Journal Pre-proof

Designing a stochastic multi-objective simulation-based optimization

model for sales and operations planning in built-to-order environment
with uncertain distant outsourcing

Rouzbeh Aiassi , Seyed Mojtaba Sajadi ,

Seyyed Mohammad Hadji Molana , Ali Zamani Babgohari

PII: S1569-190X(20)30041-1
DOI: [Link]
Reference: SIMPAT 102103

To appear in: Simulation Modelling Practice and Theory

Received date: 17 October 2019

Revised date: 29 March 2020
Accepted date: 11 April 2020

Please cite this article as: Rouzbeh Aiassi , Seyed Mojtaba Sajadi , Seyyed Mohammad Hadji Molana ,
Ali Zamani Babgohari , Designing a stochastic multi-objective simulation-based optimization model
for sales and operations planning in built-to-order environment with uncertain distant outsourcing,
Simulation Modelling Practice and Theory (2020), doi: [Link]

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2020 Published by Elsevier B.V.

 Highlights

 Contrast among two inventory management strategies: linear and fixed

policies.
 Minimizing logistics cost and improve customers experience level
simultaneously.
 Finding optimal values of decision variables (flexibility and safety stock
level).
 Replacing fixed strategy by linear one, enhances the overall performance
up to 17%.
 Backup supply cost variation has great impact on linear policy
performance.
 Designing a stochastic multi-objective simulation-based optimization model for sales
and operations planning in built-to-order environment with uncertain distant
outsourcing

Rouzbeh Aiassi1, Seyed Mojtaba Sajadi*1, Seyyed Mohammad Hadji Molana2, Ali Zamani Babgohari3

1-New Business Department, Faculty of Entrepreneurship, University of Tehran, Tehran, Iran

E-mail: [Link]@[Link]

1-New Business Department, Faculty of Entrepreneurship, University of Tehran,

Farshi Moghadam (16 St.), North Kargar Ave., Tehran, Iran, +989123850554
E-mail: msajadi@[Link]

2- Department of Industrial Engineering, Faculty of Engineering, Islamic Azad University, Science and
Research Branch, Tehran, Iran
E-mail: molana@[Link]

3- Department of Operations Management, Faculty of Management, University of Tehran, Tehran, Iran

E-mail: zamania@[Link]

Abstract

To this day, sales & operations planning (S&OP) remains one of the most critical challenges
of supply chain management in terms of improving customers’ experience level and
managing supply chain costs. Nowadays the synchronization of sales essentials and industrial
restrictions in an uncertain environment is critical for all global companies facing complexity
in supply chain networks and operations systems. Sales and supply chain structures in build-
to-order industries generally have cross-functional and conflicting objects that are not easily
comparable. The current research is an extension of an innovative S&OP model within a
more dynamically complex system in terms of demands uncertainty and unreliability of
backup supplies. The promoted model is utilized for more efficient management and
integration of industrial restrictions and sales essentials in global supply chain networks in
terms of demand and supply coordination and responsiveness toward customers’ needs.
Accordingly, this research has applied a simulation-based optimization approach to balance
the trade-off between logistics cost and customers’ experience level as the main objectives.
Two classes of strategies are offered for handling parts’ replenishment and flexibility to the
sales structure. Initially, the proposed model evaluates two main inventory strategies’
performances. Thereafter, the optimized values of flexibility degree and safety stock fraction
for corresponding inventory strategies are calculated by using a multi-objective simulation-
optimization method. The investigation begins with an overall assessment of the system’s
performance. Furthermore, it details the alterations among the strategies’ key performance
indicators (KPIs) as system parameters’ function.
Keywords: Sales and operations planning, Demands uncertainty, Distant outsourcing, Build-
to-order industries, Supply chain management, Simulation-based optimization
1. Introduction

Nowadays, the emerging growth of inter-organizational collaboration and globalization in

complex supply chains has led to significant need of enhancement via products procurements.
This phenomenon and the subsequent long procurement time has caused supply chain
systems to be more vulnerable, encountering various stochastic disruptions such as market
fluctuations and demands uncertainty [1]. In addition, globalization and industries’
deployment have steered organizations to use outsourcing in order to provide their supplies
more efficiently. This is due to many terms e.g. environmental restrictions, labor costs,
entering the new markets, etc. However, in supply chain networks settings that involve
distant outsourcing and uncertain environments in competitive markets, the existing strategies
are not adequate for managing the production and supply systems of build-to-order (BTO)
industries or generally companies within customized-based products [2, 3]. BTO industries
encounters high complexity in supply chain systems, which has led to a challenging problem
of adjusting industrial restrictions and sales essentials. Due to long nature of procurement
time in such networks, companies are obliged to order required parts and resources several
periods in advance and based on mostly unreliable estimations [4, 5]. This is one of the main
concerns of supply and purchasing departments in a supply chain network [6]; and occurs
frequently in automotive companies as a competitive business environment [7, 8]. following
the escalation of supply chains’ complexity and markets competitiveness through increasing
products variety, automotive companies’ supply chain management tactics are refined toward
BTO strategies [9]. However, the adaption of manufacturing capacities toward customer’s
requests level in unstable markets remains as a critical challenge of automotive manufacturers
and companies that functions vis-a-vis the BTO systems. Companies emphasize on more
flexibility and shorter procurement time to keep customers satisfied. While, manufacturing
systems need a steadier production plan and wider end-to-end visibility regarding future
demands. Therefore, establishing a proper trade-off between non-aligned goals is vital [10].
Sales and operation planning (S&OP) is outlined as integrated tactical planning process
that synthesizes manufacturing plans and demand management to implement demand-supply
balancing as the main objectives [11, 12]. In this subject, S&OP implementation as a function
of formal tactical planning achieves a cross-functional alignment through linking strategic
management and operations planning over the planning horizon [13]. Furthermore, the
execution of S&OP improves efficiency in the matter of supplier and manufacturer
interactions [14]. S&OP also has positive effects on overall operational performance such as
demands estimation accuracy [15], responsiveness toward customers’ needs [16], and
inventory level [17]. Based on Grimson and Pyke [18], S&OP is identified as one of essential
subjects for improving customers’ experience level and controlling operations cost in BTO
and other related industries. In competitive markets, S&OP helps companies to cope with
supply and demands stochastic changes. Firms tend to apply S&OP processes to arrange the
demands prediction and customer demands to decide on manufacturing capacity [19].
The present paper is an extension of an innovative model introduced by Lim et al. [20] to
solve S&OP challenges based on a simulation-based optimization approach. Accordingly,
this study designs a validated numerical experiment within industrial data patterns from the
prior study as inputs. The current research is an accurate representative of S&OP system in
the presence of demand uncertainty, lost sales thresholds and uncertain backup supplies’
deliveries. The present article proposes a multi-objective model to find a trade-off between
the main goals, i.e. minimizing logistics cost and enhancing customers’ experience level. For
this ambition, using a simulation-based optimization approach appears to be a logical
solution, since managing the complex supply chains and operations systems requires
sophisticated decision-support tools that can demonstrate the system’s dynamic environment
[21, 22]. it is not often recommended to use pure analytical methods for solving such complex
and dynamic issues, since is usually time-consuming and vastly costly [23, 24]. Two basic
strategies are offered to manage the products procurements and to set the manufacturing
capacity for the sales structure [20]. Based on a simulation-based optimization method, the
proposed model attempts to control system indexes, specifically flexibility ratio and
inventory level in order to balance the industrial constraints and sales essentials. Accordingly,
the drive for evaluating the system performance and deciding on the optimal strategy is
achieved by assessing the system’s operational and cost-related parameters, i.e. logistics cost
(holding and backup supply costs), and simultaneously evaluating customers’ experience
level (delayed orders and lost sales percentages).
The main contributions of this paper to the body of knowledge are threefold; firstly, is about
how S&OP performs in a BTO environment with higher complexity in matter of uncertain
distant outsourcing. Secondly, this research proposes a more realistic multi-objective
simulation-based optimization model with higher complexity, which investigates the
consequences of stochastic disruptions in terms of demands uncertainty, lost sales threshold,
unreliable distant outsourcing, and corresponding costs. Finally, the current article searches
for the optimal solutions possible to optimize the system’s key performance indicators (KPIs)
and objective functions as well as providing managerial implications in term of using S&OP
with diverse inventory strategies in BTO supply chains. This paper innovation in contrast of
prior study of Lim et al. [20] are as followings. First, in term of distant outsourcing, there
might be delay or withdrawal possibility in airfreights from the supplier to manufacturer,
which is unpredictable. Moreover, unlike other related articles, this research has considered
the fixed and variable ordering costs jointly in backup supply mode. Lastly, the evaluation of
customers’ experience level and related KPIs are different from Lim et al. [20] study.
The rest of this paper is organized as follows. Section 2 briefly reviews the literature and
related works to the article’s main scopes. Section 3 addresses the S&OP problem in the
above-mentioned industry and proposes a method to control and handle the demand and
supply integration (DSI) and responsiveness toward customers’ needs in such systems. Later
on, the hypothesis of the S&OP model, problem’s parameters, and the mathematical model
are introduced. Section 4 discusses the simulation-based optimization approach used for
modeling and solving this particular S&OP problem, via designing a validated numerical
experiment with industrial data patterns from the prior study mentioned earlier. Section 5
analyzes the numerical results of the designed experiment in terms of results’ deviations
besides the alterations among strategies’ KPIs as a function of system parameters.
Accordingly, the discussion, evaluation, and managerial implications achieved from this
investigation are presented. Finally, in Section 6, the main contributions and future works of
this research are offered.

2. Literature Review

This section reviews the literature of related research, which is categorized into six topics:
Supply chain globalization, Production planning in uncertain environment, Sales and
operations planning, Flexibility in supply chain, Build-to-order industries, and Simulation-
based optimization.
2.1. Supply Chain Globalization

The first aspect related to this research is the matter of globalization and competitive
international markets. In this regard, recently more and more companies and businesses are
seeking to integrate their operations systems with external collaborations to build a better
competitive advantages regarding managing their supply chain network [25]. This
phenomenon has led to significant growth of parts’ procurement time and also accelerates the
effect of demands’ uncertainty in supply chain planning. Consequently, this makes supply
chains more vulnerable against such dynamic changes and uncertain environments. Over the
past years, influences such as: labor cost, gas emissions and environmental issues, and the
opportunity of involving in new emerging markets in developing countries have persuade
global companies to reposition some areas of their production capacities to more economical
and cost efficiently regions around the world to remain competitive in global markets [26-29].
The study of Humphrey [30] is about the expansion and development of automotive
companies at emerging countries. In this study, complex strategies for supplying resources
have been explored in new ranges. The author claims that international companies have tend
to vertically integrate and outsource their requirements to vary their products options. This
due to achieve more market shares and decrease manufacturing costs. The study mentioned
above showed that these types of strategies and decisions are effective on improving the
overall organization performance in steady and constant conditions. Whereas, in uncertain
and dynamic supply chains, management of such systems is more complex. The study of
Pashaei and Olhager [31] suggests in competitive markets, production companies need to
design, manufacture, and distribute products to consumers; while considering the necessities
of the global operations network, such as responsiveness of customers’ needs along with
maintaining cost efficiency. In a research, Prater et al. [32] claims that considering the time-
consuming procurement, the international organizations’ supply chains, which carry out their
transportations through ships have shown an appropriate performance in responsiveness
toward customer requests and other stochastic changes at supply chains. Currently, many
organizations and companies’ sub-functions are positioned in different locations of the region
and collaborates through up/down streams links [33, 34]. In this context, it is appropriate to
mention giant automotive supply chains such as BMW (Bayerische Motoren Werke) in
Germany and GM (General Motors) in US. These two industrial giants with the purpose of
decreasing manufacturing costs, outsource and manufacture parts of the final products
components in third world countries; As it would be more cost efficient (surely under the
company jurisdictions and standards). Bear in mind, these mentioned companies, functions
based on BTO approach. For example, GM subcontracts a considerably large part of their
products platforms in Mexico, through distant outsourcing. Likewise, BMW associates in
China to produce part of its production loads.

