Smart Factory Transformation Using Industry 4.0 toward ESG Perspective: A Critical Review and Future Direction

2.1.1 Raw Material Sourcing (E1)

Raw materials are required for manufacturing products in factories. In addition to the efficiency of material usage (discussed later), the total amount of materials is an important factor in assessing the environmental impact of corporations. To assess these effects, primary raw material purchases and the amount of recycled water used in manufacturing were selected as metrics. In addition, because the impact varies depending on the type of raw material used, qualitative discussion results were used as the metric.

2.1.2 Product Lifecycle (E2)

This depends on the product manufactured by the factory, but the product lifecycle has a significant impact on the revenue of most corporations. With a short lifecycle, high revenue can be expected because consumers purchase the product more frequently; however, there is also the possibility that more waste will be discharged during this process. To assess this effect, the ratio of products certified by the product lifecycle-related institution, weight and ratio of recovered materials, ratio of reusable products among sold products, and the qualitative discussion of end-of-life management efforts were used as metrics.

2.1.3 Energy (E3)

Monitoring is a fundamental element of smart factories. There are several monitoring targets, one of which is energy consumption. When the energy consumption of each piece of equipment is monitored, the basic energy savings occurred by cutting off the standby power, additional energy can be expected by managing the equipment preheating according to the optimized factory schedule. Recently, various energy harvesting technologies have been developed and used in factories. To assess these effects, the total energy usage, ratio of alternative or renewable energy usage, ratio of off-grid power usage, total fuel usage, and the ratio of coal and natural energy usage were used as the metrics.

2.1.4 Packaging (E4)

Most products are packaged before delivery. This is for safe delivery to prevent damage or deformation during distribution, but some products are overpacked for luxury purposes. As such overpacking eventually leads to waste generation, evaluation indicators are required. To assess these effects, the weight of packaging, ratio of recycled materials used in packaging, and the qualitative discussion of the environmental impact of packaging were used as metrics.

2.1.5 Greenhouse Gas (GHG) Emission (E5)

GHGs include those that absorb and reflect infrared radiation to increase atmospheric temperature. Many studies have shown that global warming is accelerating, and efforts to manage GHG emissions continue. To assess these effects, carbon emission is the main evaluation target of ESG as CO₂ has the greatest impact on the greenhouse effect [35]. Additionally, the ratio of carbon emission-related regulatory compliance, amount of hydrocarbon emission prediction, and qualitative discussion of carbon emission management were used as metrics.

2.1.6 Waste Management (E6)

Waste is commonly generated when manufacturing products; its management is at the core of sustainable development. To assess this waste management capability, the total amount of waste discharged, the amount of incinerated waste, and the amount of electronic component waste recovered (which is especially difficult to dispose of) were used as the metrics. In addition, with the development of technologies that harvest energy from waste, such as biogas, the total amount of recycled waste can be used as a metric.

Depending on the manufacturing target, toxic waste can be generated in factories. Hazardous waste from a factory environment varies extensively depending on the process and is well classified in [36]. This hazardous waste can threaten the safety of surrounding residents beyond environmental protection. To assess these effects, the total amount of hazardous waste, number of incidents related to hazardous waste, and the qualitative discussion of hazardous waste management were used as metrics.

2.1.7 Product Efficiency (E7)

One of the reasons why additive manufacturing, such as three-dimensional printing, is attractive is because it generates smaller amounts of waste material. Therefore, even if the same products are manufactured, differences in material usage efficiency may occur because of the manufacturing processes. In addition, automation systems implemented in smart factories lead to the optimization of manufacturing and reduce unnecessary waste of resources [37]. To assess these effects, the weight ratio of the material, relative price, weight, and performance among the same industrial groups, and a qualitative discussion of the considerations for increasing efficiency were used as metrics.

