According to ISO 9000, quality is the degree to which a set of inherent characteristics meets requirements. Quality rates the output of the production process as a whole, but also of its every stage. It is affected by process parameters like, temperature or accuracy and speed. Digitalization enables operators to produce products of higher quality using several methods that improve process transparency. These methods include early detection of deviations and correction within control circuits and optimization or prevention using data analytics.
Taking up output volume, process efficiency drives the input-output ratio and production reject rates, which can be optimized by fluent, faster and real-time coordinated production flows.
Faster production flows also support lead time optimization, which describes the shortening of the time passing between initiation and completion of the production process including all preparation and waiting times.
Asset utilization provides a sophisticated view on the actual performance status, which is expressed in KPIs such as the overall equipment efficiency (OEE).
Part of planning the manufacturing processes is resource allocation. Via assigning resources and scheduling activities, businesses ensure the fluent production processes.
In order to enhance production flows, worker guidance and assistance systems, for instance machine-to-machine or machine-to-man communication and cooperation can be implemented.
Production planning and control, in the sense of digitalization, comprises monitoring, analyzing and interpreting data that is collected and transmitting orders throughout a company–wide IT network.
To identify the key use case cluster in this white paper key design principals of Industry 4.0 technologies and applications are considered. First, the development of I4.0 is driven by providing an interconnection between different objects and assets in the production ecosystem. This enables interaction like machine communication. In this context also setting communication standards and security regulation is a key aspect. Deriving from established interconnection the next key driver is information transparency. Through the interconnection of the physical and virtual world, a new level of information is provided and the complex correlation between information can be analyzed. This also acts as a base for decision making. More information available offers the opportunity to decentralize intelligence and therefore the decision-making process. In the last step, this can be used to assist workers. Through the responsible shift of workers from executive force to decision-making force information needs to be aggregated and visualized comprehensively. Also, physical assistance is implemented in this process.
Indoor Track and Trace
The characteristic of indoor Track and Trace is the localization of monitored assets by scanning the tag on the object at a certain position in the production. This predefined position can be assigned to a certain process step, machine, production area or warehouse. The technical solution to this use case consists of a tag and a scanner station or reading unit. Objects are tagged with barcodes or RFID tags to provide them with identification. Indoor Track and Trace enables the identification of assets by marking them with a uniquely assigned tag. Through the scanning at defined process stations the current position, dwell time and routes in the process chain can be tracked. Therefore, the solution increases process transparency leading to the detection of speed losses, bottlenecks and search times.
An application example for Indoor Track and Trace is the position tracking of materials in the production hall and warehouse. Also, the position and usage tracking of hand tools at different working stations and areas is possible.
Indoor tracking with barcodes and RFID tags is the most implemented identification method. The tag design is determined by the code position on the object and readability. The tracking with barcodes or RFID tags is based on visual contact with the reading unit. The position of the reading units is identified with a specific position and can either be directly connected with a computer or via a wireless internet connection.
The communication requirements for this use case subclass focus on the data transmission at defined workspaces, since the barcode and RFID scanners are located at these positions. The connection for information exchange is provided by a desktop computer with an internet connection and connection to other systems like an ERP system. For the requirement categories connection density can be rated high since multiple assets are tracked and scan stations are used. Also, high reliability is needed to ensure no asset data and status is lost.
Due to the fixed position of the scanner units these use cases are feasible with a WiFi connection, which is also in line with the current state of the art and provides a cost-efficient solution. Comparing WIFI 4 and 6 WIFI 4 is partly functional for the use case since it limits the asset number connected with the network. 4G and 5G do not limit the asset number but are not the cost-efficient solution for this use case subclass.
Outdoor Track and Trace
Outdoor Track and Trace includes all tracking activities placed outside a production hall or warehouse. Often manufacturers have multiple production sites where different steps in the production process chain are executed. It can be used to track the positions of trucks of the transportation fleet. Outdoor tracking with a continuous signal is enabled by mobile tracking based on positioning via GPS or cellular connection.
The load tracking of trucks allows production planning depending on the arrival time. Further it optimizes the handover between sites and can also be applied to control the external supply chain.
One application example is the just-in-time supply chain. The current position and arrival time will be constantly updated to the recipient. This allows improved production planning and resource allocation.
Outdoor tracking between different facilities is mainly used for the transport of valuable goods as well as for time-critical transports.
The tracking solution is only feasible with a GPS signal or localization based on the mobile phone position. Other solutions are limited to signaling inside production halls. The application of cellular tracking is also more appropriate for outdoor tracking since inaccuracies up to 45m are possible.