2.2. Production Planning in Uncertain Environment

In the field of planning environment planning prospect, frequency, and planning objects
characterizes the general planning setup factors [35]. Decisions and resolutions that affect the
system and planning environment in long-term consider as strategic planning, while plans
and decisions that are considered with near or short term impacts are established as tactical or
operation planning [12, 35]. Literature categorizes complexity into two scopes of dynamic
and detail complexity [36]. Dynamic complexity refers to the uncertainties and constraints
where, effects are not transparent and the influences surfaces gradually through planning
horizon. While detail complexity relates to numbers of variables and sub-processes in the
context which calls for further planning processes [36-38]. Regarding to the importance of
uncertainty issues in production planning, numerous researches and studies are conducted in
conjunction of production planning and the uncertainty embedded in the planning
environments. In most cases of modeling production processes with dynamic and uncertain
parameters through analytical methods, typically only one or two problem parameters are
assumed stochastic; while other assumptions on production planning have been considered
deterministic with a simple structure [39]. This is due to the lack of capability of analytic
methods to solve problems with higher complexity, which could be vastly time consuming,
and in some cases not easily applicable. One of the studies on this area has been published by
Mula et al. [39]. In this extensive literature review, authors refer to the limitation of most
analytical methods in modeling the production processes that are only able to consider one
type of uncertainty. One of the studies in the field of production planning under highly
uncertain environment is [40]. This research considers the environmental hedging point
policy to control the inventory and production plus aligning main objectives of costs, and
customers’ satisfaction. In this study, the authors show how manufacturers’ complications are
no longer restricted by economical aspects but also are effected by gas emission and customer
satisfaction issues.
The recent literature regarding the use of advanced data analysis in modern technology-
based operations networks and manufacturing systems suggests, In the event of analysis and
study of models with greater complexity, using methods like artificial intelligence, big data
and/or data-driven simulations is more applicable for evaluation and achieves knowledge that
is more influential. In this regards numbers of studies can be mentioned in which big data
analysis and other methods are used as a supporting decision-making tools to extract
beneficial information from the systems operations. The study of Zhong et al. [41] offers a
holistic big data approach to achieve frequent trajectory knowledge and data patterns from
analyzing enormous RFID-enabled manufacturing data. In this research, authors collect
enormous data from automotive manufacturer shop-floors and use it for supporting decision-
makings tasks such as production scheduling and logistics planning. In another study, Vieira
et al. [42] uses simulation as a big data validator for supply chain management and production
planning. In this paper, the authors use big data methods and technologies to provide real-
time data sets to supply chains simulation models.

2.3. Sales and operations planning

The advanced planning framework in general is categorized into three scopes comprising
strategic (long-term), tactical (medium-term) and operational (short-term) planning [43-45].
Tactical planning functions in terms of linking the long terms strategies to operational and
short term planning and scheduling. The literature argues that the core objectives of tactical
planning in complex supply chains is harmonizing the customers demand and supplies [11,
12, 14]. One of the most important challenges in managing supply chains is to subsequently
decrease the logistics cost and improve customers’ experience by considering customers’
needs and requests. The main intention of tactical planning processes are to implement a
balanced trade-off between customers’ demands (sale function) and supply capacity
(manufacturing system) in order to avoid products’ under/over capacity within medium term
planning horizon of supply chain network [12]. Inventory management has a basic and
critical role in a logistics and supply chain management. Nowadays, due to the fluctuations of
market and demand uncertainty both suppliers and manufacturers are encountered with
various problems and challenges. Suppliers are obligated to maintain the production level and
manufacturers must plan and instruct their orders in advance for future periods so the
logistics cost including holding costs, transportation costs and shortage costs reduces as
necessary as possible and simultaneously minimizes the lost sales. In addition, in each period,
manufacturer should make decisions associated with production level according to demand
uncertainty.
In recent decades, S&OP and supply chain integrations perform a crucial role in pure
competitive market structure and BTO industries. Consequently, in recent years, researchers
have put much more attention to these domains. S&OP has fundamental alterations in
contrast to operations planning in general. S&OP employ the sales objectives, allied
constraints, and industrial constraints as the main factors in planning layouts. Planning
horizon in such issues is generally around six months to three years [33]. Naturally, the
planning horizon depends on products type and context industry. primarily application of
S&OP in an industrial system is presented as an mechanism in order to reach a trade-off
between customer request level and supply plans based on production capacity [46].
Generally, S&OP put attention to interference and participation of workforce, not only in
software and model designing, but also in processes and industrial corresponding [18, 47, 48].
S&OP has been considered as an applied tool for making the supply chain system flexible
and as a supportive tool for managers to facilitate decision-making and planning processes.
On one hand, customers’ requirements and inventory level are required. On the other hand,
supply chain flexibility causes S&OP processes to become a critical for logistic systems
management in macro level with regarding to demands uncertainty. By implementation of
S&OP processes in macro level, planners are able to manage and control the cross-functional
objects properly and balance corresponding objects: decrease the logistics cost and increase
the customers experience [18, 49, 50]. S&OP is defined by Grimson and Pyke [18]: "Sales
and operations planning is a tactical production plan that connects the strategic aspects of
system management to daily manufacturing operations". Grimson and Pyke [18] study about
analyzing the S&OP processes is considered as one of most the fundamental researches in
this field. In this research, the author provided a framework for executives and directors to
evaluate the performance of S&OP processes and to integrate the activities related to the
planning processes. Thomé et al. [51] Analyzed a sample of 725 manufacturers around the
world and evaluated the impact of S&OP processes on manufacturers overall performance
level. literature suggests, supplier’s integration causes the manufacturing system
improvement and amplifies the effected of S&OP process on the overall system’s
performance [51, 52].
Study of Olhager et al. [53] identified the main factors among production strategies and
S&OP processes. The study presents two overall strategies for manufacturing based on
S&OP, which include adjustments in request level or adjustments in products supply level.
During a research review, Olhager’s study [54] has surveyed the development and progress of
operations planning and control through last fifty years. This review study has analyzed
development of production planning to more advanced levels i.e. S&OP and in general
supply chain planning, in the context of meeting the customer’s needs and linking it to
suppliers. In this paper, the author has demonstrated that it is essential to consider the
manufacturing flexibility for S&OP, according to the increasing of time procurement of
goods and intense fluctuations of demand level in markets for supply chains management. the
model in current paper has adopted a novel S&OP model that was introduced by Lim et al.
[55], to manage two cross-functional objectives i.e. sales structure and supply chain purposes.
This model can improve the trade-off between two main goals: logistics cost and customers
experience level.
2.4. Flexibility in supply chain

Flexibility in supply chain have been introduced for uncertain environments to improve
the overall performance level and agility (responsiveness to market’s demand) of an
organization. The concept of flexibility in supply chain management can be defined in many
types; from workforces, products, raw material, machines and equipment, to inventory and
routing or as a alliance of them [56]. In production processes, flexibility is crucial for
alignment the supply capacity and customers’ demands. In case of businesses with high
uncertainty in customers’ demands and high variety of products, the industries need to be
more market responsive and require more flexibility in terms of managing the supply chain
and logistics network efficiently [57]. New aspect for flexibility and sale constraints has been
presented in S&OP model by [55]. Christopher and Holweg paper [58], has claimed that
demand fluctuations and uncertainty is rising in current market. Hence, flexibility concept is
one of the key essentials to manage and control potential stochastic disorders in supply chain
management. Christopher and Holweg study [58], analyzed relations between uncertain
environmental factors, supply chain flexibility degree and its influence on industrial
environments via structural equation modeling (SEM), in a case study of German automotive
industries. This study denotes that if Synchronization of the demand uncertainty with
flexibility level in a supply chain is applied; as a result, significant progress is applicable to
improve the organization performance and to decrease the logistics cost.

2.5. Build-to-order industries

In recent decades, manufacturing companies tend to shift their approach from mass
production to the BTO approach and custom-built products [59, 60]. These industries have
changed their structural operations processes from the centralized mode to decentralized. This
renovation is through redefining the strategic principles of the supply chains and production
planning. This is due to the purpose of having more accessibility to nearby and economical
resources and emerging markets in developing countries.
Mass customization and BTO manufacturing procedures makes the possibility for
manufacturers to adjust the customers’ request rate with company’s production level more
efficiently; through being more responsive to stochastic changes and demand uncertainty in
the volatile markets [61]. Therefore supply chain networks are capable of avoiding threats of
stochastic changes [3]. In this context, industries intend to make supply chains more flexible
with the purpose of adapting production level to market fluctuations; through applying the
BTO strategies [9, 61, 62]. Automotive industries in particular face various challenges in
supply chain management. On one hand, sales representatives are seeking for taking orders as
late as possible and deliver the goods at the shortest time, to achieve maximum sales level
and customer satisfaction. On the other hand, assembly plants are obliged to declare their
orders to the allied suppliers, through unreliable requests predictions evaluated for requested
products during planning dates. Consequently, in such conditions, supply chain management
must make a trade-off between differing and interrelated goals. In these systems, sales
structure essentially seeks greater flexibility degree for production level. Beside,
manufacturers and the linked suppliers seek fewer changes and deviation in requested product
level throughout the supply chain [63, 64]. Synchronization of value chain and products
supply is considered as one of the fundamental issues in most supply chains due to the
conflicts of the main objectives, managerial issues, and coordination indexes in different units
[65]. Case in point, operations networks that are capable of manufacturing modular designed
products such as: industries via BTO approach or other methodologies like make-to-order
(MTO) and engineer-to-order (ETO). In these types of industries, the company’s products
logically have modular architecture base and parts are compatible, easily joined, and
separable [66-68]. Also these industries characteristically have many assemble plants, long
distance between the supplier and producer, long procurement time, and perform in highly
competitive and international markets [67]. Whereupon companies functions based on
customer specific demands and context-related constraints.
In 2011, Wang et al. [69] established a data-driven simulation approach that offers of rapid
examination and model-alteration capabilities. Authors were able to manage and control
efficiently the stochastic fluctuations of customer demand in markets. In the early years of
twenty first century, novel mechanisms have been introduced for Hyundai Company in order
to harmonize the supply chain and sales department in the automotive industries. Hahn et al
have a distinct attention to synchronization and integration of supply chain and sales
department [65]. Bihlmaier et al. [70] offered a deterministic and stochastic model for
automotive companies in order to optimize the tactical and strategic plans in uncertain
environments. Hereof, manufacturing methods of three major companies, including Toyota,
Nissan and Mitsubishi, have been assessed and compared which showed how respective
companies have utilized and implemented the flexible ordering systems [71]. Latest study of
Lim et al. [20] demonstrates a simulation model of a supply chain that investigates the
practicality of sales constraints and flexibility functions. Furthermore, this study develops a
decision-support tool for decision-making in S&OP. In this research, authors try to exam the
control parameters using different optimization algorithms and choose the most effective
algorithm for finding optimal decision-making variables. Regarding the studies of supply
chain management in BTO and generally modular product architecture industries, an essential
feature of BTO approach in value chains of a potential supply chain management is the
capability of reducing the procurement time of goods and services while minimizing the
inventory level for final products. One of the most vital concepts of supply chain
management in BTO is to decrease the holding cost despite of increasing the customer
experience [72]. Regarding the mass-customization of products based on customers’ desire
and availability of assembly parts, the inventory of final products significantly reduces. In
this context, supply chains stocks hold ready-to-assemble parts above all [73].
On a survey in 2012, [25] investigates the business performances of different supply chain
firms and performs a comparison based on whether the plant is operating on an MTO or
make-to-stock (MTS) basis. Demand and supply uncertainty are typically higher in BTO,
ETO, MTO Industries, in comparison to other industrial settings such as MTS settings [49].
Another comparison between traditional supply chains and BTO industries have been
presented in Table 1. The comparison has been conducted base on fundamental aspects of a
supply chain structure including marketing, manufacturing process, logistics cost and
operational functions such as customer relationship management, dealing with uncertainty,
final products inventory and suppliers involvement to name a few [74].

