Available online at www.sciencedirect.

com

ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science 00 (2022) 000–000
www.elsevier.com/locate/procedia
ScienceDirect
Procedia Computer Science 217 (2023) 376–385

4th International Conference on Industry 4.0 and Smart Manufacturing

A new procedure for spare parts inventory management in ETO

production: a case study
Marta Rinaldia *, Marcello Feraa , Roberto Macchiarolia and Eleonora Bottanib
a
University of Campania "Luigi Vanvitelli", via Roma 29, Aversa, 81031, Italy
b
University of Parma, Parco Area delle scienze 181/A, Parma, 43124, Italy

Abstract

Spare parts inventory management represents a strategic but critical process in many manufacturing companies, which have to
handle a high amount of different items. This work aims at developing a simulation model to manage the inventory level of spare
parts of an ETO (engineering to order) company. First, heterogeneous items have been analysed and classified into four different
categories. Later, a simulation model has been developed to reproduce the inventory policy currently used to monitor and control
them. Finally, a new procedure has been implemented to improve the current inventory management. Results show that the new
procedure allows for reducing both the total cost and the maximum amount of space needed to stock the items. In addition, the
simulation model could be used by practitioners as a practical tool to identify a proper inventory strategy according to the
characteristics of the different items.
© 2022 The Authors. Published by Elsevier B.V.
© 2022 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
(https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 4th International Conference on Industry 4.0 and Smart
Peer-review
Manufacturingunder responsibility of the scientific committee of the 4th International Conference on Industry 4.0 and Smart
Manufacturing
Keywords: spare parts; inventory management; spare parts classification; simulation; case study

1. Introduction

Spare parts refer to service parts, replacement parts, or repair parts that are used to support maintenance and repair
operations. The spare parts supply chain differs from the other supply chains for different reasons, mainly the

* Corresponding author. Tel.: +39 081 5010415; fax: +39 081 5010283.
E-mail address: marta.rinaldi@unicampania.it

1877-0509 © 2022 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license
(https://creativecommons.org/licenses/by-nc-nd/4.0)

Peer-review under responsibility of the scientific committee of the 4th International Conference on Industry 4.0 and Smart Manufacturing

1877-0509 © 2022 The Authors. Published by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 4th International Conference on Industry 4.0 and Smart
Manufacturing
10.1016/j.procs.2022.12.233
 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385 377
2 Author name / Procedia Computer Science 00 (2022) 000–000

uncertainty and the variability of the demand, combined with the high service level required [1]. In addition, managing
spare parts means monitoring and handling a high amount of heterogeneous items, usually differing in the procurement
lead times, lot sizes, and safety stocks. These characteristics make the forecast and inventory management of spare
parts very hard [2]. Consequently, companies often solve the problem of maintaining high inventory levels, which
leads to facing high inventory costs [3]. In this context, the identification of efficient inventory policies can help
managers in improving the system by reducing both the inventory level and the out-of-stock.
Thus, it is not surprising that managing the trade-off between the inventory costs and the service level of spare parts
is not a new topic; however, a gap between theory and practice still exists [4]. In particular, researchers are seeking
the optimization of the spare parts inventory management, i.e., the definition of an inventory policy leading to improve
the management of spare parts, by decreasing the inventory level and the related cost [5]. To this extent, the problems
addressed by the spare parts literature mainly focus on three topics [6]: 1) modelling and forecasting the demand for
spare parts, 2) classifying spare parts and determining criticality, and 3) developing stock-control techniques and
inventory policies. This paper faces the last point.
Furthermore, different research methodologies have been proposed to facilitate spare parts management, including
exact models, optimization, and simulation methods [7]. In general, it is becoming increasingly difficult to develop
analytic models to support spare parts management because of the increasing complexity of spare parts supply
systems. For such reason, simulation has been widely used for solving complex problems in spare parts management
and it can be considered a very promising method to design and plan the operations of manufacturing systems and
handle the allocation of the item [8].
Starting from the considerations above, this work aims at developing a simulation model to analyse the inventory
level of the spare parts of an ETO (engineering to order) company and to improve their inventory management. First,
the items have been analysed and classified according to their criticality to identify the relevant items, and then
according to the characteristics of their demand pattern. Later, a simulation model has been developed to reproduce
the inventory policy currently used to manage the relevant items. Finally, the model has been used to define the optimal
setting for each relevant item; each setting has been designed to improve the inventory management by defining a
proper combination of parameters. The results provided for each item have been elaborated to derive some
considerations about the relation between the optimal setting and the spare parts classification.
The proposed approach allows answering the key questions typical of spare parts management, i.e. which items
have to be stocked, when to order them and how many items to order. This model differs from the other models
available in the literature since the potential of the simulation was exploited to evaluate the specific characteristics and
constraints of each item before providing the optimal setting. Thus, this approach allows handling the characteristics
of any demand pattern and it can propose a solution considering the real requests and the practical constraints.
The contribution of this study is twofold. First, from a theoretical point of view, this paper aims at finding a proper
inventory policy to apply to different categories of spare parts, according to their characteristics. It is important to
underline that the proposed method has been validated through empirical data. Although many methods have been
proposed for managing spare parts, most of them have been tested using only example data extracted from other
contributions, and they have not been validated through empirical databases [4]. In addition, such work has also a
practical implication. The simulation tool could be used by practitioners to find a solution for real-life applications and
it could help managers to organize the inventory management of spare parts in ETO productions.
The rest of the paper is structured as follows. The next section revises the spare parts inventory management
literature focusing on studies that have proposed simulation models. Section 3 proposes the methodological approach
and describes the main logic of the simulation model. In Section 4, the numerical outcomes are presented and
discussed, and the last section summarizes the main considerations and delineates possible future developments.

