Assessment of researches and case studies on Cloud Manufacturing: a bibliometric analysis

Cloud computing technology has been studied in the context of industry 4.0 as a tool applied to manufacturing services and resources. Such concept is widely known as Cloud Manufacturing. This paper aims at mapping the current state of academic researches on this field, promoting the understanding of trends, references and practical applications in real-life conditions. A bibliometric analysis was conducted using two different databases — Scopus and Web of Sciences — and VOSviewer’s text mining tools and techniques. From a sample of 1420 papers, this study identified the countries which had the largest volume of publications, the main journals related to the subject, the most influent articles, and four clusters by keywords occurrences: (i) “Optimization of manufacturing processes”, (ii) “Collaborative networks of manufacturing resources and services”, (iii) “Industry 4.0 and cloud computing systems”, and (iv) “Data reliability and cyber-security”. Finally, this work selected and analyzed the 159 articles with applied case studies, in order to stratify and to understand the most common approaches within the four pre-established categories. This article can contribute to researchers and developers searching for successful practical applications in digitalization of manufacturing chains, as well as to those who are looking for gaps in the still unexplored fields of Cloud Manufacturing. Both the assessment and the categorization of the case studies about Cloud Manufacturing are the differentials in this article.


Introduction
The manufacturing industry, in the context of the fourth industrial revolution, must be capable of applying advanced technologies and knowledge, integrating the automation of its machines, equipment and sensors. Such integration should include not only the company's processes, systems and/or internal protocols, but also those of other companies that are part of its supply chain [1].
Academic researches and the industry's development departments have initiated to explore a set of innovative Daniel Alexandre Morelli damorelli@gmail.com Paulo Sergio de Arruda Ignacio psai@unicamp.br 1 Industrial Engineering Reasearch Center, School of Applied Sciences, University of Campinas-UNICAMP, R. Pedro Zaccaria, 1300, Limeira SP, Brazil tools, among them, a manufacturing model based on the emerging cloud computing technology, integrating different sectors and companies on the network, through the virtualization and sharing of its resources and knowledge, and the communication between machines [2].
Built over web-based architectures, these models can generate value through collaborative processes between stakeholders, cost reduction and scalability [3]. Their proposition intends to promote performance improvement in a supply chain environment over the implementation of a digital and continuous information flow [4].
The application of these standards, which associate cloud computing advantages to manufacturing resources, started to be studied in 2010 and are called Cloud Manufacturing [5]. Since then, many researches have emerged in this field. Some of them are focused on theoretical frameworks, while others propose real-life practical applications through case studies.
A bibliometric analysis about the subject of Cloud Manufacturing was accomplished in this article, in order to assess academic researches and their case studies approaches, published until January 2021.
This investigation aims at mapping the current state of a wide range of available researches in the field of Cloud Manufacturing, identifying publication trends, countries with the largest volume of papers, most relevant journals and influential articles, as well as the clusters of the keywords co-occurrences. This work then selected and analyzed the researches that included case studies, in order to categorize their approaches.
The contribution of this article is to generate relevant information to expand the knowledge about Cloud Manufacturing, contextualize trends of application tools, make successful practical solutions and relevant references available, as well as, identify gaps or improvement opportunities of studies in unexplored fields of Cloud Manufacturing.