2.2.1 Product Safety (S1)

Product safety accounts for a large part of corporate valuation as it is selected as a key issue from CSR to ESG [39]. If there is a serious failure in a product, it may lead to a product recall or the death of the consumer; these may adversely affect corporate reputation and lead to long-term damage [40]. To assess these effects, the cost paid for product safety, ratio of products that passed the volatile organic compounds (VOCs) emission regulations, ratio or revenue of products that passed the chemical and safety regulations, recall count, and qualitative discussions of compliance with chemical regulations and product safety management procedures were used as metrics. Recently, as the safety of personal information has also become important, the number of data breaches and qualitative discussions of personal information protection methods have also been used as metrics.

2.2.2 Facility Management (S2)

Facility management is not an indicator directly mentioned in the SASB ESG indicators but has emerged as an indicator for equipment management, such as pipeline management in the oil and gas industry. Aging facilities threaten workforce safety and affect product quality. The number of incidents related to facility management and a qualitative discussion of facility risk management were used as metrics to assess these effects.

2.2.3 Air Quality (S3)

In addition to product safety, air quality is also an indicator of pollutants in the manufacturing process. To assess this effect, the emissions of pollutants such as nitrogen oxides (NOx), sulfur oxides (SOx), and fine dust, the number of densely populated facilities near the factories, and the number of incidents caused by pollutant emissions were used as metrics.

2.2.4 Workforce Safety (S4)

Workforce safety is also an important indicator of CSR and has a profound impact on corporate valuation [41]. This indicator includes safety incidents and occupational diseases caused by poor working environments. To assess these effects, the ratio of the occurrence of safety incidents, number of reports of silicosis, and qualitative discussion of workforce safety management were used as metrics. It is necessary to modify these indicators to include more diverse occupational diseases.

2.2.5 Affordability (S5)

Affordability directly relates to corporate revenue. Corporations try to provide products at low prices to ensure price competitiveness in the same industry, but in special cases, such as drugs, the government restricts prices; thus, additional efforts are required to meet these prices. To assess these effects, the costs incurred to meet the price restriction and price increase compared with the inflation ratio are used as metrics.

2.2.6 Genetically Modified Organism (GMO) Management (S6)

Genetic engineering is not only used for food but also in various industries related to pharmaceuticals and healthcare [42]. Recently, products related to GMOs have been recognized as acceptable products in the market. However, in the case of GMO foods, there are business risks due to consumer health, and environmental and ethical issues. To assess these effects, only a qualitative discussion of the GMO usage strategy was used as a metric. However, more quantitative metrics are required.

2.2.7 Managing Skilled Workforce (S7)

Employment stability and job development indicators of individual employees are important factors in corporate ESG evaluation criteria [43]. As reliable ESG institutions, such as KLD, Sustainalytics, Moody’s ESG, S&P Global and Refinitiv, selected this indicator, corporations need to select their business directions to promote individual growth of their workforces [33]. Recently, as new technologies, such as AI and robotics, have replaced existing jobs, corporations have suggested various policies for employee education and job security. To assess these effects, the ratio of foreigners to overseas workers was used as a metric; however, more appropriate indicators need to be selected.

2.3.1 Systemic Risk (G1)

Corporations quantify risk factors by creating sustainability risk models. Factors that determine the value of corporations, such as the environment, society, economics, law, and politics, are intertwined and influence each other [44]. In this relationship, the extent to which changes in certain factors affect the existence of a corporation and the probability of such changes should be evaluated. To assess this effect, only a qualitative discussion is used, but quantitative metrics, which are probabilistically modeled with uncertainty, may be effective.

3.1.1 VR/AR for Training High Skilled Workforces

VR is a digital artificial environment that allows one to feel like a natural environment and immersively experience a new atmosphere. AR enables users to experience an abundance of information by overlaying a virtual environment in the real world. These mixed realities train highly skilled workers in the industry [49].