Table 1
BTO supply chains and traditional supply chains differences
Indicators BTO supply chains Traditional supply chains
Market Build based on customer order Sales from stock
Emphasis on demand level and Emphasis on scheduled
Manufacturing
supply chain flexibility manufacturing level
Logistic Agile, safe, customized Mass approach
Information management, Increase final products level in
Uncertainty control
strategic parts inventory stock
Final products inventory Low quantity in stock High quantity in stock
Suppliers Timely responsiveness Distant time procurement
2.6. Simulation-based optimization

Data analysis with simulation modeling is among the most used decision-support tools for
modeling supply chains and logistics networks as suggested by Jahangirian et al, [75] and
Pires et al. [76]. Even though, simulation is a perfect decision-support tool for improving the
system performance and reflects the uncertainty of system’s variables; nonetheless, this
method is not capable of calculating the optimal solutions for decision-making variables in
system. Simulation analysis simply investigates different setups so called "what if scenarios"
[77]. Hence, the use of an optimization model in order to calculate the decision-making
variables and system parameters is needed. Note that, the optimization models could not
solve problem’s highly dynamic parameters solely; due to high complexity degree and
uncertainty in systems’ variables, besides it may be vastly costly and time consuming [23].
Optimization and simulation techniques can be merged together by various methods [78].
Simulation models can be integrated to optimization models, known as simulation-based
optimization method. This technique is a structural method for the determination of optimal
parameters in system, whereas the objective function is measured based on a simulation
model [79]. This methodology has been utilized extensively for answering the complex and
uncertain problems in real environment. For example, it can be refer to the study in 2005
which is effectively compatible for solving uncertain and dynamic problems [80]. In order to
find the applications of different simulation-optimization methods in operations systems,
readers can refer to the study of Ghasemi [Link] [81] which has been conducted in a
semiconductor industry as a case study of ETO environments.
In current research, the focus is on simulating the S&OP processes in a BTO industry,
through the Arena software package. Furthermore, to optimize the decision-making variables
and the main objective functions of the context problem, the search of optimal and near
optimal solutions are performed through the combination of Scatter search and Tabu search
teqniques via the optimization tools of Arena, entitled as OptQuest. This optimization tool,
embedded in Arena software uses simulation-based optimization method to express system’s
inputs (controls and system’s constraints) and then executes the search procedure by running
numerous scenarios from simulation model for each set of input parameters with the intention
of finding the optimal solutions and best system outputs possible for objective functions [82].

3. Problem Statement

In current section, the details of applied model and description of the S&OP issue in BTO
industries are presented. After presenting the hypothesis of the model, the input parameters,
system variables decision-making variables are introduced. Furthermore, the problem’s
mathematical model is presented.
This research is studying a supply chain system considering uncertain requests, lost sales
threshold and uncertain distant outsourcing in a BTO manufacturing system. The S&OP
model is implemented for a single-product supply chain, as products and parts are
autonomous and not affiliate towards each other. The manufacturing system collaborates only
with a single supplier, via dual supplying methods for products’ replenishment in the
production planning scheme. The fundamental goal for this model is to find an applicable
trade-off between two main objectives of the S&OP problem; involving the logistics cost and
customer experience level. Due to the cross-functional objectives of the problem and
specified operational indexes, the operational performance level of system is assessed
through three KPIs; i.e., logistics cost (backup supply and holding costs), aggregated and
percentage of delayed and lost sales orders. Consequently, the supply structure is looking for
reducing the logistics cost whereas the sales structure struggles to develop the customer
experience level through reducing the amount of delayed and lost sales orders. The system
specifications containing input parameters, system variables, and decision-making variables
are represented in Table 2.

Table 2
Description of system specifications
Input parameters Decision-making variables
Parameter symbol Parameter symbol
Duration of planning horizon H Safety stock fraction SS t
Procurement time in normal supply L Flexibility degree FRt
Duration of Fixed horizon F System variables
Holding cost of a unit per week ch Parameter symbol
Variable cost in backup supply per
cb Inventory level at the end of week t ILt
unit
Fixed cost for ordering backup Inventory status anticipation at the
Atb IPt
supply end of week t
Number of items ordered at week
Initial inventory level IL 0 nt
t  L , which arrives at week t
Number of new requests for week j Number of items ordered at week t
W i,j bt
which have registered at week i by the backup supply mode
Random variable of actual Requests Number of requests’ prediction p
Rt rt
for week t remaining at week t
Random variable of requests Number of actual requests exist for a
Pt rt
prediction for week t the week t
max
Sales constraint for week t rt
Incoming rate of requests’ arrival vk Number of orders which have been
dot
delayed at week t
Accuracy of predictions U Number of lost sales at week t lst
Lost sales threshold G Average logistics cost per week LC

3.1. Customers arrival in system

In this S&OP study, according to the requests uncertainty, parts’ supplies are based on
requests’ prediction for upcoming periods. The requests prediction in the planning horizon is
based sales structure’s estimation and is expressed by rt p (the number of predicted requests
for week t ). Considering the uncertainty in requests, the actual requests’ volume of week t
are certainly altered from the requests’ prediction of that particular week. Customers’ orders
arrive weekly into the system and the actual requests’ volume for week t would gradually
aggregated. Moreover, arrived customers are not obligated to place their orders for the
earliest possible period in planning horizon. They are allowed to set their orders for a
particular week in advance. In order to determine the requests’ rate entering the system for
each period, every week customers arrive at the system and register orders for the future
week j , based on customer preference and the fixed horizon in planning schedule. Hence, the
model illustrates the incoming requests rate for k future weeks afterward the fixed planning
horizon with v k ; and depicts total requests entering the system at week i via R i .
Accordingly, W i , j is actual request’s level registered in week i and marked for week j . The
principles and equations related to calculate v k and W i , j are described respectively in Eq. (1)
, Eq. (2) and Eq. (3).

v j i  vk (0  vk  1)
(1)
W i , j  v j i . Ri (2)

With the arrival of each customer orders in the system, the quantity of incoming requests
modifies the requests’ prediction level for the specific period. As the result, in each period
(week), manufacturing system knows how much actual request ( rt a ) exist and how much
predicted requests ( rt p ) is remained. In this research like most production planning studies
under uncertainty, there is a fixed horizon (F weeks) wherein, manufacturing schedule is
determined and predictions are not considered for abovementioned weeks. Thereupon, final
quantity of each week’s requests is discovered, lastly at K  F weeks before ordering point.
This means, all arrived customers of week t are able to register their orders for week t  F
or later. Appendix A illustrates the variation of v k rate, which depends on the ordering date of
customer requests. Subsequently, the following equation describes, the accumulation of
incoming requests probabilities ( v k ) equals to:


H
jF
v j i  1 (3)
i j

3.2. Key factors in S&OP

 Sales constraints and flexibility:

One of the most fundamental aspects of this particular S&OP is to set up a practical sales
constraint for managing supply chain. This particular control is adopted by sales structure to
restrict the uprising of requests’ arrival into the system for prospect periods. The sales
constraint is presented by the symbol rt max . This constraint restricts the requests amounts
(includes actual and predicted requests quantities), which manufacturing system is able to
receipt for week t. This parameter is calculated based on number of requests prediction for
week t and flexibility degree at that particular week. The flexibility degree is presented as a
fraction of requests prediction quantity that manufacturing system can accept for week t at the
maximum capacity; i.e. if flexibility degree for a period is being calculated 10 %,
manufacturing system can receive up to 110% of the requests prediction for that specific
week. Simultaneously, if the number of actual requests exceeds from this limitation and this
constraint is saturated, the excess requests face postponement and risk of lost sales in the
mentioned period.

 safety stock:
In inventory management system, supply chains usually consider additional inventory
level of products vis-à-vis safety stock, in order to prevent shortage in stocks. Situations in
which, the actual requests quantity exceeds from requests prediction [4]. Otherwise, the
manufacturing system must compensate all shortages through the backup supplies, regarding
the sales constraints. Similar to flexibility degree used in sales constraints, supply chain
defines the safety stock as a percentage from requests prediction in a period as well.

3.3. Products replenishments: normal and/or backup supplies

Here, The Fig. 1 displays the dynamic flow of supplies, customers’ requests and inventory
status at the week t in planning horizon. In this figure, the supply flows (normal and backup),
products requests (actual and predicted) and inventory level are described. The planning
horizon is divided into three different sub-horizons. The corresponding durations and
descriptions are presented in Table 3.

Table 3
Three planning horizons description
Horizon type Duration Description
 Production plan is fixed and changes in
requests are not allowed.
 Request levels are fixed and predictions are
not produced.
Fixed horizon t to t  F  In case of inventory shortage, the system
orders backup supply to prevent shortage and
risk of lost sales.
 Orders which are submitted via ship-deliveries
t  L weeks beforehand, will arrive at week t.
 The production plan is managed by sales
constraints structure.
 Requests level is not fixed, and includes
Flexible horizon t  F  1 to t  L requests predictions and actual requests.
 Parts supply plan is based on pre-planned
orders through the normal supply.
 The sales constraints are not considered in this
Free horizon t  L  1 to H
horizon and orders are not planned yet.

In Fig. 1, the four first weeks of the moving planning scheme are considered as fixed
horizon. The production plans are deterministic and changes in requests’ amounts are not
allowed. The next six periods, named as the flexible horizon, sales constraints contains
predictions and actual requests. Furthermore, the arrived requests at the first week can be
registered for at least the fifth week or later, regardless of sales constraints’ bounds.
In the next figure as an illustration of the planning scheme, Fig. 2 shows the production
planning schedule at a particular period (end of week zero). In this scheme the same as the
planning horizon designed for the intended S&OP model, the fixed horizon is considered 4
weeks ( F  4 ). The flexible horizon includes next 6 weeks and thus the procurement time for
normal supply is 10 weeks ( L  10 ). Moreover, the free horizon begins from week 11th and
so on. The requests flow (predictions and actual), as well as arriving incoming customers’
requests at this week are demonstrated in this figure.
 Fixed horizon
t  F ix e d h o riz o n  t  F
Part 1. Fixed horizon

nt  bt nt 1  bt 1 nt  F  bt  F

ILt 1 ILt ILt ILt  F 1

t t 1 t F

rt a rt a1 rt a F

Flexible horizon
Part 2. Flexible horizon

t  F  1  Flexible horizon  t  L

nt  F 1 nt  L

ILt  F ILt  F 1 ILt  L 1

t  F 1 t L

rt a F 1  rt p F 1 rt a L  rt p L

Free horizon
t  L  1  Free horizon  H
Part 3. Free horizon

nt  L 1

ILt  L ILt  L 1
t  L 1

rt a L 1  rt p L 1

Inventory level at the end of period Actual requests

Normal supply (By ship) Requests prediction remaining
Backup supply (By plane)

Fig. 1. Requests flow of supply and demand at the week t of planning horizon
 Fixed horizon Flexible horizon Free horizon

2600

2400

2200

2000

1800
Amounts

1600

1400

1200

1000

800

600

400

200

Periods
1 2 3 4 5 6 7 8 9 10 11

Actual requests
Requests prediction
Sales constraints
Total prediction
New actual requests interval to system

Fig. 2. Production schedule and incoming requests at the end of week 0

After passing a week of the planning horizon, Fig. 3 displays the planning schedule at the
end of week 1. The planning horizon has moved a period forward, where the fixed horizon
starts from the second week and free horizon begins from the 12th week. In this scheme, the
customers’ requests and the dynamics of delayed and the lost sales requests in the fifth period
are illustrated. As it is obvious from Fig. 3, in the week 5, actual requests has over passed the
fifth week’s sales constraint. In consequence, a portion of customers’ orders possibly faces
postponement. Moreover, according to the lost sales threshold mechanism, the overflowing
orders are exposed to the risk of lost sales and leaving the system, as there is the possibility
that manufacturing system may not be able to satisfy all delayed orders.