Literature review
Cloud computing technology represents a service model that enables users or enterprises to access a whole set of adaptable, configurable and available on-demand IT resources, in an agile way and with a low level of effort [6].
According to Zárate and Mendoza [7], in a cloud computing architecture, all of the computing structures are treated as a service. Figure 1 shows the services delivered through cloud technology.
• IaaS (Infrastructure as a Service), also known as hardware as a service (HaaS), delivers computing infrastructures, such as storage or databases, using virtual machines (e.g., Amazon EC2, GoGrid, Flexiscale or Data Centers); • PaaS (Platform as a Service) provides a system environment called middleware, driven to software development, as well as, testing and/or hosting of applications (e.g., Microsoft Azure, Google AppEngine or Amazon Simple DB/S3); and • SaaS (Software as a Service) which offers a set of internet accessible applications to end users (e.g., Google Apps, Facebook, Youtube or Salesforce).
Cloud computing technologies applied specifically to the manufacturing sector is addressed by 3 main terms: Cloud Manufacturing, Cloud-based Manufacturing, or even Cloud-based Design and Manufacturing. By gathering the benefits of cloud computing, such as agility, flexibility, scalability and efficiency, many companies are being able not only to improve their own production processes, but also promote a greater integration and a more collaborative relationship with their business partners [9].
Based on networks, Cloud Manufacturing transforms manufacturing resources and capabilities into services performed through machine virtualization, which can be managed and operated on-demand by users for the whole life cycle of manufacturing [10].
As illustrated in Fig. 2, such process consists of three layers.
• Application Layer -This layer includes interfaces of users that need manufacturing services or resources to meet their customized demands. Those requests can be made by companies or individual users with access though the internet [12]. • Manufacturing Capability Layer -In this layer occurs the connection between the physical manufacturing machines and the cloud servers, by the virtualization of these resources [13]. Countless suppliers offer their specialized services and make their manufacturing machines available worldwide. Generally, these  The Cloud Manufacturing architecture model [11] facilities are CNC machining centers, foundries industries, 3D printing, plastic injection molding, and other component production processes. • Central Service Layer -In this layer happens the smart cloud solutions management, as matching between customers' demands and physical resources available. This assessment enables production processes evaluation, scheduling, optimal resources allocation, monitoring, quality management, big data and real time information [11].   The 5 steps applied in this bibliometric analysis [16] manufacturing companies. Cloud Manufacturing may be used to share resources and knowledge in order to be effective in the support of processes such as design, manufacturing, planning, controling, analysis and decisionmaking. [14].

Methodology
In order to explore the researches on Cloud Manufacturing in a quantitative perspective, this article conducted a bibliometric study based on the appreciation of recorded information in the scientific literature. Subsequently, this analysis used tools and statistical techniques that resulted in indexes which enabled us to recognize correlations among the investigated publications [15]. Figure 4 presents the 5 steps conducted during this study.
The research and analysis methodology is illustrated as a flowchart in Fig. 5, organizing the 5 steps mentioned, with the registration of the quantity of articles detected in each step.  [16] The 5 steps of the applied methodology are detailed as follows:

Results
Over the 1420 publications that arose from the screening process, this work built the graph presented in Fig. 6 that shows the overall trend of publications over time about Cloud Manufacturing, since the very first publication in 2010. It is possible to perceive an increase in the number of annual publications, confirming the growing relevance of this subject on both the academic and the industrial fields. Figure 6 also presents the fitted trend line, generated using the statistical tool from Microsoft Excel, as the following linear equation.
In order to avoid distortions in the trend equation, this exploration disregarded the 12 publications from January 2021, date when the bibliographic research was conducted. Table 2 shows the countries with largest volume of publications in English language, indicating China as the leader, with an annual production greater than that of all other 14 countries together. China is also the country that presents the largest number of publications and of citations in any language.
Some countries stand out in the density of publications relative to their population, as New Zealand, Sweden and Finland, with publication rates considerably higher than those of the other countries on the list. Table 3 ranks the top 10 journals by total number of published papers. Their origin countries are China, the UK, the Netherlands and the USA, with emphasis to the Chinese Computer Integrated Manufacturing Systems, CIMS, which boasts the largest number of publications, the highest number of citations, as well as the highest factor of citations per publication among the members of the list. Table 1 Summary of the applied search engines [18] Analysis of publications about Cloud Manufacturing

