Hirose: Connecting the future
Industrial Ethernet Book Issue 3 / 15
Request Further Info   Print this Page   Send to a Friend  

Ethernet in Process Control

Users in the process industries employing fieldbus-based devices and instrument management software can expect a steep rise in bandwidth requirements and therefore must have Ethernet at the host level. But, having made the choice for Ethernet, users still have to be cautious.
By Jonas Berge

System architecture

Just like H1 fieldbus (IEC 61158-2) revolutionised the field, HSE is simplifying the host-end of the system by simplifying enterprise, control and remote-I/O networking. Traditional DCS are complex, with multiple networking levels using different technologies. A system can often have as many as four different layers using different protocols: field, I/O-subsystem, controllers and business.

Ethernet 'flattens' the system networking architecture

H1 fieldbus takes the place of field and I/O-subsystem protocols, and HSE at the host-level takes the place of control network and business network. Although the host-level network in the system is kept separate from the business network using a gateway, the fact that they are using the same networking technology enables them to integrate whilst ensuring safety.

The death of H1?

In the control system architecture H1 Fieldbus is used at the field level to connect transmitters and positioners etc. HSE is used at a host level between linking devices and workstations etc. Therefore HSE will not replace H1, rather the two will complement each other. H1 has many good features that HSE does not have and vice-versa.

For example, Ethernet is limited to 100m and requires multi-core cable, which is costly. Ethernet needs a hub, which although cheap, requires one port for each field device. Ethernet provides no power and could not be used in hazardous areas. H1, on the other hand, has too low a bandwidth to be used as a backbone for the entire plant, and does not have media redundancy. In reality the two complement each other perfectly.

H1 Ethernet
Speed 31.25 kbit/s 100 Mbit/s
Distance 1900 m 100 m
Two-wire Yes No
Multidrop Yes No
Bus power Yes No
Intrinsically safe Yes No
Media redundancy No Yes
Deterministic Yes Yes

Interoperability

Try connecting two controllers using Ethernet from different manufactures together, or to the workstation of a third supplier, and you will be gravely disappointed. Why is it that they cannot talk to each other when they are using the same protocol? In fact they are not. Ethernet and TCP/IP do not eliminate the need for fieldbus standards, on the contrary, they are a celebration of what can be achieved thanks to standards. The reason is that Ethernet and TCP/IP are only half the story (4/7 to be precise).

Interoperability is more than just connecting two things together on the same wire without conflicts. Devices must talk to each other too. But Ethernet only handles the bottom two layers of the seven layer OSI stack, while TCP/IP handles the next two. In fact Ethernet and TCP/IP were specifically designed to handle many different protocols, which makes Ethernet powerful but at the same time the cause of interoperability confusion. A network technology is therefore required to have an application layer (layer 7) that is an open standard. If not, special drivers are required. Most, but not all, Ethernet networks used in control systems today are proprietary because they have their own application layer protocol. One, Modbus/TCP, uses Ethernet media and TCP/IP to communicate data using an application layer with the same register method as Modbus/RTU. Modbus/TCP therefore ensure a fair level of interoperability. Devices from several manufacturers use it. Off-the-shelf process visualisation software already supports Modbus/TCP.

But even with Modbus small variations exist. Therefore the user must configure several things manually, including data types, register numbers, significant digits and scaling. This can become rather tedious for large amounts of data. Fortunately OPC eliminates this problem as far as workstation software is concerned, but between embedded devices it still has to be made the hard way. A limitation of Modbus and other protocols is that they have no standard for programming and configuration download.

Plug-'n'-play

Today customers expect more than just interoperability. They want ease of use. That is, systems must work without manual configuration or parameter mapping. Foundation HSE fieldbus uses Ethernet and UDP/IP (the lesser known cousin of TCP/IP which is more suitable for process control) and takes interoperability to a plug-'n' -play level. Thus you can have both ease-of-use and interoperability.

HSE includes not only the application layer but also a `user' layer on top to provide a function block diagram programming language that allows users to build control strategies distributed throughout the network into devices from different manufacturers.

Industrially hardened

A shutdown can be extremely disruptive so Ethernet must also be able to handle multiple faults. Industrial grade network using several layers of redundancy and industrial-hardened components may be used in order to handle multiple simultaneous faults. Hot-standby redundancy is already part of the HSE protocol specification upper layer. In addition redundant Ethernet hardware can be used to find alternative communication routes in case one path fails.