Fixed horizon Flexible horizon Free horizon

2800
Lost sales orders
2600 Delayed orders
Sales constraints
2400 for 5th period
2200

2000

1800
Amounts

1600

1400

1200

1000

800

600

400

200
Periods

1 2 3 4 5 6 7 8 9 10 11 12
Actual requests
Requests prediction
Sales constraints
Total prediction

Fig. 3. Production schedule and dynamics of delayed and the lost sales requests at the end of week 1
 At this point, the Fig. 4 exhibits a presentation of prediction and sales constraint
percentages of fifth week’s total actual requests (used from Fig. 3). Along with demonstrating
different portions of delivered orders via normal or backup supplies. In addition, the
fragments of postponed and lost sales requests are also presented in this figure.

78 %

93 %

Sales constraints = 93%

Total prediction = 78%
Actual requests is equal to 2731 (100 % of both circles)

Fig. 4. A presentation of predictions and actual requests, normal and backup supply, delayed and lost sales ratio

3.4. Problem assumptions

 Since part procurements in normal supply is taking place by distant outsourcing, the
delivery time is considered static and equal to L week of maritime transportation. The
motive is the minor variability in time procurement (1 to 2 day) on L weeks’ scale of
transportation time, and sensibly can be ignored.
 The fixed ordering cost in backup supply is significant and cannot be ignored its value.
Because, such events do not occur frequently in every period (week). The number of
items ordered via backup supply will depend on the shortage amounts in aforementioned
period.
 Transportation in backup supply is carried out using cargo planes, which means fast
transportation logistics. As a result, procurement time in this type of supplying is less than
one week; and whenever it requested, will arrive in the same period. Consequently, it is
considered that the time procurement equals to zero for backup supply mode.
 No limitation toward suppliers’ capacity and cargo capacity is considered in this study.
Considering the fact, these restrictions has been discussed in negotiations between two
parties, this particular assumption is reasonable.
 Flexibility degree and sales constraints on the problem cause demand fluctuations to stays
under control. Consequently, requests level for products levelled in several planning
periods.
 Manufacturing system operates based on the actual requests; thus, there are no requests
predictions in fixed planning horizon. Accordingly remaining predictions are vanished
and not produced in fixed horizon.
 A lost sales threshold mechanism is considered in the S&OP model, which mostly
depends on manufacturer’s responsiveness toward customers’ requests. Consequently,
the model presumes if inventory system faces shortage and delayed orders exceed from
certain amount, the production system is not able to satisfy all the delayed orders.
Therefore, part of overflowing orders would leave the system and considered as lost
sales, where the rest of delayed orders are postponed for the following period.
 Backup supply delivery will be probabilistic; due to the environmental circumstances
and stochastic factors, unexpected and unpredictable events e.g. rough weather
conditions for airborne transportation, and other natural, economic and/or political
disruptions. Consequently, it is possible that sending this type of ordering from the
suppliers to the manufactures is disturbed and supplies are not received on time to
customers.
The odds of receiving required parts by airborne transportation from supplier to
manufacturer are assessed based on the Eq. (4), where p is the probability of occurring a
stochastic unavoidable disruption (according to Bernoulli distribution).

 p  0.1 x 1  p x  (1  p)1 x x  0,1

f ( x)     (4)
1  p  0.9 x  0  0 otherwise

3.5. Key performance indicators for S&OP in BTO supply chain

S&OP is a multi-disciplinary set of processes between two diverse structures of a supply

chain network. Both have unique goals and features and often opposed against one another’s
objectives. In view of that, assessing the overall performance of the system, using a single
category of KPI, is generally inadequate. These indexes or KPIs are divided into two main
categories. Firstly, operational costs assessing by logistics cost which is based on calculating
the holding costs and backup supply costs throughout the planning horizon. Secondly, in
order to analyze the customer experience level, aggregated amount and percentage of delayed
and lost sales orders are calculated. The main goal of this paper is to extend the trade-off
amongst logistics cost function and sales performance (customer experience level), using the
flexibility degree and safety stock fraction.
As pointed out before, the sales constraint parameter for week t will be calculated via
requests’ prediction and flexibility degree used for week t ; and it is shown by rt max . The
equation for this parameter is as follows. The value of FRt in the Eq. (5) expresses the
flexibility degree which is proposed to sales structure for week t .

rt max  1  FRt   rt a  rt p  t 0,..., H  (5)

It is noteworthy that higher flexibility degree for sales structure in a certain period makes
the manufacturing system more capable of accepting requests for the period. Even if the sale
structure estimates requests predictions poorly comparing to actual requests value in the
period, the higher flexibility degree supports manufacturing system to prevent products
shortage in the inventory. Conversely, if a lower flexibility degree is proposed to sales
structure for the intended week, the sales structure must try to decrease the differences
between actual request values and predicted request values. Otherwise, the inventory is most
likely faces shortage and consequently the potential risk of lost sales. Given that the values of
decision-making variables such as flexibility degree are calculated discretely for each week,
these values can be varied from one to another. Flexibility degree is affected by system
variables such as request amounts, operational costs, sales essentials etc.
Manufacturing system in current problem acts as a classic Material Requirement Planning
(MRP) ordering system, for normal supply of products. In order to calculate ordering amount
of normal supply for week t  L , calculation of inventory level anticipation at week
t  L will be required. This variable is indicated by IPt and is equal to differentiation of
arriving products quantity in upcoming weeks (normal supply by ship and backup supply by
cargo-plane) and requests anticipation volume (the actual requests and remaining of requests
prediction). Equations for IPt is as Eq. (6).

t L
IPt  IL 0 +  (n k  b k )  (rka  rkp )   t  0,..., H  L  (6)
k 0

Equation of normal supply is equal to requests anticipation volume for week t  L plus
added number of goods in order to cover the requests fluctuations, minus the anticipated
inventory level for previous period t  L 1. This added value of extra goods will be
computed based on the requests anticipation and safety stock percentage in period t  L .
Equations for nt  L will be calculated according to the Eq. (7).

 t L

nt  L  max 0; rt a L  rt p L  SS t  L


k t  F 1
(rka  rkp )  IPt  L 1 

 t  0,..., H  (7)

Number of products supplied by backup supply for week t, is indicated by bt . The

corresponding equation for this parameter is shown in Eq. (8).

bt  max 0; rt a  (ILt  nt ) (8)

Average logistics cost is the superior objective amongst all in current multi-objective
mathematical model in terms of decision-making preference information. This variable is
calculated by summation of average holding costs and average backup supply costs through
the planning horizon. The equation of average logistics cost is shown in Eq. (9).

 c .ILt  Atb  cb .bt 

H
1
LC  h (9)
H t 0

Other objective functions related to customer experience level, which have also known as
sales essentials, are average delayed orders and average lost sales. Respectively, are shown
by DO in which manufacturing system faces inventory shortage and is unable to recover
from it; and lastly by LS that indicates the lost customers’ orders through planning horizon.
Eventually, a multi-objectives optimization model is proposed as following equations.
  c .ILt  Atb  cb .bt 
H
1
min LC  h (10)
H t 0

Subject to:

ILt  ILt 1  nt  bt  rt a  t  1,..., H  (11)

rt p  rt a  rt max  t  0,..., H  (12)


H
dot
t 0
 DO max (13)
 ij
Wi,j


H
ls t
t 0
 LS max (14)
 ij
Wi,j

FRt  0  t  0,..., H  (15)

SS t  0  t 0,..., H  (16)

 DO max and LS max show the maximum conventional level for average delayed orders and
lost sales.
 Eq. (10) is designed to minimize the average logistics cost (average holding costs and
average backup supply costs).
 Eq. (11) expresses production flow in week t.
 Eq. (12) displays the sales constraint in order to replace the customers’ request in
production plan.
 Eq. (13) and (14) specify KPIs for sales essentials; and respectively comprises average
delayed orders and lost sales in planning horizon.
 Eq. (15) and (16) are introduced to endorse decision-making variables of problem.

The definition of decision-making variables including SS t (safety stock in week t) and

FRt (flexibility degree in week t) depend on the type of decision system and managing policy.
In following sections, the aim is to calculate aforementioned decision variables based on
linear policy and fixed policy.

3.6. Strategies for inventory management and system flexibility

In order to implement certain policies in this S&OP problem several strategies are
considered to control decision variables (flexibility degree and safety stock) and subsequently
manage the supply chain system efficiently. These policies are presented in two phases.
Firstly, in the initial phase, by adopting fixed policy, the decision variables are considered
permanent through the planning horizon. In the next phase, by applying linear policy in the
S&OP model, decision values change frequently from one period to another. The main
purpose of decision-making variables (flexibility degree and safety stock fraction) is to
decide the amount of products’ supply and sales constraints in S&OP system. In practice,
considering fixed values for decision-making variables during planning horizon is generally
simpler. Besides, it is more comprehensible for managers and decision makers in the context
industry.
This policy is determined by  s / x / y ; in which x% is the safety stock fraction and y %
is for flexibility degree during planning horizon. The equations to calculate the fixed policy
parameters are defined in Eq. (17) and Eq. (18). The fixed policy is described in four main
categories, which is shown in Table 4.

s / x /Y SS t  x % t 0 (17)
s / X /y FRt  y % t 0 (18)

Table 4
Clusters of fixed policy
Type symbol description
1  BTO  Considered based on parameters  s /0/  .
 Manufacturing system does not consider any safety stock; customers order
requests with no limitation.
 The only constraint is fixed planning horizon.
 This system performs as classic MRP system in which it uses backup supply
with boundless capacity to prevent shortage.
 Such system is more suited for factories with short lead-time and low ordering
and supply cost.
2  lock  The inventory control system do not use any safety stock or flexibility degree
(  s /0/0 ).
 More applied by factories with low demand uncertainty and are managed by
MTS strategy.
3  s /x /x 
 Values for x and y are equal ( s / x / x ).
 Automotive and generally BTO industries adopt the policy mostly for new-
stablished factories or assembly plants.
 Does not consider backup supply and constantly uses safety stock to prevent
shortage in inventory system.
4  s /x / y  Values for x and y are not identical.
 Flexibility degree is greater than safety stock ( x  y ) .

The search to find optimal values for decision variables in fixed policy includes the
finding of arrays (x , y ) . Using multiple search algorithms in simulation-based optimization
model in feasible region, in which the sales essentials (KPIs for customer experience level)
are fulfilled, as the optimization algorithm finds the optimal solution for the main objective
(minimum average logistics cost).
The main benefits of using fixed policy includes simpler implementation in planning
scheme as well as being a tangible decision-support tool to help managers for decision-
making purposes in industry. For instance, if the supply chain system of a certain automotive
company accepts 10% of flexibility degree for parts e.g. gearboxes in all planning periods. It
means the manufacturing system can accept orders up to 110% of its production capacity.
Bear in mind, although fixed policy has certain benefits such as the simplicity and fast
calculations, but assuming fixed values for flexibility degree and safety stock fraction
through the planning horizon has certain limitations and drawbacks.
 One of the critical issues of adopting fixed policy in S&OP is where flexibility degree and
safety stock fraction are represented as a percentage of requests anticipation (actuals and
predictions) in planning periods. Therefore, the number of extra products’ supply is computed
based on the predetermined flexibility degree, the safety stock, and maximum order-
acceptance constraint (sales constraint) installed by manufacturing system. Sales constraint
rate of week t are directly related to predicted request estimation for that week. As a result, it
expresses that fixed policy is affected by the phenomena titled as "Base Effect".