2. Literature review

In the past few decades, simulation has been recognized as a very useful technique in the field of operations
management and it has been widely used for managing spare parts, mainly for modelling and evaluating inventory
policies.
Persson and Saccani [9] developed a simulation model to help practitioners to evaluate the spare parts inventory
and distribution management. The simulation was used to identify the ability of a real company to transport and stock
378 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385
Author name / Procedia Computer Science 00 (2022) 000–000 3

spare parts; different scenarios have been simulated and the costs of the supply chain structure have been compared.
Van Jaarsveld and Scheller-Wolf [10] combined a heuristic procedure and a simulation model for improving the
inventory management of an assemble-to-order (ATO) system. Holding costs and backorder costs have been
considered and estimated through simulation to compare different inventory policies. The spare parts inventory
management of a real system was studied by Sun [11]; the discrete event simulation was combined with the Pareto
analysis to reproduce and improve the spare parts inventory according to the ABC classification. An inventory model
integrated with the ABC analysis has been developed by Sathish et al. [12] too. The model aimed to propose a method
for spare parts inventory management to reduce the space used to stock items, and find a trade-off between the
reduction of the inventory level and the elimination of the stock outs. An inventory model based on forecast and
simulation has been presented by Diaz et al. [13] to reduce transportation, storage, and production costs by studying
the demand pattern and considering the application of a proper inventory management policy. The proposed
methodology is based on real data from the automotive industry as well as real needs and constraints have been
implemented.
Recently, the introduction of Additive manufacturing (AM) technology in the spare parts supply chain has started
to be studied [14]. The system dynamic simulation has been proposed by Li et al. [15] to compare different scenarios
and evaluate both the economic and the environmental benefits due to the adoption of AM in spare parts supply chains.
Manufacturing, distribution, and inventory processes have been included in the analysis, however, the demand for
spare parts has been generated simply by increasing the standard deviation but without considering all the aspects that
characterize spare parts and make their management very hard (e.g. the variety of items and the unpredictability of
the demand interval). A hybrid simulation approach, combining agent-based and discrete event simulation methods
has been developed by Xu et al. [16] for investigating spare parts supply chain management. Different structures have
been considered and the performance of the system has been evaluated considering the total cost, the lead time, and
the vulnerability of the system under disruptions. In this paper, an expected demand was estimated starting from a
forecasted amount without investigating the management of different items with different characteristics. A
heterogeneous demand was modelled by Li et al. [17], who developed a discrete event simulation model for a make-
to-order (MTO) system and compared the impacts of different configurations on multiple performance dimensions.
In this case, different demand arrival rates have been tested, but the high variability in terms of quantity was not
implemented. In addition, the authors have analysed only six items, while a spare parts supply chain usually has to
handle and monitor a huge amount of different items. McDermott et al. [3] were the first who linked a realistic spare
part demand characterization to AM supply chain design using Monte Carlo simulation. The proposed approach
generates the historical demand according to the specific characteristics of four demand categories, which differ in
the volatility and the intermittency of the demand. Then an appropriate inventory policy is identified according to
the demand pattern, and the performances of the system are evaluated and compared.
Starting from the analysis of the literature, it is clear how the simulation results are a useful tool to improve spare
parts management because it allows testing of different alternatives before implementing them. However, to offer a
solution for real-life applications, the different features of spare parts have to be considered. The current literature has
proposed strategic analysis useful to solve problems at a tactical but theoretical level. The approach proposed in this
paper can find a solution for real-life systems that have to manage a huge amount of spare parts with different
characteristics and it could help managers to organize their inventory management.