Databases
Web of Science and Scopus Total number of documents 1420 (907 from Web of Science and from 1117 Scopus) General search engines terms ("cloud manufactur*" OR "cloud?based manufactur*" OR "cloud?based design and manufactur*") Specific search engine terms (for case studies identification) (("cloud manufactur*" OR "cloud?based manufactur*" OR "cloud?based design and manufactur*") AND ("case stud*" OR "case analys*" OR "simulat*")) Specific search filters (for case studies) Publication Type: Article or Review Language: English Softwares VOSviewer and Excel Types of analysis Literature review, quantitative and qualitative analysis According to the indexed categories in Scimago [20], the most relevant journals are specialized on computer science, mechanical engineering, industrial engineering, mathematics, management science, operations research, manufacturing and artificial intelligence. Table 4 shows the most influential articles on the subject of Cloud Manufacturing ranked by total citations, presenting keywords targeted for service-oriented business models, advanced manufacturing systems or platforms, internet of things (IoT), cloud computing, services optimization, real-time monitoring, big data, and distributed resources sharing.

Clustering analysis
Using the software VOSviewer, this study built Fig. 7 that reveals the keywords co-occurrences map, determinated by the documents where they are found together. Both authors keywords, and other relevant words frequently used in titles and abstracts were considered. Every keyword is represented by a circle, whose size is proportional to its frequency of occurrence. The positions indicate the relationship between two different keywords, and the lines signify connectivity and association of concepts.  Among the 1420 analyzed papers, this step found a total of 4487 keywords, of which 104 met the requirement of having at least 10 occurrences each, being then applied in the mapping process. Figure 7 shows the characterization of the four cooccurrences clusters: • the first cluster (colored red) is defined as "Collaborative networks of manufacturing resources and services", once the most frequent terms refers to manufacture, manufacturing resources and services, distributed systems architecture, and networked manufacturing environments through web-based services; • the second (green) is defined as "Industry 4.0 and cloud computing systems", once the most frequent keywords are related to computer-aided manufacturing processes in industry 4.0. This cluster encompasses the whole product lifecycle within industry 4.0, including cloud computing technologies, internet of things (IoT) and additive manufacturing; • the third (blue) is defined as "Big data reliable management in cyber-physical systems, risks and challenges", once the main occurrences are related to big data in cyber-physical systems, processes integration, interoperability, blockchain and implementation challenges; and finally • the fourth cluster (yellow) is defined as "Optimization of manufacturing processes", once the most frequent keywords are related to optimization algorithms and models of cloud manufacturing, as well as to optimal selection and allocation of resources and services.
The elements of each cluster are shown in Table 5.

Case studies analysis
The case study approach allows researchers to explore and understand complex problems in real-life applications and, therefore, is recognized by its capacity to generate value to the industrial and academic sectors. In order to assess practical applications in Cloud Manufacturing, this investigation performed the eligibility step, applying the previously described search filters and terms showed in Table 1, resulting in the selection of 159 academic papers with applied case studies out of 1219 publications in English, as shown in Fig. 8.
Afterwards, all titles, abstracts and keywords of the 159 articles were analyzed, enabling the categorization of their case studies approaches within the four pre-established cooccurrences clusters. Figure 9 presents the final stratification by cluster with the number of related articles.  This cluster contains all case studies which presented platform models or algorithms to generate optimized solutions on the following: (i) production planning [30], (ii) scheduling and selection of manufacturing resources or services [31,32], (iii) reduction of operational and/or logistic total time [33], (iv) cost reduction, (v) maximization of utilization rates [34], (vi) balancing the allocation levels of machines and tools [35], (vii) minimization of industrial resources consumption by controlling process parameters [36], (viii) energy efficiency improvements [37], (ix) productivity increase, (x) task prioritization [38], as well as other strategies aiming at improving factory performance when compared to traditional approaches. • Collaborative networks of manufacturing resources and services (52 articles): This cluster contains the models developed as tools to approach the relationship between stakeholders, aiming at assessing the following: (i) synergy of the manufacturing services and enterprises networks [39], (ii) integration of operations, information, knowledge and efforts between players [40,41], (iii) the sharing of resources capabilities [42], (iv) collaborative troubleshooting process involving multiple companies [43], (v) the dynamic decision-making process made by customers to meet their customized demands in production networks with high levels of flexibility and responsiveness [44], (vi) procedures that increase integration of capabilities in remanufacturing, recycling and recovery [45], as well as other strategies that might increase competitiveness throughout the manufacturing supply chain.   [46], internet of things (IoT) [47] and other human-machine interaction approaches in advanced manufacturing of industry 4.0; (ii) test the functionality of prototypes, comparing them to other models in order to validate benefits and disadvantages of adopting Cloud Manufacturing networks and systems [23]; or (iii) highlight implementation challenges and requirements to achieve success in the adoption of virtual manufacturing [11,29]. • Data reliability, cyber-security, risks and challenges (7 articles): This cluster contains all studies about the following: (i) modeling and simulating architectures that focus both on maximizing transparent data analysis in real-time and on improving cyber-security [48], (ii) using consensus and Proof-of-Authority (PoA) resources to improve the reliability of transaction records between various agents [49], (iii) the application of blockchain technology [50], and (iv) the evaluation of risks or potential failures in transmission of data and/or communication between companies [51].

