Content uploaded by Sinan Salman
Author content
All content in this area was uploaded by Sinan Salman on Mar 30, 2020
Content may be subject to copyright.
The Effect of Demand Variability on
Supply Chain Performance
Suzan Alaswad
College of Business
Zayed University
Abu Dhabi, UAE
suzan.alaswad@zu.ac.ae
Sinan Salman
College of Tech. Innovation
Zayed University
Abu Dhabi, UAE
sinan.salman@zu.ac.ae
Arwa AlHashmi
College of Business
Zayed University
Abu Dhabi, UAE
201509253@zu.ac.ae
Hawra AlMarzooqi
College of Business
Zayed University
Abu Dhabi, UAE
201533299@zu.ac.ae
Meera AlHammadi
College of Business
Zayed University
Abu Dhabi, UAE
201508481@zu.ac.ae
Abstract — This paper studies the impact of demand
variability on supply chain performance which is measured
in terms of operational costs, customer satisfaction, and
environmental footprint. The operations within the supply
chain have been simulated using a supply chain game
simulator. The simulation results show that demand
variability has a negative impact on all three supply chain
performance metrics but mostly on the operational cost.
The results also show that there is a significant increase in
performance measurements variability with the increase of
demand variability. The results provide managers with
insights for planning and improving supply chain
performance.
Keywords — supply chain management, simulation,
demand variability, operations management
I. INTRODUCTION
In a competitive business environment, many companies
achieve their competitive advantage by leveraging efficient
supply chain management. To manage a supply chain
successfully, it is essential that business partners in a supply
chain cooperate to manage the supply chain in the face of
constant variability and risks to meet the demand of the end
consumer. However, supply chains often operate in a volatile
and erratic environment and suffer from limited coordination
between supply chain members. Therefore, only through
information sharing companies can overcome such challenges
and improve the performance of their supply chains. Information
in a supply chain flows in both directions; demand information
flows from downstream to upstream organizations in the supply
chain, and production and product availability information flows
in the reverse direction. Other factors that can affect the supply
chain perofmance and exacerbate the effects of lack of
information sharing include market demand variability and the
accuracy of market demand forecasting [1].
This paper explores the effect of market demand variability
on the supply chain performance. A supply chain’s performance
is measured in terms of operational costs, customer satisfaction,
and environmental footprint. A supply chain game simulator
(the X-Supply Game [2]) is used to simulate operations within
the supply chain. This provides a simulation approach to
studying supply chain system behavior and evaluating the effect
of demand variability on the three supply chain performance
measures.
The remainder of this paper is organized as follows. Section
II reviews related literature. Section III represents the research
methodology. Section IV discusses the results. Finally, section
V concludes and summarizes the research findings.
II. LITERATURE REVIEW
In recent years, the supply chain area has attracted much
attention from both practitioners and researchers. The supply
chain management component has received much attention by
many reserachers [3]. Several studies confirmed that effective
supply chain management is a pre-requisite to quality and
profitability [4]. Managing the coordination and integration of
material and information flow within and across companies is
essential to achieving effective supply chain management [5].
Information sharing is basic to effective coordination in a supply
chain. Many studies have found that information sharing has
great impacts on supply chain performance [6]. Some studies
investigated the impact of demand information sharing on
supply chain performance and confirmed the significent benefit
that can be achieved through demand information sharing for all
supply chain partners [7-9]. Uncertain market demand and
variablity also affect the supply chain performance. A few
studies explore the affect of demand variability on supply chain
performance due to the complexity of using analytical
approaches to evaluate real supply chains [10]. Thus, simulation
approaches have been used instead to study real supply chain
behaviors [11]. Zhou [1] used a simulation model to examine the
impact information sharing on supply chain performance and
how demand variability and forecasting accuracy highlights the
value of supply chain information.
III. RESEARCH METHODOLOGY
In this section, the following aspects of the research
methodology are discussed: research question, simulated supply
chain, demand patterns and variability used as input, and
simulation experimental design.
