Article

Improving Equipment Maintenance—Switching from Corrective

to Preventative Maintenance Strategies
Jye West 1, * , Milind Siddhpura 1 , Ana Evangelista 1 and Assed Haddad 2

1 Structural & Civil Engineering, Engineering Institute of Technology, West Perth, WA 6005, Australia;
[email protected] (M.S.); [email protected] (A.E.)
2 Programa de Engenharia Ambiental, Universidade Federal do Rio de Janeiro, Rio de Janeiro 21941-617, Brazil;
[email protected]
* Correspondence: [email protected]

Abstract: This paper explores different building maintenance strategies in commercial buildings
in Sydney, Australia, focusing on corrective maintenance (CM) and preventive maintenance (PM).
While CM involves rectifying issues after they occur, PM aims to enhance productivity by antici-
pating potential issues. Although PM seems more logical, the decision to implement this type of
maintenance strategy are typically made based on item reliability, failure frequency, and downtime
cost, commonly found in manufacturing facilities or critical environments. However, as found in
the selected/surveyed commercial real estate buildings, CM was more frequently adopted in aged
facilities with older infrastructure, and PM was favoured for buildings without structural deficiencies;
however, operating equipment failures were common. However, in many cases, decision makers
did not consider the broader effects of downtime beyond direct financial losses, costs associated
with customer satisfaction, worker efficiency, rent abatements, and reputation damage. While each
building is unique and may require a bespoke maintenance schedule, this study’s insights may help
managers select the most appropriate maintenance strategy. Nonetheless, further research is needed
to investigate the role of innovative technologies (such as machine learning and artificial intelligence)
in enhancing maintenance efficacy and explore the influences of economic shifts, corporate and
financial objectives, and the availability of technical resources.
Citation: West, J.; Siddhpura, M.;
Evangelista, A.; Haddad, A. Keywords: building maintenance strategies; corrective maintenance (CM); downtime costs;
Improving Equipment preventive maintenance (PM); technological approaches
Maintenance—Switching from
Corrective to Preventative
Maintenance Strategies. Buildings
2024, 14, 3581. https://doi.org/ 1. Introduction
10.3390/buildings14113581
While various building maintenance strategies are utilised in Australia—such as
Academic Editor: Pedro Delgado risk-based, condition-based, corrective, and preventive—their shared aim is to optimise
infrastructure performance and lifespan and to optimise the building’s useability, reliability,
Received: 13 October 2024
and safety [1]. To achieve these goals, facility managers must establish maintenance strate-
Revised: 1 November 2024
Accepted: 5 November 2024
gies aligning with the stakeholders’ goals and specific operational objectives. This study
Published: 11 November 2024
provides a comprehensive review of corrective and preventive strategies most frequently
adopted in the building or facility management industry, as shown in Figure 1. Building
maintenance is critical for optimising the condition of these assets for current use and
preserving them for future generations [2]. This typically necessitates preventive mainte-
Copyright: © 2024 by the authors. nance (PM), thus addressing minor faults before they escalate into significant problems [3].
Licensee MDPI, Basel, Switzerland. The variety of strategies equips organisations with many options, fostering a sense of
This article is an open access article knowledge and empowerment across the more expansive facility and allied industries.
distributed under the terms and On the other hand, corrective maintenance (CM) aims to rectify problems that have al-
conditions of the Creative Commons
ready occurred, focusing on efficiency and minimal downtime and disruptions to clients [3].
Attribution (CC BY) license (https://
The selection of PM versus CM is typically determined using a decision matrix based on
creativecommons.org/licenses/by/
the equipment’s reliability, failure frequency, and downtime costs. This decision is guided
4.0/).

Buildings 2024, 14, 3581. https://doi.org/10.3390/buildings14113581 https://www.mdpi.com/journal/buildings

Buildings 2024, 14, 3581 2 of 16

by the unique building characteristics and the organisational objectives, assets, financial
position, and available resources [4]. The influence of these factors on the choice of PM or
CM is explored in this work while aiming to establish if maintenance procedures can be
more efficiently scheduled and their outputs more objectively evaluated through software
tools. Commercial Building Maintenance is critical to commercial structures’ longevity,
efficiency, and safety. The operational costs, tenant satisfaction, and commercial asset values
are essential to assess. In this approach, the study studies corrective maintenance (CM) and
preventive maintenance (PM) approaches within commercial buildings in NSW, Australia.
While CM finds out the problem after the problem, PM works proactively to avoid problems
before they worsen, cutting downtime and allocating resources optimally. However, decid-
Buildings 2024, 14, x FOR PEER REVIEW 2 of 17
ing which strategy is most effective is, at best, a delicate trade-off between item reliability,
downtime cost, failure frequency, and unique building asset corporate requirements.

Figure1.1.Maintenance
Figure Maintenancestrategy
strategymatrix.
matrix.

This assessment
On the other hand,examines howmaintenance
corrective Python programming
(CM) aims to can be used
rectify withinthat
problems Google
have
Colab
already occurred, focusing on efficiency and minimal downtime and disruptions toup
to support maintenance decision making better. This effort seeks to speed CM
clients
and PM through
[3]. The selection data-driven
of PM versusinsights. Facilitydetermined
CM is typically managers can usingalso utilise these
a decision tools
matrix to
based
optimise critical maintenance schedules, reduce unexpected failures, and
on the equipment’s reliability, failure frequency, and downtime costs. This decision is make better
decisions.
guided byThen, this research
the unique attempts
building to offer guidance
characteristics for maintenance
and the organisational managers
objectives, in a
assets,
building management environment of change, showing how tools for data analysis
financial position, and available resources [4]. The influence of these factors on the choice can
augment building operational resilience and support choosing cost-effective
of PM or CM is explored in this work while aiming to establish if maintenance procedures maintenance
strategies.
can be more Preserving
efficientlyinfrastructure
scheduled andquality and functionality
their outputs has long
more objectively depended
evaluated throughon
building maintenance. Maintenance approaches have historically evolved considerably,
software tools. Commercial Building Maintenance is critical to commercial structures’ lon-
from once used to safeguard ancient Egyptian pyramids (routine inspections) to modern
gevity, efficiency, and safety. The operational costs, tenant satisfaction, and commercial
preventive systems to minimise failures [5]. Two primary strategies dominate today’s
asset values are essential to assess. In this approach, the study studies corrective mainte-
maintenance landscape: corrective maintenance (CM) and preventive maintenance (PM).
nance (CM) and preventive maintenance (PM) approaches within commercial buildings
While CM handles problems afterwards, the PM is geared towards regular inspections and
in NSW, Australia. While CM finds out the problem after the problem, PM works proac-
maintenance to avoid the significant problems so often.
tively to avoid problems before they worsen, cutting downtime and allocating resources
As digital tools become even more advanced, maintenance is undergoing a metamor-
optimally. However, deciding which strategy is most effective is, at best, a delicate trade-
phosis into a place where data analysis and automation play a role in moving towards a
off between item reliability, downtime cost, failure frequency, and unique building asset
more individualised maintenance process. In particular, Python programming has been
corporate requirements.
simulated to handle large datasets and perform real-time monitoring. With Google Colab,
This assessment examines how Python programming can be used within Google
facility managers have access to advanced statistical analysis, predicting equipment failures
Colab
and to support
optimising themaintenance
maintenancedecision
schedule. making
This isbetter. This effort
important seeks tourban
in a complex speedcontext
up CM
and PM through data-driven insights. Facility managers
where operational continuity and tenant satisfaction are paramount. can also utilise these tools to op-
timise critical maintenance schedules, reduce unexpected failures, and make better deci-
sions. Then,
Research this research attempts to offer guidance for maintenance managers in a build-
Questions
ing management
Given the study environment
objectivesof stated
change,above,
showing thehow tools forresearch
following data analysis can aug-
questions are
ment building
formulated: operational resilience and support choosing cost-effective maintenance
strategies. Preserving infrastructure quality and functionality has long depended on
building maintenance. Maintenance approaches have historically evolved considerably,
from once used to safeguard ancient Egyptian pyramids (routine inspections) to modern
preventive systems to minimise failures [5]. Two primary strategies dominate today’s
maintenance landscape: corrective maintenance (CM) and preventive maintenance (PM).
While CM handles problems afterwards, the PM is geared towards regular inspections
and maintenance to avoid the significant problems so often.
Buildings 2024, 14, x FOR PEER REVIEW 3 of 17