3. Methodology

The proposed approach consists of two sequential steps: 1) the classification of spare parts and 2) the development
of a simulation model to reproduce the inventory policy and identify the optimal setting for the relevant items. The
classification of items has been structured into two sequential phases. In particular, the first phase allows to identify
the relevant items, i.e. those essential for the company that needs to be stocked with a specific strategy and that cannot
be managed with a just-in-time policy. Once defined the items' criticality, the relevant items have been grouped into
four categories according to the characteristics of their demand. This second phase has been inserted to understand if
it is possible to derive some general guidelines and define a particular setting to apply to the different categories of
spare parts. Then, a simulation model has been designed to reproduce the current inventory management of the
relevant items, implementing the current inventory policy with the related parameters. Later, the model has been run
 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385 379
4 Author name / Procedia Computer Science 00 (2022) 000–000

varying the parameters to find the combination that returns the minimum total cost, considering also the specific
constraints of each item. In this way, the model is able to propose a suitable solution for a real-world application.
Following, the nomenclature used to describe the model is introduced.

Nomenclature

N simulation duration [days]

EOQ Economic Order Quantity [units]

OP Order Point [units]
LT Procurement lead time [days]
O Quantity Ordered [units/day]
Norder Total number of orders placed during the simulation time [orders]
I Physical inventory level [units/day]
Imax Maximum amount of items on the shelf - constraint of the system [units]
Imaxs Maximum amount of items on the shelf - outcome of the simulation model [units]
d Daily demand of the downstream player [units/day]
co Unitary order cost [€/order]
cs Unitary cost of holding stocks [€/unit/day]
cso Unitary cost of stock-out [€/unit/day]

t Simulation day
i item

3.1. Spare parts classification

As already explained, the classification of items has been structured into two sequential phases. The first
classification has been carried out with the aim of identifying the relevant items, which have to be managed with a
proper inventory strategy. A multiple criteria ABC classification has been chosen to capture the different dimensions
of spare parts management. For decades, this approach has been widely used in the literature [18, 19].
Three criteria have been identified, according to the ETO context in which the study has been conducted:
1. Annual cost usage: it is calculated as the spare part cost multiplied by the demand volume.
2. Obsolescence: it reflects the number of handling and it is calculated considering the count of the day in
which at least one picking operation has been observed.
3. Volume: it is computed as the sum of the number of spare parts handled.
For each criterion, 20% of items cover 80% of the total value (respectively cost, number of handling, and quantity)
following in class A. Thus, the spare parts have been classified three times, according to the three criteria. Then, a
final classification has been provided considering the following rules:
• Class A+: spare parts belonging to three classes A
• Class A: spare parts belonging to two classes A
• Class B: spare parts belonging to one class A
• Class C: spare parts belonging to no class A
Finally, only the spare parts belonging to class A+ are considered relevant, i.e., each item needs the definition of a
proper inventory strategy with a proper set of parameters, while spare parts belonging to class A and B are considered
important, but the whole class could be managed according to the same strategy. Finally, spare parts belonging to
class C do not need the definition of an inventory policy, but they can be managed by a just-in-time policy.
The second classification concerns only the relevant spare parts (A+), and it has been designed to group the items
considering the characteristics of their demand. To this extent, the well-known classification proposed by Syntetos et
al. [20] has been applied. This classification allows the grouping of the spare parts into four classes by calculating the
average demand interval (ADI) and the square coefficient of variation of demand sizes (CV2).
380 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385
Author name / Procedia Computer Science 00 (2022) 000–000 5