3.7. Base Effect phenomena

The base effect definition is about the fact that, if amount of requests prediction for the
planning periods undervalues the actual requests, logically not enough capacity is considered
as flexibility amount for sales structure and as maximum acceptable orders by manufacturing
system. In addition, minor safety stock is considered for inventory system to prevent from
shortage. Conversably, if amount of requests prediction for planning periods overvalues
actual request, logically, the manufacturing system would send product orders to supplier
more than it is actually required. In which, may lead to cause a high and unused level of
inventory stocks. Consequently, the system will suffer a great deal of holding costs.
To avoid such difficulties, the model considers a new policy for determining the decision-
making variables, named as linear policy which is being shown by  L .
The linear policy reduce the influence of base effect phenomena on the value of decision
variables as well as operations management in the supply chain network. Therefore, an
average trend of previous periods’ requests will be calculated and later compared to requests
prediction level for week t  L . If requests prediction (in comparison with the actual request)
is undervalued for week t  L , the policy considers a higher flexibility degree for week
t  L and thesales structure; also the safety stock fraction for week t  L would improve.
Conversably, if requests prediction (in comparison with the actual request) is overvalued for
week t  L , the flexibility degree and safety stock fraction will decrease for week t  L . In
addition, maximum acceptable orders by manufacturing system is reduced and inventory
level considers a lower amount for safety stock.
To calculate the values of decision-making variables in linear policy individually, an
absolute value and a relative coefficient is required. The components for determination of
decision variables (safety stock fraction and flexibility degree) are as follows:

 SS  : Constant value for safety stock fraction.

 SS  : The coefficient component for safety stock fraction.
 FR  : Constant value for flexibility degree.
 FR  : The coefficient component for flexibility degree.

The average of requests trends is based on a moving average in M periods of the planning
scheme. Subsequently, in order to evaluate the validity of requests prediction level at week
t  L , the Eq. (19) for parameter t is calculated.

rt a L  rt p L
t   t  M ,..., H 
t
rkp

(19)
k t  M 1 M
 Each component in above equation is considered positive and greater than zero. If the
t is greater than 1, it means, the requests prediction estimation (in comparison with the
actual request) is overvalued for week t  L and consequently a lower safety stock fraction
and flexibility degree will be proposed for this period. As, if the t is lower than 1, it means,
the requests prediction estimation (in comparison with the actual request) is undervalued for
week t  L and accordingly a higher safety stock fraction and flexibility degree will be
proposed for intended period.
In linear policy, decision-making variables, namely safety stock fraction and flexibility
degree are calculated for each period of planning horizon based on Eq. (20) and Eq. (21).

 L :  t  m , SS t L  max 0 ; SS   SS  t  (20)

FRt L  max 0 ; FR  FR  t  (21)

Given that in linear policy, the decision variables have fours components, compared to
fixed policy with two components; and also the values of decision-making variables in linear
policy are varied from previous periods in planning horizon. As a result, the implantation and
applicability of linear policy in practice is more complex. Correspondingly, using absolute
values and relative coefficient components as a support tool for decision-making purposes are
less tangible and comprehensible to industries managers. Even so, using linear policy has
numerous benefits for optimizing S&OP systems.

4. Simulation-based optimization model

The present section concentrates on the approach used for modeling and solving the S&OP
problem. Considering the vital need of improving the S&OP problem and giving the
complexity of system and existence of stochastic parameters, the optimization procedure of
the problem through pure mathematical model is tough, if not impossible without releasing at
least some stochastic variables. The parameters and KPIs include actual requests arriving the
system, requests prediction, lost sales threshold point, and backup supply capacity. One of the
logical and common methods for modeling and solving these type of problems are
simulation-based optimization methods. Consequently, the optimal values of decision-making
variables will be obtained based on the KPIs achieved from simulation outputs [83].
The main benefit of adopting the simulation-based optimization methodology is where the
simulation model calculates the dynamics and non-deterministic variables of system. In this
approach, the purpose is to investigate the influence of decision-making variables (flexibility
degree and safety stock fraction) on BTO supply chain and modifications of KPIs in the
intended S&OP system.
According to the Fig. 5, the framework of the main simulation model is demonstrated,
which comprises the inputs and outputs parameters. Appendix B displays the default settings
and initial values of intended S&OP simulation model in current supply chain, which has
implemented and executed by ARENA simulation software.
Furthermore, finding the absolute optimal values is only possible by counting the entire
problem’s feasible space. Considering the lack of information related to KPIs in simulation
model, counting each solution is time-consuming and unjustified. Therefore, the optimization
algorithms generally uses combination of meta-heuristics search techniques. The framework
of simulation-based optimization model is also shown in the Fig. 5. The figure expresses the
relation of input data and output results of simulation model with optimization strategy.

Simulation-based optimization model framework

Simulation model framework

Inputs Outputs

 Replication length
 Number of replications
 Primary values for holding and backup
supply costs per unit  Average holding cost
 Accuracy of predictions  Average backup supply cost
Simulation  Delayed orders percentage
 Procurement time
 Fixed horizon model  Lost sales orders percentage
 Lost sales threshold
 Initial values of decision-making variables

New values for decision-making variables

Control summary Response summary

 Control variables
 Select inventory strategy
 Defining control variables criteria (Lower  Optimal values of decision-making
bound, Upper bound) Optimization variables
 Defining the mathematical model for strategy  Calculation time
optimization  Optimal value of KPIs for sales essentials
 Defining the main objective function in  Optimal value for average logistics cost
optimization model (main objective function)
 Defining the subjective constraints of
mathematical model

Fig. 5. The framework of simulation-based optimization model

4.1. Simulation model

The objective functions values of S&OP problem are evaluated through simulation model.
The model strives to calculate non-deterministic system variables by modeling the S&OP
processes of the mentioned supply chain system. The system’s KPIs include average logistics
cost, delayed orders and lost sales percentages. This study uses simulation modeling to
generate the requests’ prediction for the planning horizon, customers’ requests arrival rate
into the system, dynamics of sales essentials, measuring the delayed orders and lost sales
quantities, calculating the quantity of parts ordered through normal and/or backup supply, and
finally creating sales constraints in the planning horizon. Later in Fig. 6, a conceptional view
of S&OP simulation model is presented. The S&OP system is categorized into three key
divisions including sales structure, supply structure, and modules related to calculation of the
average requests prediction trend, to determine the flexibility degree and safety stock fraction
in each period. This division is the verge of sales and supply chain functions. Besides, Fig. 7
depicts a demonstration of the flowchart designed specifically for the intended S&OP system.

Predict
Predict Dispose
accuracy for
generating Predicts
model

Actual Updating Updating

Customers DM vars and
requests Total Predict requests sales
arrival in historic
week distribution chart schedules via constraints
average trend
for week Ni,j schedules

Inventory
Sales
update and Logistics
System face shortage ? False essentials Dispose
normal supply Cost
calculation for t+L

Backup
True supply by
plane

Fig. 6. Conceptional view of the S&OP simulation model

 Sales structure:
The sales structure is one of the fundamental sections of S&OP simulation model. Sales
structure concerns products sales volumes and customers’ experience level. It focuses on
items such as:
 Generating requests predictions for 52 weeks of every year,
 Customers’ arrival into system,
 Assigning customer orders in production planning for scheduling horizons,
 Update and apply alterations to sales constraints (replacing incoming actual requests in
place of requests prediction),
 Creating sales constraints for the planning horizon via the agreed flexibility degree
from sales structure and supply structure,
 Calculating KPIs for sales essentials in order to evaluate customers’ experience level.

 Supply structure:
Supply structure is another fundamental section of simulation model intended for current
S&OP problem. Supply structure focuses on the factors i.e. production planning, supplying
the required parts for manufacturing system and operational costs. The matters associated
with this structure includes:
 Inspect and control the inventory system with the aim of avoiding shortage in
manufacturing system,
 Updating an evaluation of the inventory level at each period,
 Ordering parts to supplier by maritime transportation for period t  L ,
  Ordering backup supply through cargo planes, to prevent shortage in manufacturing
system,
 Evaluating the logistics cost (holding costs and backup supply costs) over simulation
length.

Start

Input values of model

Requests predictions for planning horizon

Updating period (k=k+1)

Customers enter the system

Assigning customer orders in production planning

t+4, ,t+8

Update and apply changes to sales constraints

t, ,t+9

Linear policy? Yes Average predict requests trend

No Calculate SS t  L and FR t  L for period

t+L

Create sales constraints for period t+L

Shortage in system? Yes Backup supply order ( bt )

No

Deliver customers orders and calculate sales essentials KPIs

Update inventory level of period

Calculate and normal ordering ( nt ) for period t+L

Calculate average logistics cost

End of the period and enter into next period

Fig. 7. S&OP system flowchart

4.2. Simulation results

Based on the simulation model (Fig. 6), the following plots demonstrate the dynamics of
KPIs values during the planning horizon. Accordingly:
 Plot 1 displays the average of sales essentials (delayed orders and lost sales).
 Plot 2 displays the inventory changes through the simulation length.
 Plot 3 presents the average logistics cost through planning horizon, which has reached
the steady states status.

Plot 1. Average of sales essentials

Plot 2. Inventory level fluctuations

 Plot 3. Average Logistics cost

4.3. Optimization model

The purpose of simulation-based optimization method in the current problem is to identify

the effect of decision-making variables (flexibility degree and safety stock fraction) on the
overall performance of S&OP system in a BTO supply chains. The dynamics and stochastic
system variables are calculated in the simulation model. Eventually, the optimizer modules
via outputs data of simulation phase seeks to find the best possible solutions for the decision-
making variables and the main objective function in feasible solutions. In current study,
OptQuest tool is used as the main simulation-based optimization engine in ARENA, provided
by OptTek [Link], which is designed as an exclusive set of dominant algorithms and
refined exploration systems. As a result, OptQuest allows operators to move away from a
“What if scenario?” attitude to a “searching for optimal solution?” approach. The above-
mentioned optimization tool employs a combination sets of Neural Networks, meta-heuristics
and evolutionary algorithms such as Tabu Search, Simulated annealing algorithms, Scatter
Search and etc. OptQuest searches intelligently through the feasible solutions and gradually
converges the gained values to the optimal status point [84]. The optimization tool, introduced
by Glover et al. [85], tries to find the optimal values for control variables and to optimize the
problem objectives, using diverse search algorithms and meta-heuristic methods.

 Optimization model using fixed policy (  S / x / y ):

In this subsection, optimization methods search for the optimal values of flexibility degree
(FR) and safety stock fraction (SS) using the OptQuest tool. These parameters are chosen as
control variables and are limited by upper and lower boundaries. It is worth knowing that the
named parameters are considered positive and the upper bounds should be sufficiently high
enough to cover a wide range of feasible answers. The proposed values below are taken as
initial values of optimization model in Table 5.
Table 5
Proposed initial values for control variables in fixed policy
Fixed policy Initial value Lower bound Upper bound
FR 100 % 0% 200 %
SS 50 % 0% 100 %

The results obtained from fixed policy (  s / x / y ) in S&OP model and the employment of
simulation-based optimization model is illustrated in Table 6. This table displays the obtained
values for control variables (FR and SS) in order to determine the decision making variables
(flexibility degree and safety stock fraction) using the fixed policy (  s / x / y ).