VR/AR is an effective tool used to educate workers to acquire advanced skills by increasing their equipment utilization abilities. In environments wherein dangerous situations may occur during training, VR training provides safer instructions and is used to train workers to cooperate in the assembly process with a robot [50]. In an actual environment, unexpected robot moves may threaten human safety and injure workers; therefore, it is safer to experiment and verify these in a virtual environment. VR technologies are not only effective in creating skilled workers who can work with robots but also have additional effects, such as the optimization of working hours and safety training. A game-like virtual environment, referred to as a Virtual Reality Training System (“Beware of the Robot”) enables industrial robots and humans to perform simple manufacturing tasks together [51]. Attempts have been expended to train aircraft maintenance door manufacturing procedures in the MR [52].

Although AR/VR technologies can improve the skill levels of workers through immersion and provide practical safety training, there are also research results demonstrating this limitation. If the educational subject is simple and does not harm workers, they will not feel the necessity to use a virtual environment [52]. Therefore, instructors should consider the use of AR/VR technology and how it affects immersion.

3.1.2 VR/AR for Safety Education

VR/AR is an effective tool for safety education. Direct and safe training in manufacturing environments is challenging. In conventional training procedures, lecturers convey professional knowledge to workers using words. Moreover, workers could not perceive how dangerous the machines and situations were. However, the use of VR/AR can provide an immersive environment and situations that are harmful to real life; in this context, effective education is provided. There are considerable cost and safety limitations associated with worker training in the chemical industry. It creates a game-like VR program to instruct workers to be exposed to risks or take action [53]. Instructors conduct VR training on dealing with fire while operating a numerically controlled machine [54]. The VR training environment was configured to prevent incidents related to falling from stairs and pedestrian movements. In addition, both the short- and long-term memories of safety instructions were improved [55]. An AR system that works with a computerized numerical control (CNC) machine was designed to teach equipment use and safety procedures [56].

Table 1

Summary of core technologies for the men/women and their related ESG indicators

3.1.3 Wearable Devices

The working environment presents considerable potential hazards. Wearable devices, such as exoskeletons and IoT wearable devices, have been developed to protect workers from unexpected accidents or chronic stress. In addition to being exposed to hazardous environments, workers can be easily affected by work-related musculoskeletal disorders, such as lifting heavy objects or performing repetitive tasks in uncomfortable postures [57]. In particular, back disorders due to non-neutral trunk postures are common in the automobile industry [58]. Therefore, efforts are being expended to improve the health and welfare of workers in the automobile industry. A wearable mechanism, called an exoskeleton, was developed for workers in uncomfortable postures [59]. By wearing an exoskeleton, a worker can work in a more comfortable position and reduce the required force. The exoskeleton to be used in the automobile industry is divided into upper and lower limbs. The exoskeleton used in the automobile industry is either a passive structure that supports the waist or an active design that helps overhead work. Wearing the upper limb passive structure and performing the static posture task improved the duration by 31.1%; when the lower back support was worn, instability from the lower back and overall fatigue were reduced [60]. When the ABLE exoskeleton is worn, which is mainly used for rehabilitation, and during the execution of the screwing task, which is often used in the assembly process in the automobile industry, the operator’s pain is reduced, and the required torque is reduced by 38.9% [61].

Sensors have been designed and introduced to the energy industry to improve the safety of workers, jackets, shoes, and carabiners. The wearable device improves worker awareness, peer supervision, and the ability to respond quickly to emergencies [62]. A wearable device with a heart rate, accelerometer, and Surface electromyography (sEMG) can measure operator fatigue. The risk to workers can be managed by modeling fatigue using recorded data [63,64]. There have been attempts to protect workers’ safety by making a helmet composed of sensors and IoT systems, and by developing AI that can detect workers falling from high places using vision sensors [65,66].

However, there is still a barrier to the introduction of these wearable devices. These devices continuously monitor workers’ personal information so that wearing cannot be forced to do so because of due to privacy issues. Also, experienced workers experience inconvenience and discomfort with these devices. Technically and legally, these issues need to be resolved [67].

3.1.4 Working Environment Monitoring

Integrated sensor systems, IoT, and AI can manage the working environments and health. An attempt was expended to quantify workers’ stress or fatigue levels and improve their well-being using a wearable device equipped with sensors and IoT systems [68–71].