For outdoor tracking of transports between different production locations as presented in this application example the telecommunication requirements in the defined categories are low to medium. However outdoor T&T needs high connection security since its often applied to track the transport of valuable goods. Outdoor tracking with WIFI is not possible, therefore only 4G and 5G can be considered for the scoring. 4G would be the recommended technologies among the considered technologies. It provides sufficient coverage and localization with an accuracy of 45 m.
5G would not be cost-efficient for the use case presented here for outdoor T&T. However outdoor tracking with 5G could be considered for extensive logistics areas like freight depots, container terminals and ports. In these areas an application of a 5G campus network, with a localization function yet to be developed, could be conceivable.
Real-time Location Systems (RTLS)
In contrast Real-time location systems (RTLS) rely on real-time data about assets tracked in the production facility. RTLS consist of tags placed on the monitored assets and antennas providing the areas with even signal coverage to enable wireless communication. The exchange of information and data between transmitting and reading units is not only possible in spatial proximity or in the case of targeted encounters between transmitters and receivers like in the subclass “Indoor T&T”. The communication can be based on different tag variations. These not only enable map-based location tracking of assets, but offer the possibility to track further information. This includes the asset status (e.g. battery status), condition (e.g. progress, scrap), history and detailed product information (e.g. order number, final product properties, and characteristics). Also, the tracing data saved on the advanced tags can be edited and further information added by the worker.
The solution allows a holistic real-time visualization of material and process flow throughout the production facility. This provides the basis for process optimization and intelligent just-in-sequence production leading to reduced costs and lead times.
The RTLS can be used to track the processing status of a product route through the manufacturing processes. Also, real-time AGV tracking with information about loaded orders, materials or tools and real-time arrival notification for the worker is possible.
RTLSs are applied in a highly automized manufacturing surroundings, for standardized processes and in areas with decentralized process control.
For the data transmitting different communication standards are available on the Track and Trace market and are implemented in production halls. The most popular solutions for wireless communication between tags and antennas are Bluetooth, Ultra-wideband (UWB) or ZigBee. Telecommunication technology based on the current 4G standard is not common. Further the localization within a 4G network is not precise enough to track assets inside a production site.
In order to implement real-time tracking of different assets like orders, material, AGVs or tools special requirements have to be considered. Firstly, production halls have special requirements according to the signal quality and even illumination is needed to ensure real-time tracking. Signal disturbance can be caused by metal objects like shelves and machines. Secondly, location tracking based on different zones requires accurate geofencing for clear differentiation of process steps.
The requirements for data throughput and connection density are driven by the number of assets to be tracked, the higher the number of assets tracked the higher the requirements. Since it’s a real-time application the latency requirements are high, but latency also needs to be considered with regard to information and data recipient. The latency requirement for a worker is lower than for the handover of information between machines, because of different reaction times. Furthermore, the integration of larger numbers of assets tracked and the handover between process steps results in a high security requirement.
State of the art for data transfer and communication in RTLS systems are standards like Bluetooth, UWB or ZigBee. These technologies provide high accuracy in localizing the objects and can be expanded modular to provide even signal coverage. Looking at cellular communication, location tracking with 4G is not feasible for this use case subclass since the localization is too imprecise.
However, 5G could be the enabling force for cellular RTLSs. Its technology aims to provide new localization possibilities which enable a tracking function with a cellular network. Also, 5G technologies can build a reliable and secure network allowing high connection and data density to provide a holistic T&T solution.
Defined path transportation
A fixed or designed path is an actual physical guide path for the vehicle. They can move by following a path defined by inductive wires that are buried or embedded into the ground. Other options are surface mounted magnetic or optical strips for vehicle guidance. These paths are defined during the layout planning of the production hall. Predefined transport paths offer restricted but secure transport. Set within separate areas, paths ensure independent transport without interfering with other assets and workers in the facility. The goal is to implement a lean and efficient intra-logistics to realize cost and time savings.
A possible application for defined path transportation is the transport of heavy and bulky goods to discharge workers and improve occupational safety. Another example is speed line transportations, where smaller goods and batches are transported with increased speed.
Defined path AGVs based on magnetic tapes, wires or laser orientation are in common use in different industries and realizing transport in hazardous areas to improve efficiency and worker safety. The market also offers a variety of different AGVs carrying different technical features to lift, grab, hold and load goods. Defined paths AGVs provide a cost-efficient solution for intralogistics. Also, WiFi mesh solutions allowing communication of these vehicles are available on the market.
