Knowledge Base /
White Papers
Industrial Ethernet for Process Automation
By Martin Berutti
Product: DeltaV Virtual IO Module - VIM2

This white paper discusses the use of Industrial Ethernet for plant floor integration of process plants. In addition comparion of Industrial Ethernet protocols and OPC DA is covered.

Industrial Ethernet is emerging as the fieldbus of the future for integration of plant floor data in the process industries. While much of the industry attention is focused on wireless, there is a definite trend away from proprietary fieldbus networks towards Industrial Ethernet. Industrial Ethernet networks are increasingly being used to connect Process Automation Systems with PLCs, ESDs, MCCs, Plant-floor HMIs, weigh scales, analyzers, and remote IO. Properly implemented Industrial Ethernet networks have been proven to meet the needs of the process industries in: Meeting Process Control Requirements Providing good Network and Data Integrity Providing new levels of Operational Flexibility

Most articles about Industrial Ethernet in automation trade journals focus on the use in the discrete industries. Because MYNAH provides products for the process industry, this article will focus on the use in these applications. It will also address confusion about Industrial Ethernet protocols in the marketplace.

Process Control Requirements

The requirements for plant floor communications are quite different from discrete control to process control. In discrete control, speed of response is a primary concern. Data transmission times of 10-50 millisecond are required. Because of this need for speed, Ethernet implementations often use custom network interface cards built with high-speed ASICs. EtherCAT, Sercos III, Ethernet Powerlink, and Profinet all utilize custom ASIC network cards in order to achieve the fast data transmission. It is interesting to note that Ethernet/IP protocol claims high speed transmission capability without the use of custom hardware due to the structure of Class 1 CIP messaging. This is a clear advantage that promotes the use of this protocol.

For process control the need for high speed is not as important. Update times of 500 milliseconds are often adequate. Data integrity, data flexibility and integrated time-stamps are more important to the process industries. The use of custom hardware is not required and undesirable. From a practical level, the field device used in the installation will dictate the protocols that can be used.

The dominant protocols in use for process application include the following:

  • Modbus TCP/IP, www.modbus-ida.org - If the strength of the protocol was determined by the number of different devices that support it, Modbus TCP/IP would be the winner. Market surveys show that about 25% of the Industrial Ethernet installations use this protocol. Within the Modbus TCP/IP protocol are three variants of communication based upon encapsulated Modbus messages in TCP/IP packets: (RTU via TCP), Modbus over UDP(RTU via UDP), and Open Modbus TCP/IP (RTU TCP). Modbus TCP/IP provides a high amount of data flexibility for process control applications supporting discrete, analog, and floating point data. It is easy to implement, reliable, and easy to troubleshoot. The protocol however has the limitations of not providing any mechanism for passing associated data quality or time stamps. Modbus TCP/IP communication rates easily meet the needs of process control.
  • Ethernet/IP, www.odva.org, - Although, fewer vendors support Ethernet/IP than Modbus TCP/IP, Rockwell's support of the protocol has allowed it to also take about 25% of the Industrial Ethernet market. The name of the protocol is not descriptive. Ethernet/IP is an implementation of the Common Industrial Protocol (CIP) over the TCP/IP transport and network layer. Within the Ethernet/IP protocol there are two variants of communication: Class 1 Implicit (IO) messaging, and Class 3 Explicit messaging. Both message types provide adequate speed for process control. Ethernet/IP is a much more complex and capable protocol than Modbus TCP/IP, but it also more difficult to implement for the device manufacturer. Setup of Ethernet/IP networks and communication can be more difficult for the user as well. Ethernet/IP provides the same level of data flexibility as Modbus TCP/IP. Both Class 1 and Class 3 messaging can be defined by the device vendor to include data status information. Time stamps could also be associated with the data in a Class 3 message however this is not covered in the specification (the method for designating the data would be vendor proprietary).
  • PROFINET, www.profibus.com, The VIM2 is implemented as a PROFINET I/O Controller, and behaves as a gateway between DeltaV and plant floor devices residing in a PROFINET network. Specifically, real-time (RT) data exchange between the VIM2 and other I/O controllers (PLC, etc.) and I/O devices (field devices) is implemented. Isochronous real-time (IRT) communication is not supported. Field device communications comprise I/O data, alarms and diagnostics. Other aspects of field data, e.g., data records and logbook information, are not supported. Alarm and diagnostic information is reported to PC based VIMNet Explorer tools. Only I/O data and corresponding status are reported to DeltaV. This firmware conforms to v2.2 of the PROFINET protocol standard.