Manufacturing work environments contain noise, dust, and hazardous gases that threaten the safety of workers. Low-cost sensors can be integrated to monitor and measure particles, carbon monoxide, ozone, nitrogen oxide, and noise. Multiple monitoring systems are installed in heavy vehicle manufacturing factories to monitor gaseous emissions and noise and provide a safe working environment [72,73].

3.2.1 Machine Monitoring System

Machines are elements directly involved in manufacturing products in factories. During various processes, machines inevitably generate additional outputs in various forms [77].

The most significant characteristic of machine outputs is that they take various forms. These include vision-type information that can be easily acquired with a camera, heat, and sound generated during processing and the amount of electricity used by machines; this information is easily collected. To collect data on a machine or phenomenon in a smart factory, one should consider which signal is the most effective. If the most suitable signals and techniques for processing are determined, useful monitoring information that cannot be obtained using intuitive correlation alone, such as in existing factories, can also be obtained.

The machine monitoring system consists of various scales; these range from the monitoring of the entire system (consisting of machines) to detailed monitoring to diagnose component failures. The primary objective of large-scale machine monitoring is to monitor simultaneously the status of each machine in a complex factory environment. In particular, the current is frequently used for this type of monitoring because it is an intuitive, low-noise signal [78] that indicates the on/off states or normal operation of each machine. Considering the smart factory technology applied to clothing factories in 2019 [79], it was possible to monitor the entire process line with high accuracy by collecting simple electric energy consumption information and by introducing convolutional neural network techniques. There is also a precedent for determining the operation of each machine without installing a complex sensor in a complex factory environment achieved by acquiring the operating sound of the machine, which was previously considered as noise [80].

Various signals can be used to monitor not only the entire process but also detect single-component machine problems. Typically, bearing condition monitoring occurs inside a machine and cannot be easily assessed as in the cases of visual inspection [81]. Mechanical engineers have developed bearing-monitoring techniques based on current and microscopic acoustic emissions [82,83]. Similar systems are applied to smart factories and used in monitoring systems that predict the lifespan of each component and diagnose problems. The remaining life of rolling bearings and other tools can be predicted by combining relative features and adapting multivariable support vector machines [84,85]. The lifecycle of the entire system was also predicted using these small unit predictions [86].

3.2.2 Machine Operation Design

If machine monitoring is a passive core technology that recognizes and predicts problems, then machine operation design is an active technology that sets machine configurations for optimal processes based on data. Optimization of the machine based on the information acquired through the sensor, referred to as numerically controlled (NC) optimization, is based on a technology that existed previously, and a smart factory is the best environment for its easy use.

Operational design is often used to improve the conditions and lifetime of machine parts. This is achieved by changing the figure of one element of the machine operation based on the data obtained by the sensor [87]. One of the most representative examples is speed control based on spindle load. The spindle feed rates of various machines can be optimized by executing adaptively an NC program [88].

Machining condition selection is another way to use machine-operation design. This is the most commonly used example in CNC milling machines, where the cutting transfer parameter is automatically set according to the processing type. The process of minimizing the surface roughness is also designed by analyzing the information to be processed in a digitized form [89]. Attempts have also been expended to reduce the grinding time of the grinding machine and improve the surface quality by measuring the cutting speed, wheel speed, and feed rate using sensors [90].

3.2.3 Machine Energy Saving

Another advantage of monitoring and properly controlling machines inside smart factories is that energy consumption can be minimized. Various attempts have been expended to monitor the energy consumption of smart factories and use it as a basis to predict data to induce continuous energy consumption [91,92].