Table 6
Values of control parameters (FR and SS)
Fixed policy LC DO max LS max FR SS
1 12001 8% 6% 10 % 2%
2 10964 9% 7% 5% 2%
3 10812 9% 8% 4% 1%

From this table, it is obvious, the values of logistics cost as the main objective of S&OP
model, are in conflict toward the sales essentials (customers’ experience level) KPIs.
Accordingly, by reducing the maximum sales essentials level, the holding cost and backup
supply cost grow and consequently the logistics cost will incline rapidly. Whereas, by
releasing the sales essentials level, the holding and backup supply cost will be reduced and
consequently the logistics cost would drop significantly.
The average logistics cost is the main objective function of the problem, according to the
desired information from the decision maker. The effect of other objective functions related
to sales essentials, including delayed orders and lost sales percentage on the objective
function is depicted in the Plot 4; which are obtained using the efficient response sets from
the ɛ-constraints method in the multi-objectives optimization problems. Appendix C displays
the identical statistics shown in the Plot 4 in spreadsheet viewpoint.

LS max
DO max

Plot 4. The 3D plot of changes in KPIs in S&OP model, using the fixed policy
 Optimization model using the linear policy (  L / x ):
0 / x1/ y 0 / y1
In this subsection, considering the S&OP model, utilizes the linear policy (  L / x )
0 / x1/ y 0 / y1
to enhance the overall performance of the system; it considers decision-making variables to
be varied during the planning horizon from each period to another. As the result, the current
simulation-based optimization model will be more complex comparing to fixed policy model.
In such circumstances, the event-driven simulation-based optimization model generally
requires more time to find the best values nearby the absolute optimal solutions.
Respectively, each decision variables in the main problem is defined as dual control
parameters (fixed value and relative coefficient). At this point, for linear policy is obligatory
to find and determine four control parameters for the optimal decision values in comparison
of fixed policy (  s / x / y ), which is defined based on searching for two control parameters.
The proposed values below are considered as initial values of optimization model
(  L / x / x / y / y ). Moreover, the related boundaries and initial values for control parameters
0 1 0 1
are demonstrated in Table 7.

Table 7
Proposed initial values for control variables in linear policy
Linear policy Initial value Lower bound Upper bound
FR  100 % 0% 200 %
FR  100 % 0% 200 %
SS  50 % 0% 100 %
SS  50 % 0% 100 %

The results obtained from implementation of linear policy (  L / x ) in S&OP

0 / x1/ y 0 / y1
model and the employment of simulation-based optimization model is illustrated in Table 8.
This table displays the obtained values for control variables ( FR  , FR  , SS  , SS  ), in order
to determine the decision making variables (flexibility degree and safety stock fraction) using the
linear policy (  L / x / x / y / y ) and by Eq. (19) and Eq. (20).
0 1 0 1

Table 8
Values of control parameters (FR0, FR1, SS0, SS1)
Linear FR  SS 
policy
LC DO max LS max FR  SS 
1 10691 8% 6% 81.65 % 91.39 % 59.04 % 58.48 %
2 9977 9% 7% 97.88 % 115.8 % 69.23 % 45.65 %
3 9610 10 % 8% 92.93 % 197 % 68.4 % 58.55 %

As stated before, the influence of customer experience KPIs (sales essentials) which,
include delayed orders and lost sales percentages on the objective function is perfectly visible
in the Plot 5; which are obtained using the efficient response sets from the ɛ-constraints
method in the multi-objectives optimization problems. Appendix D displays the identical
statistics shown in the Plot 5 in spreadsheet viewpoint.
 LS max
DO max

Plot 5. The 3D plot of changes in KPIs in S&OP model, using the linear policy

4.4. Numerical results

In current section, the results’ deviation among both fixed and linear policies are
expressed in Table 9. The main distinction is about the ratio of minimizing average logistics
cost (LC) as the main objective and relative control parameters.
As it can be observed from the following table, the linear policy (  L / x / x / y / y ) has led
0 1 0 1
to find better optimal values for LC in comparison of deploying the fixed policy (  S / x / y ). It
is conspicuous, the result of using the linear policy is more tangible as the objective value is
significantly altered from the fixed policy objective results. Although, considering the fixed
policy’s simplicity and being more straightforward for managers and decision-makers, this
policy is less time consuming and easily applicable.

Table 9
Values of LC and control variables in both fixed and linear policies
Policy Decision variable value LC DO max LS max
FR  81.65 %
FR  91.39 %
Linear 10691 8% 6%
SS  59.04 %
SS  58.48 %
FR 10 %
Fixed 12001 8% 6%
SS 2%

At this point, a more subjective and exclusive evaluation of S&OP model performance
considering fixed and linear policy is made. In the following, the main different aspects of the
chosen polices are expressed. Table 10 displays the overall performance level of fixed and
linear policy in terms of cost objectives and average calculation time.
Table 10
Performance comparison of fixed and linear policy
Policy LC Compared to least cost Calculation time Compared to least time
Fixed 12001 +12.25 % 900 secs 0%
Linear 10691 0% 2400 secs +167 %

The optimization model via fixed policy has performed more agile, trying to find the best
solutions possible. This result seems obvious; since, the number of control parameters
required in fixed policy is two time less than the number of control parameters required in
linear policy (to calculate decision-making variables). The average calculation time needed to
detect optimal objective values is significantly longer in linear policy compared to fixed
policy. Regarding to the objective function value (average logistics cost), the S&OP model,
has a better overall performance in linear policy rather than fixed strategy. For instance, by
using  L / x / x / y / y , S&OP model is able to reduce the logistics cost up to 17% compared
0 1 0 1
to  S / x / y result. As it is presented in Table 11, the policies’ performance rate depends on
the dynamics and variation of requests nature (requests’ prediction accuracy).

Table 11
Difference of two policies in accuracy and variation of requests’ prediction
requests’ prediction accuracy Fixed policy Linear policy Variation ratio
(-100,100) 9296 10125 -8 %
(-500,500) 10691 12001 -11 %

According to the Table 11, it is plausible that reduction ratio of logistics cost using the
linear policy performs more efficiently, whereas requests’ prediction is less accurate rather
than the scenario where the requests accuracy is high. This outcome is sensible, since the
primarily purpose of linear policy in determination of decision variables is to decrease the
influence by Base Effect phenomena. In addition, the results show us, changes in customers’
requests uncertainty will not have a noteworthy influence on the improvement of linear
strategy performance contrary to fixed policy. Table 12 describes the results obtained from
two strategies in the matter of sales essentials variability effects on S&OP model.

Table 12
Changes effect in the sales essentials
Sales essentials Backup supply
Fixed policy Linear policy Variation ratio
LS max
 DO max
cost
50 $ 12954 11369 -12.2 %
 5%  7% 100 $ 16596 13830 -16.7 %
50 $ 12001 10691 -11 %
 6%  8% 100 $ 14808 12646 -14.6 %
50 $ 10964 9977 -9 %
 7%  9% 100 $ 13551 12127 -10.5 %
50 $ 10024 9610 -4.3 %
 8% 10% 100 $ 12283 11602 -5.5 %

As demonstrated in Table 12, the enhancement ratio of implementing linear policy

compared to fixed policy is more visible; when the supply chain system requires higher
customer’s experience level (sales essentials are in more restricted). This means, if
organizations seek to gain a higher customers’ experience level, the benefits of using linear
policy would have a more tangible impact. In such conditions, the feasible solution space is
more constrained; since, the manufacturing system requires more backup supplies to prevent
the risk of stock shortage.

LC ($)
Backup supply variable cost per product (type 1) = 50 $
Backup supply variable cost per product (type 2) = 100 $
18000 Linear Policy
Fixed Policy
16000

14000

12000

10000
50 $

50 $

50 $

50 $
8000

100
100

100

100

$
$

50 $
50 $

50 $

6000 50 $
100
100

100
$
$

4000 $

100
$
2000

(5,7) (6,8) (7,9) (8,10)

4444max max 4444
(LS , DO )
4444
(Percent)
Plot 6. Linear and fixed policy results due to changes in variables backup supply cost and sales essentials

Plot 6 demonstrates the significance of backup supply cost and sales essentials variability
respectively in both fixed and linear policies. As a result, this examination indicates that by
increasing the backup supply variable cost, the importance of using linear policy instead of
fixed one would be more practical.

5. Discussion and managerial implications

In this article, authors have designed a multi-objective model and explored solutions’
feasible space for nearby optimal values in order to manage and control the inventory level
and flexibility ratio in sales and supply structures, using a simulation-based optimization
approach. The results obtained from this study indicate that managers and decision-makers,
whom practice the S&OP process in supply chain networks can benefit from substantial
advantages of using the linear policy instead of considering the flexibility degree and safety
stock fraction permanent in the entire planning horizon. These advantages mainly depend on
the system’s features, including prediction accuracy, requests variability and backup supply
costs. Also the company size can play a major role in utilizing S&OP processes for supply
chain management, as larger companies tend to obtain greater advantages than small and
medium-sized enterprises (SMEs) [18, 86]. In most manufacturing settings, particularly in
BTO industries e.g. automotive companies and other related industries, aligning the S&OP
processes to find appropriate decision variables are performed frequently for each product
value chain. In such circumstances, it is possible that S&OP process is executed for up to 30
products on every occasion; depending on the location and the environment of the
manufacturing system.
In practice, an hour of calculation time is acceptable and justified for implementing the
S&OP model for a single product. The S&OP model via fixed policy calculation time,
regardless of product type and systems parameters, is mainly short and tolerable (see Table
10). As it is been explained before, in this type of optimization processes, there is only need
to calculate two control parameters to find the optimal values of decision variables (flexibility
degree and safety stock fraction). In contrast, modeling and optimization of S&OP model
with linear policy can be challenging due to the potential extensive calculation time process.
As mentioned before, in this setting in order to find the decision variables optimal values via
linear policy, the simulation-based optimization model needs to search the feasible space for
four control parameters. Given the fact that decision variables are considered dynamic
throughout the planning horizon in the mentioned setting.
The subject of deciding on applying certain inventory strategies in the automotive
industries is critical and common, especially for the modeling of S&OP processes for new
products, which are exposed to severe fluctuations in demands and other internal/external
factors. Companies must adopt a suitable policy to manage the intended S&OP processes.
The impact of S&OP processes is not equal among different contexts. For example accessing
adequate supply capacity is crucial throughout new product launches [87]. However, for
manufacturing systems with less diverse products and steadier customers’ orders uncertainty,
the adoption of linear policy by managers and decision-makers can achieve major advantages
beyond the fixed policy capability. In supply chains with less accurate requests prediction and
less end-to-end supply chain visibility i.e. companies contributing to highly competitive
markets, using linear policy compared to fixed policy performs more efficiently in terms of
reducing logistics cost and improving customers’ service level (see Table 11). Based on the
outcomes achieved from the numerical experiment, when the context supply chain
management desires higher level of customers’ service (lower percentage of lost sales and
fewer delayed orders), execution of linear policy is more influential compared to its
counterpart (see Table 12). It is recommended, if the industry setting comprises capital-
intensive products with inflated transportation cost, establishing and execution of linear
policy in supply chain planning has more priority over fixed policy (see Table 12 and Plot 6).
This is due to the dominant performance of such policy in supply chain management, as it
decreases the destructive influence of base effect phenomenon in terms of over/under supply
capacity difficulties; along with superior enhancement of objective functions from the
production planning perspective.
To resolve the challenges of choosing appropriate policy in the intended supply chain
network, authors propose a setting that the system should apply S&OP processes via fixed
policy in planning horizon. To be precise, assigning static values for flexibility degree and
safety stock fraction in the early phases of new products. This is due to the simplicity of fixed
policy implementation for new processes of S&OP. In the next phase of planning, in case of a
more S&OP maturity level, it is recommended to modify the system’s policy to linear setting
in order to achieve better coordination of customers’ requests and supply capacities,
respectively in sales and supply structures. Accordingly, the intended manufacturing system
and supply chain network gains superior performances in terms of operational cost and
customers’ experience level. This takes place by aligning flexibility ratio and inventory level
shaped by flexibility degree and safety stock fraction for each period individually.
6. Conclusion and future works