The announcement of Schneider Electric becoming a principal member of ODVA is of great interest to the future of Industrial Ethernet in process control. Shortly after the announcement in April 2007, a standard committee was formed within ODVA to "provide compatibility of Modbus TCP devices with networks built on the Common Industrial Protocol". This extension to the Ethernet/IP protocol plus the promotion of ODVA by Schneider and Rockwell should propel Ethernet/IP (with the Modbus extensions) to be the dominant protocol for Industrial Ethernet installations in the process industries.

Network Integrity

Industrial Ethernet networks provide a new valid form of determinism. The deployment of switched networks adds a level of messaging integrity and traffic control required for mission-critical process control communications. In addition, high bandwidth (as much as 20 times the speed of the fastest proprietary networks) provides a margin of error for message transmission that supports secure data delivery. On top of that, the effective use of SNMP tools can increase the security and integrity of the network. The cost of bandwidth investment and management tools is insignificant due to the COTS nature of the equipment.

Redundancy for industrial Ethernet communications is proven and in wide spread use. The redundancy schemes may not be as comprehensive as the redundancy implemented in a proprietary, single-vendor environment. However, most are adequate to prevent a single point of failure from shutting down the plant floor communications.

Troubleshooting and addressing issues in Industrial Ethernet networks can be handled by technicians with networking skills using inexpensive tools. In particular, Wireshark software, www.wireshark.org , is an open source, powerful tool for troubleshooting Industrial Ethernet networks and messaging.

Operational Flexibility

One of the real benefits of using Industrial Ethernet for process plant floor communications is the flexibility it provides to plant operations and maintenance. Since communications is enabled over a network, adding additional nodes to the network that act as a consumer or producer of information is as simple as connecting to the switch. Producers or slave devices can write data to multiple devices, while consumers or masters can read from multiple devices. If it is decided that data needs to be routed to multiple locations, it can be easily accomplished within the network.

In addition, maintenance or configuration applications can manage devices across the network just by being connected to the network and having access rights. This frees up the automation system architect to not have to account for local configuration or maintenance stations or bus addresses when the installation is being planned. Maintenance staff can take advantage of device and asset management tools from any place with a connection to the network.

Protocol Confusion

Confusion about Industrial Ethernet protocols is rampant in the industry and is at times perpetuated in the trade journals. One article in a well known automation industry magazine, stated that the respondents in a survey claimed the use of Modbus TCP/IP, TCP/IP, UDP, and Ethernet/IP for the protocols in their industrial Ethernet networks. Our experience shows a great deal of confusion with users thinking that any TCP/IP device will talk to any other TCP/IP device at the application level. In addition we often see the belief that any device that uses Ethernet must also support the Ethernet/IP protocol. Both Modbus TCP/IP and Ethernet/IP use TCP/IP (which by definition includes TCP and UDP messaging) as the transport and network layers of communication (following the 7-layer OSI model). Both Modbus TCP/IP and Ethernet/IP provide the data and message definition for the application, presentation, and session layer of communication. However, just because a device supports Ethernet and TCP/IP, it may not support the Modbus TCP/IP or Ethernet/IP protocol.

The term Industrial Ethernet is also misleading. ; It is true that the dominant Physical and Data Link layer used for these installations is Ethernet (IEEE 802.3). However, both protocols can be implemented on any TCP/IP compliant physical layer including fiber, wireless, and cellular.

Predictions of the Future

While very few users are installing true 'proprietary' fieldbus, we do see a movement away from 'open' busses like DeviceNet and ProfibusDP towards Industrial Ethernet. In addition, users are more frequently implementing hybrid networks where the Industrial Ethernet backbone is bridged to a local lower speed bus like DeviceNet. Over a period of time, we expect that Industrial Ethernet will erode the use of DeviceNet and ProfibusDP.