Energy monitoring eliminates leakage power by monitoring energy consumption and optimizes the process itself based on the acquired data. One advantage of digitizing the various elements inside a smart factory is that the factory system can be digitized. This is also helpful in terms of energy conservation. By creating an energy consumption model, it is possible to increase the energy consumption efficiency of machine tools [93]. Additionally, it is possible to quantify the energy consumption characteristics of a machine; thus, a power-saving strategy can be devised [94]. In a 2014 study, a method was proposed to minimize energy consumption by optimizing the selection and operation order of machine tools based on the situation [95]. This may be due to the unique technology utilized in a smart factory, which can collect different types of information in real-time to draw the desired conclusion.

3.3.1 Material Flow Analysis

The technologies applied to factories in the early days were developed to increase productivity and efficiency. However, as each technology affects production, economy, environment, and society as a whole [96], interest in sustainable development has increased. Moreover, government regulations and standards, such as international standardization organization 14044, have been established [97]. Research on lifecycle assessment (LCA) has been conducted to evaluate the impact of each process on the economy, environment, and society based on sustainability throughout the entire product lifecycle and value chain [98–100]. LCA attempts to quantify the overall environmental [101–104] and economic impact [100,105,106] in terms of material and energy consumption [107] and carbon footprint [101]. LCA has emerged as a valuable decision-support tool for policymakers and industries to assess the cradle-to-grave impacts of a product or process.

Table 2

Summary of core technologies for the Machine and their related ESG indicators

3.3.2 Integration of Materials with Product Lifecycle Management (PLM)

In conjunction with quantitative indicators, such as resource utilization (including material and energy), waste emissions, and costs, that had been previously derived through LCA, the field of PLM has been extensively studied for methodology development to improve resource sourcing and utilization [108–111], energy efficiency [112], and cost reduction [113] throughout the resource flow path [114]. In particular, in the product lifecycle, resource flow is subdivided into six stages: reduction, reuse, recycling, recovery, redesign, and remanufacturing (which is called 6R) [115], and studies have been conducted to improve each stage [112,116,117].

3.3.3 Material Processing Development

In smart factory production, processes, including manufacturing, must be reproduced if quality standards are not met, such as the defects that occur during the manufacturing process [118]. Therefore, in-process quality monitoring [119–122] is one of the research fields that aims to improve the efficiency of manufacturing processes. In addition, efforts are being expended to use new eco-friendly materials (green materials) [123] to satisfy the criteria on the increase of the conversion rate of resources, such as materials and energy, and the reduction of waste emissions. In addition, the development of a new process to process these materials is an elemental technology of smart factories [124–126].

3.4.1 Industrial IoT (IIoT)

The IIoT is a technology that optimizes the IoT for industrial sites. Traditional manufacturing equipment at industrial sites efficiently collects and manages data based on the manufacturing execution system (MES); however, IIoT is more efficient, manageable, and accessible through organic connections between each piece of equipment and sensor. Technologies for fast, accurate, and universal data exchange among manufacturing equipment are being developed using IIoT-based networks [140–142].

The IIoT is used in various manufacturing industries. It is possible to collect data accurately and quickly and analyze them in real-time to predict the state of equipment by applying IIoT technology to sensors installed in individual manufacturing equipment [143,144]. In addition to manufacturing equipment, Heating, Ventilation, and Air Conditioning (HVAC) facilities essential for factories can use IIoT technology to monitor a factory’s atmospheric environment in real-time, thus enabling a rapid response to gas leaks and fires. [145]. Using the IIoT, the safety and situation data of workers in a factory environment are monitored in real-time to improve the productivity of the factory and respond rapidly to crises [146,147]. In addition, the reliability and stability of smart factories are ensured through IIoT data exchange efficiency and security research [148,149].

3.4.2 Big Data Analytics

Big data analytics is a technology that can extract features and improve target performance based on vast and unstructured data collected through various sensors in a factory environment. It can learn an AI model based on big data collected in a smart factory environment and predict the conditions of equipment or process results based on the learned model [150–153].