Research Questions
Buildings 2024, 14, 3581 3 of 16
Given the study objectives stated above, the following research questions are formu-
lated:
•• How can
How can Python
Pythonprogramming
programmingbebe utilised to to
utilised objectively evaluate
objectively the efficacy
evaluate of con-
the efficacy of
temporary building management strategies, mainly preventive and corrective
contemporary building management strategies, mainly preventive and corrective man-
agement?
management?
•• To what
To what extent
extent can
can the
the use
useofofPython
Pythonprogramming,
programming,especially
especiallyinin
thethe
Google
Google Colab
Co-
environment,
lab contribute
environment, to thetoseamless
contribute integration
the seamless and execution
integration of modern
and execution mainte-
of modern
nance approaches
maintenance withinwithin
approaches building management
building systems?
management systems?
Focus is
Focus is given
given to
to Python
Python programming
programming language
language andand the
the Google
Google Colab
Colab environment,
environment,
as software
as software tools
tools are
are increasingly
increasingly usedused in in different
different domains
domains to to automate
automate or or streamline
streamline
operational processes. Thus, their integration into the building
operational processes. Thus, their integration into the building maintenance decision maintenance decision
making is expected to yield beneficial results [6]. Moreover, as Python can process large
amounts of data in real real time
time and
and thus
thus proactively
proactively identify
identify maintenance
maintenance and and breakdown
breakdown
problems before they become more severe [7], it can can significantly
significantly reduce downtime by
incorporating Python into the maintenance processes and optimising resource allocation. allocation.
However, introducing
However, introducing an an interactive
interactive component
component to remote and collaborative maintenance
cloud-based Google
solutions requires the cloud-based Google Colab
Colab platform,
platform, necessitating
necessitating thatthat companies
companies
possess sufficient
possess sufficient IT
IT capabilities
capabilities and
and technical
technical knowledge
knowledge to benefit from from this approach
approach to
maintenance
maintenance fully.
fully.Their
Theirmaintenance
maintenanceteams teams need
needtoto
bebeable to analyse,
able understand,
to analyse, understand, andand
act
on
act maintenance
on maintenance data. As Google
data. As GoogleColab provides
Colab providesa shared
a sharedworkspace,
workspace, it enables
it enablesseamless
seam-
communication
less communication and recording of all of
and recording actions. This is
all actions. particularly
This usefuluseful
is particularly in a preventive main-
in a preventive
tenance environment,
maintenance wherewhere
environment, immediate actions
immediate are essential.
actions This model
are essential. can becan
This model extended
be ex-
to multiple
tended teams working
to multiple at different
teams working geographical
at different locations,locations,
geographical making itmaking
indispensable and
it indispen-
highly cost-effective for organisations with significant building portfolios.
sable and highly cost-effective for organisations with significant building portfolios.
However, PM may may only
only bebe suitable
suitable forfor some
some situations,
situations, as downtime
downtime cost, upkeep
frequency,
frequency, andand equipment
equipment dependability
dependability may may necessitate
necessitate CM CM adoption
adoption [8]. In practise,
[8]. In practise,
such
such decisions
decisions areare made
made by by considering
considering the the distinctive
distinctive characteristics
characteristics ofof an
an organisation’s
organisation’s
assets
assets and
and overall
overall corporate
corporate objectives
objectives while
while recognising
recognising thatthat all
all maintenance
maintenance methodsmethods
will
will inevitably evolve in response to changes to industry standards, corporate
inevitably evolve in response to changes to industry standards, objectives,
corporate objectives,
and
and technological
technological improvements
improvements [9]. Figures 22 and
[9]. Figures and 33 provide
provide examples
examples of of HVAC
HVAC Chiller
Chiller
Plantroom and Chilled Water Pumps that have completed their
Plantroom and Chilled Water Pumps that have completed their DLP (Defect Liability DLP (Defect Liability
Pe-
Period)
riod) and and
areare entering
entering the first
the first phase phase of routine
of routine maintenance,
maintenance, as theas the operators
operators have
have imple-
implemented
mented a CM aprocess
CM process
during during the early
the early phasephase
of theofasset
the asset lifecycle.
lifecycle.

Figure 2.
Figure 2. HVAC
HVAC chiller
chiller plant.
plant.

Organisations thus require a sophisticated yet flexible approach [10]. They should
conduct a cost and benefit analysis to determine the most optimal and cost-effective
maintenance plan for their unique building assets considering their corporate goals and
objectives. To demonstrate how this is achieved in practise, the incorporation of Python
tools and the cloud-based Google Colab platform into the contemporary building mainte-
nance methods adopted in Sydney, Australia, is examined in the following sections. The
findings can benefit other researchers and organisations aiming to identify the most im-
pactful variables driving the decision-making process regarding their assets and building
maintenance strategies.
Buildings 2024,
2024, 14,
14, 3581
x FOR PEER REVIEW 44of 17
of 16

Figure 3.
Figure 3. Chiller
Chiller water
water pumps.
pumps.