Since the AGVs move along a defined path, the communication requirements are lower than those of other use cases in this cluster. Communication is based on the position and status of the AGV. An exchange between the vehicles is not necessary. Workers receiving deliveries from the AGV have access to a computer in order to confirm tasks which can then be sent to the AGV without critical requirements. Also, paths for AGVs are defined at the planning stage of a site so signal coverage can be considered. The communication and data transfer is feasible with all the presented technologies, but the application is most cost-efficient and practicable with a WiFi 4 communication.
Autonomous independent routing
In this subclass, AGV transport is carried out on varying routes within defined areas. AGVs are equipped with additional sensors and technology facilitating navigation on production sites or in warehouses and reacting to obstacles and crossing workers. Autonomous routed vehicles have functions of environment awareness, real-time decision-making as well as behavior control and execution. It is one of the key equipment of flexible modern production lines, assembly lines and warehouse automation systems. AGVs can take on multiple tasks in one transport and routes can be adjusted according to the altering needs in production. The implementation of these AGVs can be retrofitted without defining restricted areas. The AGVs have advanced equipment and are therefore more expensive but can be applied in a set system, providing automated material flow and worker support.
Mostly autonomous AGVs are used to transport smaller goods which can be navigated through the plant faster and more reliable than through manual transport by the workers. Destinations of transports can be adjusted to deliver goods to workers. Furthermore, goods can be transported from the warehouse to the ordering worker on demand.
Regarding the state of the art of independent routing inside production sites and warehouses, different technologies for the AGV navigation and sensors for surrounding recognition are implemented. A geo-guided AGV recognizes its environment to establish its location. Without any infrastructure, the AGV equipped with geoguidance technology detects and identifies columns, racks, and walls on the shop floor. Using these fixed references, it can position itself, in real-time and determine its route. Vision-guided AGVs can be installed with no modifications to the environment or infrastructure. They operate by using cameras to record features along the route. The vision-guided AGV uses 360-degree images and builds a 3D map to follow a trained route without human assistance and additional special features, landmarks or positioning systems.
Navigation without retrofitting of the workspace is called Natural Features or Natural Targeting Navigation. One method uses one or more range-finding sensors, such as a laser range-finder, as well as gyroscopes or inertial measurement units to understand its whereabouts and dynamically plans the shortest permitted path to its goal. The advantage of such systems is that they are highly flexible for on-demand delivery to any location. AGVs can handle failure without bringing down the entire manufacturing operation since they can plan paths around failed devices. To enable these functions the AGVs need a real-time connection to send their location, receive new tasks and plan their routes accordingly.
The communication requirements for AGVs with an independent routing function are higher compared to defined path solutions. Providing independent routing, they need a continuous connection to the control system to receive their task and send the current status. In order to enable a dependable system and ensure continuous material flow, requirements for reliability and security are higher. Furthermore, the AGVs need to react to changes faster, which results in higher dependency on the latency. Due to the mobility of the AGVs a reliable cell handover is required. For WiFi 4 a fast cell handover is not feasible. This function is improved with WiFi 6, but still handover speed depends on the number of devices connected to a cell. To enable seamless cell handover 4G would provide the needed reliability and connection density among the options considered. However, similar to RTLSs there are wireless cost-efficient communication technologies like Bluetooth or UWB available.
A use case implementation with 5G is also conceivable, since 5G will provide new localization functionalities as well as providing higher reliability and security. Further a higher number of AGVs can be connected and the integration of different use cases with one communication technology can be considered here.
Intelligent (central) guided routing
Other transport solutions follow the trend shifting towards a centralized intelligence which the AGVs access and act upon. The route planning of the AGVs is organized for the whole transportation fleet in mutual dependence. Therefore, the AGVs need to communicate and cooperate with each other, so that material transportation and loads are optimized. The basis is distributed data management and handling between the whole fleet. This central data is used to perform intelligent road mapping, plan the shortest path based on the AGV position and status, as well as realize the smartest dispatch possible. Permanent communication regarding routes, orders and current position is necessary.
Further functions added to AGVs also enable observation of production grounds and find missing items with camera-based object recognition.
The intelligent routed AGVs provide continuously improving autonomous transport that enables just-in-sequence deliveries with a smaller fleet compared to the other.
Using intelligent guided AGV fleets, various goods from materials to hand tools can be transported through the production hall and warehouse. Orders placed by workers can be delivered to their destination on the shortest way. Also, efficiency and reliability can be ensured.
Because of the enormous complexity and required IT infrastructure a decentralized intelligent fleet is challenging to implement into a production environment. The needed IT infrastructure is partly still in development for broad application.
Through the decentralized routing intelligence, the AGVs are more compact, mobile and cheaper because of the minimization of needed on-board computing power. Further the system allows updates on the go.