Table 3

Summary of core technologies for the Material and their related ESG indicators

Big data are analyzed using mathematical methodologies, such as machine learning, and are used for various purposes within a smart factory [154]. Detecting process conditions or anomalies based on power or vibration data from manufacturing equipment [155–158]. Power or vibration information in the manufacturing industry is in a time-series format that contains considerable noise, and it is difficult to extract characteristics using a heuristic method. Big data can be trained with an artificial neural network-based ML algorithm, such as a recurrent neural network (RNN), and meaningful results can be extracted from vast amounts of data by creating an AI model suitable for the characteristic. Thus, the operational stability and product quality of the smart factory can be improved.

3.4.3 Cloud Computing

Cloud computing is a technology that can flexibly store and manage the data generated by manufacturing without loss. In a smart factory, vast and diverse types of data are collected and processed in real-time, and existing local servers cannot respond flexibly. Cloud Computing enables efficient and stable data processing because it can utilize computing and other system resources related to the cloud according to circumstances and requirements [159,160].

In a smart factory, where the process sequence or work method changes in real-time according to the demand or form of the product, the amount or form of collected data also decreases or increases depending on the situation [161]. Cloud computing enables efficient and flexible processing of vast amounts of data [162–165]. In addition to cloud-computing technology, edge computing, and communication protocol technologies have been developed. Data collected from the manufacturing floor are preprocessed in the edge computing system layer and uploaded to the cloud server to improve computing efficiency [161,165,166]. In addition, the open platform communications unified architecture (OPC-UA) communication protocol is used to simplify the communication and data collection processes between heterogeneous devices [167,168]. Cloud computing and additional technologies can help address unnecessary costs or operational risk problems caused by local servers.

3.4.4 DT

DT replicates the characteristics of a physical system in a virtual environment and feeds back the results obtained in simulations to a real system [169]. Real data obtained from core technologies, including manufacturing equipment and robots, can be simulated in a DT environment, and the optimization results obtained for the target work performance can be reflected in an actual system in real-time, thus enabling organic and efficient factory operations [12,170,171]. DT technology requires all the element technologies of Industry 4.0 because it analyzes big data obtained from the IIoT communication with a learned AI model, and the system itself must be able to operate the entire factory system based on the obtained results. In smart factories, DT can improve factory productivity based on optimized results by simulating the processing status or logistics of individual manufacturing equipment [172]. After establishing various processes in a DT environment, DT can improve the quality of the workpiece or reduce waste materials by determining the optimal parameters based on simulations [176,179,180]. It is possible to improve the overall productivity by optimizing the complex process sequence with DT [174,175,181], or by automatically controlling numerous logistics robots in the factory based on deep reinforcement learning algorithms, thus enabling safe and stable factory operations [173,177,178].

4.2.1 Advancement in Machine/Material/Method for the Workforces (S4 and S7)

The findings in Fig. 6 confirm that technologies on workforce safety (S4) and managing a skilled workforce (S7) receive less attention in terms of machines, materials, and methods compared with men/women. To create an advanced safe factory, it is necessary to consider how men and other smart factory elements can affect worker safety. As workers process objects using machines in factories, they are exposed to the potential dangers of machines. Specifically, the use of legacy equipment without protective devices is dangerous. However, most SMEs continue to operate their legacy equipment for many years (for example, equipment older than ten years may be used), and worker accidents would easily lead to major accidents. Therefore, appropriate smart-factory technology is needed that can be configured to legacy equipment to a) ensure worker safety, and b) to study equipment using the latest safety devices [48]. In addition, as the material is processed in a step-by-step manner, the byproducts and gases that emerge can affect workers’ health. For example, factories produce numerous harmful substances, such as byproducts from the oxidization of organic substances, ozone gases from welding, and materials used for semiconductor etching and deposition. Many efforts have been made to monitor these harmful factors using sensors and the IoT and to control safely the environment of the plant [72,73,145]. However, research on creating safer materials should continue because the fundamental problem lies in the materials. For example, polylactic acid, a biodegradable plastic, has poorer physical properties than acrylonitrile butadiene styrene; however, its demand is increasing because of its simplified physical and chemical recycling processes and the reduced toxic gas emissions during processing. There are great possibilities for the development of safe materials that have smaller impacts on the environment. The development of new materials can affect the outcomes of these methods. For example, suppose that a material can be processed at a lower temperature or does not produce dangerous byproducts. In that case, the process will also be simplified and developed in a way that can protect the safety of workers. Manufacturing processes have been developed to improve the productivity and quality of products. In factories, noise from cutting and hitting materials can cause hearing and irreversible lung damage due to fine particles generated during the process. Therefore, the use of process development to reduce harmful effects on workers is also required in terms of workforce safety.