The four categories are:

1. Smooth: low variability of the demand size and regular frequency of the demand.
2. Intermittent: low variability of the demand size and irregular frequency of the demand.
3. Erratic: high variability of the demand size and regular frequency of the demand.
4. Lumpy: high variability of the demand size and irregular frequency of the demand.
As already asserted, this second classification has been inserted to study the possibility to apply different inventory
strategies to different classes of spare parts.

3.2. Simulation model

The second step of the study consists of the development of a simulation model to reproduce the inventory policy
used to manage the relevant spare parts (A+) in an ETO context. A discrete event simulation model has been designed
using the computational potential of Microsoft Excel. The aim of the simulation model is to answer two key questions,
i.e. when to order and how many items to order. In addition, the model is able to provide a specific answer for each
item, allowing implementing of an ad hoc solution. This aspect is very important since each item has different and
very specific characteristics that make the identification of a general strategy very hard.
First, the simulation model has been developed to reproduce the inventory policy used to currently manage the
relevant items. The input data for the simulation model is the daily demand of the i-th item. It is important to underline
that the historical data of the last two years have been used to reproduce the behaviour of the system, without any
forecast or hypothesis on the distribution of the demand. The reason is the particular pattern of the spare parts demand,
which makes the generation of behaviour that reflects the real characteristics of the demand very difficult.
The company currently adopts an EOQ policy, thus the following rules and formulations have been inserted into
the model. At the beginning of each day, the demand is checked and the model determines whether the available
inventory is sufficient to fulfil such demand, i.e., whether 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡−1 ≥ 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 . Whenever the order cannot be fulfilled, the
model will identify an out-of-stock situation for 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 − 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡−1 ; that quantity will be back-ordered, and a stock-out cost
will be computed. This cost reflects the customer dissatisfaction due to the unavailability of items against demand,
and the unitary cost of stock-out is quantified starting from the economic value of the item [21]. Overall, the total
stock-out cost will be calculated as follows:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 = ∑𝑁𝑁𝑁𝑁
𝑡𝑡𝑡𝑡=1 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 ∙ � 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 − 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡−1 � (1)

To avoid warm-up effects of the simulation, it is assumed that the system owns an initial (t = 0), not null inventory,
obtained as the 10% of the OPi value (which, in turn, is different for each item). Once the order has been fulfilled, or
the quantity has been backlogged, the model updates the inventory level and applies the reorder process:

𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 = 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡−1 − 𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 + 𝑂𝑂𝑂𝑂𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡−𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 (2)

Then, the model checks whether the inventory level is lower than the order point, i.e. 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 ≤ 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑖𝑖𝑖𝑖 . Whenever the
stock is found to be lower than the order point, an order is placed to the supplier (𝑂𝑂𝑂𝑂𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 = 𝐸𝐸𝐸𝐸𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 ); otherwise, no orders
are placed. The total order cost can be computed as follows:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑑𝑑𝑑𝑑𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 ∙ 𝑁𝑁𝑁𝑁𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 (3)

Where the number of orders is a direct outcome of the simulation and it is computed as the total number of orders
placed during the simulation time.
The cost of holding stocks is computed as follows:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 = ∑𝑁𝑁𝑁𝑁
𝑡𝑡𝑡𝑡=1 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 ∙ 𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖,𝑡𝑡𝑡𝑡 (4)

Finally, the total cost of inventory management for the i-th item accounts for:
 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385 381
6 Author name / Procedia Computer Science 00 (2022) 000–000

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑑𝑑𝑑𝑑𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 (5)

Once developed the model, the current inventory management of each relevant item has been simulated (AS-IS
scenario). To this end, for each item, the company has provided the value of the two parameters that characterize the
inventory policy, i.e. EOQ and OP, and the data related to the historical demand, the unitary costs, and the LT. Then,
the results of the AS-IS scenario have been compared to the real outcomes to validate the model, before developing
the TO-BE scenario. In particular, the new scenario aims at identifying the optimal combination of parameters, i.e.
the one that provides the minimum total cost of inventory management. A particular procedure has been implemented
with the purpose of varying EOQ and OP meanwhile checking the constraints of each item. The structure of the
procedure is shown in Figure 1. The procedure is iterated until all the items have been simulated and the best
combination of parameters has been identified.