2. Materials and Methods

Organisations thus require a sophisticated yet flexible approach [10]. They should
conduct
The ahistory
cost and benefit maintenance
of building analysis to determine
can be traced theback
most to optimal and cost-effective
ancient civilisations, where
maintenance building
fundamental plan for their unique building
maintenance forms were assets considering
practised to ensuretheir
thecorporate
structuralgoals and
integrity
objectives.
of To demonstrate
architectural marvels. In how this isEgypt,
ancient achieved in practise,
workers the incorporation
regularly inspected andofrepairedPython
monumental structures such
tools and the cloud-based as pyramids
Google and temples
Colab platform to prevent
into the deterioration
contemporary building caused by
mainte-
environmental
nance methods factorsadopted like
in erosion
Sydney,and weathering.
Australia, Similarly,
is examined thefollowing
in the Romans implemented
sections. The
maintenance practises
findings can benefit to maintain
other researcherstheirand
roads, public buildings,
organisations aimingand infrastructure,
to identify the most there-
im-
fore recognising
pactful the importance
variables driving of a sustainable
the decision-making processmaintenance
regarding their practise to and
assets prolong the
building
longevity
maintenance of public infrastructure [11]. Throughout the medieval period, maintenance was
strategies.
a vital aspect of building management, particularly in the upkeep of castles, cathedrals,
and fortified and
2. Materials structures.
Methods However, it was not until the Industrial Revolution that building
maintenance
The history became a systematic
of building discipline.
maintenance can beWith
traced the arrival
back of mass
to ancient production
civilisations, and
where
urbanisation, regular maintenance became necessary to address the
fundamental building maintenance forms were practised to ensure the structural integrity challenges of rapidly
expanding citiesmarvels.
of architectural and theInassociated supporting
ancient Egypt, workers infrastructure [12]. Since
regularly inspected and then, building
repaired mon-
maintenance has evolved into a highly complex, multidisciplinary field encompassing
umental structures such as pyramids and temples to prevent deterioration caused by en-
preventive, corrective, and predictive strategies, supported by technological advance-
vironmental factors like erosion and weathering. Similarly, the Romans implemented
ments, sustainability, and the Internet of Things (IoT), providing rich data sources for
maintenance practises to maintain their roads, public buildings, and infrastructure, there-
ongoing analysis.
fore recognising the importance of a sustainable maintenance practise to prolong the lon-
The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)
gevity of public infrastructure [11]. Throughout the medieval period, maintenance was a
diagram in Figure 4 has been used to undertake a systematic review and visualise the
vital aspect of building management, particularly in the upkeep of castles, cathedrals, and
literature-screening process deployed in this research assignment. The flowchart illustrates
fortified structures. However, it was not until the Industrial Revolution that building
how researchers narrowed the initial search results to the final journals and publications
maintenance became a systematic discipline. With the arrival of mass production and ur-
in multiple screening stages. The process began with identification, where databases are
banisation, regular maintenance became necessary to address the challenges of rapidly
searched, and records are collected. Followed by screening, where duplicate records are
expanding cities and the associated supporting infrastructure [12]. Since then, building
removed, and titles/abstracts are reviewed. During eligibility assessment phase, full-text
maintenance
articles has evolved
are evaluated against into a highly
inclusion complex,
criteria. multidisciplinary
Finally, the included studies field are
encompassing
determined
preventive, corrective, and predictive strategies, supported
and documented. Each stage is quantified, showing the number of studies retained by technological advance-or
ments, sustainability,
excluded. and the Internet
This standardised approach of enhances
Things (IoT), providing rich
reproducibility and data sources
allows for on-
readers to
going analysis.
assess potential selection bias in the review process. The diagram’s structured format
Theconsistent
ensures PRISMA reporting
(PreferredacrossReporting Items
different for Systematic
systematic reviews Reviews and Meta-Analyses)
and meta-analyses, making
it easier for researchers to compare methodologies and evaluate the and
diagram in Figure 4 has been used to undertake a systematic review visualise the
thoroughness of
literature-screening
literature searches. process deployed in this research assignment. The flowchart illus-
tratesThis
howsystematic
researchers narrowed
review the initial
identified records, search results
journals, andto the final journals
publications from 87and publi-
databases
cations in multiple screening stages. The process began with
and sources including Google Scholar, Scopus, Academia, ResearchGate, and Semantic identification, where data-
bases are searched, and records are collected. Followed by screening,
Scholar and involved publications from Frontiers, Buildings, Emerald, Elsevier, IEEE, where duplicate rec-
ords are removed, and titles/abstracts are reviewed. During eligibility
Springer, Sage, and ScienceDirect. Title searches included corrective maintenance, preven- assessment phase,
full-text
tive articles are
maintenance, evaluated against
maintenance inclusion
strategies, AI andcriteria.
buildingFinally, the included
maintenance, BIM and studies are
building
determined and
maintenance, etc.documented.
The first levelEach stage is removed
of screening quantified, showing thedue
5 publications number of studies
to duplications,
retained or
resulting in excluded.
a balance of This standardised
82 records. approach
The second enhances
screening reproducibility
phase involved removing and allows
doc-
readers to
uments and assess potentialthat
publications selection
did not bias in thethe
address review process.
research theme Theordiagram’s structured
scope regarding the
format ensures consistent reporting across different systematic
building maintenance strategies. This resulted in 13 records being removed and a balance of reviews and meta-
69 remaining journals and publications. Finally, 16 records were removed as they were not
Buildings 2024, 14, 3581
Buildings 2024, 14, x FOR PEER REVIEW 55 of
of 16
17

considered academically appropriate or sufficiently peer-reviewed journals. The balance of

analyses, documents,
reference making it easier forand
journals, researchers to compare
publications methodologies
used in this and evaluate
research assignment the
consisted
thoroughness of literature
of 53 publications. searches.

Figure 4.
Figure 4. Prisma model for research
research assignment.
assignment.

Table 1 providesreview
This systematic examples of documents
identified records, scrutinised
journals, and during the literature
publications from 87 review,
data-
highlighting the emergence of computerisation and digital data.
bases and sources including Google Scholar, Scopus, Academia, ResearchGate, and Se-
mantic Scholar and involved publications from Frontiers, Buildings, Emerald, Elsevier,
Table Major reference
IEEE,1.Springer, Sage, journals and articles. Title searches included corrective maintenance,
and ScienceDirect.
preventive maintenance, maintenance strategies, AI and building maintenance, BIM and
Publication
Author building maintenance, etc. TheTitle
Article first level of screening removed 5 publications due to du-
Year
plications, resulting in a balance of 82 records. The second screening phase involved re-
K. Fraser moving documents and publications
Facilities management: the strategic selection of athat did not address
maintenance system the research theme 2014 or scope
[1]
D.S. Pugh
regarding the building maintenance strategies. This resulted in 13 records being removed
Maintenance
and a management
balance of 69 and regulatory
remaining compliance
journals and strategies
publications. Finally, 16 records 2003 removed
were
[12]
Fan et al. as they were
Deep learning not considered
and artificial intelligenceacademically appropriate
in sustainability: A review or sufficiently peer-reviewed jour-
of SDGs,
2023
[6] renewable energy, and environmental health.
nals. The balance of reference documents, journals, and publications used in this research
CIBSE Maintenance Engineering
assignment consistedand of
Management, Chartered Institution of Building
53 publications. 2018
[13] Services Engineers
Table 1 provides examples of documents scrutinised during the literature review,
highlighting the emergence of computerisation and digital data.
The quantitative research component included surveys and interviews with facil-
Table
ity 1. Major reference
managers, maintenance journals and articles.and maintenance engineering consultants. The
contractors,
collected data included the frequency of various maintenance techniques, the degree to
Publication
Author which they were effective, Article Title affecting the decision between PM and CM, and the
variables Year
respondents’ familiarity with Python and Google Colab. To supplement this information,
K. Fraser
annual maintenance
Facilities management: the strategic costs, budgets,
selection of arepair downtimes,
maintenance systemfailure rates, and system reliability
2014
[1]
statistics were subjected to statistical analysis to detect trends, patterns, and connections to
D.S. Pugh
Maintenance determine
management if PM
andorregulatory
CM wouldcompliance
be a better strategies
option for each case. To quantify item reliability,
2003
[12] mean time between failure (MTBF) was adopted [13]. At the same time, onsite maintenance
Fan et al. Deep learninglogsandorartificial intelligence
data retrieved in sustainability:
from the A review
CMMS were utilised of SDGs, the
to establish re- maintenance frequency
2023
[6] newable energy,fromandtheenvironmental
reference dataset health.
buildings. Moreover, downtime cost was calculated by applying
Buildings 2024, 14, 3581 6 of 16

the formulas for the direct costs of unplanned maintenance, including lost production and
labour costs over 12 months. This quantitative information was crucial in determining the
maintenance strategy that would be most optimal for each analysed site [14].

3. Data Analysis
Preventive maintenance (PM) practises ensure efficient operation, longevity, and safety
of built environments across various residential, commercial, industrial, and infrastructure
sectors. With the advancement of technology, increasing complexity of buildings, and
growing awareness of sustainability, preventive maintenance has developed into a sys-
tematic and proactive approach aimed at minimising breakdowns, reducing downtime,
and optimising asset performance. One of the critical principles of preventive mainte-
nance is the regular inspection and servicing of building systems and components before
they fail or malfunction. This proactive approach allows maintenance teams to promptly
identify potential issues, address them, and minimise the need for costly repairs or replace-
ments. For example, in HVAC systems, regular cleaning or replacing air filters, lubricating
bearings and rollers, and calibrating building automation controls can prevent energy
inefficiency and system failures and improve indoor air quality [15]. Preventive mainte-
nance practises increasingly leverage data-driven strategies and predictive analytics to
anticipate equipment failures and optimise maintenance schedules. Using sensors, IoT
devices, and predictive maintenance software, building owners and facility managers can
monitor equipment performance in real time, detect anomalies, and schedule maintenance
activities based on actual conditions rather than fixed intervals [16]. This proactive ap-
proach minimises downtime, maximises asset lifespan, and reduces maintenance costs. In
addition to technological advancements, the adoption of PM practised is usually driven
by regulatory requirements, evolving industry standards, and government sustainability
targets or objectives.
Conversely, corrective maintenance (CM) strategies focus on addressing equipment
failures or malfunctions after they occur, aiming to restore the required building func-
tionality and minimise downtime. Despite its reactive nature, corrective maintenance is
still frequently practised in building management, particularly for unexpected failures
or emergencies and non-critical or less essential equipment, where the cost of proactive
maintenance outweighs the benefits. However, overreliance on CM can lead to increased
downtime, productivity losses, and higher repair costs in the long run [17]. As facility man-
agers must establish efficient maintenance workflows, prioritise critical equipment, and
maintain adequate spare parts inventory to implement CM strategies effectively, using a
combination of CM and PM when managing complex building portfolios is not uncommon.
Increasingly, companies are also leveraging technologies such as Building Management
Computer System (BMCS) and Computerized Maintenance Management System (CMMS)
to streamline work orders, track maintenance history, and optimise resource allocation [18].
Some practitioners also advocate for the run-to-fail approach to maintenance, prioritis-
ing cost-effectiveness and operational efficiency and giving precedence to other market
demands such as customer service, contractor management, and financial and administra-
tive tasks. This strategy is particularly relevant for non-critical or efficiently replaceable
assets [19].