S&OP in supply chain management is known as one of the most critical subjects,
concerning the challenges associated with customers’ requests and supply capacity
coordination and the system’s overall performance in enhancing the customers’ experience
level and controlling operational costs. The expectant growth of globalization and natural
stochastic environments of supply chains has impelled companies and organizations to
emphasis on promoting the coordination and collaboration among sales and supply structures.
This matter carries various complex challenges for managing global supply chain networks.
Given the fact that, sales and supply structures often contain conflicting and at times
incomparable goals. This paper extends a novel S&OP model, which has been presented by
Lim et al. [20] before; by designing a validated numerical experiment within the industrial
data patterns from the prior study as inputs, and with the intention of managing and
controlling S&OP issues that various industries such as BTO and other related industries
encounter on regular basis. The main objective of this particular simulation-based
optimization model for the S&OP issues in BTO industries are to minimize the logistics cost,
while simultaneously increase the customer experience level as much as possible. This
theoretically will prevent losing potential and enduring customers and market’s sales shares.
This research considers issues associated with demand and supply coordination, uncertain
distant outsourcing, operational costs, and customers’ experience level.
The main purpose of this paper to expand knowledge is as follows. Initially, is about how
S&OP processes are executed in a BTO industry with highly complex environment,
concerning the uncertain distant outsourcing. In addition, this research designs a more
accurate stochastic simulation-based optimization model and investigate the consequences of
stochastic disruptions in terms of demands uncertainty, lost sales thresholds, unreliable
distant outsourcing, and relative costs. Finally, yet importantly, the current article enhances
the system’s KPIs and objective functions plus providing managerial insights in term of
selecting appropriate inventory strategies concerning fixed or linear policy in S&OP process
of BTO supply chains. This research has three main modifications, in terms of innovation and
the novelty regarding to Lim et al. [20] research. Firstly, in this research, the backup supply
costs are calculated based on counting both fixed and variable costs simultaneously. Given
that, the occurrence of parts backup supplies is non-deterministic and would not happen
frequently in every period. Secondly, parts delivery are considered probabilistic and uncertain
in the backup supply situation. Due to the long distance of suppliers’ allocations, it is
possibility that unpredictable events take place outside the system boundary, which the
system lacks the capability to predict and/or manage such events adequately. Hence, it is only
logical to enhance the accuracy of S&OP model; thus, the probability of delay or withdrawal
of backup supply transports is considered. Finally, the designed model for intended system is
a unique model based on simulation-based optimization approach and is original to previous
related researches. The evaluation of customers’ experience level and related KPIs are unlike
to the previous inspired paper mentioned above.
This S&OP model has been laid out for companies and industries with unreliable demands
predictions and distant outsourcing; and can be utilized for companies that function with
BTO, MTO, ETO approaches, and generally all global operations networks, which are
capable of developing modular product architectures. These companies characteristically
have various assemble plants, long distance between the supplier and producer, long lead
times, and perform in highly competitive markets [67]. In conclusion, this specific S&OP
model is applicable for all capital-intensive industries that experience such conditions; i.e.
demands uncertainty, unavailable resources and lost sales thresholds established by the
context manufacturing system.
 As a final point, this research puts forward few pointers and prospects intended for future
studies on the subject of S&OP in BTO and other related industries as mentioned in the
context.
 In this study, procurement time from the supplier to manufacturer is static. For further
studies, the outcomes maybe different by considering the lead-time as an uncertain
variable and a source of complexity.
 To this point, the authors has not found any research in this particular S&OP setting
literature, which considers constraints related to supplier’s responsiveness capacity.
Future researcher may consider the idea of applying certain restrictions in the terms of
supplier and the manufacturer cooperation via primarily negotiations amongst parties of
the supply chain network.
 Future researchers can improve the requests anticipation in such supply chains, using the
meta-heuristic algorithms, machine learning and deep learning techniques as supporting
decision-making tools.

Acknowledgements

The authors would like to express their gratitude toward chief editor and anonymous
reviewers for their support and constructive comments and suggestions from their
perspective, which led to significantly improving the excellence of this publication.
Appendices
Appendix A
Values v k for deployment of interval requests
k  j i vk
j i  F 0
j i  F 0.37
j  i  F 1 0.31
j i  F  2 0.17
j i  F 3 0.12
j i  F  4 0.03
j i  F  4  0.1 0

Appendix B
Default settings and Initial values of the simulation model
Parameter Value Parameter Value
H 3650 days (520 Ws) min Rt  500
L 70 days (10 Ws) max Rt  2000
F 28 days (4 Ws) min U  -500
ch 10 $ per part max U  500
cb 50 $ per part vk Appendix A
Atb 1000 $ per order SS t 20%
IL 0 0 unit FRt 5%
Pt Rt  U
Replication 50
G 100 delay orders units

Appendix C
The identical statistics shown in the Plot 4 in spreadsheet viewpoint
Number DO max LS max LC FR SS
1 11 9 8926 1.05 0.17
2 10 8 9904 5.72 2.85
3 9 8 10599 8.63 1.43
4 9 7 10674 9.65 1.53
5 8 6 11487 14.91 1.84
6 7 6 11950 18.21 6.3
7 7 5 12522 26.43 6.03
8 6 5 12862 30.71 0
9 5 4 13267 44.22 0
10 4 3 13829 75.26 0
11 3 3 14327 114.71 0.14
12 3 2 14569 185.9 0.13
13 2 2 15088 195.55 3.22
14 2 1 18288 185.93 68.56
15 1 1 19145 200 77.68
Appendix D
The identical statistics shown in the Plot 5 in spreadsheet viewpoint
Number DO max LS max LC FR  FR  SS  SS 
1 11 9 8693 41.68 173.67 39.47 25.38
2 10 8 9098 45.94 54.88 41.67 39.1
3 9 7 9858 82.21 94.52 50.72 91.17
4 8 7 10438 92.88 99.77 39.53 49.82
5 8 6 10680 100 100 50 50
6 7 6 10819 119.45 125.38 45.71 85.55
7 7 5 10959 140.71 142.95 80.31 74.44
8 6 4 11323 176.46 182.78 98.18 84.79
9 5 4 11809 193.53 200 100 100
10 4 3 12813 200 188.7 42.35 29.14
11 3 3 13475 191.65 154.32 44.69 34.58
12 3 2 14183 170.93 75.42 68.77 67.72
13 2 2 14876 200 0 100 100
14 2 1 18343 200 60.93 97.85 20.31
15 1 1 18815 200 8.3 95.55 15.82