Fig. 1. TO-BE scenario - scheme of the implemented procedure.

382 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385
Author name / Procedia Computer Science 00 (2022) 000–000 7

The first step of the procedure is the data entry for each item analysed (blue box). Then, the range of values of EOQ
and OP is defined (green box): one operating leverage (OP) was expressed as a function of the k-th percentile and it
depends on the historical demand. In particular, once the value of k is set, the procedure puts in ascending order all
the quantity of the historical demand and returns the value that corresponds to the k-th percentile. Thus, the higher is
k, the higher OP. The other operating leverage (EOQ) was varied according to the current lot size and expressed as a
multiple of it.
Then, for each set, two constraints are checked (yellow box):
1. The EOQ value has to be higher than the OP value.
2. The EOQ value has to be lower than the maximum amount of space assigned to the item for its storage
(Imax); this is a constraint of the real system.

If both the constraints are respected, the simulation model is run to calculate the total cost of the inventory
management for that specific combination of parameters, after verifying an additional constraint: the maximum
amount of space for stocking the items returned by the simulation (Imaxs) has to be lower than the maximum amount
of space assigned to the item for its storage (Imax).
Finally, the total cost of inventory management of each combination is obtained, as output, and a couple of
parameters, which minimize it, are identified (red box).

4. Results and discussion

The next sub-sections provide the results, distinguishing between the two steps of the proposed methodology.

4.1. Spare parts classification: results

The data of the last four years have been used to classify all the spare parts according to the multiple criteria ABC
and the Syntetos classifications. Figure 3 presents the results of the spare parts classification. Results show how 75%
of items should be managed by a just-in-time strategy (class C), while the remaining 25% (A+, A, B) needs the
definition of a proper inventory strategy. In addition, 6% of items (538 items belonging to class A+) should be
managed considering a proper strategy and a specific combination of parameters, according to the items’
characteristics and constraints. Thus, the second classification aims at grouping the 538 items belonging to class A+
into four categories, according to the characteristics of their demand. From Figure 3, it is interesting to note that the
erratic and the smooth classes are the most numerous, meaning that the most of items have a regular frequency of
demand (77%) but they differ in the variability of the demand size.

Fig. 2. Results of the spare parts classification.

 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385 383
8 Author name / Procedia Computer Science 00 (2022) 000–000

4.2. Simulation model: results

In the first analysis, the results of the AS-IS and TO-BE scenarios have been compared. The aim is to understand
if the best combination of parameters identified by the implemented procedure is able to provide better results with
respect to the current strategy adopted by the company. To this end, two different analyses have been performed to
test the ability of the new procedure to 1) reduce the total cost of inventory management, and 2) reduce the maximum
amount of space needed to stock the items. It is important to underline that the comparison is possible only if both the
models provide a feasible result and respect all the system’s constraints detailed in the previous section.
Results reported in Table 1 confirm that the new procedure could help managers in reducing the total cost in about
80% of cases, while 13% of items return a lower cost if managed with the current combination of parameters. For the
rest of the items (6%), the AS-IS and the TO-BE models provide the same outcome.
Concerning the amount of space needed to stock items, results show that the new procedure could improve the
inventory management for 61% of items, by reducing the maximum amount of stock.

Table 1. Simulation results: identification of the best scenario.

Key Performance Indicator AS-IS TO-BE Same result
Minimum 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 13% 81% 6%
Minimum Imaxs 33% 61% 6%

The second analysis of this section considers only the results provided by the TO-BE model. In particular, the aim
is to study the possibility to apply different inventory strategies to different classes of spare parts. As already explained,
the 538 items belonging to class A+ have been grouped into four categories, which differ in the variability of the
demand in terms of size and frequency. In addition, the new procedure implemented into the TO-BE model also sets
one operating leverage as a function of the demand by varying k, which represents the k-th percentile used to calculate
OP. Figure 3 shows the results of the simulation related to the identification of k, i.e. the k-th percentile that allows
returning the minimum total cost. The vertical axis represents the frequency count, i.e. the number of items that fall
into the related frequency category. Six frequency categories have been defined, covering all the k values tested (from
0,5 to 1 with a step of 0,1) plus one category that represents the number of items for which no feasible solutions have
been found (named “no”).