As constant connection and collaboration of the AGVs are required, connectivity requirements are very high. The central intelligence is based on continuous data streams and data analytics. Different mathematical models, which are adopted from optimization algorithms, are implemented to serve as the basis for route calculation.
Resulting from dependent task assignments, dispatching, and coordinating the AGV fleet requires a permanent connection and communication. Vehicles need the ability to react rapidly, therefore latency requirement is high. Also, the amount of data exchanged between clients is higher compared to the other AGV options presented. This also results in higher data throughput as well as a higher requirement for the user experience data rate. As the intelligent AGV fleet is designed to optimize efficiency and material flow in the manufacturing process, it needs to be highly reliable. Following the requirements and the mobility assumption as stated in the use case subclass before implantation with WiFi 4 and 6 is not feasible. To ensure a permanent and reliable connection between the vehicles 4G can be recommended.
As this use case subclass further follows the trends of I4.0 for interconnection, transparency and decentralized decision-making value could be added through 5G on different levels. With 5G more assets could be tracked and more data could be recorded. A 5G network could also enable updates on the go for AGVs and provide them with continuously improved models for the routing calculation. Further, as AGVs regulate the process flow 5G provides higher security levels. Another aspect of the value creation of deploying 5G could be the integration of other use cases, for example, an RTLS into one network without having multiple technologies present in the production site.
VR as support for engineering and product development
In production, VR is mostly used in production hall design as well as supporting product development and engineering processes. VR content is displayed via a VR headset. Played video content is specially designed and rendered for VR applications.
Through the visualization of product and construction data, development can be made more accessible. Engineers have the possibility to engage virtually with products or experience work surroundings in the planning phase. These showcases are also often used in marketing. Through VR different visualizing data like CAD files can be streamed and made perceptible. Therefore, the planning process can be supported and enables early detection of planning errors.
An application example for VR in manufacturing is shop floor planning. A manufacturer plans a new assembly line, which includes workstations for human workers and process steps which will be performed by robots. To simulate the workflows, check collaboration and security between robots and humans, data about machines and future workplaces can be experienced in the production hall.
Currently, VR is focused on development actions as stated above. VR elements are technologically mature and the market offers different products. To further engage VR in a manufacturing process heterogeneous VR element technologies have to be connected efficiently, and research for further application in manufacturing has to be carried out. The development of standards and the extension of existing standards for dynamic integration are needed. Also, techniques that enable handling the whole factory at an instance, analyze it rapidly, and provide fast feedback to the shop floor are very important.
The content presented in a VR environment can be saved on the VR device directly or streamed from a central platform. If videos will be streamed to the end device special requirements apply. In VR environments, stringent latency requirements are of most importance for providing a pleasant immersive VR experience. The human eye needs to perceive accurate and smooth movements with low motion-to-photon (MTP) latency, which is the lapse between a movement (e.g. head rotation) and a pixel frame shown to the eyes. High MTP values send conflicting signals to the vestibulo-ocular reflex (VOR), a dissonance that might lead to motion sickness. There is broad consensus in setting the upper bound for MTP to less than 15-20 ms. To achieve this latency requirements streamed content is often buffered on the device. Crucial for the provision of video data in this regard is the data downlink. This results in very high requirements for the user experience data rate.
Comparing these findings with the considered telecommunication technologies a VR implementation on the factory level is feasible with WiFi, if the application is set in a static surrounding. This means, if the user streams content in one set position. If the user is moving through the production hall to experience the VR content, cell handover could be difficult to implement, and inconsistent transmission could result. Since the requirements for latency are very high a use case realization with 4G could be limited due to available properties. These obstacles are not present with a 5G communication and also a seamless handover will be possible.
AR for worker guidance
The key feature for AR in manufacturing is the support of workers with manuals and additional information about processes and machines. The information can be provided by using wearables (e.g. smart glasses) or content can be accessed via a mobile device like a smartphone or tablet. This offers portable access to all relevant documentation (e.g. manuals, videos, photos) and an overview of required tools and materials is provided. It can also provide workflow guidance with the help of 3D animations. In advanced systems editing of existing documentation during the task by taking notes or pictures etc. is enabled.
Through AR support of workers, the demanded quality is assured, learning processes and training is improved as well as non-value adding process time can be minimized.
Another possible application area could be maintenance support. The access to live telemetry data of specific machine parts while being present at the regarded machine. Also, cross-referencing of prior maintenance cases to reuse solutions established in similar cases could make the maintenance process more effective and quality-driven. This also demands to record statistical machine data for future planning of similar tasks.
An application example is worker guidance in an assembly line. The worker is equipped with safety glasses that provide AR features. Thus, he can access an assembly manual according to his position in the assembly line and access specific information about parts he has to process.
