The factory of the future will be built upon cyber-physical systems—the integration of computation, networking and physical processes. An example of such a system is the smart grid, a modernized version of the electric power grid that uses communications technology to gather and automatically act on information.
Today, automation in factories is organized in a clearly delineated hierarchy – from the factory floor to the back office. Systems at each level perform specific processes related to the functions at each level:
- Field level: Sensors and actuators;
- Control level: Control devices, I/O modules and operator terminals;
- Process management level: Engineering, supervisory control and data acquisition (SCADA) systems and manufacturing execution systems (MES); and
- Enterprise level: Enterprise resource planning systems
In the factory of the future, the field level will be connected but will also continue to perform a wide array of processes autonomously. The major change at this level is that the systems will embed more and more intelligence.
The other three levels – control, process management and enterprise – are likely to move to high-performance servers located in a cluster, data center or in a cloud. Virtualization, the separation of specific function and processing hardware will find its place in the factory as well.
Essential to driving the enormous increases in efficiency in this model is advanced communication technology that can eliminate the obstacles for communication between the server and field levels.
Designing for More Devices
It’s estimated that the number of connected devices in the factory of the future may double or triple to achieve the objective to collect as much real-time data as possible. Meeting the increase in both devices and the amount of data collected will drive several shifts in network topology, including:
- All communications will be based on IP protocol families and Ethernet will be the underlying communication protocol for consistent and unified communication.
- The network of a large number of devices should be hierarchical. Today, the field level is divided into logical cells, but as the amount of data generated in the cells will be significantly higher, this model will simplify network management and operation.
- The use of star topologies will increase due to the advantages of lower latency and higher reliability. Simulations have shown that the use of one larger switch for connections delivers higher Mean-Time-Between-Failures (MTBF).
- Extensively meshed network topologies will also increase.
Combined with the use of new protocols, these networks will be easier to manage, make better use of resources and reduce costs.
Wireless Technology Advances
In most industrial environments, communications have been almost exclusively based on wired networks, primarily due to the stringent need for reliability.
But with increasing need for and interest in use of industrial wireless, technology suppliers have recognized the need to address some of the gaps in reliability found in wireless products. Enhanced electrostatic discharge protection for hazardous environments, wireless mesh technology for quick network reconfiguration and service assurance, and redundancy protocols are among the advances improving the suitability of wireless devices for industrial environments.
In the factory of the future, the dominance of wired communications will continue. However, especially in hazardous areas, the flexibility afforded by wireless connectivity, will increase the use of industrial wireless products designed for such environments
In most factory applications, Fast Ethernet with 100 mega bits per second (mbps) is the standard. Compare this to Gigabit Ethernet – 1000 mbps – which has been state-of-the-art in the IT world for some time. As the volume of data grows, driving the need for higher data rates, the appetite for Gigabit in industry will only escalate.
Several trends will accelerate the adoption:
- Lower costs for faster connections, as new chip developments integrate Gigabit Ethernet;
- Lower power consumption driven by advances in semiconductor processes; and
- Simplified cabling which will allow Gigabit Ethernet to run on a single pair of copper wires (as opposed to the four pairs required today)
A Real-Time Standard for Data Transfer
In the factory of the future, deterministic behavior and the maximum latency guarantees for data transfer will be essential. No longer confined to data transfer within a single cell at the field level, these data transfers will take place between production cells and possibly to locations outside the plant.
The timing requirement for these transfers is primarily driven by the process, but a guaranteed maximum latency from the source to the destination and back is imperative if the vision of a data-driven industrial operation is to be achieved.
Closing the gap between what exists today with some real-time Ethernet protocols and an Ethernet standard will be achieved through one or more of these communication infrastructure elements:
- Time Synchronization, based on IEEE 1588 Precision Time Protocol (PTP) allows decentralized synchronized clocks to run on all components with an accuracy of less than one microsecond, making it possible to separate processes from communication and allow actions to be time-driven rather than event-driven.
- Increased data rates, made possible by the adoption of Gigabit Ethernet, deliver significant improvement for real-time applications. Lower latencies of data packets and improved data forwarding inside high-performance switches mean that switches are blocked only a fraction of the time.
- The Time-Sensitive Networking (TSN) work group, under the umbrella of IEEE 802, is working to define a deterministic vision of Ethernet. Early work focuses on several technical concepts, including ways to control the flow of real-time data packets from within the switch, a bandwidth reservation protocol for all required resources on the network, and a pre-emption framework which would allow high-priority packets to pre-empt low priority packets in the queue.
Ensuring Network Reliability
While it’s impossible to prevent disruption in the network completely, it is possible to design the communication network so that traffic can be redirected to an alternative path in the case of failure.
A basic requirement for each Ethernet network is that there be only one active path between the source and destination at any given time. Alternative paths provide redundancy, and a control protocol resolves the inherent contradiction—ensuring one logical path, while the others are on standby. In the case of failure, a switch is made to an alternative path. There are several protocols in the market based on this principle; they differ based on time required to switch paths and the topology they support:
- Rapid Spanning Tree Protocol (RSTP) is effective for a variety of topologies; it’s restricted by the number of switches between sender and receiver;
- Media Redundancy Protocol (MRP) is limited to ring topologies, but offers very fast and deterministic characteristics; and
- Parallel Redundancy Protocol (PRP) and High Availability Seamless Ring (HSR) are based on networks with two independent active paths between devices to avoid any downtime in the event of failure.
For applications in the Smart Factory, the first step is to determine what redundancies are required and then choose the protocol best suited for the environment. In most cases, a mix of segments exists, resulting in the use of PRP for those requiring full redundancy and other protocols for remaining segments.
The Journey Begins Today
The communications requirements of the factory of the future are ambitious. Significant progress has been made to close the gaps on what’s available and what will be needed to streamline and optimize the entire value chain. Organizations that begin to plan their communications networks today will be in a good position to capitalize on the coming revolution.