Conclusions
From the bibliometric analysis this research conducted on the emerging technology of Cloud Manufacturing, it was possible to map the current literature conditions and to verify that, since 2010, when the first studies appeared, there has been a growing trend in annual publications, suggesting an increasing interest about the subject. The results obtained reveal that researchers from China are those who generate the highest volume of publications, even when only English language articles are taken into account. Another interesting evidence concerns countries such as New Zealand, Sweden and Finland, which presented considerably high publications rates relative to their populations.
It was not only possible to identify the most relevant journals, with the top editions coming from the USA, the UK, the Netherlands and China, but also a list of the top-ranked articles in terms of influence relative to their number of citations. Those lists can be useful references for researchers interested in the literature about Cloud Manufacturing.
By using the software VOSviewer, this paper elaborated a visualization of bibliometric maps and created four clusters of keywords co-occurrences: "collaborative networks of manufacturing resources and services"; "industry 4.0 and cloud computing systems"; "big data reliability, cybersecurity, risks and challenges"; and "optimization of manufacturing processes".
In order to understand the current situation about reallife applications already developed and studied in Cloud Manufacturing, this investigation selected and analyzed all 159 articles written in English that contained case studies. All of them were assessed and categorized according to the four previously cited clusters. It became evident that, while most of the models laid within three of the four clusters -"optimization of manufacturing processes", "collaborative networks of manufacturing resources and services", and "industry 4.0 and cloud computing systems" -only a few of them were related to "big data reliability, cyber-security, risks and challenges".
This cluster analysis of publications containing case studies about Cloud Manufacturing revealed diverse webbased applications that may generate potential gains in production planning, scheduling of services, cost reduction, minimization of industrial resources consumption, synergy of enterprises networks, integration of operations and scalability. However, this analysis also suggests a potential research gap of practical application in the topic of data management and cyber-security, indicating the need of both a reliable data registration, and an agile and traceable information flow, in order to increment the success of this business model.
The great majority of the practical models assessed throughout this study needed a centralized third-party to manage transactions and communication between users. Such practice may reveal weaknesses in data reliability and system cyber-security, resulting in eventual loss, damage, or even manipulation of information and contents. In this context, such centralized digital platforms must apply resources and technologies to increase data management reliability and security against cyber-attacks.
The absence of depth in such an important matter is also an opportunity for development of future works and researches about the application of technologies for the evolution of cyber-security in cloud manufacturing, improvements of transparency in data traceability, utilization of consensus resources in validation protocols, elimination of the need for a third-party in transactions management, as well as, standardization of systems in order to eliminate communication conflicts between stakeholders of the entire supply chain.
Those key recommendations for further research could potentially reduce breaches or vulnerabilities in cloud manufacturing networks, and contribute to technological advancements in such a new, yet underexplored field.