Another application example could be a navigation feature in AR glasses supporting the worker navigating his way through the production hall or warehouse.
Another possible application area could be maintenance support. The access to live telemetry data of specific machine parts while being present at the regarded machine. The worker can scan a QR-Code on the machine with his phone and see the parameters and characteristics of different parts of the machine. Also, cross-referencing of prior maintenance cases to reuse solutions established in similar cases could make the maintenance process more effective and quality-driven. This also demands to record statistical machine data for future planning of similar tasks.
Quality insurance for final product approval for cars could be conceivable. At the final approval cars could be checked by AR supported workers who final check e.g. the cap dimensions of the parts.
The application of wearables for staff in production is an important trend in Industry 4.0 technologies. The usage of mobile devices to display information is common. Since the hardware is constantly developing and standards are defined by research, hardware with a suitable performance for the industry application is developed. For a broad industrial application, it is desired that workers will be equipped with an AR compatible device that provides support and information in regards to daily changing tasks. Therefore, the content and relating instructions need to be streamed to the devices and be continuously updated.
For the worker guidance with AR, virtual content is layered over the real-world perception. Therefore, the data amount streamed to the AR devices is smaller than for VR applications. Latency and user experience data rate requirements are still very high. Further for the application in manufacturing surroundings, the number of devices connected to the system is high, which results in a high connection density demand. Compared to VR, worker guidance with AR not only needs to provide a downlink, but further needs to enable a fast uplink to react to the worker's action. For example, in quality control with AR the worker needs sufficient feedback from the system about parameters, values and status of the part and the defined quality limits. Therefore, the user must be equipped with a channel towards the server hierarchy to signal its position and what view it is observing in the virtual environment. He also must be equipped with a channel in the other direction to receive the data pertaining to this virtual environment. For application examples of AR for navigation also a seamless handover and reliable connection are needed. Hence as a result the presented AR use cases are partly functional with WiFi and 4G, depending on the mobility demand of the specific application and the latency required. An integration with 5G is fully functional and provides advanced security to ensure data safety for generated expertise and knowledge.
Data storage / Data Lake
In order to be able to use advanced analytics in production, all generated data has to be stored. In the field of industry 4.0, the collected data is characterized by high volume, velocity, and variety. This kind of data is called Big Data. In the past, data has been stored in relational databases with a certain syntax. These databases are limited in speed and flexibility. Therefore, they do not meet the requirements for dealing with Big Data. To offset these limitations data lakes are implemented. In data lakes, any kind of raw data can be saved and used flexibly for data analytics. The data is stored in a cloud and can be accessed from everywhere. Therefore, one data lake can be used for more than one production site.
As an example, the results of AI-based optical quality control can be correlated with sensors and process data to understand the impact of certain parameters on the final product. By using a data lake, these totally different types of data (e.g. image data, KPIs and process parameters) can be stored in the same data storage and processed much quicker than in a relational database.
Data lakes are increasingly common in manufacturing enterprises. The demand for analysis of enormous data amounts occurring in Industry 4.0 manufacturing sites makes data lakes an attractive solution for pattern recognition and process optimization. Nonetheless, as the usage of data lakes is spreading out the requirements concerning data transparency, quality and compliance rise. The demand for analysis of enormous data amounts occurring in Industry 4.0 manufacturing sites makes data lakes an attractive solution for pattern recognition and process optimization. Since there are no uniform connectivity demands, specifications need to be formulated according to the highest standards considered.
In an ideal environment, all company data is collected in a data lake. Cybersecurity standards of a data lake are extremely strict. Therefore, the requirements for the security of the telecommunication technology are very high. The collected data can be used for various process-critical data analytics, resulting in a very high need for the network. Today data lakes are realized by using WiFi. For all other criteria, this is sufficient. But as the result of data leaks is increasingly expensive for companies, only the implementation of a campus network covers reliability and security fully. Therefore, it can be viewed as an enabler to the full potential of this use case.
Statistic Data Analytics (descriptive analytics)
Statistic data analytics adapts performance characteristics like the process capability index, which is a statistical measurement of the process capability: the ability of a process to produce output within specification limits. Framework conditions, parameter limits and alternative courses of action are defined. Typically, data is being sent to a centralized database (relational database) or a cloud (e.g. data lake) and processed by a central algorithm. The processing of the data often involves manual process steps and the interference of a responsible human being. Therefore, descriptive analytics is not time-critical and is usually performed periodically. Descriptive analytics creates knowledge out of raw data leading to improved process transparency and lays the foundation for improvement measures for action.