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)
978-1-5386-7684-4/19/$31.00 ©2019 IEEE
A. Research Question
This study focuses on effects of increased demand variability
on supply chain network performance. To examine this effect,
three hypotheses were tested in this study: 1) increased demand
variability has a significant influence on the supply chain cost,
2) increased demand variability has a significant influence on
the supply chain environmental footprint, and 3) increased
demand variability has a significant influence on the customer
demand fulfilment level.
To test these hypotheses and answer the research question, a
supply chain simulation network is setup and a set of
experiments with increasing levels of consumer demand
variability is simulated. The following subsections goes into the
details of this approach.
B. The Simulated Supply Chain Network
This paper uses the X-Supply Game (XSG) engine to
facilitate simulation of the operations of a four-stage supply
chain. The XSG simulates on a small-scale system dynamics
found in real supply chains. The game simultaneously deal with
multiple competing objectives involving operating costs (i.e.
inventory, backorder, and transport costs), fulfillment rate, and
environmental impact. The game showcases the difficulty and
importance of supply chain integration to achieve higher
performance in the supply chain. Although the XSG was
conceived as interactive educational tool, its automated
capability for station ordering and shipping decision making
renders it suitable for fully automated supply chain simulation.
The simulated supply chain includes eight stations
consisting of two manufacturers, two wholesalers, two
distributors, and two retailers (Figure 1). Each supply chain
station targets the satisfaction of consumers 1 and 2’s demands
while reducing costs and minimizing environmental negative
impact. The two manufacturers produce the same product.
Figure 1. Supply chain network
Table I summarizes the game operational parameters setup.
In the simulation game, truck utilization is used as a proxy
metric for the environmental footprint of the supply chain.
The simulation methodology utilized here is turn-based. The
simulation horizon of one year is divided into 52 weekly
intervals, or turns, with each interval including station decisions
for ordering quantities and shipping quantities. At the beginning
of each week, station inventories are adjusted with deliveries
arriving from suppliers, and backorders, if any, are adjusted with
orders from stations’ immediate customers. Next, an automated
ordering and shipping decisions logic is executed for each
station to determine the week’s orders to suppliers and
shipments to customers. The ordering logic calculates total
orders needed by subtracting inventory and outstanding orders,
if any, from current backorders. If total orders are positive, it
allocates them to suppliers proportionally according to their
outstanding orders inverse ratio; those with lower outstanding
orders receive a larger share of the total order. The shipping
logic will attempt to satisfy all current orders and backorders if
sufficient inventory exists. If current inventory is not enough,
available inventory will be shipped to customers proportionally
according to their backorder count; customers with higher
backorders will receive a larger share of the shipped inventory.
At the end of each week, operational cost variables (holding
backorder, and transportation costs) and operational delay
queues (shipping and ordering) are adjusted in preparation for
the following week.
Table I. GAME STATION PARAMETERS
Parameter
Value
Inventory holding cost
$1 per item-week
Backorder cost
$2 per item-week
Transport cost
$220 per truck
Transport size
200 items per truck
Shipping delay
2 weeks
Ordering delay
2 weeks
Queue initial value
100 items
Beginning inventory
100 items
Order minimum
0 per week
Order maximum
1000 per week
C. Demand Patterns
To study the impact of demand variability on supply chain
performance, different demand patterns are generated at
consumers’ points and the simulation game is run for each
pattern to measure the system performance. The demand
patterns are generated using Equation 1, which is a modified
version from the one that is extensively used in supply chain
simulation studies [8, 12-13].
!"#$%&'$(&()*
+
,-
.
'/0
1
2
3
$4&5./6761
(1)
where D is the demand placed on a retailer by a consumer in
week t; B is the baseline demand; m describes the decreasing or
increasing trend of the demand; S represents the mangnitude of
seasonal variation of demand; p is the phase shift of the demand
in time; c is the cycle of the seasonal variation of demand; N is
the magnitude of random variability which is assumed to be
uniformly distributed noise with a magnitudee of 10% of the
base demand for low demand variability, 30% for medium
demand variability, and 90% for high demand variability. When
negative demand is encountered, it is replaced with zero
demand.
The characteristics of the demand patterns generated at the
two consumer stations and placed on each of the retailer stations
are summarized in Table II. Figure 2 illustrates the average and
one-standard deviation of the variability range for the resulting
demand for both consumers.
