Network architecture of the future: It’s now
We have all seen the situation: a new piece of equipment has to be connected to some network right now, because the facility manager has labeled it critical and it must appear on the monthly reports. So it gets tied into the nearest network, or perhaps an odd gateway is installed to connect it to an existing system, or maybe we have to install an extra communications module to let it talk over some other protocol. Next month, we go through a similar process when a new interlock between unconnected equipment needs to be installed, so we rig something up to get from A to B. Follow this process for a few years and the process control network diagrams begin to look like spaghetti.
It doesn’t have to be that way. You can future-proof your networks by building a detailed network plan and then only adding systems that fit the plan into the network.
What kind of network?
Every facility has a unique set of needs for its process control network. For some, data loading is very light, but security is paramount. For others, security is less important, but it must handle huge amounts of data. Do you have a safety network running alongside the process network and are there any common failure points? The balance point between these various needs will likely differ between parts of the facility.
The ISA 95 framework is a commonly referenced model that defines multiple levels of process control. As you work through the exercise of defining network needs, consider what level that network falls into. Most process control networks are at Level 1 or Level 2 and most business networks cover Levels 3, 4, and 5.
How many networks?
As you identify the requirements for each of your various networks, often you will find some that conflict. You may even have some that are diametrically opposed. One solution is to choose a middle way that tries to serve all, but does it poorly. The better solution is to break your data paths into two or more networks. One network with low bandwidth and ultra-high availability might be perfect for the safety interlock system. A different network with very high throughput but occasional packet loss might be the one to use for HMI process monitoring. Ultimately, the most thorough approach is a three-layer network.
Layer A is the I/O network, used by the PLC or DCS to communicate with its field devices. This layer is at Level 1 in the ISA95 model. Frequently, Layer A is broken into many small networks serving different areas or zones. There might be an RS485 serial network on one piece of equipment, remote I/O over Ethernet in an adjacent room, and Profibus DP in the next building.
Layer B is the interconnect network, used for communication between PLCs, DCS nodes, or other intelligent control devices. This layer may be at Level 1 or Level 2 in the ISA95 model, depending on exactly how it is used. This layer may also be segmented, but less so than at Layer A, due to the need for different supervisory equipment to share information with each other.
Layer C is used by the HMI or SCADA system to collect process data and present it to the user. This layer is generally at ISA95 Level2.
Additional layers that correspond to Levels 3, 4, and 5 of the ISA95 model are generally the business LAN and WAN and are outside the control of process engineers.
Develop the plan
Consider the most complex or data-intensive project you have implemented in the past five years. With the rapid pace of advancement in automation, that project is likely to become your baseline standard project in the not-too-distant future. Plan for your worst expectations five years down the road, and then triple it.
• Lay out all the networks at Layers A, B, and C in all areas of the facility, even if you currently have no equipment that needs a particular layer in a particular area.
• Identify addresses or address ranges, preferred hardware, cable specifications, and all the other small details that will be needed when the time comes to install something new in that area.
• Specify lists of acceptable control hardware that works with the planned networks.
Once the network plan is complete, in all its detail, review it periodically. The overall architecture should not change often, but might need minor adjustments. The hardware lists should be updated to address newly available functionality, while also paying attention to backwards compatibility. Interconnect requirements may change as the product process is reconfigured.
Plan the work
Once you have a plan for future expansion and utilization of the process control network, stick with it. Document the plan as an official policy or institutional guideline rather than remaining as a department preference. Spread the information as widely as possible, so that all hands can contribute to maintaining its integrity.
Every time there is a new project to implement, reference it against the network plans and associated standards documents. You will find that having all the details pre-planned dramatically simplifies the networking portions of the project development process.
Work the plan
When implementing a project, do not make modifications to the plan for the sake of expediency. Deviating from the standards ”just this once” can quickly turn into a standard which is generally ignored.
Include the network standards in project scoping documents so that vendors are obligated to follow them. If purchasing approves a new piece of equipment that is incompatible with the plan, does engineering have the power to refuse it when it shows up on the receiving dock? More importantly, how can you work with purchasing to prevent the incompatible purchase in the first place?
Plan comprehensively, build incrementally
As we continue to move farther into the data-driven future of automation, we can no longer deal with our process networks as afterthoughts. They have become as critical to our success as the production equipment they are attached to. Few facilities have the luxury or the funding to build the perfect network today, but almost all have the ability to plan it today. By developing a detailed plan, and building it incrementally over time, you have to power to make tomorrow’s network a success with your efforts today.
This post was written by Robert Henderson. Robert is a principal engineer at MAVERICK Technologies, a leading automation solutions provider offering industrial automation, strategic manufacturing, and enterprise integration services for the process industries. MAVERICK delivers expertise and consulting in a wide variety of areas including industrial automation controls, distributed control systems, manufacturing execution systems, operational strategy, business process optimization and more.