by each manufacturer, the
optimal quantity of goods that need to be transferred
from each manufacturer to each distribution center, and
the optimal quantity of goods that need to be sent from
the distribution centers to the stores in order to satisfy
demand and minimise the costs.

References

[1] Reza Alikhani, S Ali Torabi, and Nezih Altay. Strategic

supplier selection under sustainability and risk criteria.
International Journal of Production Economics, 208:69–82,
2019.
[2] G. Blanchard and R. Loubere. Fashion industry, 2021.
Available from: https://www.britannica.com/art/fashion-
industry.
[3] Kasarin Charnsirisakskul, Paul M Griffin, and Pınar
Keskinocak. Pricing and scheduling decisions with leadtime
flexibility. European Journal of Operational Research,
171(1):153–169, 2006.
[4] Naila Fares, Maria Lebbar, and Najiba Sbihi. Quick
response in fast fashion retail: An optimization supply chain