Fig. 3. Simulation results: OP setting as a function of k

 Author name / Procedia Computer Science 00 (2022) 000–000 9

384 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385

Results show that the TO-BE model is able to find a feasible solution for classes with a regular frequency of demand
(7% and 5% of no feasible solutions respectively for smooth and erratic). In addition, for both these classes, the higher
frequency is obtained by the 0.9 category (42% and 33% respectively) followed by the 0,5 category (27% and 28%
respectively). This means that for most items, the best strategy for improving inventory management is to increase the
value of OP, but, at the same time, it is clear how the choice of k strictly depends on the specific characteristics of the
items. Thus, it is not possible to define a general strategy, but the new procedure can help managers to identify a
proper strategy for each item.
Concerning lumpy and intermittent, outcomes show that it is more difficult to find a feasible solution, especially
for intermittent (51% of items return an unfeasible solution). This result is in line with the characteristics of these
special classes, whose inventory management is very hard. However, reducing the value of OP seems to be a good
strategy to manage their inventory and reduce the total cost, especially for lumpy (61% of items belong to the 0,5
category).
The last analysis aims at identifying what constraint does not allow the procedure to find a feasible solution.
Overall, the new procedure is not able to identify an acceptable solution for 14% of items (77 items over 538), and
the reasons are listed in Table 2.

Table 2. Simulation results: reasons for unfeasible solutions

Type of Constraint Number of items Percentage of items
𝐸𝐸𝐸𝐸𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 < 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑖𝑖𝑖𝑖 40 52%
𝐸𝐸𝐸𝐸𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 > 𝐼𝐼𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚,𝑖𝑖𝑖𝑖 7 9%
𝐼𝐼𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚,𝑖𝑖𝑖𝑖 > 𝐼𝐼𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚,𝑖𝑖𝑖𝑖 30 39%

5. Conclusion

This paper has proposed a simulation model for spare parts inventory management of an ETO company. First, the
items have been classified according to their criticality and demand pattern. Later, a simulation model has been
developed to reproduce the inventory policy used to currently manage the relevant items. Finally, a new procedure
integrating the simulation model has been developed to define the optimal setting of each relevant item; each set has
been identified with the aim of minimizing the total cost of inventory management by defining a proper combination
of operating leverages (EOQ and OP).
The proposed approach allows answering the key questions typical of spare parts management, i.e. which items
have to be stocked, when to order them and how many items to order. In addition, such work has also a practical
implication. In fact, the simulation tool could be used by practitioners to find a solution for real-life applications and
it could help managers to organize the inventory management of spare parts in ETO productions. In fact, the results
show that the new procedure is able to improve the system by reducing both the total cost of inventory management
and the maximum amount of space needed to stock the items.
Furthermore, some considerations have been derived about the possibility to define a general strategy for improving
the inventory management of items that have the same demand pattern. In particular, the TO-BE model is able to find
a feasible solution for classes with a regular frequency of demand even though the best setting strictly depends on the
specific characteristics of the single item. On the contrary, finding a feasible solution for classes with irregular
frequency results to be very hard, especially for intermittent. However, reducing the value of OP seems to be a good
solution to manage such items and reduce the total cost.
In future developments, a comparison of the proposed model results with existing methods could be carried out,
and other inventory policies could be tested to evaluate the possibility to improve the spare parts inventory
management and reduce the total cost.
 Marta Rinaldi et al. / Procedia Computer Science 217 (2023) 376–385 385
10 Author name / Procedia Computer Science 00 (2022) 000–000

Acknowledgments

The authors wish to express their thanks to the Company where the present study was conducted.