4. Data Collection and Preparation

Data collection related to maintenance activities was collected from twenty-one (21)
different commercial building types in New South Wales, Australia, including high-rise,
low-rise, retail, government, and residential apartment complex buildings while consid-
ering various floor areas (sizes), building ages, and tenancy configurations. This dataset
included variables like downtime cost, failure frequency, item reliability and maintenance
strategy. This allowed analysis of both corrective maintenance (CM) and preventive main-
tenance (PM) strategies [20]. The first thing to do with data preparation was to clean the
Buildings 2024, 14, 3581 7 of 16

data to be consistent and compatible with Python’s data analysis libraries. In this process,
the column names must be standardised, and extra data must be removed.

4.1. Descriptive Analysis

With the data ready, the analysis started by calculating summary or descriptive statis-
tics to summarise the key variables and give us an idea of the baseline. The statistics
reported include central tendency and variation measures for downtime cost, failure fre-
quency, and item reliability [21]. For example, mean downtime cost and average failure
frequency were defined, indicating the everyday demands and costs for maintaining each
strategy. Admittedly, these descriptive insights give us the foundation to understand
how maintenance strategies affect operational costs and paint an initial portrait of the
maintenance space.

4.2. Correlation Analysis

After that, an analysis of the correlation between the variables was performed. Find-
ing the associations between downtime cost, failure frequency, and item reliability was
computed using Python’s Pandas library, and a correlation matrix was created. Among
other things, it revealed strong negative correlations (e.g., between failure frequency and
downtime cost), indicating that as prevention maintenance reduces failure frequency, and
it may increase downtime cost due to the proactiveness required.

4.3. Data Visualisation

Python’s Seaborn and Matplotlib libraries were used to generate visualisations to
improve the interpretability of the data. A box plot of downtime cost between maintenance
strategies showed that PM generally incurred higher costs than CM because of PM’s
proactive nature [22]. Finally, a scatter plot of failure frequency versus downtime cost
of indicated that PM assets tend to have fewer failures but higher initial costs, whereas
CM assets have the opposite pattern. Finally, a correlation heatmap demonstrated these
findings by visualising the relationships between key variables.

4.4. Regression Modelling

Downtime cost was quantified using a regression analysis, in which failure frequency,
item reliability, and maintenance strategy were considered. A multiple regression model
was built using Python’s Scikit-Learn library and was found to do well with an R-squared
of 0.838. Through this model, it was found that a considerable amount of variance in
downtime costs was explained by the selected variables [23]. The model achieved a Mean
Squared Error of approximately AUD 80,504, which can serve as a basis for predicting
downtime costs and showed that a data-driven approach for maintenance strategy selection
is possible.
Owing to the growing computerisation, BMCS and CMMS are increasingly used
to analyse large amounts of data from multiple sources in real time [24]. Thus, these
systems can be harnessed for collecting, storing, and analysing maintenance-related data,
including work orders, equipment history, and asset performance metrics, allowing facility
managers to streamline maintenance workflows, prioritise tasks, and allocate resources
more efficiently [18]. As big data analytics can identify patterns, trends, and anomalies
in equipment behaviour, the generated outputs can facilitate PM strategies and informed
decision-making [25].
In recent years, the use of Python software 3.13.0 to analyse maintenance data in build-
ings has gained traction due to its versatility, efficiency, and extensive data manipulation
and analysis libraries. The wide range of Python software tools and packages enables
engineers to perform various tasks such as data cleaning, visualisation, statistical analysis,
and predictive modelling, allowing them to make data-driven maintenance decisions to
optimise resource allocation [26]. Furthermore, Python’s open-source nature fosters collab-
oration and innovation within the maintenance community, facilitating the development
Buildings 2024, 14, 3581 8 of 16

of customised solutions tailored to specific maintenance environments. Using Python as

the data analytics tool makes gathering and examining maintenance-related data easier.
The Python library, which includes NumPy and Pandas, has extensive dataset processing
capabilities, improving the software’s machine learning (ML) capabilities. Python’s MQTT
and CoAP libraries facilitate a seamless connection between sensors and maintenance
management systems, ensuring real-time updates on the state of assets. Additionally, by
accessing its Monte Carlo simulation tools, Python offers operators extended capabilities
with reliability-centred management. TensorFlow and Scikit-learn also offer frameworks
for developing ML models based on specific maintenance requirements in Python [27].
After importing the historical maintenance data, Python algorithms were used in
this study to identify trends and abnormalities that facilitate equipment failure prediction
and preventive maintenance scheduling. The code was created and run using the Jupyter
Notebook framework-based Google Colab platform [28]. The program uses essential
Python libraries for numerical calculations, data manipulation, and visualisation, such as
Pandas, NumPy, and Matplotlib, and was thus adopted to conduct maintenance scheduling
analyses [29].

5. Data Preparation and Cleaning

The obtained dataset was pre-processed to remove whitespaces from column names
and make it suitable for further work with tools in the Python environment. The primary
columns incorporated in the analysis were downtime cost, failure frequency, item reliability,
and maintenance strategy. Also, the dichotomy of the maintenance strategy, where 1 repre-
sents preventive maintenance (PM) and 0 represents condition-based maintenance (CM),
was introduced for regression modelling. The cleaned dataset proved to be appropriate for
meaningful and accurate analysis.

5.1. Descriptive Analysis

Buildings 2024, 14, x FOR PEER REVIEW 9 of 17
The descriptive statistics given in Figure 5 were computed for significant study vari-
ables, enabling researchers to identify critical features.

Figure 5. Descriptive statistics.

Figure 5. Descriptive statistics.
5.2. Downtime Cost
5.2. Downtime
DowntimeCost
costs per asset were averaged at AUD 1959 and ranged between AUD 1000
Downtime
and AUD costs per asset
3000, suggesting were differed
that costs averaged at AUD
based 1959 and ranged
on maintenance between
needs and AUD
approaches.
1000 and AUD 3000, suggesting that costs differed based on maintenance needs and ap-
5.3. Failure Frequency
proaches.
With an overall mean of 20, this variable fluctuates across assets, occurring at least
5.3. Failure
10 times and Frequency
a maximum of 31 times.
With an overall mean of 20, this variable fluctuates across assets, occurring at least
5.4.times
10 Item and
Reliability
a maximum of 31 times.
The mean reliability of the equipment is 0.85 on the 0–1 scale, which implies that most
of the
5.4. Itemequipment
Reliabilityis highly reliable. This supports the notion of a good PM strategy. These
statistics give areliability
The mean starting point
of theofequipment
normal maintenance cost
is 0.85 on the 0–1and relative
scale, whichdependency
implies thatsomost
that
we can compare the relative cost of each strategy and its impact.
of the equipment is highly reliable. This supports the notion of a good PM strategy. These
statistics give a starting point of normal maintenance cost and relative dependency so that
we can compare the relative cost of each strategy and its impact.