As an example, quality data can be used to calculate the mean and standard deviation. The data is processed by statistical analytics software. This can be used to calculate different key performance indicators (e.g. OEE) and improve the process based on the results manually. Statistic data analytics is performed disconnected from the production hall and the results are used time-delayed.
Statistic data analytics has been used for the last decades. Through the creation of massive amounts of data (Big Data) in production, statistic data analytics is as effective as it has ever been. The challenge is being able to filter and process all existing data. A relational database is sufficient for statistical data analytics, but a data lake can also be used. More allocated data sets higher demands for the provided connectivity.
Statistic analytics is based on historical data and used for non-real-time decisions. Therefore, it is not time-critical to transmit the data. The amount of data of each machine, that is used in statistic analytics, is low. This leads to a low need in latency, range and connection density. The security requirement is high, because of the transfer of process data. Therefore, WiFi 4 can be used for statistic data analytics. The more data is collected and used the more sensible it is to switch to 4G or 5G. They provide higher security standards, lower latency and higher data rates.
Intelligent Real-time Data Analytics
New technologies provide more data. To use the full potential of big data, this data needs to be processed in real-time. Smart algorithms use machine learning to generate deep insights and find unknown correlations beyond the human analytics capacity to improve the process design and explore external influences not directly linked to the processes and machines. The more data is created and the longer these algorithms are in use, the better the results will be.
By implementing intelligent real-time data analytics of a production process, unknown correlations can be discovered. When using data analytics to improve the quality of a product, all parameters of the production process are analyzed. The result of the analysis shows, that a parameter needs to be adapted, which has not been associated with the production process of the product before.
The intelligent data analytics software which allows complex data analytics is expensive as well as implementing the needed IT-infrastructure. The results of the algorithms are only as good as the collected data. The amount of data that is collected differs immensely in various factories. It depends on the digital maturity of the machines. The degree of connectivity of older machines can be improved by retro-fitting, which describes adding new features, sensors or control panels.
This use case poses high requirements for fitting telecommunication technology. The real-time analysis of complex machine, process and company data (Big Data) requires low latency, a high connection density and data throughput per area. Unique and complex AI-algorithms help to get deep insights. Used in the right way, this leads to a competitive advantage over other competitors. This is the reason why security standards have to be high. Up to a certain degree, WiFi 4 and 6, as well as 4G, can meet the demands. But in a fully digitalized and connected factory only 5G can elevate intelligent data analytics to reach its full potential. It allows a very high data throughput per area, ultra-low latency and provides very high security.
The enhancement of intelligence analytics is predictive analytics, which not only provides insights on current and past data but allows statements about future developments. In digitalized manufacturing, maintenance must take a key role to reduce the risk and minimize the consequences of unplanned stops and disruptions. Predictive analytics can help to predict deviations in data, to trigger preventive actions to avoid failures or to reduce their consequences resulting in increased machine uptime and higher productivity. In order to do so, data is collected in real-time and compared to historical data. This is an infinite process ensuring, that parts are only changed when it is needed and before a machine breaks down.
Long-time milling machine data is used to predict the need for a tool change. This ensures that quality requirements are reached, and scrap is minimized. Furthermore, downtimes and maintenance costs are reduced. In this example data shows that a tool change is needed in a certain time frame. The right tool can be ordered beforehand. The service worker who changes the tool can be scheduled for the right time and change the tool in a designated time slot. Therefore, the only downtime occurring is the time it takes to change the tool.
This technology is increasingly used. The issue with predictive maintenance is, that the prediction is only as good as the collected data. Predictive models based on machine learning have to be efficiently programmed, continuously updated and provided with data
The telecommunication requirements are identical to the requirements for intelligent data analytics. The required data throughput per area and security are very high. The effectiveness of the predictive analytics algorithm depends on the amount of collected data. In the vision of the factory 4.0 all machines are fully connected. Nowadays, in order to retro-fit machines all sensors need to be connected to a PLC. By using 5G technology, all sensors could be connected wirelessly to an IoT-Platform. This allows an easier retrofit. For example, a moving part could be equipped with an acceleration sensor without worrying about the cable. Therefore, 5G helps in enabling a more efficient and effective way of performing predictive analytics.
Machine monitoring represents the simplest use case in the use case cluster dashboarding & monitoring. In order to monitor a machine, process and condition, data is collected. Therefore, a digital copy of the current state is created. If the digital copy represents the current state of a machine fully, it is called the digital shadow. It is the basis for all data visualization and analytics. A connected sensor network providing all key parameters and values of the machine or asset is a necessity. By using a Manufacturing Execution System (MES) or IoT-platform, this data can be accessed remotely. Key performance indicators (KPIs) are defined and visualized on a dashboard to monitor the data using a mobile device or stationary screen. This increases process transparency and the data can be used for manual process interventions to ensure performance and efficiency.
