The legacy control system dilemma
For those of us who have been in the control industry for some time, we’ve seen a lot of changes in controller platforms and HMI packages over the years. I remember back in the late 1990s when Wonderware, FIX32, Citect, US Data, Iconics, and others were really beginning to gain a lot of traction on the plant floor and in control rooms. Some of those companies have come and gone, and some have been rebranded under new names, but there are a lot of legacy PC-based HMI products still running on plant floors across the country.
Many of those original systems are still active front-ends on equally mature control platforms. The Moore APACS controller used Wonderware extensively as its front-end and repackaged it as Process Suite. This was basically Wonderware v7.1 with some added integration functions that made it pair up with APACS more seamlessly. Over time, we’ve seen upgraded controller and HMI offerings for new installations as a matter of progress.
Now, certainly nobody denies the benefits of progress, but sometimes there can be too much progress in the PC operating system world. To put it bluntly, Microsoft likes putting out new software versions about as often as we have national elections, so we have a large catalog of various versions of Microsoft Windows in just the past 30 years.
Figure 1: Chronology of Microsoft Windows Operating Systems (source: Wikipedia.org)
Considering that control systems for process control facilities like chemical plants are usually designed with a 20+ year lifetime, this high frequency of operating system turnover has had unforeseen consequences in the control world. Most notably, when Microsoft stops supporting an operating system it stops releasing security patches, and any replacement computers are compelled to use more current operating system versions of Windows.
Software isn’t the only culprit in this dilemma because hardware advances have also caused headaches. When a circa 1998 PC running Windows 95 finally dies and the controls engineer needs to reload software on a computer that no longer has an ISA slot (or RS-232 port for that matter), he/she can be in for a long night. Hardware incompatibilities like these have forced control system providers to come up with exotic and expensive methods of maintaining the viability of legacy control systems in order to make them compatible with new operating systems and PC hardware.
Extending the life of your legacy installation
Considering the above scenarios, you are left with a couple of options as a control engineer with a legacy control system and HMI platform. You either rip and replace everything and start with a new control installation, or you use current technology to extend the life of these existing systems. I would like to concentrate this discussion on the latter option to look at how we can save some money and retain our previous programming.
In truth, upgrading control systems doesn’t always guarantee that your company is going to make more money in the process. What it will do is prevent a costly downtime incident in your plant when an obsolete component goes belly up and nobody can find a replacement. So the risk equation has to be balanced with the cost of upgrades and the cost of lost production. I want to share a couple of ideas that I have used on existing systems that reduced the cost of this upgrade and extended the life of the control system indefinitely.
Thin client technology and their new role in HMI applications
Let’s consider the following example: we will presume you are a process or manufacturing plant with several Allen Bradley SLC 5/04 PLCs which are distributed across a Data Highway Plus network. These PLCs are connected to twelve PCs running Wonderware v9.5 (each with its own runtime license) on a Windows XP OS with KTX cards or some equivalent DH+ interface inside each PC.
Perhaps the time has come where you can no longer procure PCs that have the correct ISA or PCI bus architecture, or maybe corporate IT has decided it needs all PCs to be upgraded to Windows 7 or later for security purposes. In either case, the decision is made to upgrade the PLCs to a single ControlLogix PLC and now the decision turns to how to best deal with Wonderware and the legacy PCs and operating systems.
One major technological advancement that has emerged in the PC market has been the server/thin client model of delivering information or content to remote terminal stations. In the above installation, your fictitious plant had twelve PCs running Wonderware each with its own license. If this model is retained during the upgrade, you have the added burden of having to load each station individually with software and keep up with it for patches, upgrades, graphics changes, etc.
With thin client architecture with a central server (or redundant servers), you can make application or graphics changes at just one location: the server. This saves a lot of time and bookkeeping for the engineer and IT department and assures that all active users running Wonderware will be on the same version with the same set of graphics.
There are additional products (such as ACP’s Thin Manager) that add a lot of flexibility for managing thin client systems, but the end result is that the system is far easier to maintain. One of the greatest benefits of this type of thin client architecture is manifested when there is a failure of a thin client machine. In the old model, if one of your Wonderware PCs suffered a failure (power supply, motherboard, etc.), you were forced to find a new PC, load the OS, load Wonderware, activate the license, and get it configured to talk to the PLC.
Now, if a thin client fails during operation, simply pull a new thin client (any model will typically work as long as it has PXI boot capability) from the storeroom and plug it into the Ethernet cable. In a matter of seconds, the server will recognize that there is a new thin client on the network and that an old one is missing. It will assign a new IP address to this unit and will cache the Wonderware screens onto the new unit. You are now up and running in less than five minutes and your operation can continue where it left off with minimal disruption.
Server virtualization in the HMI world
There has been a lot of discussion and articles written about server virtualization via VMware and its applications for the process control market. For those who are not familiar with server virtualization, it can seem daunting at first, but what it allows you to do is create multiple PCs or servers in virtual space on a server class machine like a Dell PowerEdge server.
So, your first question might be why anyone would want to do that. The short answer is that you can create multiple virtual computers with any type of operating system and compartmentalize it on a real server machine. This eliminates many ‘real’ machines by consolidating them into virtual space. Most of today’s servers have a lot of processing power available and can run a multitude of virtual machines without difficulty.
In terms of advantages, there are many that I won’t get into today, but consider the following: in our aforementioned example, let’s say that instead of a single Wonderware application loaded onto all of the thick client computers, we instead have two or three different applications based on different areas of the plant (maybe a heat treat application, a wastewater plant application, and a welding application).
One option during this upgrade is to consolidate these applications into one single large application, but that requires additional engineering time and testing to avoid tag collisions and other various scripting problems. However, with server virtualization, we could take the PCs that are running the heat treat application and replace them with thin clients. We could then create a virtual server that only runs that heat treat application and tie it in with a small thin client network. The same could be done for the wastewater plant and welding areas and the operators would be none the wiser to what is going on in the background.
Now we have several independent thin client networks running on several virtual servers all housed on a single ‘real’ server. If you ever needed to make changes to one of these Wonderware apps, you could do so and even reboot that virtual server without affecting the other networks.
Is indefinite HMI support really a reality?
In closing, I hope I have given you some food for thought on ways to extend the life of your HMI software development efforts. Many of our customers have been supporting HMIs with thousands of tags and screens and are loathe to give them up purely for the sake of staying ahead of the Microsoft Windows OS wave. With thin clients and server virtualization you can extend the life of older controllers and HMIs without compromising the manufacturing process.
Thanks for reading!