5.5. Correlation Analysis

The correlation matrix, Figure 6, was employed to investigate relationships between
the numerical data.
 5.4. Item Reliability
The mean reliability of the equipment is 0.85 on the 0–1 scale, which implies that most
of the equipment is highly reliable. This supports the notion of a good PM strategy. These
Buildings 2024, 14, 3581 statistics give a starting point of normal maintenance cost and relative dependency so9that
of 16
we can compare the relative cost of each strategy and its impact.

5.5.
5.5.Correlation
CorrelationAnalysis
Analysis
The
The correlationmatrix,
correlation matrix,Figure
Figure6,6,was
wasemployed
employedto
toinvestigate
investigaterelationships
relationshipsbetween
between
the numerical data.
the numerical data.

Figure 6. Correlation analysis.

Figure 6. Correlation analysis.
5.6. Downtime Cost and Failure Frequency
5.6. Downtime Cost and Failure Frequency
The results showed a strong negative relationship between the two (−0.86), which
The results
confirms showed
that while a strong
restricting negative
the failure relationship
frequency between
(typically the two (−0.86),
by implementing which
prevention
confirms that while restricting the failure frequency
techniques), the costs for downtime are on the rise. (typically by implementing preven-
tion techniques), the costs for downtime are on the rise.
5.7. Item Reliability and Downtime Cost
5.7. Item Reliability
Both andand
reliability Downtime
downtimeCost costs have a positive correlation coefficient of 0.78;
hence, improvements
Both in reliability
reliability and downtimemay behave
costs attained at the expense
a positive of costly
correlation maintenance.
coefficient of 0.78;
hence, improvements in reliability may be attained at the expense of costly maintenance.
5.8. Item Reliability and Failure Frequency
5.8. Item
TheReliability
negativeand (−0.91)
Failure
sign Frequency
corresponds to Pearson’s r, indicating that reliability is
minimised
The negative sign (−0.91) corresponds tothis
with high failure frequencies; is evidence
Pearson’s that preventive
r, indicating methods
that reliability can
is min-
improve reliability. These correlations shed light on the costs usually involved
imised with high failure frequencies; this is evidence that preventive methods can
Buildings 2024, 14, x FOR PEER REVIEW in choosing
im-
10 of 17
a maintenance
prove strategy,
reliability. providing quantitative
These correlations evidence
shed light on that,
the costs although
usually PM isin
involved more reliable,
choosing a
it is also more
maintenance costly. providing quantitative evidence that, although PM is more reliable,
strategy,
it
6. is also more costly.
6. Visualisation
Visualisation
6.1. Box
6.1. BoxPlot
PlotofofDowntime
Downtime Cost
Cost by
by Maintenance
Maintenance Strategy
Strategy
Thiscomparison
This comparisonofofthe thecosts
costsofofpreventive
preventivemaintenance
maintenance (PM)
(PM) and
and corrective
corrective main-
mainte-
tenance (CM) used in Figure 7 established that PM is more costly than CM in
nance (CM) used in Figure 7 established that PM is more costly than CM in most cases.most cases.
Such a result could be attributed to implementing a proactive strategy typical
Such a result could be attributed to implementing a proactive strategy typical for PM, for PM,
which entails
which entails typically
typically higher
higher costs
costs than a reactive strategy to prevent failure.

Figure7.
Figure Downtimecost
7. Downtime costby
bymaintenance
maintenancestrategy.
strategy.

6.2. Scatter Plot Failure Frequency vs. Downtime Cost by Maintenance Strategy
The scatter plot in Figure 8 indicated that the observed assets under PM exhibit fewer
failures but higher downtime costs. On the other hand, the CM strategies are coupled with
higher failure frequencies but have lower downtime costs than any other strategies.
 Buildings 2024, 14, 3581 10 of 16
Figure 7. Downtime cost by maintenance strategy.

6.2.
6.2. Scatter
ScatterPlot
Plot Failure
Failure Frequency
Frequency vs.
vs. Downtime
Downtime Cost
Cost by
by Maintenance
Maintenance Strategy
Strategy
The
The scatter
scatter plot
plot in
inFigure
Figure 88indicated
indicatedthat
thatthe
theobserved
observedassets
assetsunder
underPMPMexhibit
exhibitfewer
fewer
failures but
failures but higher
higher downtime
downtime costs.
costs. On
On the
theother
otherhand,
hand,the
theCM
CMstrategies
strategiesare
arecoupled
coupledwith
with
higher failure
higher failure frequencies
frequencies but have lower downtime costs than any other strategies.

Buildings 2024, 14, x FOR PEER REVIEW 11 o

Figure
Figure
6.3. Failure frequency
8. Failure
8.
Correlation frequency
Heatmap vs. downtime
vs. downtime cost.
cost.

6.3. Correlation
The heatmapHeatmap
in Figure 9 was used to validate the correlation analysis result and c
firm The
the heatmap in Figure of
high magnitude 9 was
theused to validate
correlations the correlation
between downtimeanalysis result
cost, and frequen
failure
confirm
and itemthereliability.
high magnitude of the correlations between downtime cost, failure frequency,
and item reliability.

Figure 9.Correlation
Figure 9. Correlationheatmap.
heatmap.
6.4. Regression Analysis
6.4. Regression
The currentAnalysis
study used a regression analysis, see Figure 10, to model the relationship
between
Thefailure
currentfrequency,
study useditem reliability,
a regression maintenance
analysis,strategy, and 10,
see Figure downtime
to modelcost.the
Therelations
model gave a coefficient of determination, or R2 , of about 0.838, which signifies that these
between failure frequency, item reliability, maintenance strategy, and downtime cost.
factors can explain 83.8% of the variance in downtime2costs [30]. The Mean Squared Error
model gave a coefficient of determination, or R , of about 0.838, which signifies that th
factors can explain 83.8% of the variance in downtime costs [30]. The Mean Squared E
(MSE) of around AUD 80,504 represents the tendency of errors in the downtime co
Altogether, these findings affirm that when it comes to downtime costs, the frequenc
failure and item reliability are indeed cardinal in driving the costs, as reflected in the
 The current study used a regression analysis, see Figure 10, to model the relati
between failure frequency, item reliability, maintenance strategy, and downtime co
model gave a coefficient of determination, or R2, of about 0.838, which signifies tha
Buildings 2024, 14, 3581 11 of 16Squared
factors can explain 83.8% of the variance in downtime costs [30]. The Mean
(MSE) of around AUD 80,504 represents the tendency of errors in the downtime
Altogether, these
(MSE) of around AUDfindings
80,504 affirm thatthewhen
represents it comes
tendency to downtime
of errors costs,costs.
in the downtime the freque
failure andthese
Altogether, itemfindings
reliability arethat
affirm indeed
when cardinal
it comes toindowntime
driving the costs,
costs, as reflected in
the frequency
search findings
of failure byreliability
and item selectingarethe right
indeed maintenance
cardinal in drivingstrategy
the costs,through theindata
as reflected the collec
research findings by selecting the right maintenance strategy through the data collected.

Figure 10. Regression result.