A. Simulation Experimental Design
The simulations experiments included a total of 300
experiments representing different levels of demand variability
(low, medium, and high), with each level having 100 simulation
runs with randomly generated demand as previously described
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)
in section III.C. All experiments shared the same supply chain
setup described in section III.B.
Table II. CHARACTERISTICS OF DEMAND PATTERNS
Characteristic
Consumer 1
Consumer 2
B
100
100
m
1
1
S
50
50
p
65
77
c
13
25
Figure 2. Consumers demand patterns
IV. RESULTS AND DISCUSSION
The simulation results provides an insight into demand
variability impact on supply chain performance. Figures 3, 4,
and 5 illustrate the impact of increased demand variability on
cost, fulfilment rate, and environmental impact (proxied through
the transport resource utilization, or GreenScore), respectively.
The figures show the average performance measure for all
simulation experiments at given demand variability level (the
blue line). It also shows the 95% confidance interval
surrounding the average via the light blue band surrounding the
line.
Figure 3. Impact of demand variability on supply chain total cost
Figure 4. Impact of demand variability on supply chain
average fulfilment rate
Figure 5. Impact of demand variability on supply chain average
environmental impact
Analyzing the graphs one can conclude that although
demand variability has a negative impact on all three supply
chain performance metrics, the severity of the impact is not the
same. While increased variabililty from 10% to 90% leads to an
increase in total supply chain costs from $1.113K to $1.170K
(or 5.14%) the effect is not as pronounced as one would expect.
The cost increase is even less pronounced at 0.4% when medium
demand variability level is simulated.
Fulfilment rate also shows a negative trend with increased
demand variability, however, it is barely noticeable with 0.3%
and 1.6% loss in average fulfilment rate when medium and high
demand variability levels are respectively simulated.
Similarly, the environmental impact, as measured via the
GreenScore transport resource utiliazation, shows a negative
trend with increased demand variability. The loss in GreenScore
is also marginal compared to the change in demand variability.
Medium variability leads to a loss of 0.4% while high demand
variability leads to a loss of 0.6% in GreenScore.
These results can be attributed to the way ordering decisions
are automated in the XSG game; where previous weeks’
customer backorders and outstanding orders to suppliers are
taken into consideration producing a dampening effect, and in
effect averaging out some of the increased demand variability
effects on the supply chain. Of course, real supply chains
managed by people behave in more unpredictable manner
increasing the impact of variability on them, as seen in the case
of the bullwhip effect [14].
It is noteworthy that the impact on the mean of all supply
chain performance measurements (cost, fulfilment, and
GreenScore) when comparing low and high demand variability
proved to be statistically significant when subjected to the t-test
with a threshold of p = 0.05. However, the impact of the change
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)
from low to medium and medium to high demand variability on
the results was not always statistically significant. This can be
deduced from figures 3, 4, and 5 by comparing the 95%
confidence intervals between demand variability levels for each
performance measurement. The presentation of the numerical
statistical results is omitted here for brevity sake.
Of more interest here is the significant increase in
performance measurements variability with the increase of
demand variability. This can be observed through the increased
95% confidence intervals shown in figures 3, 4, and 5. In fact,
the increased uncertainty in the supply chain performance as
demand variability increase can be of more concern to supply
chain mangers as their ability to predict the outcomes of
operational strategies and decisions is negatively impacted.
Table III shows the extent of the impact to the supply chain
performance measurements, measured through the standard
deviation. For example, the standard deviation for cost in a high
demand variability scenario is more than 7.5 times its counter
part in a low demand variability scenario. On one hand, cost and
fulfilment rate showed a significant negative impact in this area.
On the other hand, Green score showed little impact in this
regard.
Table III. IMPACT OF DEMAND VARIABILITY (via Standard Deviation)
ON SUPPLY CHAIN PERFORMANCE MEASUERMENTS VARIABILITY
Cost
Fulfilment
GreenScore
Low
14,838
0.3
0.9
Med
42,450
1.0
0.8
High
112,509
2.7
1.2
V. CONCLUSION
This paper investigates the effect of market demand
variability on supply chain performance. The supply chain’s
performance is measured in terms of operational costs, customer
satisfaction, and environmental footprint. A supply chain game
simulator is used to simulate operations within the supply chain.