Petrobras's Namorado-I platform off the coast of Brazil uses a redundant HSE fieldbus network to achieve high availability. The network uses wire in the control room and at the field locations and has a hub-based star topology which ensures only one device per wire so that they may be disconnected without disrupting communication or control in other parts, and any wire fault has a very limited effect on the system.

Fibre optics is used to link the switching hubs in the different areas over longer distances. All parts of the network have media redundancy, including the hubs, i.e. there are two independent networks. Industrial grade DIN-rail mounted switching hubs from Hirschmann have redundant power supplies and media redundancy. They are also very rugged and have a wide operating temperature range suitable for harsh environments. The redundancy switchover time is very short and recovery is much faster than for the traditional spanning tree algorithm, so operations continue without any loss of data. The fibre-optic part is also connected in a ring topology where communication can flow either clockwise or counter-clockwise. This means that the network can sustain multiple faults but still continue to function.

Ethernet software

Users are looking for homogenous software environments. Therefore Ethernet must be functionally part of a single application, not an Ethernet software `island'. For instance, Smar's SYSTEM302 uses Foundationš H1 fieldbus at the field level, and HSE at the host level. The engineering and maintenance tool SYSCON not only lets the user build the control strategy and configure devices, but also manage both networks.

Using this application the user can optimise network parameters to suit devices from different manufacturers and meet the control objectives. The software auto-detects and identifies new devices, making commissioning and expansion easy. Bridging between H1 networks is managed and H1 networks can be connected, disconnected and rearranged to other ports. Address assignment is automatic eliminating human errors. It is also possible to visualise and manage the HSE redundancy, even of secondary devices that are not active. The software runs in the familiar Windows NT environment and integrates with Windows NT networking making the setup and maintenance easy. The user can navigate the networks in the Fieldbus Explorer's hierarchical tree-like structure, from HSE, through linking devices to H1 and thence on to instruments. One advantage of IP that users particularly like is the ability to Ping nodes to diagnose the network, particularly when there are many gateways and subnets. A log of communications errors can be reviewed any time at the click of a button.

OPC

The point where Ethernet and software really meet is at OPC (OLE for Process Control). A modern control system needs more than just configuration and monitoring software, more than inflexible `closed' applications and inaccessible data associated with DCS. Of all the special software that exists today none `speaks' HSE or Ethernet. The SYSTEM302 universal Fieldbus bridge, called DFI302, is among other things a H1 to HSE linking device, and it comes with OPC server applications running in redundant server stations that makes all the process and device data available to any OPC client software. Since OPC is built on DCOM (Distributed Common Object Model) any workstation on the Ethernet can access this data. Thus information can be disseminated throughout the enterprise.

In addition to OPC the configuration tools uses an OLE client-server architecture to fill the gap left by the OPC data access. OLE functionality is used for control strategy configuration download and other functions required for device management but not provided by OPC data access. Actually, it can be seen as a precursor to the emerging OPC complex data access intended for such functions. The OLE and OPC effectively integrate with the rest of the system. Once a function block is configured in a device from the engineering tool, all its parameters are automatically made available via the OPC server to all clients, where they may be selected for display, trending and reporting etc. For example, Petrobras' Namorado gas platform has a SYSTEM302 with a dual redundant Ethernet where the devices, controls and network are completely managed by the SYSCON application.

Conclusion

Every protocol is migrating towards Ethernet including Modbus/TCP, Ethernet Profibus and Ethernet ControlNet. But don't expect the different flavours to work with each other. Select one carefully. For process control, Foundation HSE Fieldbus may be a good choice since it is part of the IEC standard. Using Ethernet is a no-brainer but don't forget to check what application layer protocol is being used. Don't write off H1 until Ethernet provides power, is intrinsically safe, multidrops and runs 2 km!

Jonas Berge can be contacted at jberge@smar.com.sg

Source: Industrial Ethernet Book Issue 3 / 15
Request Further Info    Print this Page    Send to a Friend  

Back

Sponsors:
Tailyn EN50155 Ethernet switch
DINSpace fiber optic and Cat 6 patch panels
Accelerate your HART data at the speed of Ethernet
Panduit IndustrialNet 8-port DIN Rail Mount Patching Solution
iManufacturing Shanghai 2019

Get Social with us:



© 2010-2019 Published by IEB Media GbR · Last Update: 19.06.2019 · 48 User online · Privacy Policy · Contact Us