Machine data from a milling machine is collected and sent to an IoT-Platform. In the programming environment of the IoT-Platform, a dashboard is created. This dashboard visualizing machine data can be displayed on the smartphone of the production manager. The manager can act manually if the displayed data shows abnormalities. Generating this data and visualizing it is typically the first step in a digitalized production.
Monitoring machines is common in today’s manufacturing environment. Usually, the created data is created using sensors collected centrally at the machine. Then the data is transmitted to the MES or IoT-Platform via cable or WiFi. The collected data is processed and visualized in a dashboard. The dashboards can be displayed directly at the machine or accessed on different mobile devices.
Machine monitoring is one of the first use cases implemented on the way to a fully connected factory. Therefore, the telecommunication requirements which have to be met to implement machine monitoring are low. The most important category is the user experience data rate because of a fluent display of data. WiFi 4 covers this use case completely. WiFi 6, 4G and 5G could be used as well, but the application does not add any value in this use case and is more costly.
Mapping of Target and Actual State
The use case of comparing the target with the actual state builds on the monitoring use case. The monitored process and machine data are collected and immediately compared to predefined targets. These targets are stored centrally, e.g. in an Enterprise Resource Planning System (ERP). If anomalies are detected, an automatic notification is sent to the responsible person. Anomalies are for example quality issues, process stops or shortages of material. The immediate comparison between actual and target state requires real-time data and remote data access. This leads to quicker response times and allows the comparison between the current and desired state of process efficiency. Therefore, cost and time management is improved and process quality ensured.
Building on the example of the milling machine in the monitoring use case, the collected milling machine data is automatically compared to other data sources from the cloud in real-time to evaluate the current machine status. The findings can lead to adaptions of the following process steps. They will be carried out by the responsible worker.
The mapping of the actual and target state is performed regularly in manufacturing companies for manual process control. It is often limited by the connectivity of heterogeneous systems, the lack of processing power and data creation. Therefore, data is often compared periodically and not in real-time.
The mapping should be executed in near real-time. It needs to be reliable and secure because confidential company data is compared to the actual state of the machine. This requires high security, high reliability and low latency. The basic requirements are met by WiFi. But if more machines are connected and the more diverse data with high velocity needs to be processed and compared in a secure way, a campus network using 4G or 5G enables reaching the full potential of the use case.
Closed Control Loops for Machines
In a digitalized production, all kind of machine and product data is collected during the manufacturing process. This data is stored and processed in a cloud. The results are used to optimize machine usage, parameters, and output. This optimization is called a “closed loop”. The optimization is performed autonomously without a human being (e.g. an operator) interfering. The goal of implementing closed control loops for single machines is to improve the stability and lead time of the machine as well as the quality of the process step. The reaction time is reduced immensely by eliminating worker involvement. Each machine earns the ability to react to errors or anomalies immediately. Data-based parameter optimization can be viewed as more efficient and effective than manual optimization.
All sensors and the control of a milling machine are connected to an IoT-Platform. Through applying machine learning algorithms and models, the process can be optimized. The result is then looped back to the machine controls. With adapted parameters the milling process becomes more efficient. This loop is repeated continuously.
Closed control loops for machines go beyond process control as it has been the norm in manufacturing for decades. The status quo of connecting machines is the usage of hard wires and WiFi. This limits the flexibility, the minimum latency and the maximum amount of data that can be transferred. An IoT-Platform with the right algorithms needs to be established. Once the IT-infrastructure is installed, closed control loops can be implemented in production which is technologically possible nowadays.
The usage of an automated control loop of a machine poses very high requirements for the categories latency, reliability and security. If the connection is interrupted, the production process is affected. In the worst-case scenario, the entire production process is stopped. Due to these requirements, WiFi 4 & 6 as well as 4G would limit the effectiveness of a real-time control loop for a machine. A 5G network meets all the requirements.
Closed Control Loops for Process Line
In this use case, more than one machine communicates to a centralized platform e.g. an IoT-Platform. The collected data is not only monitored but also used to control the process. This bidirectional communication is fully automated. In order to control the process, all machines in the production line have to be interconnected. It is necessary that the implications of changes at the beginning of the process are known for the following steps. Therefore, mathematical models and algorithms describing the process and implications have to be implemented. By combining the digital shadow with those models and algorithms, a digital twin of the process is developed. The continuous comparison of the actual and target state combined with optimization models leads to autonomous improvement of process flows. Ultimately a full interconnection of machines and closed control loops over the production process would result in the spectrum of automated process control as well as automated utilization and performance adjustments.