4.2.2 Advancement for Air Quality (S3)

Based on this review, the research that has been conducted on the air quality of machines and materials has been limited and its progression has been relatively slow. A few studies related to air quality have been conducted on the environment inside factories and (in most cases) on the urban environment scale of the surrounding factories. According to a recent study [182] that examined air quality in a furniture factory, it was demonstrated that air quality changes dramatically depended on the progress of the task. This suggests that the machine or material used for processing can significantly affect air quality.

To study how machines and materials can affect air quality, it is necessary to quantify the amount of byproducts or dust generated in each process. As shown in [182], a large amount of data is required to quantify the amount of harmful dust generated by assembly, drilling, and surface treatment methods. Therefore, it is necessary to determine the correlation between the effects of machines and materials on air quality. Because there is no research in this area, the first challenge pertains to the determination of the part of the machine that needs to be improved to generate less dust and the identification of the physical properties of the material which should be improved. Appropriate criteria representing air quality in a factory should then be defined. This approach has a high probability of a tradeoff depending on various indicators, thus indicating the efficiency of the entire process and the machine and material used. In addition, it is important to establish an appropriate balance between the machine and the material.

4.2.3 Technology for Linking Smart Factory Technologies to Governance (G1)

As summarized in Section 2.3 and Fig. 6, the government indicator related to smart factories is much smaller than that of the environmental and social sectors. Although research on sustainable management has been conducted for a long time, as it has been mainly conducted on the environment and workers, it has become an issue of governance owing to the recent emergence of ESG. However, research on this topic is lacking. In addition, the values and interests pursued by each stakeholder are different, and various factors are interconnected in a complex relationship when measuring governance, thus making it difficult to identify and quantify this relationship compared with that related to environmental and social factors. To address this concern, more research on data-driven modeling based on big data should be conducted on the complex relationships related to governance. Furthermore, based on DT technology, it is possible to gather all the information generated within a smart factory transparently. In addition, efforts at the government level should be expended collectively so that the government of a smart factory can evaluate whether it is doing well.

4.2.4 ESG Adoption in SMEs

By reviewing smart-factory technology and ESG indicator cases, it was found that these technologies were designed and developed around large companies that led the market. As mentioned in the introduction, the latest smart technologies have been used in lighthouse factories, and have demonstrated the best performance in terms of quality and productivity indicators. However, SMEs with limited investment and research capabilities cannot easily adopt the latest technologies [18,48]. The adoption of ESG indicators by SMEs is also similar. Various measures have been proposed that focus on the ESG management of large companies with large investments and operations. These changes require SMEs in the supply chain to improve their ESG factors. However, similar to the previous smart factory case, it is difficult for SMEs to respond to the additional requests in addition to producing low-value-added products. This can be referred to as the appropriate manufacturing technology recently proposed for SMEs in smart factories. Appropriate manufacturing technology can lower the barriers and facilitate the adoption of smart technologies through affordable and easy-to-use concepts that can be accommodated by manufacturing SMEs. Similarly, ESG improvement can be achieved by developing appropriate ESG implementation technologies that can be easily introduced in the field by analyzing the current statuses of SMEs and related smart factory technologies. Thus, new possibilities for ESG management in the manufacturing industry are presented. For example, if safe smart-factory technology is introduced to reduce industrial accidents related to worker safety, it will improve the social sector in ESG and the external reliability of SMEs [183].