The results from the simulation show the following key findings.
Demand variability has a negative impact on all three supply
chain performance metrics, but the severity of the impact is not
the same: cost is most sensitive performance measure while
GreenScore is the least sensitive. Supply chain performance
impact in average values is limited and significantly less than
the magnitude of the demand variability change; a change of
90% to see some impact on results. The results also show that
there is a significant increase in performance measurements
variability with the increase of demand variability. Variability
change is significant and poses a higher risk as supply chain
professionals lose some of their predicting capabilities of the
scenario or operational strategy outcome.
Future work could include investigation of the impact of
additional supply chain design parameters (e.g. delays,
topology/layers, costs) on the study outcomes. The
interdependency between supply chain performance measures
with regard to input variability level is also an interesting
research question to investigate. In addition, a discrete event
simulation model can be built on top of the XSG to model the
randomness of the supply chain. Applying this investigative
approach to an example supply chain based on a real example
would also be of additional value.
REFERENCES
[1] Q. Zhou, G. M. Lee, T. Lee. “Demand Variability,
Forecasting Accuracy, and Supply Information Sharing,”
upublished.
[2] S. Salman and S. Alaswad, “The X-Supply
Game,”. Proceedings of Institute of Industrial & Systems
Engineers Annual Conference & Expo, 2018.
[3] K. Burgess, P. J. Singh, R. Koroglu, "Supply chain
management: a structured literature review and
implications for future research," International Journal of
Operations & Production Management, vol. 26, issue 7,
pp.703-729, 2006.
[4] S. Li, B. Ragu-Nathan, T. S. Ragu-Nathan, S.S. Rao, “The
impact of supply chain management practices on
competitive advantage and organizational performance,”
Omega, Vol 34, issue 2, pp. 107-124, April 2006.
[5] I. Baihaqi, N. Beaumontand, A. SohalInformation,
“Sharing in Supply Chains: A Survey of Australian
Manufacturing,” International Review of Business
Research Papers, vol 4, issue 2, pp.1-12, March 2008.
[6] W. Zhu, S.Gavirneni, R. Kapuscinski, “Periodic flexibility,
information sharing, and supply chain performance,” IIE
Transactions, vol 42, pp. 173-187, 2010.
[7] Z. Yu, , H. Yan, T. C. E. Cheng, “Benefits of information
sharing with supply chain partnerships,” Industrial
Management & Data Systems, vol. 101, issue 3, pp. 114-
119, 2001.
[8] X. D. Zhao, J. Xie, “Forecasting errors and the value of
information sharing in a supply chain,” International
Journal of Production Research, vol. 40, issue 2, pp. 311-
335, 2002.
[9] H. Choi, “Information sharing in supply chain
management: a literature review on analytical research,”
California Journal of Operations Management, vol. 8,
issue. 1, pp. 110-116, 2018.
[10] A. M. Law and W. D. Kelton,. Simulation modeling and
analysis (3rd ed.). Boston: McGRAW-HILL, 2000.
[11] Z. Huang and A. Gangopadhyay, “A simulation study of
supply china management to measure the impact of
information sharing,” Information Resources Management
Journal, vol. 17, issue 3, pp. 20-31, 2004.
[12] L. Abuhilal, G. Rabadi, A. Sousa-Poza, “Supply chain
inventory control: a comparison among JIT, MRP, and
MRP with information sharing using simulation’
Engineering Management Journal, vol. 18, issue 2, pp. 51-
57, 2006.
[13] R. Schmidt, “Information sharing versus order aggregation
strategies in supply chains. Journal of Manufacturing
Technology Management,” vol. 20, issue 6, pp. 804-816,
2009.
[14] H. Lee, V. Padmanabhan, S. Whang, “The Bullwhip Effect
in Supply Chains.” Sloan Management Review, vol. 38,
issue 3, pp. 93–102, 1997.
2019 8th International Conference on Modeling Simulation and Applied Optimization (ICMSAO)