Interconnected machines in one production line enable optimized and dynamic route planning in case of machine failure as described in the following scenario:
The production line has three production steps and two machines for each step. When Machine 1 in step one has an error, there is an automatic feedback to the other machines and the production capacity and speed can be changed and counter-measurements (e.g. producing other parts in steps two and three) can be taken.
Fully autonomous production lines are the holistic vision of Industry 4.0, which is admired to reach in the future of manufacturing. In order to reach this state in a factory, many steps have to be taken beforehand. By creating a digital twin of the entire factory, the amount of data being transferred and processed will increase exponentially compared to the status quo. Today’s networking technology is not capable of fulfilling all requirements of ultra-low latency, high-reliability and speed.
If the vision of a fully connected factory becomes reality, a 4G network, which can be used for implementing a control loop for a machine, is insufficient. It does not offer the needed connection density, latency and stability. Additionally, WiFi 4 and 6 are not secure enough. Only a 5G network enables a closed control loop for the entire production process. Machine control and communication between all assets is highly challenging for telecommunication technologies. 5G serves as an enabler for realizing a completely smart factory.
Robotic Application with Predefined Motion Control
The standard robotic application possesses predefined motion control. The motion of the robot is programmed in so-called motion scripts before the usage (offline). The robot arms etc. move in predefined repetitive motions. Those motion programs can be stored locally on the robot’s CPU or remotely to enable agility of the motion scripts. Those robots can be working for years without changing the script or the operation place. They are a cost-efficient and reliable way to automate certain manufacturing steps. Robots can replace manpower and minimize the dependency on human beings in the process.
An assembly robot in a car production line is a typical robot with predefined motion control. Standard tasks of these robots are pick and place movements. Only one type of car can be produced on these lines also because of the lack of flexibility of the robot. Changing the process and therefore the field of application of the robot is complex and expensive.
These kinds of robots are the status quo in many factories. They are the standard way to automize a process. They have been the norm for many years and still will have an important role in the future. One of the biggest disadvantages of these robots is the lack of collaboration. Robots with predefined motion control cannot work in the same manufacturing cell as a human being. Safety measures have to be taken into consideration. Often robotic cells are a fenced-off area in a production hall.
Once these robots are implemented, there is no need for a time-critical, highly stable connection. The data transfer is only one way: from the robot to an IoT-Platform or MES. This data can then be used for various analytics. Robots are, as previously mentioned, implemented in critical processes. The transferred data could hold confidential process information. Therefore, the security of the telecommunication technology is rather important. This use case is covered by all considered telecommunication technologies.
Robotic Application with Reactive Motion Control
Robots with reactive motion control are based on the standard robots with predefined motion control. Additionally, they can react directly to errors or abnormalities in the process and adjust motions or tasks in real-time. Sensors integration might be needed. This can be helpful if the robot is used as a mobile robot. The reactive motion control also supports the usage of robots in a collaborative manner. They can work without additional safety measures like a fence because they are able to alter their motion reacting to their dynamic surroundings. It is also possible to have these robots interacting with other robots or human beings. Also, their operation location can be adjusted to current needs and flexible human support is possible.
A mobile cobot can be used in many ways. For example, for half of the day, the robot can support a worker with a simple pick and place use case. Then it can change its gripper/end-effector and be used to operate on its own during night shifts. In order to do that, it positions itself autonomously in the environment and fulfills the task in a smart way without the need for manual calibration or teaching.
This use case is rather new and not common yet. But with the increasing automation and digitalization of factories, robots will take over more complex and diverse tasks. Mobile applications will become the norm. The goal will be to construct the robots as lean and as flexible as possible. Therefore, the processing unit will be centralized, and the robot control will be streamed.
A consistent connection with other robots in line and a data hub is needed to ensure harmonized work as well as a proper IT-infrastructure needs to be implemented. Cloud-computing, machine learning algorithms and adaptive control are the key enabler for this use case.
The collected data of a robot needs to be sent to a cloud, be filtered and used by machine learning models to update robot control and sent back to the robot in real-time. Entire fleets can be managed and automatically optimized centrally. Therefore, this use case depends on an ultra-reliable low-latency network with a high data throughput per area. If the network fails to meet the requirements, then the entire robot fleets and factories will not be able to function properly. Very high security standards are required because any malicious external interference with cloud-based robot control can possibly shut down the entire production site. Only a 5G campus network meets the requirements of this use case. It is an enabler for the vision of fully digitalized and self-regulating robot applications in a smart factory.