7. Results
Figure 10. Regression result.
The literature review, survey, and document analyses revealed that CM is performed
more frequently in established buildings with older infrastructure. This raises questions
7.about
Results
how well contemporary building materials and methods can reduce the need for
The literature
unscheduled review,
maintenance. survey,
It also prompts and document
us to reconsider analyses revealed
the durability that CM
of structures and is perf
whether more effective preventive measures are needed. The results
more frequently in established buildings with older infrastructure. This raises qu showed a relation-
ship between the frequency of preventive actions and the overall operational lifespan of
about how well contemporary building materials and methods can reduce the ne
building components, highlighting the significance of preventive measures for extending
unscheduled
the infrastructure maintenance. It alsoitprompts
utility. Furthermore, highlights us
the to reconsider
need theresearch
for additional durability
and of stru
and whether
analysis as to themore effective preventive
decision-making measures
processes associated withare needed.
either Thetheresults
extending decisionshowed
to maintain or implement CM strategies and whether short-term financial
tionship between the frequency of preventive actions and the overall operational li cost savings are
ofanbuilding
influencing factor.
components, highlighting the significance of preventive measures for e
The results of our regression analysis highlighted the importance of creating a main-
ing the infrastructure
tenance plan that proactivelyutility. Furthermore,
addresses it highlightsunderscoring
potential interruptions, the need fortheadditional
need re
and analysis
for routine as to the decision-making
maintenance processes
of critical assets. This associated
analysis illustrated thewith either
diversity extending
among
businesses stemming from their unique assets and corporate objectives
cision to maintain or implement CM strategies and whether short-term financial co while emphasising
the delicate
ings are an balance between
influencing PM and CM. Our regression analysis results also indicated
factor.
that downtime costs, maintenance frequency, and item dependability significantly influence
whether an organisation opts for preventive or corrective maintenance. Specifically, higher
downtime costs and lower item dependability are associated with a higher likelihood
of choosing CM strategies, underscoring the cost implications and reliability considera-
tions that guide decision making in maintenance management. Although links between
“downtime costs”, “failure frequency”, and “reliability measures” may be implied by ob-
servable associations, more thorough causal modelling and experiments are necessary to
show causality.

7.1. Pair Plot and Box Plot Analysis Findings

Using computational tools and platforms to implement current building maintenance
methods optimises the decision-making processes. Maintenance executives can use Google
Colab and Python programming to perform statistical analyses that effectively direct
their course of action [9]. The “pair plot” analysis indicated that managers prioritised a
CM strategy for properties with lower downtime overheads and expenses. In contrast,
they gave PM precedence for properties with high-risk exposure or higher overhead
expenses (rent abatements, etc.). In both contexts, using Python and Google Colab can
help organisations make wiser strategic choices by optimising their maintenance methods
according to specific building asset characteristics, technological dependability, financial
objectives, and resourcing considerations. They can also rely on “pair plots”, “box plots”,
and other visualisation tools during the decision making [31].
The pair plot was used to visually show the relationships between the critical elements
affecting CM and PM choices. This involved dependability data, failure frequencies, and
downtime costs. “Box plots” were also utilised to compare the probability distributions of
CM- and PM-related characteristics. Analysing the downtime costs linked to each method
and correlating the distribution of these expenses offered essential insights into the financial
 Buildings 2024, 14, 3581 12 of 16

implications of each maintenance technique. Four themes were identified based on the
research and analyses performed, as discussed below.

7.2. Performance in Correlation with PM and CM Strategies

The findings indicate that CM is only appropriate for unforeseen emergencies and
items that have a limited impact on building operations. Indeed, PM outperforms the
CM in most settings, owing to its cost-effectiveness and long-term benefits [32]. Thus, to
achieve comprehensive and long-term maintenance objectives, decision makers should
consider an integrated and balanced plan that leverages the advantages of both techniques.

7.3. Consideration of Downtime Costs

The choice between CM and PM is typically based on the expected downtime. How-
ever, decision makers often neglect the broader effects of downtime beyond the direct
financial losses, such as rent abatement, tenant satisfaction, worker efficiency, and repu-
tation, which may suffer due to frequent or prolonged disruptions [33]. Thus, a hybrid
Buildings 2024, 14, x FOR PEER REVIEW
approach that balances CM and PM would reduce the likelihood of such incidents. By 13 of 1
addressing present issues and investing in long-term stability, efficient resource alloca-
tion may enhance operational resilience, reduce downtime expenses, and boost overall
economic optimisation.
7.4. Equipment Failure Frequency
7.4. Equipment Failure frequency
The historical Frequency of equipment failures can be examined to optimise the bal
anceThebetween CM
historical and PMofstrategies.
frequency equipmentBusinesses with
failures can be a history
examined of equipment
to optimise failure typ
the balance
between CM and
ically employ PM strategies.
reactive Businesses
maintenance with adespite
strategies history growing
of equipment failurethat
evidence typically
this increases
employ
costs and reactive maintenance
operational strategies in
interruptions despite growing
the long termevidence that this increases
[34]. Conversely, proactive costs
strategie
and operational interruptions in the long term [34]. Conversely,
are favoured in firms where equipment failure has traditionally been uncommon. proactive strategies are To bet
favoured in firms where equipment failure has traditionally been uncommon. To better
ter inform their decisions, companies can rely on statistical methods such as regression
inform their decisions, companies can rely on statistical methods such as regression analysis,
analysis,
pair plots, pair plots,and
box plots, boxcorrelation
plots, andmatrices
correlation matrices
[35]. By [35]. By
establishing establishing
connections connections
between
between maintenance
maintenance plans and
plans and variables suchvariables suchcost,
as frequency, as frequency, cost,
and reliability anddowntime,
during reliability during
logistic regression helps determine the ideal balance between PM and CM for eachPM
downtime, logistic regression helps determine the ideal balance between and CM fo
specific
each specific
context. Figurescontext.
11 andFigures 11 and
12 provide 12 provide
examples examples
of a building of aperformed
retrofit building retrofit
in 2023 performed
to
improve the property’s energy efficiency rating, as the facility management
in 2023 to improve the property’s energy efficiency rating, as the facility managemen teams will
transition
teams will to transition
a CM strategyto aafter
CM completing the DLP.
strategy after completing the DLP.

Figure11.
Figure 11.Hot
Hotwater
water service.
service.
Buildings 2024, 14, 3581 13 of 16

Figure 11. Hot water service.

Figure12.
Figure 12.HVAC
HVAC plantroom.
plantroom.

8. Ethical Considerations
8. Ethical Considerations
Ethical considerations in modern building management strategies are critical to ensur-
ing theEthical
safety, considerations in modern
reliability, and overall healthbuilding management
and well-being strategies including
of all stakeholders, are critical to en
suring the
tenants, safety,
service reliability,
providers, and overall
contractors, health
visitors, andand
thewell-being
people in the of all stakeholders,
broader commu- includ
nity [36]. The survey and interviews with service providers and facilities managers revealed commu
ing tenants, service providers, contractors, visitors, and the people in the broader
anity [36].
strong Theofsurvey
sense and interviews
professional with
responsibility, service and
diligence, providers
desire to and facilities
ensure managers re
equipment
functionality,
vealed a strong safety, andofoperability.
sense professionalTheir responses to multiple
responsibility, diligence, questions underscored
and desire to ensure equip
the facilities
ment managers’safety,
functionality, vigilance
andtowards occupational
operability. Their health and safety.
responses The stakeholders
to multiple questions under
also
scored the facilities managers’ vigilance towards occupational health andfrom
demonstrated a strong focus on sustainability targets and objectives ranging safety. The
green energy solutions to waste reduction and minimising carbon emissions impacting
stakeholders also demonstrated a strong focus on sustainability targets and objective
society, suggesting that they were committed to ensuring the long-term reliability of their
ranging from
maintenance green [37].
systems energy solutions to waste reduction and minimising carbon emission
impacting society, suggesting that they were committed to ensuring the long-term relia
9. Limitations
bility of their maintenance systems [37].
When modern building maintenance strategies are analysed, a complicated decision-
making context emerges, as companies need to weigh any chosen approach’s costs and
benefits [38]. Increasingly, they rely on software applications to streamline every aspect of
the maintenance process, thus achieving strategic objectives more cost-effectively. While
this work offers insight into the factors that drive the decision-making process, as it was
based on a limited number of buildings at one location and the input provided by pur-
posely selected industry practitioners, the findings cannot be generalised to other contexts.
Specifically, additional or altered perspectives might have been attained if the research
scope was extended to include different commercial buildings, a broader range of operating
and maintenance strategies, and multiple geographical regions and building occupant
types. Furthermore, the fluid nature of choosing a maintenance plan indicates the need for
an advanced understanding of the constraints and external influences that play a role in
the maintenance strategy determination process.
A further limitation arises due to the reliance on quantitative methodology when
assessing the benefits and drawbacks of CM and PM, which may only partially capture
qualitative factors such as corporate culture, interpersonal dynamics, and external in-
fluences. Additionally, how maintenance strategies evolve in response to technological
advancements, economic shifts, or changes in corporate objectives was not explored in this
research. As these factors inevitably influence the adaptation and effectiveness of mainte-
nance practise, these study limitations provide a solid foundation for further research in
this domain.