References

[1] Tang, C.S., Perspectives in supply chain risk management. International Journal of
production economics, 2006. 103(2): p. 451-488.
[2] Elias, S., New car buyer behaviour. Research Survey Report. Cardiff University,
Cardiff, 2002.
[3] Holweg, M., S.M. Disney, P. Hines, and M.M. Naim, Towards responsive vehicle
supply: a simulation-based investigation into automotive scheduling systems. Journal
of Operations Management, 2005. 23(5): p. 507-530.
[4] Jonsson, P. and S.-A. Mattsson, An inherent differentiation and system level
assessment approach to inventory management. The International Journal of Logistics
Management, 2019.
[5] Iakymenko, N., A. Romsdal, M. Semini, and J.O. Strandhagen, Managing
engineering changes in the engineer-to-order environment: challenges and research
needs. IFAC-PapersOnLine, 2018. 51(11): p. 144-151.
[6] Ganjavi, N., R. Tavakkoli–Moghaddam, and S.M. Sajadi, Modelling and solving of a
lot sizing problem for multiple items and multiple periods with shortage. International
Journal of Services and Operations Management, 2015. 20(1): p. 102-117.
[7] Elkins, D.A., N. Huang, and J.M. Alden, Agile manufacturing systems in the
automotive industry. International Journal of Production Economics, 2004. 91(3): p.
201-214.
[8] Childerhouse, P., S.M. Disney, and D.R. Towill, On the impact of order volatility in
the European automotive sector. International Journal of Production Economics,
2008. 114(1): p. 2-13.
[9] Miemczyk, J. and M. Holweg, Building cars to customer order—what does it mean
for inbound logistics operations? Journal of Business Logistics, 2004. 25(2): p. 171-
197.
[10] Stevenson, M. and M. Spring, Supply chain flexibility: an inter-firm empirical study.
International Journal of Operations & Production Management, 2009. 29(9): p. 946-
971.
[11] Jonsson, P. and J. Holmström, Future of supply chain planning: closing the gaps
between practice and promise. International Journal of Physical Distribution &
Logistics Management, 2016.
[12] Shurrab, H., Tactical Planning in Engineer-to-Order Environments. 2019, Chalmers
Tekniska Hogskola (Sweden).
[13] Ivert, L.K., I. Dukovska-Popovska, A. Fredriksson, H.C. Dreyer, and R. Kaipia,
Contingency between S & OP design and planning environment. International Journal
of Physical Distribution & Logistics Management, 2015.
[14] Oliva, R. and N. Watson, Cross-functional alignment in supply chain planning: A
case study of sales and operations planning. Journal of Operations Management,
2011. 29(5): p. 434-448.
[15] Wagner, S.M., K.K. Ullrich, and S. Transchel, The game plan for aligning the
organization. Business Horizons, 2014. 57(2): p. 189-201.
[16] Muzumdar, M. and J. Fontanella, The secrets to S&OP success. Supply chain
management review, v. 10, no. 3 (Apr. 2006), p. 34-41: ill, 2006.
[17] Bower, P., How the S&OP process creates value in the supply chain. The Journal of
Business Forecasting, 2006. 25(2): p. 20.
[18] Grimson, J.A. and D.F. Pyke, Sales and operations planning: an exploratory study
and framework. The International Journal of Logistics Management, 2007. 18(3): p.
322-346.
[19] Sorensen, K., G. Janssens, K.;, M. Lasgaa, and F. Witlox, Sustainable Chemical
Warehousing, in Management Principles of Sustainable Industrial Chemistry. 2013,
Wiley Online Library.
[20] Lim, L.L., G. Alpan, and B. Penz, A simulation-optimization approach for sales and
operations planning in build-to-order industries with distant sourcing: Focus on the
automotive industry. Computers & Industrial Engineering, 2017. 112: p. 469-482.
[21] Jonsson, P., L. Kjellsdotter, and M. Rudberg, Applying advanced planning systems for
supply chain planning: three case studies. International Journal of Physical
Distribution and Logistics Management, 2007. 37(10): p. 816-834.
[22] Sajadi, S., M.R. Taghizadeh Yazdi, and A. Aghaee, Multi objective robust and
simulation-based optimization approach for convergent supply chain with continued
production despite the fraction of some part. 2019.
[23] Araúzo, J.A., J. Pajares, and A. Lopez-Paredes, Simulating the dynamic scheduling of
project portfolios. Simulation Modelling Practice and Theory, 2010. 18(10): p. 1428-
1441.
[24] Rad, M.H., S.M. Sajadi, and M.M. Tavakoli, The efficiency analysis of a
manufacturing system by TOPSIS technique and simulation. International Journal of
Industrial and Systems Engineering, 2014. 18(2): p. 222-236.
[25] Olhager, J. and D.I. Prajogo, The impact of manufacturing and supply chain
improvement initiatives: A survey comparing make-to-order and make-to-stock firms.
Omega, 2012. 40(2): p. 159-165.
[26] Beckman, S.L. and D.B. Rosenfield, Operations strategy: competing in the 21st
century. 2008: McGraw-Hill/Irwin.
[27] Hanson, G.H., The rise of middle kingdoms: Emerging economies in global trade.
Journal of Economic Perspectives, 2012. 26(2): p. 41-64.
[28] Dicken, P., Global Shift: Mapping the Changing Contours of the World Economy.
2011: Guilford Press.
[29] Kinkel, S., Trends in production relocation and backshoring activities: Changing
patterns in the course of the global economic crisis. International Journal of
Operations & Production Management, 2012. 32(6): p. 696-720.
[30] Humphrey, J., Globalization and supply chain networks: the auto industry in Brazil
and India. Global Networks, 2003. 3(2): p. 121-141.
[31] Pashaei, S. and J. Olhager, The impact of product architecture on global operations
network design. Journal of Manufacturing Technology Management, 2017.
[32] Prater, E., M. Biehl, and M.A. Smith, International supply chain agility‐Tradeoffs
between flexibility and uncertainty. International journal of operations & production
management, 2001.
[33] Stadtler, H. and C. Kilger, Supply Chain Management and Advanced Planning:
Concepts, Models, Software, and Case Studies. 2010.
[34] Affonso, R., F. Marcotte, and B. Grabot, Sales and operations planning: the supply
chain pillar. Production Planning and Control, 2008. 19(2): p. 132-141.
[35] Jonsson, P. and S.-A. Mattsson, Manufacturing planning and control. 2009.
[36] Bozarth, C.C., D.P. Warsing, B.B. Flynn, and E.J. Flynn, The impact of supply chain
complexity on manufacturing plant performance. Journal of Operations Management,
2009. 27(1): p. 78-93.
[37] Senge, P.M., The leader’s new work. Leading organizations, 1998. 1: p. 439-457.
[38] Ivert, L.K. and P. Jonsson, When should advanced planning and scheduling systems
be used in sales and operations planning? International Journal of Operations &
Production Management, 2014.
[39] Mula, J., R. Poler, J.P. García-Sabater, and F.C. Lario, Models for production
planning under uncertainty: A review. International journal of production economics,
2006. 103(1): p. 271-285.
[40] Afshar-Bakeshloo, M., A. Bozorgi-Amiri, S.M. Sajadi, and F. Jolai, A multi-objective
Environmental Hedging Point Policy with customer satisfaction criteria. Journal of
Cleaner Production, 2018. 179: p. 478-494.
[41] Zhong, R.Y., G.Q. Huang, S. Lan, Q. Dai, X. Chen, and T. Zhang, A big data
approach for logistics trajectory discovery from RFID-enabled production data.
International Journal of Production Economics, 2015. 165: p. 260-272.
[42] Vieira, A.A., L.M. Dias, M.Y. Santos, G.A. Pereira, and J.A. Oliveira, On the use of
simulation as a Big Data semantic validator for supply chain management.
Simulation Modelling Practice and Theory, 2020. 98: p. 101985.
[43] O’Reilly, S., A. Kumar, and F. Adam, The role of hierarchical production planning in
food manufacturing SMEs. International Journal of Operations & Production
Management, 2015.
[44] Miller, T., Hierarchical Operations and Supply Chain Planning. 2012: Springer.
[45] Hans, E.W., W. Herroelen, R. Leus, and G. Wullink, A hierarchical approach to
multi-project planning under uncertainty. Omega, 2007. 35(5): p. 563-577.
[46] Jacobs, F.R., Manufacturing planning and control for supply chain management.
2011: McGraw-Hill.
[47] Wallace, T.F., Sales & operations planning: the" how-to" handbook. 2004: TF
Wallace & Co.
[48] Bower, P., Guideposts, directions, and scenarios: connecting S&OP to detailed
planning. The Journal of Business Forecasting, 2012. 31(2): p. 20.
[49] Olhager, J., The role of the customer order decoupling point in production and supply
chain management. Computers in industry, 2010. 61(9): p. 863-868.
[50] Thomé, A.M.T., L.F. Scavarda, N.S. Fernandez, and A.J. Scavarda, Sales and
operations planning: A research synthesis. International Journal of Production
Economics, 2012. 138(1): p. 1-13.
[51] Thomé, A.M.T., R.S. Sousa, and L.F.R.R. Scavarda do Carmo, The impact of sales
and operations planning practices on manufacturing operational performance.
International Journal of Production Research, 2014. 52(7): p. 2108-2121.
[52] Hulthén, H., D. Näslund, and A. Norrman, Framework for measuring performance of
the sales and operations planning process. International Journal of Physical
Distribution & Logistics Management, 2016.
[53] Olhager, J., M. Rudberg, and J. Wikner, Long-term capacity management: Linking the
perspectives from manufacturing strategy and sales and operations planning.
International journal of production economics, 2001. 69(2): p. 215-225.
[54] Olhager, J., Evolution of operations planning and control: from production to supply
chains. International journal of production research, 2013. 51(23-24): p. 6836-6843.
[55] Lim, L.L., G. Alpan, and B. Penz, Reconciling sales and operations management with
distant suppliers in the automotive industry: A simulation approach. International
Journal of Production Economics, 2014. 151: p. 20-36.
[56] Farsi, M., J.A. Erkoyuncu, D. Steenstra, and R. Roy, A modular hybrid simulation
framework for complex manufacturing system design. Simulation Modelling Practice
and Theory, 2019. 94: p. 14-30.
[57] Fisher, M.L., What is the supply chain for your product. Harvard Business Review,
1997. 75(2): p. 105-106.
[58] Christopher, M. and M. Holweg, " Supply Chain 2.0": managing supply chains in the
era of turbulence. International Journal of Physical Distribution & Logistics
Management, 2011. 41(1): p. 63-82.
[59] Brabazon, P.G., B. MacCarthy, A. Woodcock, and R.W. Hawkins, Mass
customization in the automotive industry: comparing interdealer trading and
reconfiguration flexibilities in order fulfillment. Production and Operations
Management, 2010. 19(5): p. 489-502.
[60] MacCarthy, B., P.G. Brabazon, and J. Bramham, Fundamental modes of operation for
mass customization. International Journal of Production Economics, 2003. 85(3): p.
289-304.
[61] Volling, T. and T.S. Spengler, Modeling and simulation of order-driven planning
policies in build-to-order automobile production. International Journal of Production
Economics, 2011. 131(1): p. 183-193.
[62] Howard, M., P. Powell, and R. Vidgen, Automotive industry information systems:
from mass production to build-to-order. Journal of Cases on Information Technology
(JCIT), 2005. 7(2): p. 16-30.
[63] Zijm, W.H., Towards intelligent manufacturing planning and control systems. OR-
Spektrum, 2000. 22(3): p. 313-345.
[64] Melchiors, P., R. Leus, S. Creemers, and R. Kolisch, Dynamic order acceptance and
capacity planning in a stochastic multi-project environment with a bottleneck
resource. International Journal of Production Research, 2018. 56(1-2): p. 459-475.
[65] Hahn, C.K., E.A. Duplaga, and J.L. Hartley, Supply-chain synchronization: lessons
from Hyundai Motor Company. Interfaces, 2000. 30(4): p. 32-45.
[66] Fine, C.H., Clockspeed: Winning industry control in the age of temporary advantage.
2010: ReadHowYouWant. com.
[67] Pashaei, S. and J. Olhager, The impact of global operations on product architecture:
an exploratory study. International Journal of Operations & Production Management,
2017.
[68] Voordijk, H., B. Meijboom, and J. de Haan, Modularity in supply chains: a multiple
case study in the construction industry. International journal of operations &
production management, 2006.
[69] Wang, J., Q. Chang, G. Xiao, N. Wang, and S. Li, Data driven production modeling
and simulation of complex automobile general assembly plant. Computers in Industry,
2011. 62(7): p. 765-775.
[70] Bihlmaier, R., A. Koberstein, and R. Obst, Modeling and optimazing of strategic and
tactical production planning in the automotive industry under uncertainty, in Supply
Chain Planning. 2009, Springer. p. 1-26.
[71] Tomino, T., Y. Park, P. Hong, and J.J. Roh, Market flexible customizing system
(MFCS) of Japanese vehicle manufacturers: An analysis of Toyota, Nissan and
Mitsubishi. International Journal of Production Economics, 2009. 118(2): p. 375-386.
[72] Stevens, S.S., On the psychophysical law. Psychological review, 1957. 64(3): p. 153.
[73] Bowersox, D.J., T.P. Stank, and P.J. Daugherty, Lean launch: managing product
introduction risk through response‐based logistics. Journal of Product Innovation
Management, 1999. 16(6): p. 557-568.
[74] Gunasekaran, A., The build‐to‐order supply chain (BOSC): a competitive strategy for
21st century. Journal of Operations Management, 2005. 23(5): p. 419-422.
[75] Jahangirian, M., T. Eldabi, A. Naseer, L.K. Stergioulas, and T. Young, Simulation in
manufacturing and business: A review. European Journal of Operational Research,
2010. 203(1): p. 1-13.
[76] Pires, B., J. Goldstein, D. Higdon, P. Sabin, G. Korkmaz, S. Shipp, S. Keller, S. Ba,
K. Hamall, and A. Koehler. A Bayesian simulation approach for supply chain
synchronization. in 2017 Winter Simulation Conference (WSC). 2017. IEEE.
[77] Glover, F., J.P. Kelly, and M. Laguna. New advances for wedding optimization and
simulation. in WSC’99. 1999 Winter Simulation Conference Proceedings.’Simulation-
A Bridge to the Future’(Cat. No. 99CH37038). 1999. IEEE.
[78] Ladier, A.-L., A. Greenwood, and G. Alpan. Modeling issues when using simulation
to test the performance of mathematical programming models under stochastic
conditions. in Proceedings of the European Simulation and Modelling Conference
2015. 2015.
[79] Swisher, J.R., P.D. Hyden, S.H. Jacobson, and L.W. Schruben. A survey of simulation
optimization techniques and procedures. in 2000 Winter Simulation Conference
Proceedings (Cat. No. 00CH37165). 2000. IEEE.
[80] Rosen, S.L. and C.M. Harmonosky, An improved simulated annealing simulation
optimization method for discrete parameter stochastic systems. Computers &
Operations Research, 2005. 32(2): p. 343-358.
[81] Ghasemi, A., C. Heavey, and G. Laipple. A review of simulation-optimization
methods with applications to semiconductor operational problems. in 2018 Winter
Simulation Conference (WSC). 2018. IEEE.
[82] Soroush, H., S.M. Sajjadi, and S.M. Arabzad, Efficiency analysis and optimisation of
a multi-product assembly line using simulation. International Journal of Productivity
and Quality Management, 2014. 13(1): p. 89-104.
[83] Ólafsson, S. and J. Kim. Simulation optimization. in Proceedings of the winter
simulation conference. 2002. IEEE.
[84] Kelton, W.D., Simulation with ARENA. 2002: McGraw-hill.
[85] Glover, F., J.P. Kelly, and M. Laguna. New advances and applications of combining
simulation and optimization. in Proceedings of the 28th conference on Winter
simulation. 1996. IEEE Computer Society.
[86] Thomé, A.M., R.S. Sousa, and L.F. do Carmo, Complexity as contingency in sales
and operations planning. Industrial Management & Data Systems, 2014.
[87] Kaipia, R., J. Holmström, J. Småros, and R. Rajala, Information sharing for sales and
operations planning: Contextualized solutions and mechanisms. Journal of Operations
Management, 2017. 52: p. 15-29.