Analog devices like transmitters and control valve positioners are a different story. Foundation Fieldbus in particular is a unique technology with its distributed control function block standard and has unique, significant advantages for analog process control. We do not see Industrial Ethernet encroaching on the use of Foundation Fieldbus, HART, or Wireless HART. These protocols are uniquely suited to meet the needs of these devices. They will continue to be integrated with different methods than PLCs, ESDs, Drives, and MCCs in process plants.

Comparison of Industrial Ethernet Protocols and OPC DA

While Industrial Ethernet protocols are simple, they are also effective for real-time, plan-floor data transmission. The advantages of Modbus TCP/IP and Ethernet/IP include:

  • They provide high-speed data communications. Testing in our labs at MYNAH has shown that 5,000 16-bit values can be read easily in less than 500 msec.
  • These protocols were not developed around Microsoft standards, so they are ideal for embedded systems and non-PC hardware. This provides the ability to architect a more stable communication platform.
  • The cost for vendors to implement these protocols (Modbus TCP/IP in particular) is low.
  • Redundancy schemes for both Modbus TCP/IP and Ethernet/IP are available and proven.

But of course, there are disadvantages:

  • Both protocols are designed for data transmission. There is no inherent ability to communicate quality or a time-stamp with the data. Data reads of quality and diagnostic data are transmitted separate from the actual values.
  • These protocols have no inherent ability to browse tags or identify data sources (Ethernet/IP does support some limited tag browsing).
  • Both protocols incorporate open data sources and messages so any security must be designed by the implementer. Physical isolation of networks is recommended.

In comparison, OPC DA is a much more sophisticated protocol by design. It incorporates many of the capabilities missing in the open Ethernet protocols:

  • OPC supports the data discovery by tag browsing capability.
  • Data quality and time stamps can be available as an attribute of the value.

Although there may be some who will argue with me, my experience is that the COM architecture (that OPC is built upon) is proven and sufficient for real-time data communications. However, when applied to real-time, plan-floor network communications between different vendors, OPC DA runs into big limitations such as:

  • The performance and stability of communications is many times dependent upon the characteristics and design of the OPC Server. The OPC Client application almost needs to be tuned or designed for the characteristics of the OPC server that it is addressing.
  • OPC is too big and difficult to implement for embedded systems. Because it is built upon Microsoft technology, it is not easily ported to other operating systems.
  • The configuration of DCOM between networked nodes is tricky and easily broken. OS service packs can undo DCOM settings breaking the communication links.
  • Because data sources are implemented as Servers, passing data between two data sources requires a middleware of some type of two-headed client. Often these applications add another layer of complexity and unreliability.
  • Redundant OPC applications require proprietary components and implementation. Out-of the-box redundant OPC applications are just not available.

Applying OPC DA and Open Ethernet Protocols

Although there are many limitations with OPC DA, there are some good applications to apply this technology such as system to system communications of almost-real-time applications. These would include SCADA, Advanced Control, PIMS implementation and other applications where occasional reboots (although not desired) could be tolerated. Our recommendation for these implementations is to avoid DCOM by having the OPC Client and Server on the same machine. In addition, the use of OPC for a communication standard between process simulators and off-line control systems is proven and appropriate. We have several automation system interfaces for our Mimic Simulation Software based upon OPC DA. Redundant OPC installations and distributed OPC installations using DCOM should be avoided.

In comparison, the open Ethernet protocols of Modbus TCP/IP or Ethernet/IP are excellent choices for plant floor data communications such as PLC to PLC, PLC to DCS, IO, drive, and analyzer integration. Just remember, network isolation is a necessity; the best approach is physical isolation.

The Users' Experience on OPC DA

Across the process industries users have expressed frustration with OPC when aplied to critical real-time data information exchanges. Process users are moving away from OPC towards open Ethernet protocols, in particular Modbus TCP/IP. OPC has proven to be a risky choice in mission critical applications with little or no tolerance for automation system downtime. In the refining and petrochemical industry, redundant systems are a given. In the full range of process industries their needs for uptime and reliability may not be necessary in all cases. Food processing plants, for example, are much more tolerant of process computer reboots. However, in terms of capital investment in process automation, the petrochemical and energy production industries are the source of the majority of the global spend. The OPC Foundation has stated that the problems with OPC DA will be addressed in the still to be released Unified Architecture protocol (OPC UA) Until that standard is released and delivering on its promises, Industrial Ethernet protocols are the best integration method for plant floor data.