10. Discussion
Although maintenance is an essential and critical workplace practise within the facility
management industry, academic research insufficiently recognises the need for compressive
classification and comparative evaluation of different approaches [39]. This gap in the
Buildings 2024, 14, 3581 14 of 16

literature has motivated the present study, which aims to identify the most appropriate
maintenance selection processes in different contexts [40]. The obtained findings support
those reported in the extant literature, highlighting the benefits of PM, such as enhanced
efficiency, reduced downtime, and less frequent malfunctions.
Conversely, CM is typically seen as a viable strategy when potential failures are rare
and do not impact operational efficiency. As businesses must evaluate a wide range of
complex factors (including repair delay cost, abatement expenses, market perception, and
the impact of deteriorating operating conditions on the asset’s lifecycle and longevity) to
achieve a successful maintenance plan, they would benefit from the assistance of software
tools such as BMCS and CMMS to streamline the decision-making processes. Integrating
reliability-centred maintenance (RCM) approaches like TPM-RBM (total preventative and
risk-based maintenance) adds a new layer of complexity, emphasising the need for extensive
planning if an organisation adopts this advanced maintenance strategy. As the building
asset ages and its operating equipment begins to deteriorate, a thorough re-evaluation
of the initial maintenance strategies is required to ensure that the equipment lifecycles
are optimised (or extended). The present study’s findings have extensive applications in
several sectors, as they can aid organisations in customising their maintenance strategies
by considering item reliability, failure frequency, and downtime costs [41].
As the property and facility industries are experiencing a rapid digital transition due
to the advances in information and communication technology [42], authors of future
studies in this domain could investigate innovative technological approaches, such as a
combination of machine learning (ML) and artificial intelligence (AI), to enhance the efficacy
of maintenance strategies. It would also be beneficial to investigate how intelligent IoT-
enabled sensors might be used for predictive maintenance and real-time monitoring [43].
Moreover, this study can be replicated in other settings and industries, thus obtaining more
comprehensive insights into the benefits and drawbacks of different maintenance strategies
in diverse contexts.

Justification of Python in the Research

The choice of a maintenance strategy in commercial buildings determines the reliability
of equipment used, the operational time lost, and the cost. This paper evaluates corrective
and preventive maintenance on assets in New South Wales (NSW), Australia. Python
illustrates for the data analysis aspect of the research, and frequencies and costs of failures
are analysed to understand the reliability standards of the system and patterns in failure
frequencies, costs, and reliability indices, and decision support for optimum maintenance
consequentially are interpreted by employing data analysis tools in Python language [44].
Due to high efficiency, flexibility, and ease of use, Python has become a preferred program-
ming language for data management and analysis of massive dataset maintenance.
Using Python’s Pandas library enabled missing value management and categorical
data encoding steps. The presence of libraries such as Scikit-Learn helped create a sound
regression model that can predict downtime costs effectively. Both Seaborn and Matplotlib
are visualisation libraries in Python that offer comprehensible, easily deciphered diagrams
and graphs. The strength of using Python for this type of analysis is that Python is an open-
source environment that can enjoy multi-usages or replication of such analysis for other
datasets or regions. Each of these tools is unique. Together, they make Python invaluable
in performing statistical, prognosis, and visual analyses that would not be easily possible
or efficient using the formats of the spreadsheeting program.

Author Contributions: J.W.: conceptualization, data curation, formal analysis, investigation, method-
ology, project administration, resources, software, validation, visualisation, writing—original draft,
and writing—review and editing; M.S., A.E., and A.H.: supervision. All authors have read and
agreed to the published version of the manuscript.
Funding: This research received no external funding.
Buildings 2024, 14, 3581 15 of 16

Data Availability Statement: The original contributions presented in the study are included in the
article; further inquiries can be directed to the corresponding author.
Acknowledgments: The author acknowledges the technical, academic, and practical support and
encouragement received from his supervisors, Milind Siddhpura, Ana Evangelista, and Assed
Haddad, which was instrumental in the completion of the work presented here. The authors would
like to acknowledge the support of Conselho Nacional de Desenvolvimento Científico e Tecnológico
(CNPq 304726/2021-4) and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio
de Janeiro (FAPERJ E-26400.205.206/2022(284891).
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Fraser, K. Facilities management: The strategic selection of a maintenance system. J. Facil. Manag. 2014, 12, 18–37. [CrossRef]
2. Abdelrazik, H.; Marzouk, M. Investigating parameters affecting heritage building maintenance in Egypt. Int. J. Build. Pathol.
Adapt. 2021, 39, 734–755. [CrossRef]
3. Qi, W.; Sun, M.; Hosseini, S.R.A. Facilitating big-data management in modern business and organisations using cloud computing:
A comprehensive study. J. Manag. Organ. 2023, 29, 697–723. [CrossRef]
4. Khaleel, M.; Yusupov, Z.; Yasser, N.; El-Khozondar, H.J. Enhancing microgrid performance through hybrid energy storage system
integration: ANFIS and GA approaches. Int. J. Electr. Eng. Sustain. 2023, 1, 38–48. Available online: https://ijees.org/index.php/
ijees/article/view/34 (accessed on 22 July 2024).
5. Weeks, D.J.; Leite, F. Minimising facility corrective maintenance: Benchmarking preventative-to-corrective maintenance ratios
using maintenance data and building age in dormitories. J. Manag. Eng. 2022, 38, 04021086. [CrossRef]
6. Fan, Z.; Yan, Z.; Wen, S. Deep learning and artificial intelligence in sustainability: A review of SDGs, renewable energy, and
environmental health. Sustainability 2023, 15, 13493. [CrossRef]
7. Rahman, U.; Mahbub, M.U. Application of classification models on maintenance records through text mining approach in an
industrial environment. J. Qual. Maint. Eng. 2023, 29, 203–219. [CrossRef]
8. Polese, M.; Bonati, L.; D’Oro, S.; Basagni, S.; Melodia, T. Melodia, Understanding O-RAN: Architecture, interfaces, algorithms,
security, and research challenges. IEEE Commun. Surv. Tutor 2023, 25, 1376–1411. [CrossRef]
9. Liu, W.; Hua, M.; Deng, Z.; Meng, Z.; Huang, Y.; Hu, C.; Song, S.; Gao, L.; Liu, C.; Shuai, B.; et al. A systematic survey of control
techniques and applications in connected and automated vehicles. IEEE Internet Things J. 2023, 10, 21892–21916. [CrossRef]
10. Ma, S.; Chou, Y.C.; Zhao, H.; Chen, L.; Ma, X.; Liu, J. Network Characteristics of LEO Satellite Constellations: A Starlink-Based
Measurement from End Users. In Proceedings of the IEEE INFOCOM 2023—IEEE Conference on Computer Communications,
New York, NY, USA, 17–20 May 2023; pp. 1–10. [CrossRef]
11. Jones, M.J. A Companion to Roman Britain; Wiley Online Library: Hoboken, NJ, USA, 2004. [CrossRef]
12. Pugh, D.S. Maintenance Management and Regulatory Compliance Strategies; Industrial Press Inc.: New York, NY, USA, 2003;
Available online: https://books.google.com/books?hl=en&lr=&id=ZlWOp0ciwQsC&oi=fnd&pg=PR7&dq=Pugh (accessed on
22 November 2023).
13. Jeang, A.; Ko, C.P.; Chung, C.-P.; Chen, Y.-J.; Lin, I. Optimal availability for determining choice and repair policy of system
components. Int. J. Qual. Reliab. Manag. 2019, 36, 347–357. [CrossRef]
14. Ismail, A.A.; Mbungu, N.T.; Elnady, A.; Bansal, R.C.; Hamid, A.K.; AlShabi, M. Impact of electric vehicles on the smart grid and
future predictions: A survey. Int. J. Modell. Simul. 2023, 43, 1041–1057. [CrossRef]
15. CIBSE. Guide M: Maintenance Engineering and Management, Chartered Institution of Building Services Engineers. 2018.
Available online: https://www.amazon.co.uk/ (accessed on 22 November 2023).
16. Feldman, S. The role of predictive maintenance in facility management. Facilities 2020, 38, 456–470.
17. Ganesan, R.; Swarnalatha, P. A review on corrective maintenance. Int. J. Sci. Eng. Technol. Res. 2015, 4, 144–148.
18. Sajadi, S.M. Evaluation of computerised maintenance management systems for maintenance management. Int. J. Eng. Res. Afr.
2018, 36, 100–109.
19. Matoka, A.; Nadjamuddin, H.; Salama, M.; M. Arsyad, T. Analysis effect of sedimentation at MHP type turbine open flume on
irrigation channel. Int. J. Eng. Res. Appl. 2017, 7, 50–57. [CrossRef]
20. Gan, S.; Song, Z.; Zhang, L. A maintenance strategy based on system reliability considering imperfect corrective maintenance and
shocks. Comput. Ind. Eng. 2022, 164, 107886. [CrossRef]
21. Mol˛eda, M.; Małysiak-Mrozek, B.; Ding, W.; Sunderam, V.; Mrozek, D. From corrective to predictive maintenance—A review of
maintenance approaches for the power industry. Sensors 2023, 23, 5970. [CrossRef]
22. Martins, L.; Silva, F.J.; Pimentel, C.; Casais, R.B.; Campilho, R.D.S.G. Improving preventive maintenance management in an
energy solutions company. Procedia Manuf. 2020, 51, 1551–1558. [CrossRef]
23. Zhao, J.; Gao, C.; Tang, T. A review of sustainable maintenance strategies for single-component and multi-component equipment.
Sustainability 2022, 14, 2992. [CrossRef]
24. Galán, M.H. Evidence-based asset management applied to maintenance function control. J. Qual. Maint. Eng. 2019, 25, 635–644.
[CrossRef]
Buildings 2024, 14, 3581 16 of 16