References

[1] Saalmann, P., Wagner, C., & Hellingrath, B. (2016). Decision support for a spare parts supply chain coordination problem: designing a tactical
collaborative planning concept. IFAC-PapersOnLine, 49(12), 1056-1061.
[2] Syntetos, A. A., Keyes, M., & Babai, M. Z. (2009). Demand categorisation in a European spare parts logistics network. International journal of
operations & production management.
[3] McDermott, K. C., Winz, R. D., Hodgson, T. J., Kay, M. G., King, R. E., & McConnell, B. M. (2021). Performance tradeoffs for spare parts
supply chains with additive manufacturing capability servicing intermittent demand. Journal of Defense Analytics and Logistics.
[4] Hu, Q., Boylan, J. E., Chen, H., & Labib, A. (2018). OR in spare parts management: A review. European Journal of Operational Research,
266(2), 395-414.
[5] Armenzoni, M., Montanari, R., Vignali, G., Bottani, E., Ferretti, G., Solari, F., & Rinaldi, M. (2015). An integrated approach for demand
forecasting and inventory management optimisation of spare parts. International Journal of Simulation and Process Modelling, 10(3), 233-240.
[6] Bacchetti, A., & Saccani, N. (2012). Spare parts classification and demand forecasting for stock control: Investigating the gap between research
and practice. Omega, 40(6), 722-737.
[7] Zhang, S., Huang, K., & Yuan, Y. (2021). Spare parts inventory management: A literature review. Sustainability, 13(5), 2460.
[8] Mourtzis, D. (2020). Simulation in the design and operation of manufacturing systems: state of the art and new trends. International Journal of
Production Research, 58(7), 1927-1949.
[9] Persson, F., & Saccani, N. (2009). Managing the after-sales logistic network–a simulation study. Production Planning and Control, 20(2), 125-
134.
[10] Van Jaarsveld, W., & Scheller-Wolf, A. (2015). Optimization of industrial-scale assemble-to-order systems. INFORMS Journal on Computing,
27(3), 544-560.
[11] Sun, X. (2013). Simulations on the Spare Parts Inventory Management of Equipment Maintenance in manufacturing enterprise based on
Flexsim. In Advanced Materials Research (Vol. 712, pp. 3181-3186). Trans Tech Publications Ltd.
[12] Sathish N, Anbuudayasankar S, Deepan M, Narassima M. Procurement Methodologies to Optimize the Inventory Levels of Spare Parts.
International Journal of Recent Technology and Engineering. 2019; 8(3).
[13] Diaz, D. A. B., Hennequin, S., & Roy, D. (2019). Spare parts management in the automotive industry considering sustainability. In World
Congress on Global Optimization (pp. 1109-1118). Springer, Cham.
[14] Abbate, R., Manco, P., Macchiaroli, R., & Lambiase, A. (2022, August). A Survey on the Integration between Additive Manufacturing and
Maintenance. In Macromolecular Symposia (Vol. 404, No. 1, p. 2100452).
[15] Li, Y., Jia, G., Cheng, Y., & Hu, Y. (2017). Additive manufacturing technology in spare parts supply chain: a comparative study. International
Journal of Production Research, 55(5), 1498-1515.
[16] Xu, X., Rodgers, M. D., & Guo, W. G. (2021). Hybrid simulation models for spare parts supply chain considering 3D printing capabilities.
Journal of Manufacturing Systems, 59, 272-282.
[17] Li, Y., Cheng, Y., Hu, Q., Zhou, S., Ma, L., & Lim, M. K. (2019). The influence of additive manufacturing on the configuration of make-to-
order spare parts supply chain under heterogeneous demand. International journal of production research, 57(11), 3622-3641.
[18] Flores, B. E., & Whybark, D. C. (1986). Multiple criteria ABC analysis. International Journal of Operations & Production Management.
[19] Hu, Q., Chakhar, S., Siraj, S., & Labib, A. (2017). Spare parts classification in industrial manufacturing using the dominance-based rough set
approach. European Journal of Operational Research, 262(3), 1136-1163.
[20] Syntetos, A. A., Boylan, J. E., & Croston, J. D. (2005). On the categorization of demand patterns. Journal of the operational research society,
56(5), 495-503.
[21] Bottani, E., Ferretti, G., Montanari, R., & Rinaldi, M. (2014). Analysis and optimisation of inventory management policies for perishable food
products: a simulation study. International Journal of Simulation and Process Modelling 11, 9(1-2), 16-32.