Virtual Factory: The Physical and Virtual Collide
In the intensely competitive world of manufacturing, global enterprises face a variety of pressures, including frequent product updates, price wars, cost reductions, resource optimizations, and enhancements to energy efficiency. To protect their bottom line and meet a growing number of customer needs, manufacturing enterprises must enrich their product offerings, improve activities, and flexibly adjust production to adapt to frequent market changes.
Industry 4.0, a concept that promotes the computerization of the manufacturing industry, provides a blueprint for the future of industrial manufacturing. For example, some tobacco companies are currently building virtual factories to help achieve flexibility, efficiency, traceability, and self-optimization. A virtual factory is a combination of physical manufacturing and virtual presence, which uses sensors positioned throughout the factory that collect real-time data on the manufacturing processes. This data is then used to analyze, in fine detail, each element and link involved in production. Based on the real-time data, a virtual factory leverages a diverse range of technologies, including digital application modeling, Big Data analytics, and 3D Virtual Reality (VR), to simulate, assess, and optimize entire processes in a virtual computer environment. By doing so, a virtual factory seamlessly combines physical and simulated manufacturing, which ultimately promotes the efficiency of production facilities.
A virtual factory must meet certain requirements. Physical and virtual factories must operate in tandem, with the virtual factory collecting data from the physical factory in real time; then, the virtual factory calculates ways to refine the current operation with immediate implementation back into the physical factory. Physical and virtual environments must combine seamlessly. The virtual factory uses 3D VR technologies to display the physical manufacturing environment and performs real-time monitoring and analytics of devices. And finally, the Big Data analytics platform analyzes and mines large amounts of data stored in clouds. This analysis helps factory operators make more accurate decisions, enhance efficiency, and improve product quality while reducing costs.
Real-Time Data Collection
To seamlessly combine physical and virtual manufacturing, a virtual factory must collect precise and intact data in real-time. A tobacco factory may have tens of thousands of data collection points, as well as dozens of device interface types and communication protocols. In the case of legacy production systems, the time it takes for a virtual factory to obtain manufacturing data after performing necessary protocol conversions is too long to be effective. For example, Cigarette Wrapping Machine (CWM) data must be transmitted to the CWM industrial control system and, in turn, to a real-time database via the Modbus protocol and Manufacturing Execution System (MES) over the Profibus protocol. Industry 4.0 and Internet of Things (IoT) scenarios have been conceived to overcome precisely this type of latency.
Huawei’s AR Series IoT Gateway solves this challenge. In addition to open Ethernet interfaces, the AR IoT gateway provides rich, industry-specific interfaces and bus standards to directly interact with various sensors. The AR IoT gateway converts numerous sensor protocols into upper-layer protocols, seamlessly converging manufacturing and information networks to build a flattened, intelligent IoT architecture.
When deployed at the controller layer, the AR IoT gateway connects a variety of sensors and controllers to a virtual factory in real-time implementing network-wide data sharing that guarantees timeliness and integrity of manufacturing data. Robust computing capabilities allow the virtual factory to aggregate network-wide data to the cloud and eliminate information silos. As manufacturing applications and services are migrated to the cloud, enterprises readily optimize their plans or activities depending on customer needs and market changes, creating custom, small-scale, and flexible manufacturing processes.
Big Data Analytics — Scheduling Made Easy
Traditional manufacturing scheduling schemes must be manually formulated and programmatically executed by automation systems. While this mode applies to large-scale manufacture of mass-market products, it does not apply to small-scale, custom products, let alone Industry 4.0.
A virtual factory collects very large amounts of real-time and historical manufacturing data. Precision analysis and mining of these types of structured and unstructured data is a key factor for determining enterprise-scheduling policies.
The Huawei FusionInsight Big Data Platform enables tobacco enterprises to discover valuable information from large amounts of real-time and historical data. FusionInsight can be programmed to detect potential risks and support proactive decisions. Based on the results of Big Data analytics, a virtual factory formulates manufacturing schedules, models manufacturing outcomes, and uses the 3D VR system to simulate an environment to execute high-fidelity models. If problems are detected during the process, enterprises can, for instance, adjust scheduling schemes that target optimal results. Virtual factories can shorten transient adjustment cycles from days to minutes — a 98 percent improvement in scheduling accuracy. Scheduling schemes are tested in virtual environments first — at no risk to the physical plant or perishable resources — at a millions of dollars savings in annual resource consumption.
In the Industry 4.0 sphere, physical and virtual worlds become blurred. Connections between people, machines, and information are neatly managed and manufacturing data aggregated to the CPS, or virtual factory with computer, network, and automation technologies. Virtual factories establish factory-wide 3D models to display the running status of devices. In a 3D virtual factory, users can perform multi-axis point-of-view operations — i.e. push, pull, rotate, and zoom — to check manufacturing data from different viewpoints.
The server on which the virtual factory software is running must possess a robust 3D graphics processing capability from which to deliver the lifelike graphics. FusionSphere is Huawei’s proprietary cloud operating system that leverages Graphics Processing Unit (GPU) pass-through technology and dedicated hardware video cards to accelerate 3D graphics rendering. The chip-based GPU pass-through technology surpasses software-only 3D graphics processing.
As a growing number of manufacturing applications and services have shifted to the cloud, so too, the virtual factory server must also operate on the private cloud of the manufacturer. The GPUs operating on ordinary Virtual Machines (VMs) are typically able to update images at the rate of one frame per second, which does not reflect the real-time running status of activity under study.
The hardware GPU pass-through technology is engineered to display high temporal detail of the manufacturing activities that allow personnel to easily transpose the synthetic image with the real-world. The net effect is that better models, and better connectivity, lead to improvements in production efficiency. In the example of preventive inspections that will take a full day to conduct manually, it may take only a few minutes in the “virtual factory” when the equipment on the factory floor is fully wired with a sensor network. Fault location or points requiring repair and maintenance are quickly identified. This combination of real-time status monitoring and 3D device visualization for monitoring and interrogation greatly enhance staff productivity.