25. Lee, C.K.M.; Cao, Y.; Ng, K.H. Big data analytics for predictive maintenance strategies. In Supply Chain Management in the Big Data
Era; Chan, H.K., Ed.; IGI Global: Hershey, PA, USA, 2017; pp. 50–74. [CrossRef]
26. Brownlee, J. Machine Learning Mastery with Python: Understand Your Data, Create Accurate Models, and Work Projects End-to-
End, Machine Learning Mastery. 2016. Available online: https://books.google.com/books?hl=en&lr=&id=BgmqDwAAQBAJ&
oi=fnd&pg=PP1&dq=Brownlee. (accessed on 22 July 2024).
27. Akrami, N.E.; Hanine, M.; Flores, E.S.; Aray, D.G.; Ashraf, I. Unleashing the potential of blockchain and machine learning:
Insights and emerging trends from bibliometric analysis. IEEE Access 2023, 11, 78879–78903. [CrossRef]
28. Qadir, J. Engineering education in the era of ChatGPT: Promise and pitfalls of generative AI for education. In Proceedings of the
IEEE Global Engineering Education Conference (EDUCON), Kuwait, Kuwait, 1–4 May 2023; pp. 1–9. [CrossRef]
29. Lange, R.T. evosax: JAX-Based Evolution Strategies; Cornell University: Ithaca, NY, USA, 2023. [CrossRef]
30. García, F.J.Á.; Salgado, D.R. Analysis of the Influence of Component Type and Operating Condition on the Selection of Preventive
Maintenance Strategy in Multistage Industrial Machines: A Case Study. Machines 2022, 10, 385. [CrossRef]
31. Beltrán, E.T.M.; Pérez, M.Q.; Sánchez, P.M.S.; Bernal, S.L.; Bovet, G.; Pérez, M.G.; Pérez, G.M.; Celdrán, A.H. Decentralized,
federated learning: Fundamentals, state of the art, frameworks, trends, and challenges. IEEE Commun. Surv. Tutor. 2023, 25,
2983–3013. [CrossRef]
32. Duggineni, S. Impact of controls on data integrity and information systems. Sci. Technol. 2023, 13, 29–35. [CrossRef]
33. Juffinger, J.; Lamster, L.; Kogler, A.; Eichlseder, M.; Lipp, M.; Gruss, D. CSI: Rowhammer—Cryptographic security and integrity
against Rowhammer. In Proceedings of the 2023 IEEE Symposium on Security and Privacy (SP), San Francisco, CA, USA, 21–25
May 2023. [CrossRef]
34. Crothers, E.N.; Japkowicz, N.; Viktor, H.L. Machine-generated text: A comprehensive survey of threat models and detection
methods. IEEE Access 2023, 11, 70977–71002. [CrossRef]
35. Ali, S.A.; Zafar, M.W. Istio service mesh deployment pattern for on-premises. Int. J. Comput. Sci. Inf. Technol. 2021, 5, 95–110.
36. Welch, D.; Napoli, A.; Back, J.; Buggaveeti, S.; Castro, C.; Chase, A.; Chen, X.; Dominic, V.; Duthel, T.; Eriksson, T.A.; et al. Digital
subcarrier multiplexing: Enabling software-configurable optical networks. J. Lightwave Technol. 2023, 41, 1175–1191. [CrossRef]
37. Ali, I.; Golgeci, I.; Arslan, A. Achieving resilience through knowledge management practices and risk management culture in
agri-food supply chains. Supply Chain Manag. Int. J. 2023, 28, 284–299. [CrossRef]
38. Logeswaran, K.; Suresh, P.; Ponselvakumar, A.P.; Savitha, S.; Sentamilselvan, K.; Adhithyaa, N. Exploration of assorted
modernisations in forecasting renewable energy using low power wireless technologies for IoTSG. In Smart Grids Internet Things
Energy Perspect; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2023; pp. 157–191. [CrossRef]
39. Garg, A.; Deshmukh, S.G. Maintenance management: Literature review and directions. J. Qual. Maint. Eng. 2006, 12, 205–238.
[CrossRef]
40. Khazraei, K.; Deuse, J. A strategic standpoint on maintenance taxonomy. J. Facil. Manag. 2011, 9, 96–113. [CrossRef]
41. He, Z.; Huang, H.; Choi, H.; Bilgihan, A. Building organisational resilience with digital transformation. J. Serv. Manag. 2023, 34,
147–171. [CrossRef]
42. Kunju, F.F.; Naveed, N.; Anwar, M.N.; Haq, M.I.U. Production and maintenance in industries: Impact of Industry 4.0. Ind. Robot.
2022, 49, 461–475. [CrossRef]
43. Yun, G.; Ravi, R.V.; Jumani, A.K. Analysis of the teaching quality on deep learning-based innovative ideological political education
platform. Prog. Artif. Intell. 2023, 12, 175–186. [CrossRef]
44. Alsyouf, I.; Hamdan, S.; Shamsuzzaman, M.; Haridy, S.; Alawaysheh, I. On preventive maintenance policies: A selection
framework. J. Qual. Maint. Eng. 2021, 27, 225–252. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.